'Ax).& yuv-~Xs£ps~a,.r:~8ou F&yc~={5~' &psXo5~ But the woman took off the great lid of the jar with her hands and scattered all these and her thought caused sorrow and mischief to men. (HEs*oI~: Works and Days)
Just over four years have passed since a group of research workers of several American universities who had joined in a prospective and cooperative study program of diabetes (The University Group Diabetes Program or UGDP) issued a report on correlations existing between diabetes treatment, metabolic control, and complications of the disease. The results which appeared to amount to an indictment of oral hypoglycemic agents caused embarrassment, surprise, discussions and official statements by Associations and Committees aII over the world: a modern edition of the mythicaI Pandora's box. As a resuh, the comptex etiological relationship between diabetes and atherosclerosis has been further complicated by the intricacies of methodological, statistical, and moral questions, thus adding to the difficuhies of understanding the problem and drawing practical conclusions. As we shall see, the UGDP research workers were forced to defend the results of their study everywhere since they appeared immediately open to criticism and in blatant contrast to the findings of a number of other investigations. Much has been said and written since the St. Louis Meeting until the recent Milan S~.~'aposium. The present review is intended to give a historical account of the problem and at the same time to focus the most significant criticisms.
HISTORICAL ACCOUNT T~rE ST. Louis MEETt>rG, JUNE 1970 About 1960, two 'longitudinal' studies were started, one in the U.S.A. and the other in Great Britain, ~ t h the object of assessing whether metaboIic control obtained by current methods of treatment were ap~ to influence the onset of vasc~lar complications of diabetes. Apparently, for a iong time 357
the two groups of research workers were unaware of the aims and results of their respective studies. The American investigation had been planned and carried out by a number of Clinical Centers which had joined the University Group Diabetes Program (UGDP). The British one was superintended by H. KEEN (London) and its results have been published in 1968 46 and communicated in September 1969 to the 2nd International Symposiunz on At,berosderosis, Chicago, the Proceedings of which appeared in 1970 ~a. A further study, conducted along different lines but the significance of which was anaIogous to that of the foregoing ones, was started in t963 by J. Pa.~sxKIw (Stockholm); its results were published in 1970 58. Thus in the early summer of 1970, the UGDP research workers already had knowledge of the results obtained by KEeN and PAASIKIw. Since the 30th annual Conference of the American Diabetes Association was to be held on June 13-14 of that year the organizers planned a special session with the object of giving the research workers of the three groups an opportunity for meeting and discussing the reasons why their results differed so widely. It should be pointed out that both KE~N and PAAsI~xvI had knowledge of the UGDP results from a certain number of 'confidential bulletins' the UGDP had issued at the end of their investigation. The meeting, during which LEWN~ (New York) acted as moderator, was attended by a limited number of diabetologists among whom M*~BLE (Boston) and G,RcIA (Framingham) who took occasion to repeat personal data already published in the medical press and which were in open contrast to the UGDP results. T~m UGDP ReFO~T The twelve Ctinical Centers where the research was carried out are listed in tab. 1. The data collected in these twelve Centers were eIaborated by an Executive Committee and by several specialized Technical Committees (tabs 2 and 3). The research had been planned in 1960 with the foIiowing three maior objectives: "1) evaluation of the efficacy of hypoglycemic treatments in tl~e prevention of vascular complications in a tong-term, prospective, and cooperative clinical trial; 2) study of the natural history of vascular disease in maturity onset, non-insulin dependent diabetics; 3) development of methods applicable to cooperative clinical trials" 7~ As a matter of fact the investigation was modified to some extent in the course of the years. It had in .fact been started in five Climcal Centers in 1960 (Baltimore, Boston, Cincinnati, MJnneapolis and New" York). When it got under way in 1961, two more Centers were added (Cleveland ~md Williamson). In I962, it was decided to extend the scope of the study to phenformin as well. During the same year three more Centers joined (Birmingham, Chicago, and St. Louis), and two additiona! ones (San Juan and Seattle) in 1963. Every Clinical Center selected a certain number of diabetics (77 to 94) presenting certain well defined features (tab. 4). Altogether 1,027 diabetics were studied; they were divided into 5 groups according to the management applied. Tab. 5 shows the size of these groups and the treatment they received. 358
clinic
PLBO
TOL]3
ISTD
[
-IV---~-R
PI~rEN
dl
1. Baltimore
24
22
20
20
0
86
2, Cincinnati
23
22
24
21
0
90
3. Cleveland
19
18
20
20
0
77
24
24
0
94
4. Minneapolis
22
24
5. New York
22
21
22
22
0
87
6. Wiliiamson
23
22
23
24
0
92
7. Birmingham
13
11
13
12
38
87
8. Boston
15
17
t6
15
23
86
9. Chicago
11
12
I2
i1
35
81
10. St. Louis
I0
11
12
11
35
79
11. San Juan
12
t3
13
13
40
91
12. Seattle
11
11
I1
11
33
77
all clinics
2O5
204
210
204
204
1,027
Table 1 . Number of patmnts assigned to treatment by Clinic k: UGDP study 7~
The PLBO group was given lactose tablets or capsules who's appearance was indistinguishable from that of tolbutamide tablets or phenformin capsules respectively. Tablets were used in Clinical Centers I to 6 (see tab. t) and capsules in Centers 7, 9, 10, 11, and 12. At the Boston Center (no. 8) both capsules and tablets were given. The trial was run on a double-blind basis. Dietaw management was the same in all groups and consisted in limiting ca!oric supply in such a way as to maintain body weight within + 15% of the ideal value. The carbohydrate, protein and fat content of the diet was 45, 2 0 , and 35% respectively. Fasting blood glucose was assayed quarterly, m~ oral glucose tolerance curve and fasting serum cholesterol tevel were checked every year, and in one Center the triglyceride level as well ALl subjects underwent complete study of their vascular condition. In some Centers the vibratory perception threshold of fingers and toes was also measured. Tab, 6 shows the study routine for every patient. SubsequentIy it was pointed out a~ that a short oral G T T had been performed at four-monthly inte~-als in a timited number of subjects, In this test, glucose (50 g) was administered 30 rain after the patient had received his medication (insulin or tolbutamide or phenformin or p!acebo). It must be stressed here that at the begLn±ng of the study some of the selected subjects atready presented cardiovascular changes or changes contributory to the pa~hogenesis of or consequent on vascular disease (tab. 7). The resuhs communicated and discussed a~ S~. Louis were published a supplement to the ADA journal 7~, ~7 and concern the study of 823 359
partidpadng institutions
principal investigators
co-principal investigators
Clinical Centers
Appalachian Regional Hospital West Virginia School of Medicine Witliamson/W. Va
C. A. JoN~S
W.H. JACoBs
The Jewish Hospimt and Medical Center of Brooklyn Brooklyn/N. Y.
M. G~ GOLDNER
S. WEtsEN~eLD
The Johns Hop~ns School of Medicine Battimore/Md
T.E.P~ovT
Massachusetts General Hospital Boston/Mass.
R, K. OsuoRsz
D. B. MAR~IN
Presbyterian-St. Luke's Hospital Chicago/Ill.
T,B. ScHwARTz
F. O. BECKER
University- of Alabama Medical Center Birmingham/Ala
B,R. Bos~LL
J. C, BARRETT
University of Cincinnati Medical Center Cincirmati/O.
H.C. KNOWLESJr,
K. KKeINES
University Hospitals of Cleveland Cleveland/O.
M. MILLER
W. B. NzWBE~eYJr.
University of Minnesota Hospitals Minneapolis/Minn.
F.C, GoE~z
M. E. JacossoN
University of Puerto Rico School of Medidne San Juan/Puerto Rico
L. HA~DOCK
L, A. VZGA
The Virginia Mason Research Center Seattle/Wash.
R.L. Reews
A. y.BowzN
Washin~on University Schooi of Medidne St. Louis/Mo.
W.H. DAUCHADaY M. E. LEVlN C. KILo
Coordinating Center
University of Mary]and Baldmore/Md
C. R. KL~MT
LipM Laboratories
Universh-y of Minnesota LaboratoD" of Physiological Hygiene 2v~nneapotis/M~nn.
H. BLAC~URN
G. L. K.'~'t~T'rEt
West Virginia University Medical Center Morgantoven/W. Va M, 7. ALSRI,~K
Tab& 2 *. Study personnd for each pardcipatLng Institution in UGDP 76
360
administrative committees
chairmen
Executive Committee
M.
Editorial Committee
H. C, KNOWLESJr.
MILLER
technical committees
chairmen
BOSt-IELL
Diabetic Control Commhtee
B, R .
Eye Committee
T. 1~. PRovT
Heart Committee
R, K. OSBORNE
Kidney Committee
F. C. GOETZ
Medical Technolog7 and Quai;~ty Control Committee
H. C. KNOWLESJr,
Mortality Committee
R. K. Ossomv~
Neurological Committee
R. L, REEVES
Per~pheeal Vascular Committee
M. G. GOEDNSt
S~atistics Committee
C, R. KLI~IT
Weight Committee
T, B, SCHWARTZ
"Feb& 3 -
UGDP Committees ~,
patient selection criteria
1 -
maturity-onset diabetes
2 - diagnosis of diabetes within past 12 months 3 - diagnosLs con.qrmed by glucose tolerance test 4 - absence of keconuria on diet alone 5 - Ere expectanw estimated to be at least fi years 6 - the padenFs adhesion to the study-program
Tabde 4 -
Paden~ se!ecdon crke~ia in UGDP rese~_~chTL 361
groups
abbreviation
no, of patients
tst
PLBO
205
diet + placebo
2nd
TOLB
204
diet + toIbutamide (1.5 g/die)
3rd
ISTD
210
diet + standard insutin (10-16 U / d i e lenin insulin at a fixed dosage depending on body surface area)
4th
IVAR
204
diet + variable insulin (dosage of lente insulin suNcient to control blood glucose)
5th
PHEN
204
diet + phenformin (100 rag/die)
therapy
Table 5 - Treaunent groups in UGDP research :~.
examinations performed quarterly general
general cIinicaI examination body weight
blood
fasting blood glucose
urine
glucose kemnes examinations performed annually
blood
fasting serum cholesterol fasting serum trigIycerides fasting serum creatinine
eye
fundus photography visual acuity tonometry
heart
cardiac size (X-ray) blood pressure electrocardiogram
kidney
creainine clearance proteinttia (qumntitative)
peripheral vessels
calcification (X-ray) oscitiometry
neurology
vibratory sense (quantitaive)
Tab[.e 6 - Exam;nations peKormed quarterly or annua!ty in patients of UGDP research ~'t
362
abnormalities
% values
cardiovascular hypertension history of angina pectoris electrocardiographic abnormalities (significant) arterial calcification
3L5 5.8 4.1 16.8
ophthalmic visual acuity (either eye < 20/200) fundus abnormalities (excluding ex~ddates) retinal exudates
5.4 16.5 42.0
blood fasting serum cholesterol (.'& 300 rag/100 ml) obesit;z (+ 25%) Tabte
7 -
1.3.4 56.0
Abnormalities in the patients at the beginnke,g of UGDP research=.
patients of whom 654 had been followed for 5 years. Data concerning group 5 (PHEN) were not reported to the St. Louis Meeting in view of the belated start of this trial. The most interesting findings were those concerning: 1) the degree of fasting blood glucose control by various treatment protocols; 2) the onset of vascular disorders in the course of the study; 3) the number of deaths and their causes. As for fasting blood glucose (fig. 1) a drop was observed in all four groups, even in the placebo-treated one, at the start of the study. Subsequently, however, blood glucose levels increased in all groups except the one treated with insulin doses continuously adapted to the patient's metabolic requirements. It is worth noting that during the last two years average fasting blood glucose in the subjects of groups TOLB and ISTD reached levels higher than those found at the beginning. The same was observed in. the placebo-treated group with the sole difference that the return to starting levels occurred sooner (about 30 months). it is truely remarkable that the lines shown in fig. 1 should represent the mean fasting blood glv.eose leve!s e f a limited number of patients (434 out of 823)~ Fig. 2 shows the average behavior of blood glucose t h after the adrc~inistration of 50 g glucose. These data concern only 396 out of 823 patients. From this point of view, too, the best metabolic control was obtained in the IVAR group, the worst in the PLBO and ISTD groups. Tbe TOLB 363
PLBQ (n = I06) /
//
5i
2
4
-
s~
5
/
I~
//
•--154 " i
--20]
...'"
/TQLB (n = 102)
,/
!
//
_..................... ' " /"
[•,,
"
/
/....
/"
"
/
/
",,z / .-'-.
: /.,X.--,'/ V" ',, / --2~J
/
',.z. . . . /~. ,.i¢~ ISTD (n = 1~6) ./
, A , a . // g// 10 12 /'14'X/-'16" ///follow-up examination /'~/
6
I
"~
A\
/
.WAR (n = ~t0)
"'"'",. ............ " ' " " " " " . . . . . . . . . . . "', ..... "................. :'"
%
F i g . 1 - Percent change in fasting blood glucose tevets from baseline to each follow-up examination for the cohort of patients foIlowed through i9th follow-upexaminationin the study of UGDpT:.
group had the best control initialiy but subsequently blood glucose increased progressively to the extent that after about 4 years this group had a less satisfactory control than the IVAR group. The degree of metabolic control was assessed by the authors on the basis of the mean fasting blood glucose levels found during periodic follow-up examinations. The distribution of patients belonging to the various treatment groups according to their degree of metabolic control and the criteria for the evaluation of this controI are shown in tab. 8. From this it would seem that the best control had been achieved in the IVAR group and the poorest in the PLBO group which is comprehensible if it is kept in mind that the evaluation of metabolic control was based upon baseline glucose leveI obtained by a short GTT performed, as we have already seen, 30 rain after the administration of insulin, tolbutamide, phenformin, or placebo. A point of particular interest is the mean body weight in the four groups under study (fig. 3). This concerns about 60% of the subjects invob;ed. It will be seen that the ISTD and TOLB groups experienced only transient and imperceptible initial weight toss ( - 1%) followed by a subsequent gain. In the IVAR group the initial weight loss was more marked ( - 2.2%) and lasted longer but the subsequent gain was equally greater than in the other groups. On the other hand in the PLBO group a more marked weight reduction was obtained ( - 3 g0) and maintained throughout the observation period. FinalIy, patient compliance was evaIuated but this was limited to drug treatment. Tab. 9 shows the criteria for the evaluation of patient compliance and the distribution of the subjects according to these. Compiiance was least 364
=so
ISTD (n =
,A.
I08)
/..-/. pLBO I,,= ~81
J ¢¢¢¢~
\~j"
/
/
.//
E
.~
g [/
'/
w.
g
~°i
/4 ///...~--m.'4: " .,_,,j
_]
--&
/
I
21o-i
/
-
/.
".~"\ \
/
v
./
/
/,'/
7
!'k
/ I
i
-_............,'"
..."
•
-~
.."
/ ~
........ tv,,,~(~=,los;
........../,.
"
..
!
\ ,i
I"
\./
~-
6
i/
\
'
"\Xvl. i
i-------+
o
r +~
2
"
i
¢
s fo!1ow-up
r
~o
~'2
/4
I'~
,'---f----"
Is
examination
F&. 2 - Mean value of 14 blood glucose levels at baseline and each follow-up examination for cohort of patients followed through t9th follow-up examination in the study of UGDP ~v
levd o{ control
PLBO
TOLB
ISTD
IVAR
good
27.0
fair
32.3
37.4
38,5
49.0
37.0
33.2
39.4
poor
40.7
25.6
28.3
i1.6
no. of patients ~
204
203
205
198
* Only padents with at ]eas+~one follow-up examLu~Monwith a short GTT were induded in this analysis T a b l e 8 - Percent distribution of patients ku the {our treatment groups according eo lmzet of btood glucose control in UGDP study rL Good: 70% or more of aR fasting values of short GTT (30 rain before the start of short GTT, the patients had taken their assigned medication) observed in each subject during the follow-up study < I10 rag/100 ml; fair: neither 'good' nor 'poor'; poor: as above: > 130 mg/I00 mI. 365
3q
4 J
'""',. /\.,~[qx/~ ./
t t?,
!~k
/
z_/k / / l
/i
//
"~:
, WAR (n = 131)
- v : . /% x." /
I TOLB (n = 1 I D
~
/ ISTD (n = 137}
xj
:
:
:'
8
follow-up examination
i j }i~> j/-~ ___j
./r-~\._
// ,
PLBO <~= ~_~s)
Fig, 3 - Percent change in retative body weight from baseIine to each follow-up examination for cohort of patients followed through 19th follow-up examination in the study of UGDP a,~.
satisfactory in the IVAR group and practically identical in the PLBO and TOLB groups. The d e n o v o occurrence of cardiac, ocular and renal complications in the four groups is shown in tab. 10, Statistical evaluation did not provide evidence of significant differences among the groups~ Fig. 4 and tab. 11 show global death rate and cardiovascular mortaliW in the 4 groups for a total of 89 deaths in 823 patients. As wilt be seen, the two indices are practically identical for the PLBO, ISTD and IVAR groups but higher for the TOLB group. Statistical analysis of the data with the Xa method shows the differences concerning global mortality to be non significant while the differences for cardiovascular deaths between the TOLB group and the others are significant (p = 0.005 for the comparison TOLB/ PLBO but hardly significant (p = 0.02) for TOLB/tSTD and TOLB/tVAR. On the basis of these findings the UGDP Executive Committee reached the following conclusions vT: --- ': the combination of diet and tolbutamide therapy is no more effective than diet alone in prolonging life"; "tolbutamide and diet may be less effective than diet alone or than diet and insulin at least insofar as cardiovascular mortality is concerned"; -
366
-
level of adherence
PLBO
TOLB
tSTD
IVAR
tow
10.2
1.0.3
12.8
14.8
intermediate
20.0
15.7
29.6
39.9
high
69.8
74.0
57.6
45.3
no, of patients
205
204
2t0
204
Percent distribution of patients in the four treatment groups according to IeveI of adherence to prescribed therapy in UGDP study % Low: patient was judged to have satisfactorily complied with therapeutic directives for < 2595 of all follow-up periods; intermediate: as above for 25 to 7495 of {oIlow-up periods; Ngh: as above for >_ 7595 of follow-up periods. (Adherence was considered satisfactory if patient took medication in the form and dosage prescribed for at least 75 days in each 3-month foliow-up period). Table 9 -
- - "any concIusions reached in this study pertains only to the type of patients studied, and to the specific hypoglycemic agents and dosage schedules used. ExtrapoIadon of findings obtained in the U G D P to other dosage schedules of the same drug or to other chemically related hypoglycemic agents not included in this study must be made on a judgmental and non statistical basis"; - - "insulin variable treatment was more effective than any of the other therapeutic modalities studied in maintaining fasting blood glucose control, The mortality results for the two insulin treatment groups are nearly the same as those for p]acebo-treated patients".
nonfatal event
PLBO
TOLB
ISTD
IVAR
6.7 (179)
9.4 (181)
5.6 (rig)
5.7 (i75)
significant ECG abnormality
14,1 (t98)
82 (196)
8,2 (i95)
7.9 (190)
hypertension
36,8 (125)
44.4 (135)
41,3 (138)
44.3 (t40)
visual acuity <_
20/200 in
either eye
serum creatinine > t.5 rag/ t00 mI urine prote[r~ k i g/I arterial caldCcadoo amputation
6.5 (184)
5.9
(t88)
3.6 (I92)
5.4 (i86)
0.5 (204)
2.5 (204)
0.0 (204)
2.0 (198)
li.3 (168)
14.8 (I55)
I6.8 (16t)
17.2 (157)
t.0 (202)
0.0 (202)
0.0 (209)
0.0 (204)
T 7 . ao~e 10 Percentage of nonfatal events in treatment groups of UGDP research 77,. Denominators {n parentheses,
367
k~ O~ CO
Fig. 4 -
g
f~
F,
/
//
•
// "
z/
/
'
z"
/
years of fellow-up
,."" """ / / / / /" //" /
A. all causes
i
./,>
/
z
/"
/'f
/
,'/
/
../'
I
l1
,/
/
"*
:,
r---t:*-
../
~,;," J-~ /! /
/
/
i / ,/
I
/"
/
.-
/'/
ISTD PLBO
: WAR
TOLB
0
10"
I
1
2
•
/
/' i
,/
/
/"
/
/
/
/
4
5 years of follow-up
3
6
/
/
/
!
z
//
/
/
/
/
7
.,,IVAR ISTD
/. TOLB
8
- - - - - PLBO
/ - ----,,"
f"
. cj>/~,~.-'"
/
, J /
/
./ ........y / ...... / ..................... ,~" ~
B, cardiovascular causes
Cumulative mortality rates per 100 population at risk by year of follow-up in the study of UGDP 'tL
10-
15
20 t
PLBO
TOLB
ISTD
IVAR
210
204
12 (5.9/%)
cardiovascular causes
10 4 5
1 - myocardial infarction
2 - sudden death 3 - other heart disease 4 - extracardiac vascula* disease non
cardiovascular causes
5 - cancer 6 - cause ofiqer than 1-5 7 - unknown cause no. at risk of death
Tab!e
205
204
alI cardiovascular causes
lO (4.9%) 26 (12.7%)
13 (6.2%)
a~ causes
21 (10.2%)
20 (9.5~)
1I
-
30 (14.7%)
Number and per cent dead by cause of death in treatment groups of UGDP
research a.
THE SUPPLEMENTARY U G D P
P,.EPORT ON PHENFORMIN
As has been mentioned above, the phenformin trial was started i8 months later than that of the other treatments and was carried out (see tab. 1) in a limited number of Clinical Centers. This is why the data concerning this drug were not discussed in 1970 at St. Louis and only the genera! raethodological lines of this trial were known at the time. The data appeared in the shape of a prelimiv.ary repor~ :a in August 1971, published by the ot~idal organ of the AMA. If we refer to them at this point of our review it is because they are an inseparable pare of the UGDP study. In the six ClinicaI Centers in which phenforrain was tried (and which were soon to be referred to as Phenforrain Clinics, so as to distin~fish them flora the Non-Phenform_in C!irics) the phenforraimtreated groups comprised a larger number of patients than the other groups (see tab. 1). This arrangeraent was made in order to have an equal totat nuraber of subjects in all five treatment groups, i.e. PLBO, TOLB, PHEN, ISTD, tVAR. This is why in the six Phenformin CIinics PLBO, TOLB, ISTD and tVAR treatments were appiied on a reduced scale. Selection criteria for tl-,e patients in group P H E N were the same as those described for the whole study. DurMg the first week 50 rng phenformin were administered daily, subsequently the dose was raised to 100 nag. For the evaluation of drug eff_ects on the cardiovascular system onIy data concerning the Phenformin CIinics were used. But since, as has been pointed out above, a smaller number of subiects was distributed to the other 369
groups in these Centers, the authors had great diNculty in elaborating the data owing to the differences in the size of the five treatment groups. They were thus compdled to create two pools (see tab. 12), one including groups tSTD and IVAR (called INS) and the other including groups ISTD, IVAR and PLBO (called P + IN). During the 8 years of the trial 47 deaths occurred of which 3t in group PHEN. The causes of death in the various groups are shown in fig. 5 and tab. 13. Tab. 14 shows the statistical elaboration of the data from which it may be seen that differences were fairly signi.ficant for the comparison P H E N / INS for deaths from nil causes and for the comparisons P H E N / P L B O and P H E N / P + IN for deaths from cardiovascular causes. Nonfatal cardiovascular events occurred with significantly greater frequency in the PHEN group as compared to the two pools (tab. 15). As to the effect of phenformin on fasting blood glucose Ievels, the drug proved promptly effective in that it induced an average drop of 20 to 25%. This hypoglycemic activity was maintained during the first 3 months after which a gradual rise of blood glucose teveIs was observed which was, however, slower than in groups PLBO and ISTD. After 57 months (19th follow-up examination), glycemia was still lower (about - 6 % ) than baseline (fig. 6); whereas in the other two groups it had already risen to leveIs exceeding baseline. The authors therefore come to the conclusion that, although phenformintreated patients had lower blood levels than subjects treated by diet alone, the incidence of nonfatal events during the trial was higher in the phenformin group which goes to show that there is no proof of the beneficial effect "of blood glucose control in the prevention of nonfatal vascular complications" 7~
THE BEDFORD 't'RIAL
As has already been pointed out, at the time of the St. Louis Meeting KE~N and Ja~ea~Ttr had aIready performed a similar trial at Bedford 4a, 4~. The results were discussed at St, Louis, The study had been initiated in 1962 following an investigation on the correlations of diabetes to atherosderosis, performed at Bedford, The study tasted 7 years and was carried out not on overt diabetics but on 248 adults defined as borderline cases because their blood glucose 2 hrs after a glucose load (50 g p.o.) was t20 to 200 rag/100 ml. The selection of these subjects in whom glucose tolerance was not outspokenly diabetic had been suggested by the observation that cardiovascular risk is particuIar!y high in this group. The subjects were divided into four groups as shown in tab. 1.6. Distribution of ~ e patients into these four groups was at r~dom, Throughout the study period patients were submitted to periodic cliMcaI examination in addition to electrocardiogram, blood pressure measurement and blood sugar assay 2 hrs after a glucose load. The cardiovascular complications observed dur'mg the 7 years of the stud>" are listed in tab. t7, together with their distribution in the four experimental groups. As wilt be seen from the table, fatal and nonfatal 370
kaa ~d
(64) (64) (62) (63) (60)
(63)
59.4 (64) 43.8 (64) 0.0 (56) 1..6 (64) I8.8 (64)
31.3 4.8 6.3 8.1 7.9 45.0
.37.5 (64) 40.6 (64) 37.5 (64.)
65.7 53.9 7.4 2.0 13.6
2g.4 4.6 7.4 5.0 11.1 43.5
(204) (204) (t88) (2o3) (198)
(20t) (197) (202) (20t) (I98) (191)
49.5 (204) 3t.9 (204) 41.7 (204)
PHEN
61.8 48.5 3.2 1.5 20.9
20.9 6.0 6.0 4.5 14.9 43.8
(68) (68) (62) (68) (67)
(67) (67) (67) (67) (67) (64)
60.3 (68) 32A (68) 42.6 (68)
ifSTD
63.~ 462 7.0 0.0 15.9
23.4 1.6 3,t 4.7 15.6 39.1
(65) (65) (57) (64) (63)
(64) (64) (64) (64) (64) (64)
52.3 (65) 27.7 (65) 32.3 (65)
IVAR
'1'able 12 -Baseline characteristics of patients in Phenformi~. Clinics of UGDP 7a. Denominators in parentheses.
fasting blood glucose ,> !.10 rag/100 ml relative body weight > 1.25 visual acuity (either eye < 20/200) serum creatinine > 1.5 rag/100 m] arterial calcification
other baseline characteristics
hypertension history of digitalis use history ot! angina pec~oris significan~ ECG abnormality d~olesterol 2"_ 300 mg/I00 ml one or more cardiovascular risk factors listed above
cardiovascular risk factors
~onwhite
I~ale
age > 55
demographic cbaracmristics
PLBO
62,4 47.4 5.0 0.8 18.5
22.1 3.8 4.6 4.6 15.3 41.4
(t30)
(133) (133) (119) (132)
(131) (t3t) (:[3I) (1.31) (13i) (128)
56.4 (133) 30.1 (I33) 37.6 (I,33)
INS
61.4 46.2 3.4 1.0 18.6
25.1 4.t 5.1 5.7 12.9 42,6
(197) (197) (175) (1.96) (t94)
(i95) (194) (i95) (1.93) (194) (188)
50,3 (197) 33.5 (197) 37.6 (t97)
P + IN
A. all causes
/ PHEN
20
// / / /
//"
16
/
/ PLBO
// /
/ 12-
j-/
. . ~ ' J " 7 /'~
/
I / / P + tN
,/~ :~.............. INS
/,/ /
/
4
/
/
. ......
...- ........ / 5:; .. .... _-~,/
m
>"
, ..,-.Lr J-~ 0
1
..J 2
¢~ >
s
4
3
years of follow-up B. CardJogsscufaF causes
,f-
i
/
/
/.
/
PHEN
./
/ / / ./
J
1/ /.
q
....... l
"J
'*"~
..................
,........... /"**. . . . . .
INS P + IN
/.-~---PL~O
,
i
........ ~-"<: ................ -.....""
7
.... ~ /
y~ars o,~ follow-up
Fig. 5 - Cumu[ative death rates per 100 population st risk by yeaz of follow-up durhng the
~esearches of Pher~ormin Clinics of the UGDP s~udy ;L
372
~a --d ,,,~
-
myocardial infarction sudden death other h e a r t disease extracardiac vascular disease
4 (6.25'~)
(8.8%)
3i (15.2%)
6 (9.4%)
aI1 causes
Table
13
~
N u m b e r and percent dead by cause of d e a t h as of J a n u a r y 6, t971 in P h e n f o r m i n Clinics of U G D P 7~.
6
3 (4.6%)
6 (8.896)
65
26 (12.7%)
no. at risk of death
2 (3.1%)
68
IVAR
all cardiovascular causes
lactic acidosis cancer cause other t h a n 1-6 u n k n o w n cause
ISTD
204
~
PHEN
64
5 6 7 8
n o n cardiovascular causes
I 2 3 4
cardiovascu[ar causes
PLBO
10 (7.5°/6)
9 (6.896)
t33
INS
~6 (8.I%)
11 (5.6%)
197
P + IN
PHENPLBO
PIIEN. INS
PHENP + IN
ISTDPLBO
IVARPLBO
number dead from all causes difference in % dead
5.8
7.7
7.1
p value
0.33
0.05
0.04
t,00
0.72
difference in /% dead
9,6
5.9
7.t
5,7
1.5
p value
0.04
0.11
0.02
0.32
1.00
--3.2
--0.6
number dead from cardiovascular causes
I 4 - ProbabiIky o f obtaining observed distribution of deaths in two groups of U G D P study being compared calculated with the 'Fisher exact test' 78
Table
t0-
5 ~
0. . . . . . . ~
,
-r--...........,
4
6
i - s- ,~
.
.
8 I0 12 fOIIOW-Up examination
////
ii "\
-- 20-
.
.
14 16/ ~ X. / /,A ",/
/
"//~\
-- v iSTD (n =: 42)
/L--w--* Pt.BO (n = 41)
? ', /
18 / ,,/ / ///
PHEN
(n = 134)
\
/
ii~ , L>"-,
I ~
-,
_
/
.
..' ...... Ivan {n = ~ )
--25 q
J
I - 35 .i
Fang. 6 - Percent change in fasting blood glucose levels from baseline to each follow-up exaumination for ~he cohort of patients tot[owed d'lrough 19,th follow-up examination in the researches o~ Phenformin Clinics of the U G D P study %
374
(57) (59) (51) (43)
3,3 (60) 14.9 (47) 12,5 (56)
t.6 (62) 6,6 (61)
0.0 (55) 9.6 (52) 15.s (38)
15,8 J.7 4.9 32,6
I
I
(19i) (!86) (194) (143)
1.o (19s) 22.6 (168) 16.2 (185)
4.0 (198) 6.7 (195)
5.8 (I73) 5.1 (i77) 35.4 (99)
9.4 9.1 5.2 47.6
PHEN
(64) (5s) (ss) (49)
0.0 (50) 17.0 (47) 13.8 (58)
1.7 (59) 11.7 (60)
5.5 (55) I0,7 (56) 24.3 (37)
6.2 3,4 1.7 34.7
tSTD
(6i) (60) (61) (47)
0.0 (61.) 16.0 (50) 14.3 (56)
5.0 (60) 6,7 (60)
3,9 (5t) 9,8 (51) 32.4 (37)
4.9 0,0 0,0 34.0
1VAR
(125) (11.9) (96)
(118)
o.o (121) 16.5 (97) 14,0 (t14)
3.4 (119) 9.2 (120)
I0.3 (1.07) 28.4 (74)
4.7 (io6)
5.6 1.7 0.8 34.4
INS
(182) (177) (180) (,39)
1.1 (181) 16,0 (144) ) , 9 (170)
2.8 (18t) 8.3 f181)
3.1 (161) 1.0.1 (t59) 24.1 (112)
8,8 1,7 2.2 33.s
P-+ IN
v,a.,4 Table 15 Percentage of patients wiflJ first occurrence of specified nonfataI everlt identified at follow-up examination as of January 6, 1971. in va~ Phenformln Clinics of UGDPVL Denominators in parentheses.
-
amputadon of all or part of eitbel: limb arterial calcification claudicatio interrnittens
peripheral vascular examination
urine prot:ein > i g/1 serum creadnine > ],5 rag/100 ml
kidney examination
visual ac~dty :-~ 20/200 (either eye) opacity (vitreous, tens or corneal, either eye) fundus photo abnormalities excluding exudates
eye examination
significant ECG abnormality use of digitalis l!ospitalization for heart disease hypertension
hear~: examhtation
PLBO
nO. Of
group
patients
treatment
1st
65
diet: reduction of only sugar + placebo twice daily (containing 3 mg of tolbutamide)
2nd
57
diet: reduction of alt the carbohydrates (t20 g/die) + ptacebo twice daily (containing 3 mg of tolbutarnlde)
3rd
62
diet: reduction of only sugar + tolbutamide (1 g/die)
4th
64
diet: reduction of all the carbohydrates (120 g/die) + tolbutamide (1 g/die)
Table I6 - Patients grouping according to treatment in KEEN and JARPd~T~r'Sresearch ,5
cardiovascular accidents were Iess frequent in the tolbutamide-treated subjects as compared to the placebo groups but the differences were not statistically significant. The same may be said of global mortality and cardiovascular mottalky (fig. 7). Both types of mortality were lower in tolbutamide-treated
PLBO group (no. 54)
placebo + only sugar restriction (no. 28)
placebo + alI carbohydrates restriction (no. 26)
cardiovascular death
6
7
myocardial infarction
3
2
angina (onse0
6
9
cIaadicado (onset)
3
I
stroke
P
")
ECG abnormalities
8
3
event
TOLB group (no. 42)
tolbutamide + only sugar restriction (no. 24)
tolbutamide + all carbohydrates restriction (no. 18)
T~M.e J7 - Fatal and nonfata! events in treatment groups o( KEEN'S research 42.
376
25 ~
,#q. ai~ causes / / / /
/ i tt /
/ /'
20/ 2
/ /
/
-i
/ /'
15~
/
p,/ • cardiovaac@af / ,/
. i{ C~k;s~$ GJ
./ /i
/
/
10<< / 1 /
/
xT<" o~ " ""
•
0
Q,
4
5
5~
' /'
•
o"
., /
o
o
o 1
2
3
6
7
years cf [oIIow-up e =
PLBO
o
=
TOLB
7 - Toni mortality end numbers of deaths certified as due to cardiovascula'2 cause by year of follow-up in to!butamide and placebo treated borderline diabedc subjects in the &year period I962-1970 (KE~N44),
Fig.
groups than in the pIacebo ones but the differences were not statistica]Jy sig_uifc~a~t. It should be remembered that subsequently KeEN 4~_ reelaborated these data by dividing the population according ¢o age at entry into the study, the dividing line being drawn at the age of 60. This was done in order to isolate the highest risk group (age > 60) since cardiovascular risk is kno\vn to increase with age. The upper panel of fig. 8 shows vascular disorders to be less frequent in the younger group of toIbutamide-treated subjects as compared to placebo-treated ones of the same age group, although the dikerence is barely signi~cant (0.1 > p > 0.03). The author further divided the same subjects ~4 according to the cardiovascu!a~ risk as judged for every inc~vidud on the basis of the presence or absence of vascular disorders or arterial hypertension prior to entry into the study. The upper panel of Eg, 9 shows the R~cidence of new vas~lar 377
lOOq
age < 6 0 {n = 136) 80 -
J
~oJ
J I
1
0.I > p >" 0,05
!'~!i i~,i{:{ $
0
--
- - " ' - -
...................
"7_ ca ! O0 I
~.
t
age ~ 60 (n = 112}
804
60-1
! 40~
I,,,
"i~
" i!iiiil iiiii li{ ,
ii~!
TOLB CHO
}
Ii:: !!iii iiliil
qlI 0
0
0
+
4-
O,
+
0
+
+ 0
+-
Fig 8 -, Arterial events according to different ages and to different therapy in the study of KEEN 44.
events to be tower in the 'tow risk' tolbutamide-treated group than in the corresponding p!acebo group. This permits to exclude an effect of tolbutamide ~avouring cardiovascular disorders; on the contrary, it raises the possibility that the drug may have a protective action againsc such disorders. 378
~oo] low risk (n =1173}
I 80-4
[
e0 i
p < 0.05
4° I
30.4
l:ii < =
t00
i~;i~
1 N g h risk
n = 75)
j3
iiill i
~i;~
fi~ilit
a!
liiil
~:'
!!i
....
,.,
iii'i"
l~:!'~
i!!?i]
ii~i{i
TOLB
0
0
4"
÷
0
,il ii
J.
+ 0
+
Fig. 9 Arteriai events accordhng to different cardiovascu!ar risk and to ~diKerent therapy i.,tthe study of KEEN 4~
THE STOCKHOLM TRIAL A*~ the time of the St. Louis Meeting, the results of a research car~qed out by PAAStKIVI ss at the Department of Medicine, Karotinska Inst#~tetSera#me~'fasaret4 Stockholra Bad just been published in a supplement to 379
controi group
TOLB group (1 g/die)
no. of patients
83
95
males
67
78
females
16
17
59 _+ 11
59 ± 9
1.14
1.I4
characteristics
mean age (years) mean k-value anticoagulants
Table t8 * Composition of patients group in PaAsmIv/s study ss. A c t a M e d i c a Scandinavica. This study, which had been started in 1963, concerned .178 patients of either sex (145 men with an average age of 58, and 33 women with an average age of 68) out of 262 patients who had survived an episode of ascertained myocardial infarction. The subjects were randomized into two different management groups: totbutamide and placebo, Tab. 18 shows the characteristics of both groups. Tolbutamide was administered cautiously in small quantities at first; subsequently dosage was raised to 1 g/die,
examinations general
general clinic~l examination body weight blood ery~&rocyte sedimentation
ra~e
white blood ceil count fasting blood glucose rasing serum cholesterol fasting serum triglycerides glucose tolerance test i.v, heart blood pressure electrocardiogram dectrocardiographic exercise ~est he~rt volume Table 19 - Examinations performed at regular intervals in patients of PAASI~
380
Overt diabetics and subjects with a family history of diabetes were excluded from the study. An intravenous glucose toIeranee test was performed in every patient at the beginning, i.e. 3 to 4 weeks after the first episode of myocardial infarction, and was repeated at regular intervals. Clinical
normal
I',;GTT
100-~
abnormal IVGTT
I", ```%- .... v,,
'X
9oj
~oi',
I! I'
8o-!
~
I
j
7oj
', ~--~.
'x
i
b---o~
--<--- - < > - - - . c ~ ---{;
70
io
#--%-
~
observation time (years)
diabetic IVGTT
90-
.,OOT__~__~\
I
\
i!
\
gc ii
i
,
borderline IVGTT
i
,
\
80-
I
7o! 0
x ,
O
\,
I
i ~,~> , - - < ~ - - ¢ , -
1
2
-.<>-
-o - -.~>
.--~- -
3
4
~-" -' ~ 5
70/j
# o
~1
2
5
4
5
observation time (years) o .......~a ccntrol g~oup
F%
10
-
Cumulative survival in relation to I V G T T
~-.------~ "FOL~ g r o u p
i'_,~the study of l°,~.:~sI~(Ivx 5a
381
1.50-!
abnormaf IVGTT
totaI
!
normal t V G T T
1.40]
,so-I 1.20 ; " 0 0 ~:i:i:i:' ~ } :5:)1:1:{:~ : :~:i:i:?i:!:):?i:)i;??!:i:i:)i:!:i:!:i:i:) ...
i:~:::::::;:::?g~:~:~
::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::$::?:?::-~!:i:5:!:i:i:~:i:!:i$?i:i: :)i: ~!i!:i:?!:5:i:#:%5:iii:iiS::iiiii/i~:!~i# ~ ~?i?}#?~2i~/i~i~ t:i::?::?!;!)::i:{:?:i:{:!i!i!!?i?ii?~ ? # ~iii%':Ni~ i? i'{
i',iiiii!ii!;)i!i;iii :ii{iii#ii' :ii i!iii!':ii' i{ii{i{{i:,iiii{i{ili
'- "/! iii#ii!!!~{{ii{!?i!:i!!~{i!~ ji~i:i~!:j:ii?![:{):~i!ii:!:{:!:?:!i!# ?3~ N ===========================================================================~=~:!$i:~:!:!:i:!:!:5:i:5:75~?i!!:!ii:i:i:!:!$:8 ====ilg~:!~!~!ii~iii>~s:~i~i!si~zs!~::>::~ = iii!:!:!:i:Si!i}}~#i: :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
0
1
2
3
4
5
0
I
2
3
4
5
::
".. :~:~; ! ::~:::. •:!:i!~
......,...,,,::{:i:!:i:~,:*-.'::::.~!:i:!:i:i:i
~:i:~#::#i:~:.~:~i~::~:~.:``:.~#i:;::i~i~i::,i{:~i#i~i!i::i~:)~?~i~ig 0
1
2
s
4
5
observatbo time (years) o---~---s control group
Fig.
11 -
TOLB group
Changes of k values in tolbutamide ~oup and in controI group in the study of
PAASIKIVI 58
examinations and instrumental and laboratory tests listed in tab, 19 were equally performed in every subject at regular intervals. The research was aimed at studying the effects of tolbu ~~armoe: "' 1) on subjects survival; 2) on the glucose assimilation coefficient k; 3) on. fasting serum cholesterol and triglyceride levels. Fig. 10 illustrates the results as regards the first point. If patients are separated into those having normal and those having abnormal carbohydrate tolerance at the start, the difference between tolbutamide-treated and non to!butamide-treated subjects is quite clear. At some points of the curves, statistical analysis reveals these differences to be at the limit of significance. The time elapsed betvreen the beginning of treatment and ne\v episode of infarction (fatal or not) also proved Ionger in the sulfonylurea-treated group. Totbutamide treatment did not appear to influence the gIucose assimilation coefficient k (fig. 1t), But if subjects having a baseline coeNcient k < 1,i0 are considered separately it is found that an increase took pIace in the tolbutamide-treated patients while the placebo-treated ones experienced a further reduction. Finally, tolbatamide seemed to have had a favorable action on the fasting serum triglyceride levels in the group of subjects with k < 1.t0, as may be gathered from fig. t2. Serum choIesterol did not undergo any variation. These 6mdings strongly suggest a protective acdon of ~olbutamide as regards coronary heart disease. 382
f
abnormal I V G T T
normal I V G T T 3.0-i
3.0!
//° J / t /
/ 1
% 2.0" E
I-oi
<
1. . . .
i.o4, ~--,
,year c> . . . . . .
year
o control group
~,. . . . . ~ - ~ TOL8 group
Mean serum trigiyceride lcve!s it~ 46 patients at discharge from hospital and one year tater in the Pa.aSIK~W's study~%
Fig. 12 -
THE
FRAMINGHAM STUDY
In the course of the St. Louis Meeting, GARCIa et el. ~ recalled some data which were part of the investigation carried out at Framingham aa on factors bNuencing the onset of cerebral, cardiac and peripheral atherosclerosis. These data., too, were known to the diabetologists present at the Meeting. The authors of this study had followed 5,209 subjects of either sex residing at Framingham (Mass., U.S.A.), during a 16-year period. The age of these subjects varied from 30 to 62 years, tn the course of the study 111 men and 1t4 women developed overt or chemical diabetes; of these, 93 were treated with insulin, 87 with oral hypoglycemic agents and 45 by diet alone. In the course of the observation, 55 of these 2 2 5 diabetics died, thus showing the death rate of diabetics to be twice and one-half as high as that of the non diabetic population. The highest death rare was registered among insulin-treated diabetics (32), above aI1 in the women of this group. The excess mortality among diabetics had to be attributed mainly to cardiovascular causes. The main cause of death, coronary heart disease, was three times more frequent in insu!in-treated diabetic women than in non diabetic subjects, g r i t in the group of insulin-treated diabetics coronary heart disease was fatat i~ a larger proportior~, o~ c~ses than in non diabetic subjects, FURTHER COI~,!PARATIVE
TRIALS
. . .0ah, . . AT ALR£ADY T." x~NO
ST, L O U I S
A t St. Louis; during ~ e session devoted to the UGDP report, a discussion too!; p]ace during which )XARBLE reported clinicai and p o s t morce~,v 383
.la.
kaa OO
55
i9
2nd
3rd diet 69.5
69.6
70.9
hypoglycemic sulfonylureas insulin
mean age (years)
therapy
I1
33
33
male
8
22
22
female
patI~otogicat findings (%)
5.8
11.3
10.8
63
62
75
t
21
25
t6
16
33
1t.
!mean duration cerebral peripheral of diabetes myocardiaI { infarction t vaseuiar (years) infarction ]or hemorrhage . disease
Cl]nlcal characteristics and pathological findings of 1.29 diabetic subjects (BALomMos et al.S).
55
1st
Table 20 -
~o. of patients •
group
clinical cI~aracteristics
va~
ka,}
1.7
50
1st
2nd
Table
21
-
ott~er therapies
hypoglycemic sulfowIureas
therapy
0.05 < p < 0.1
34
65
myocarditis
0.8 < p < 0.9
60
53
myocardial infarction
0,9 < p < 0.95
58
53
congestive heart failure
pathological findings (~;)
Pathological findings of 67 diabetic st~bjects in Bzool~\voR'rH and HaMWI's study 'q
statistical analysis
no, of patients
group
0,8 -< p < 0.9
36
29
cerebral vascular accidents
CO O",
57
30
2nd
3rd diet
insulin
hypoglycemic sulfonytureas
therapy
14
25
47
64.8
65.8
64.7
I
male I no, mean age
[6
32
54
no.
64A
66,1
69.8
mean age
~emale
26.7
38.6
30.7
myocardial infarction
3.3
5.3
9.9
cerebrovascular accidents
10.0
24.6
18.8
peripheral vascular disease
cardiovascular diseases (~))
Prevalence of cardiovascular disease and composition of three treatment groups in BALODIMOSet al.'s study4.
10t
tst
T a b l e 22 -
no, of patients
group
clinical &aracteristics
data from the rich material of the Joslin Clinic (Boston)5, part of which had already been pubIished. As far back as 1968, the Boston diabetologists had reported on post morte,ea findings in 129 diabetics of whom 55 had been treated with hypoglycemic sulfonylureas, mainly totbutamide, for 24 to 112 months; another 55 had had insulin and the remaining 19 had been managed by diet alone. Tab. 20 summarizes the main features of the three groups and the prevalence of post r~ortem findings reladng to vascular disorders. Statistical analysis showed the differences between the three groups to be not significant. These findings are in accordance with previous ones reported by WELLmAa.W et al. ~2, while they are at variance with those of BLOODWO~TH and HAMWI 7 who, in a smaller series, had observed the more frequent post morteru finding of granulomatous myocarditis in diabetics having undergone suIfonylurea treatment as compared to a control group of patients. But apart from this isolated observation of visceral microgranuloma, findings concerning myocardial infarction, congestive heart failure and vascular accidents are strong evidence against a negative influence of sulfonytureas on vascular disease. As shown by the data listed in tab. 21, differences concerning vascular disorders are not significant. Also at the St. Louis Meeting, Mm
The "lay' press showed signs of alarm already three weeks before the St. Louis Meeting as a result of the interview granted by one of the UGDP study organizers to a reporter of the Miami Herald Tribmne with advance information on data and conclusions reached by the UGDP. On the following days, the press agencies of various countries started to broadcast the deplorable news that totbutamide had a detrimental effect on the heart. In I t J y , the news arrived after considerable delay and did not cause much of a stir in the non medical press.
OFFICIAL REACTIONS OF NATIONAL ORGANIZATIONS
Unb.'ed States of Awerica
The UGDP results were presented and discussed at the 30th Annual Meeting of the American Diabetes Association (ADA) wb2ch took place at St. Louis on June 13, 1970. As soon as the meeting was closed, the 387
ADA handed out its first press release on the subject :0, which read: "New data have been presented, some of which raise questions about the et~cacy and safety of oral therapy, tlowever, it is diKicult to generalize from these unpublished data. Careful evaluation of the complete data and further study will be necessary to reach final concIusions" ADA at once nominated an Ad Hoc Editorial and Advisory Committee, and on the basis of studies carried out by this Committee during several months, published an Editorial Statement s4 dated October 7, 1970, in which the salient and most controversial aspects of the UGDP study were discussed and the following recommendations concerning therapy were made: "Tolbutamide, as well as other oral hy-pogtycemic agents, has no place in the routine treatment of chemical or latent diabetes, suspected diabetes, or prediabetes. Such therapy has never had a place in diabetic ketoacidosis or in those prone to it. The clearest indication for oral agents is diabetes of mild or moderate severity in a patient who proves to be poorly controlied with diet and who is unabte or unwilling to take insulin". "In adult-onset diabetes", the statement continued, "with hyperglycemia and glycosuria, symptomatic or not, and in the absence of ketosis, a triaI with an appropriate diet should come first, if tl~is does not establish satisfactory control, insulin is to be preferred to other therapeutic agents because it is more uniformly effective in controlling hyperglycemia and the UGDP study indicates that it may be safer". On October 27, 1970, ADA sent out a circuIar letter to all its members to inform them in advance of the above statement and particularly to make known that the Ad Hoc Editorial and Advisory Committee and the Executive Committee "agree with the conclusion that death rates from cardiovascular disease in the study patients receiving tolbutamide were significantly l~Jgher than in study patients receiving other treatments". And this in spite of the admission that "interpretation of the data is di~cult because of the possibility of an increased predisposition to cardiovascular disease in the tolbutamide-treated patients"; and that "European studies on the use of tolbutamide, but not strictly comparable, have reached different conclusions". Almost at the same time, viz. on November 2, 1970, the American Medical Association (AMA) stated its position through the Council on Drugs. This statement v1 recognized that the interpretation of the UGDP data was not simple in view of certain statistical diNculties deriving from the lack of uniformity- of the ~.ndings in various clinics and from the hi~ner baseline incidence of cardiovascular risk factors in the tolbutamide-treated group. Nevertheless, "every effort should be made by the physician to control the symptomatic, maturity-onset diabetic with diet done. Should this fail, treatment with insulin or oral hypoglycemie agents shouId be undertaken". "The consideration of treatment with oral hypoglycemic agents should be secondary to the use of insulin". The Food and Drug Administration (FDA), which Iogicalty was the American organization most directly involved issued a report ~7 at the end of October 1970 containing the recommendation that %ral hypoglycemic agents should be used only in diabetics with adult onset, stabie disease which cannot be controlled by diet alone atad for whom insu!in is unacceptable or impractical". 388
After some comment which was substantially similar to that of the two American medical corporations referred to above concerning the UGDP resrdts, the FDA report continued: ~The Food and Drug Administration recommends that the use of Orinase (totbutamide) and other sulfonylurea type agents, Dymelor (acetohexamide), Diabinese (chlorpropamide), Tolinase (tolazamide) should be limited to those patients with symptomatic adult onset nonketotic diabetes mellitus which cannot be adequately controlled by diet or weight loss alone and in whom the addition of insulin is impractical or unacceptable. The oral hypoglycemic agents are not recommended in the treatment of chemical or latent diabetes, in suspected diabetes, or in prediabetes, and are contraindicated in patients with keto-addosis'. The t'~A has recentIy confirmed this attitude in the find tabeling of hypoglycemic agents an, s~ressing that this was in accordance with. the recommendations of the ADA. On the occasion of the 32nd ADA Meeting, on June 23, 1972, the Board of Directors of the ADA voiced its criticism of the FDA and stated that no data are avaiIable to iustify the attribution of the untoward consequences ascribed to tolbutamide (and phenformin) by the UGDP report to other hypoglycemic agents, tt also denied that the ADA had been consulted with regard to the statements contained in the labeling of these drugs ~_9
Australia The Australian Drug Evaluation Committee took sides with the three great American organizations, tn its statement it took note of the results reached by the UGDP study and of subsequent comments, analyzing the two most debated points: the greater baseline cardiovascular risk in the tolbutamide-treated group, and the discrepancies between the ~dings of the UGDP and those of similar studies carried out at Bedford and in Sweden. In agreement with the three American bodies the Committee declared itseK inclined to accept t~he UGDP results and convinced that "the recommendation of the Americae Diabetes Association. presents a Balanced view, in accord with the original concept of the use of this type o~ drug" a
United Kingdom Immediately after the first news of the UGDP data had appeared in advance in the non medical press, the British Diabetic Association (BDA) published a press release dated May 22, 1970, in which it wished "to reassure diabetics in the United Kingdom about the resuhs of the United States University Medical SchooIs' tdals of toIbutamide". Certain conclusions were considered premature, because "the patients involved in the trials should not Gbe r%ardeo as representative of the British alaoeac '" * ~" population, and it is moreover clear that the results vary considerably between t)le different Uz,Ated States Centres ...TriaIs in Britain since t~e 1962 BecFord survey, and in Scandinavia come to quite different conc!usions"; diabetics were therefore advised to continue tolbutamide treatment without making any changes. A few months lar.er and in possession of fulier information, the BDA confirmed its position in a further and more con:.pre~hensive statement, dated November 5, 1970, in w}'~ich doctors were advised to "continue their usuaI vigilance in the care of individua! patients taking ora! antidiabetic drugs". 389
in any case, the Association did not see "reason to recommend any systematic transfer to other forms of therapy". It its Armual Report for 1969 and 1970,0 the authoritative Committee on Safety of Drugs took sides in a chapter headed: "Oral Anti-diabetic Agents" from which we quote the following: "Publication of the results of studies carried out in the United States of America which suggested a possible relationship between the administration of tolbutamide and cardiovascular complications, caused the Committee to examine the findings which had led the Food and Drags Achministration in the United States to issue a bulletin to all doctors in that country expressing a view that oral hypoglycemic agents should be used only in diabetics with adult onset, stable disease which could not be controlled by diet alone and for whom insulin was unacceptable or impractical. The Committee concluded that the evidence was unconvincing in relation to the reported hazards from tolbutamide and other oral antidiabetic drugs, and that as used by physicians in the United Kingdom there was no need to recommend any &ange in the use of such drugs. The Committee decided, however, regularly to review data from long-term observations with this group of drugs in this country, and to encourage those concerned to undertake studies to provide those data". The attitude of the two authoritative British medical periodicals, the British Medical Journal and The Lancet, is remarkable. The issue for 21 November i970 of the British Medical Journal = had a leading article which, after briefly summarizing the purpose, findings and conclusions of the UGDP study, went or~ to say "the case against tolbutamide is far from overwhelming, and several criticism must be answered before the programme's conclusion can be accepted". The article lists some of this criticism (differences in the death rate at various Clinical Centers, insuf~ciencies in the diagnoses of the causes of death, excess number of white women in the group of subjects treated with tolbutamide) and rejects the conclusions of the FDA and ADA. About two months later, The Lancet ~a, too, took sides in a leading article published in its January 23, i971 issue. After summarizing the UGDP findLrtgs the article ends: "The recommendation that oral antidiabetic drugs should be used only as a last resort will be received with dismay by those who have been impressed by their ease of administration, freedom from apparent side-effects and continued e~cacy for as long as fifteen years. Moreover, the dangers of insulin even in small doses are not negligible and hypoglycaemia can be a serious risk in the elderly and infirm. What should be the decision in Britain? t-Iere it is impossible to mount a study of similar structure: apart from the enormous cost, the delay in achieving figures of any signiAcance would make it most unlikely that the results would be of any value. At this time it does not seem right to abandon this group of welt-tried drugs".
South djrica The Council of the Society of E.~-docrinology, Diabetes and Metabolism of South Africa issued a short statement which was published in the 4th September 197t issue of the Medical Proceedings, as editorial in bilingual version =6 390
The results of the UGDP were listed and the following five objections raised: "1) The patients were ethnologicatly and socially heterogeneous; 2) the 'fixed-dose' tolbutamide regimen was unclinical (if not unethical); 3) the 4 groups differed considerably in initial cardiovascular 'risk factors'; 4) the higher incidence of cardiovascular mortality in the tolbutamide group arose from resuhs in only 3 of the 12 clinics; 5) autopsies were performed on only one third of the patients who died" In view of these reasons for perplexity and keeping in mind that "other countries have not followed the FDA recommendations and in several ... the official bodies have advised diabetics to continue taking sulfonylurea without fear ... the Council supports this advice, but stresses that sulfonylureas should be used with the proper caution and supervision due to any potentially powerful drug whose actions and effects are incompletely understood".
Canada The official statement of the Canadian Diabetic Association appeared in a bultedn in May 1971 ,a It openly disagreed with the FDA and the other American societies that had endorsed its stand. According to the Canadian organization "there is not acceptable evidence at present that tolbutamide is harmfuI and therefore insufficient reason to issue warnings or restrictions on its use in this country". The Association suggested that tolbutamide and other oraI hypoglycemic agents be used at the physician's discretion in appropriate cases in which diet alone was inadequate to control the metabolic disorder.
Gereegany A few days after the St. Louis Meeting, the Deutsche DJabeles-Gesellst,bait (DDG) issued its first statement n based on careful evaluation of the UGDP findings, It pointed out that "the American study, although representing an interesting contribution to the problem of the causes and evolution of cardiovascular disorders occurring in diabetes - - in view of its lay-out and performance - - was not apt to elucidate the problem it had set out to soIve and its conclusions could not withstand scientific criticism". Certain discrepancies concerning s~atistical evaluation and data collection were pointed out. FinaI!y, the DDG whi!e agreeing with the American authors' contention that a case of diabetes that could be controlled effectively by diet atone should not be treated with drugs for reasons of convenience to the doctor or the patient, stressed that "a diabetic for whom his physician considers drug therapy necessary in addition m dietary restrictions, can follow such ~reatment wkhojat apprehension". 591
The DDG's position was reconfirmed subsequently, in February 1971, by a declaration e0 stating that "even after renewed scrutiny" the Association saw "no reason to recommend a change in its therapeutic approach to diabetes ",
France The comment of the Association des Diabdtologues de Langue FranGaise (ADLF) a on the results of the UGDP study was expressed officially at the General Assembly of May 8, ],971, during which the following resolution was adopted: "In view of the spread of information of medical content among the public, the Association des DiabftNogues de Langue Fran~,aise is of the opinion that at present there is no decisive reason to change the indications of various oral hypoglycemic agents". This resolution was fotIowed shortly by a Round Table ~ at which French diabetologists and experts in statistics took part, for the purpose of analyzing as objectively as possible the UGDP problem in the light of previous reactions and of the stand taken by the ADLF. At the end of the meeting the round table participants found themselves in agreement that the findings of the UGDP research did not warrant the consequences drawn by the ADA, AMA and FDA but that nevertheless they had the merit of obliging "diabetologists to reconsider the position as far as oral hypoglycemic agents and their use in the management of non-insulin dependent diabetes was concerned". The need for early diagnosis and treatment was underlined, treatment "consisting first and foremost, in cases with excess weight, in dietary restrictions and return to ideal weight before other forms of therapy were undertaken". Short-term treatment with oral hypoglycemic drugs "was justified only in case of failure of dietary management and on condition of a close watch on weight and g!ycoregulation. In the absence of ready and complete success of such treatment, change-over to insulin therapy is called for".
Italy The Societa Italiana di Diabe~ologia did not issue an ol~cial statement. The Ministero della Sanita, having consulted the Consiglio Superiore di Sanita, ruled that the following remark should be added to the leaflets contained in packages of oral hypogtycemic drugs: "The use of blood-gIucosedowering agents of the sutfonylnrea class shouId be limited to cases of symptomatic adult onset non ketosis-prone diabetes mellitus that cannot be controlled by diet and in which insulin admiristration proves impracticable. OraI hypogtycemic drugs are not indicated in the treatment of l~ten.t or suspect diabetes or in prediabetic states; they are contraindicated in patients with keto-acidosis". The term 'impracticable' referring to insuJin therapy is at ieast surprising. The remarks obviously are nothing but the verbatim translation of the above mentioned statement by the American FDA, and the Italian word imp~,'aticabile is intended to translate and contract the two English expressions 'impractical' and 'unacceptable'. It so happens, however, that the Italian impraticabile has a different and much more restrictive meaning. 392
THe UGDP RePoxT aT MEDICAL CONYERF,NCES
The Boston Conference The most vocal protest against the of~cial position taken by the FDA was raised immediately by more than thirty American diabetologists ~z, at a meeting he!d at Boston on November 30, 1970. They defined the FDA recommendations premature and devoid of yard critical scrutiw, and demanded reexamination of the UGDP report holding the interpretation o£ the findings to be faulty. They maintained that the resuIts of the study were invalidated by obvious mistakes in assessing the patients' clinical condition, by the use of fixed ~olbutamide dosage and by the lack of homogeneity of the various groups of subjects, all of these being factors which alone were apt to nullify any statistical analysis.
The Williamsburg Conference At the Williamsburg Meeting of the American College of Physidans 7.9 held during the same period, the problem was discussed by several diabetoiog~sts with the participation of PROUT. Moss, of Georgetown University, Washington/D.C, maintained that the different degree of cardiovascular risk factors in the various treatment groups had profoundly impinged upon the results of the UGDP. He pointed out that 27% of the patients in the tolbutamide group suffered from heart disease as against 17% in the placebotreated group. Similarly, 14.2,°7o of the tolbutmmide-treated subjects had high Blood cholesteroI levels, as against onty 6.2/% in the placebo group; electrocardiographic abnormalities were present in 24 and 15% respect~vely, and finally, 7.6% of the former and 4.5% of the latter had digitalis therapy. Moss therefore came to the conclusion that ~when you have two groups with as many differences as those, you can't say the result is due to the drug". Reporting his own 15-year experience in the treatment of 2,300 diabetics, Moss pointed out that of 3'10 deaths (13.6%) 14.3'~ occurred in insulin-treated patients, 13.8% in those receiving phenformin, 14.2% in those treated with tolbutamide, and 10.6% in those controlled by diet alone. The reason for the low mortality in this last group was that it included the patients with the least severe forms of diabetes. The author added that he world have submitted at least hail the UGDP patients to diet alone because they had fasting blood glucose values Below 130 rag/ 100 mt and were obese. Only 1/8 of his own patients had required daily tolbutamide doses of 1.5 g/die all the others being controlled with 0.5 to ! .0 g. Lie stressed the fact thaa the administration of large doses to subjects who do not requ~e them may not be devoid of a&erse consequences, tn reply to t~d~s PROO-T underlined the fact teat during the trimestred fo]J.ow-up exa~ninations instances of h~Toglycemia were very rare; in his turn he criticized the lack of a control group in Moss's study. Of the other participants at the Meeting, M.CXBLZ,of Ha~ward Universiw, stressed the discrepancy between the findings of tee 7os]in CU.nic where death from cardiovascular complications of diabetes is extremely frequent 393
(76.6%) and the findings of the UGDP study in which not a single death due to myocardial infarction was recorded in the placebo group. On the same occasion, DOLGE~, of the Mr. Sinai School of Medicine (New York), asserted on the basis of his personal experience of 220 patients treated for t4 years, that good control of glycemic homeostasis, irrespective of the treatment applied, is the best guarantee of success. The same author had occasion to express his views on the UGDP study more explicitly and with a partioalarly controversial accent in a lecture, held at the Universita Cattolica del Sacro Cuore, Rome, October 22, 1971, of the meaningful title "An American Fiasco' in which, having once more underlined some of the criticism already referred to, he openly charged the UGDP and particularly the present director of research of the FDA, Clint, with having literally wasted the sum of 7 million dollars for an unsuccessful study that had yielded extremely few results. He also mentioned other authoritative comments on the UGDP resuhs, particularly the critical study by Feinstein.
The V Capri Conference This Conference which was held at Capri in April 197i was devoted to 'Blood Vessel Disease in Diabetes Mellitus', L~NDBmK and KEeN = acting as coordinators. The presence of MAx MT_Hm.% one of the most committed members of the UGDP Executive Committee, and of KE>;N and PAASlKiVt, the two workers who had reached different conclusions, caused the discussion to touch repeatedly upon the most controversial results of the UGDP study. A particularly lively and interesting exchange of views took place during the panel discussion on 'Factors Influencing the Progress of Atherosclerosis in the Diabetic', PF:eIFrE~ ~* having stressed the important role played by overeating and reduced physical activity and KAPPeRT 4~ having reported his results showing the importance of tobacco, MILLE~{s5 summarized and commented the UGDP results. Next, P&aStKtW ~9 illustrated the. findings of his already known research which .had shown tolbutamide to be beneficial in the management of myocardial infarction; and KEeN ~'~, presenting a new elaboration of the Bedford data showed even more clearly that tolbutamide does not add to the cardiovascular risk associated with diabetes, albeit in its latent
forKrl, The discussion was opened by the critical observation by Keen concerning the astonishing fact that in the PLBO group the UGDP research had not registered a single instance of death from myocardial infarction, a fact which by itself casts some doubt on the whole study since death rate from coronary disease is 20 to 3098 in the normal population and 40 to 70% m diabetics. MILLER answered this criticism by saying that ~it is the role of the dice". KOHNel< took exception at the decision of the UGDP to stop experimenting with totbueamide: the likelihood of the absence of deaths from myocardial infarction in the PLBO group being a mere coincidence was much smaller than that Of the greater number of deaths in the TOLB group being due to chance. tt had been a mistake to stop the experiment in the TOLB group since there was the possibility of the resuh being a statistical freak. M~LLE~_answered that "it would be a subject for a congress of psychiatry as to why we have stopped it. It has beer, a very complicated decision to make, and some of us were very unhappy about it". 394
LUNI)B.~K and PFeI~FE~ criticized the small proportion of cases (40%) which underwent post mortem examination; consequently, a large proportion of the deaths considered due m cardiovascular causes were so only by 'hear-say'. POZZA drew attention to the differences between the various groups as regards electrocardiographic signs of and deaths due to myocardial i~arction. If the two figures are added, relevant differences are no Ionger found between the two groups but the dea~s due to myocardial infarction are actually more numerous in the TOLB group. It could therefore be thought that the management of infarction had interfered with this astonishing distribution of results. Findiy, Ct, tXD censured the fixed tolbutamide dosage (i.5 g/die) used by the UGDP. In his opinion the findings of the study should apply only to that dosage while we do not know the effects of the customary daily dose o~ I g.
12...... /ourndes de l'Hdtel Dieu A montl:~ later, during the 12...... yourn6es Annuelles de Diabdlo[ogie de I'HdteI Dieu, the last morning was entirely taken up by a panel discussion 2 on the subject: Les hypoglycFmfants de synthhse sont-ils nFfastes? During the lengthy debate, in which research workers from various countries took part, most of the speakers were in agreement. A very comprehensive criticism was set out by PIRART sa who, after having expressed his approval of the study protocol followed by tee UGDP, pointed out the errors committed in its performance: these were of such import as to lead to most debatable results. The same opinion was voiced by BASTENIe. CONSTA~vIand JARRE'TTreminded the audience of the results of their own studies which are in disagreement wkh those of the UGDP as shown in another part of the present review. Finally, LouB.#rIkxEs stressed the di~culty of establishing the precise cause of death in subjects suffering from other disorders Lq addition to diabetes, tt is also important to keep m mind that metabolic control in the TOLB group could not have been op~,ma, ~" ' since their blood glucose levels kept on rising as if these subjects had no~ undergone any treatment at all. At the end of the discussion, the General Assembly of the Associaaion des diab6toIog~es de langue frangaDe, chairman GmaeT, unanimously adopted the following resolution ~: "In view of the spread of information of medical content among the public, the Association des D~ab~toIogues de Langue Franqaise is of the opinion teat at present there is no decisive reason to change the indications o[ various oral hypoglycemic agents. The D~abdtologues de Lang,.~e Frangaise stress once more that the ,prescription of these drugs in no way dispenses the patient from strictly complying with the usual &e, .
8~h Conference of the Deufsche Diabete>GeseIlscha/t A few days later, on May 15, 1971, during the 6th Conference of the
Deutsche Diabetes-Gesellschaft F~e]d at Dfisseldorf, SCH6FFLI~-C~s thoroughly discussed the UGDP study both as to its approach and operatiom His state395
ments were based mainly on a non-open meeting called by the German health authorities (February 12-t3, 1971) at which the subject was discussed with international experts on diabetes and statistics, including some of the UGDP staff, tn his opinion particularly serious mistakes had been made in the choice and randomization of the subjects: many suffered from retinopathy which raises the suspicion that they had been diabetic for more than a year, many others were not diabetic at all. In these, any treatment was obviously 'abusive' and liable to have triggered counterregulation factors that may have played a role in the occurrence of cardiovascular lesions. Further aspects which laid the study open to criticism were the fact that smoking habits in the various groups had not been taken into account, necropsy had been performed in a larger proportion of the TOLB group and the research had been stopped prematurely in this group without valid justification.
The .[erusalesa Con/erence In Jerusalem, too, in the course of the International Symposium (October 1971) held in order to commemorate the 50th anniversary of the discovery of insulin and dedicated to the subject 'Impact of Insulin on Metabolic Pathways' a panel discussion deah with 'Oral Substitutes for Insulin: Modern Appraisal of their Action and Indications'. GOLDN~ as, one of the principal workers of the UGDP research, presided over the meeting which he opened by giving a detailed account of the conduct and findings of that study. Having repeated the data already known, GOLDN~ came to the conclusion that "there may be some indications for the oral hypoglycemics as substitutes for insu!in, but none for them or for insulin as substitutes for dietary restrictions". Foilowing some presentations on the mechanism of action of sulfonyiureas and biguanides, CANAANIlz took the floor speaking on 'Some Evidence for the Innocence of Long-Term Tolbutamide Treatment on the Frequency of Cardiovascular Disease'. The author expressed the surprise of Israeli diabetologists at the drastic conclusions o£ the UGDP research and reminded the audience of the divergent results reached in the studies by BALODt~mS, PAASlKIVI and I~EN, concluding that the Israeli diabetologists sided with their British colleagues and medical bodies. Subsequently, RA>!EJAsa (Bombay) reported a personal study in which 296 diabetics had been followed from the cardiovascular point of view for an unspecified period of time. The results are shown in tab. 23. As will be seen from the tabIe, the patients were divided into five groups according to treament. Unfortunately, the groups are no~ homogenous as to number of patients and duration of iltness. Above all, the lack of a controI group of diabetics managed by diet alone is not compensated by the 'irregular' treatment group which does not supply a ;¢alid term of comparison. Lastly, the risk factors of the different groups do not appear to have been assessed at the start of the investigation. The highest death rate and progressiveness of cardiac complications were registered in the insulin-treated groups while the tolbt~tamide group differed in this respect from the so-called control group by a factor that barely reached statistical significance. 396
toa ~D "--I
1"10. O~
23 35
insulin + suKonylurea
biguanide -I- inst~lin/suKonylurea
Table
2.3 -
74,2
100.0
84.2
67.1
51,2
deterioration in cardiac status (%)
Cardiac status and mortality in 296 diabetics divided according to treatment i n }~AHEJA'S s t u d y 6a.
6.2
9.83
9.31
4.84
7.87
mean duration of diabetes (years)
* Deterioration ir~ cardiac status as compared to irregnlar treatment group
19
143
sulfonylurea
insulin
76
patients
irreguI~r treatment
therapeutic group
43.5
31.6
18.2
10.5
mortality
ValUe ~'¢
< 0,01
< 0.001
< 0.01
< 0.05
p
KD OO
Table
17 0 18
38 18 29
9 years
10 to 15 years
total
24 -
13
67
0
diet
TOLB
16 41 21
INS
30 41 36
50 to 59 years
18
60
12
diet
40 27
51
20
29 64
TOLB
INS
~3. 60 years
34
36
33
diet
38 24
41
19
TOLB
48
32
INS
all ages
Percentage of cardiovascular complications in diabetics treated with insulin, tolbutamide, and diet in F:mEL'S study ~2.
4 to
TOLB
40 to 49 years INS
diabetes duration
age and treatment
25
45
20
diet
FID~L a= (Jerusalem), presented a study on 456 subjects with adultonset diabetes who were separated into three groups according to treatment: diet, tolbutamide, insulin. Tab. 24 shows the prevalence of cardiovascular complications in the three treatment groups, at the same time dividing the patients by age and duration of diabetes. As shown in the table, cardiovascular complications were more frequent in. the insulin-treated group whereas the tolbutamide group did not differ from the group treated by diet only. The round table ended with F~LIG a~ (New Haven) insisting upon the need for appropriate dietary restrictions and WEISENFELD~0, another of the UGDP research workers, defending the results of the study. LOUBATIk~S's remarks during another part of the Jerusalem Symposium mus~ certah~ly not be forgotten. His report which was subsequently pubtished ~0 is the very model of an analysis of the difficulties the research worker has to overcome in evaluating the results and consequences of various antidiabetic therapies. Having pointed out the mistakes incurred in by the authors of recent studies, the Montpellier pharmacologist gave a britiiant description of how to conduct an investigation into this subiect with due methodologicaI exactitude.
The Evian Symposium During a coherence of French language scientists held at Evian in Sept,ember t971, the UGDP study was thoroughly discussed in a Iucid and accurate analysis by FAJANS~8 and in the debate which followed. It was concluded that the study should have satisfied the following conditions: diagnosis and assessment of the duration of diabetes by tolerance tests, etc.; selection of younger patients (age below 40) so as to reduce the prevalence of vascular changes present at the start; homogeneity of the various groups; largest possible numbers of subjects; perfect metabolic control as shown by ideal fasting and postprandial blood glucose values. These requirements were not entirety met in the UGDP study, hence its limitations and the doubKut validity of the statistical calcuIations on which its conclusions are based. In particular, the lack of homogeneity of the clinical populations was underlined: this alone was su~cient to account fo~ the discrepancies in the findings. Thus e.g. mortality from cardiovascular causes was nil in four of the twelve Clinical Centers, 24-32% in three and 50% in another two. Strange though it may seem, not a single fatal myocardial infarction was registered among the placebo-treated patients° Further, the excess mortality due to myocardial infarction in the tolbutamide group contrasts with the lack of excess incidence of nonfatal vascular complications in this group. A statistical comparison between groups as b.eterogeneous as these is obvious!y inadmissible, The Evian Symposium tIaerefore came to the unanimous conclusion that however useful from certain points of view, the UGDP study did not answer two questions, viz. whether tolbutamide was actually harmful, and whether good blood glucose control was apt to prevent the occurrence of cardiovascuIar comp!ications. 399
treatment
no. of patients
angina pectoris and/or claudicatio i intermittens
myocardial infarction
CVC (%)
none
5O
12
diet only
38
6
2
21
diet + placebo
34
5
3
23
diet + tolbutamide
35
4
t
14
non diabetic s~d~jects
49
3
0
6
30
Table 25 - DeveIopment of cardiovasc~lar complkations (CVC) in borderline and control groups during 10 years of fotlow-np in CAI~Ls'r~O~aet al.'s study ~L The Milan Meeting Finally, in May 1972, in the course of the 8th Meeting of the European Study Group for Diabetes EpidemioIogy held in Milan, the problem of cardiovasc~alar hazards during tolbutamide therapy was again debated, the occasion being provided by a report of CAaLSW~OM et aI. is of Lund University. By screening 228,833 subjects aged above 16 these authors had isolated a group of 157 individuals with borderline glucose tolerance. The group was divided into four subgroups according to management, and was followed for 10 years together with a group of non-diabetics, in order to study the incidence of cardiovascnlar complications. The findings of this reseat& are shown in tabs 25 and 26 from which it appears evident that in diabetics cardiovascular disorders are more frequent than in non-diabetics and that totbutamide treatment (1 g/die) has a protective effect on the occurrence of angina pectoris, myocardial infarction, claudicatio intermittens and isc.hemic electrocardiographic alterations.
group
no trea~ent diet only
at rest
during work
7.1
45
12.1
immediately after work
4 rain a{ter work
9.5
35.7
40
4.1
32
diet + tolbutamide
4.6
30.2
23
23
non diabedc subjects
4
t6
4
16
Tab2e 26- !schemic electrocardiograpi{c changes in borderline and control groups during !0 years o c follow-up in CAe.LST.'{6Me~ aL's study ,s
400
RESeaRCH CARRIED OUT atTaR TVE UGDP REPORT After the St. Louis Meeting and as a result of the heated debate that followed, some research workers fdt stimulated to undertake studies intended to ascertain whether oral hypog!ycemic agents were actually apt to cause cardiovascular disorders. Most of these studies were of the ,'etrospecrive type. Thus, CONSTAM1~ (Ziirich) reexamined the records of 651 diabetics who had died: is_ 236 death was attributable to cardiovascular causes. By separating these cases into t~ve groups (tab. 27) according to the treatment applied during the years preceding death, the author found cardiovascular fatalities to have been more frequent among patients treated by diet only as compared to those who had received stdfonyIurea drugs. An unexpected finding was the low cardiovascular mortality among patients with brittle diabetes especially in view of the longer duration of the disease in this group. Another retrospective study was carried out by I-~ADDEN e t a l . as at B'Nfast where the causes of death of 670 women with maturity-onset diabetes were examined. The findings are shown in tab. 28. Clearly, the mottalky due to myocardial infarction was higher in the women treated with oral hypoglycemic agents, and higher stitl in those receiving insulin. However, the authors themseives admit that their investigation had some weak points, such as the longer duration of the disease in the group treated with h.~4?oglycemic drugs and the less satisfactory control and greater severity of the metabolic disorder in the groups treated with hypoglycemic agents and insulin. At the same Be.fast hospital BoYsz et al. 10 undertook a 'longitudinal' study ranging from 1965 to 1971 in order to examine various aspects of the natural disposition of diabetics towards cardiovascular disorders: 186 patients entered the research (96 men and 90 women), ages ranging from 35 to 75. They had developed diabetes in 1965 and were treated either by diet alone or with oral h)>ogtycemic agents, according to customary the> apeutie criteria. A!I were constantly kept under supervision as regards cardiovascular pathology. In the course of the six years considered, 25 cases of myocardial infarction occurred (10 in women and 15 in men)° Tab. 29 shows the results separated by treatment groups wkh their corresponding percentages as obtained by the Xa method. The data reported in this paper are too scanty to permit evaluation of the results. The difference between the groups is barely within the limits of statistical significance. The authors themselves admit the possibilky that there were variations between the two groups as to severity of disease and cardiovascular risk. Recently, a retrospective study appeared in this journal by Dx~uxY and TIMONEY ~" relating to the causes of death in 284 maturity-onset d~abetics. If these cases are divided into three groups according to the treatmenc applied consistent ~fferences are not found, as may be seen in tab. 30. Lastly, TZAGOU~'~IS and R~-ExTsoN 7s went into the problem of cardiovascular risk connected with phenformin. Starting in 1965, the authors studied t37 subjects who were neither diabetic nor obese and had survived • an episode of myocardb'Jl infarction am~nosed chnma_,v, electrocardio~raDhica!Iy and by laboratory tests. The majority of these patients (104) were as.
~
~
401
O I-,3
T a b l e 27 -
8 30
38
69 t56
225
94 I48
242
26 37
63
39 44
83
no. o[ dead patients
diet + insulin labile diabetes
diet + insulin in larger doses
diet-t-insulin in small doses
diet + oral hypogIycemic agents
diet alone
therapy
Groups of patients and causes of death in CONS'rAM'S research ~7
cardiac deaths o~her deaths
5 total
cardiac deaths other deaths
4. total
cardiac deaths other deaths
3 total
cardiac deaths other deaths
2 total
cardiac deaths other deaths
1 total
groups
age
nleatl
35 22
24
45 50
48
53 55
54
55 62
28 22
23
20 15
17
I6 15
62 50
53
68 54
59
54 51
52
69 59
11 10 t6
63
11.
77 77
11 1,3
58 57 59
77
12
(%)
male
57
dm'ation at onset af diabetes (years) (years)
meall
sex
38 .50
47
32 46
41
46 49
48
31 41
37
23 23
23
(%)
female
10
54 4I
45
33 34
33
57 43
48
32 54
44
(%)
ovel-
weight
17 52
45
51 47
49
61 55
58
5O 57
54
67 58
62
(%)
hype> tension
oral hypoglycemic agents
diet only
causes of death
insulin
total
myocardiM infarction
23 (25~8)
54 (31.0'%) 13 (35%)
other cardiac death
12 (I3%)
24 (13.8%)
3
(8%)
90 (29.6%) I 39 (I2.8%)
(2%)
8 (4.6%)
2
(5%)
12 (3.9%)
other thrombotic events
2
cerehrovascniar accident
22 (24%)
34 (t9.5%)
cancer
14 (15%)
t4
o~hers
20 (22%) t 40 (23.0%)
total T a H e 28 -
therapy.
(8.~%)
66 (21.7%)
10 (27%) 2
(5%)
30 (9.9%)
7 (19%)
67 (22.1%)
93 (100%) 1i74 (100%) 37 (100%)
304 000%)
Causes of death in the patients of HADDEN et al,'s research 39 according to
signed at random to two treatment groups: diet only (40% carbohydrates, 2 0 % proteins, 4 0 % fats) or diet plus phenformin (100 r a g / d i e ) . Twentyfive of these patients died. Fig. 13 shows the respective survival rates during the years of follow-up of the two groums (on the te~t) and of the 104 randomized subjects (on the right). The difference noted daring the last three years is not statistically significant. In these non diabetic coronaropathic patients, phenformin did not exert any effect, either negative or positive, on survival after a first episode of infarction.
oral hypodiet oniy lgiycemic" dru~
Z"
m~!es no. of patients myocardial infarction
57
39
7 (12,2%11 8 (20.5%
0,7 < p < 0.8
0.64777
females no. of patiens n~yocardial infarction
58 4
32
6 (18.7%)
(6.8%)
1.85631
0.10
<
p
<
0.20
both sexes no. of patients myocardial infarction
Ii5 it
71
(9~6%)I 14 (t9~7%) !
3.0658
1
0,05 < p < 0,t
Table 29 - Percentage of myocardial infarction in padents of BoYL~ et aI.'s study ~0 403
cardiac deaths
cerebrovascutar deaths
renovascutar deaths
all cardiovascular causes
diet only
19 (50.0%)
7 (18.4%)
1 (2.6%)
27 (71.1%)
38
oral hypoglicemic agents
60 (44.495)
30 (22.2%)
4 (2.9¢8)
94 (69.6%)
135
insulin
58 (52.2%)
17 (15.395)
8 (7.2%)
83 (74.8%)
111
groups
TabIe 30 -
aU causes
Causes of death iv. the patients of Dr~uRY and TIMONEY'Sstudy = according to
therapy.
t003-..... `
~
[
oo! !
iOOq--~
<
7'<
\
"b.-:
80
! I i " ~ -) i 1 i ro 1 i
g t to4 E { ,
!
6o~ {
e0-i
:r 4
5
years of fol o w - u o
......
diet
.........
PHEN
Survival of aR 137 padents (at the left side) and of 104 randomized patients (at the right side) with coronary disease trea:ed wir~ phenformin plus diet or diet alone in the TzAc~o~m>~tsand ~EYNERTSON'S study ~L Fig. I 3 -
C R I T I C I S M OF T H E U G D P R E P O R T From what has been said so far the discrepancies between data contained in the U G D P report and those of a number of simukaneous, previous and subsequent studies appear most dearly and the consequent refusal of the conclusions drawn by U G D P by the majorky of diabetologists becomes understandable. However, it cannot be denied that the U G D P research was the first attempt at a pluricentric qongkudinal' study on a chrop2c disorder. Inevitably its organization could not be perfect. This is why in the course of ~he last three years a true 'error hunt' has taken place vdth ~he object of de~ecdng where a mistake has been made 404
in this first mu!ticenter trial. Out of the large number of papers dealing with this criticism the following should be quoted: LOUP~ATIgaES~0, PmART ~z FEINSTEIN ao SCHOR ~, SELTZER o7, ~a TEUSCHER ~a. >~ and ~t~RAUSE 48 ~or having gone most deeply into the problem. The principal criticisms to be Ievdted at the UGDP report concern essentially fore" aspects: - - the general tay-out of the research; its operation; data processing; - - inteqoretation of the data.
-
-
-
-
MISTAKES IN THE LAY-OUT OF THE STI_FDY
1) Excessive elasticity in evaluation of cardiovascular risk. The research protocol of the UGDP concerned adult-onset diabetics (average age: 52.7 years; range 20 to 79); inevitably therefore part of the population had a certain baseline risk of cardiovasc~iar disorders and it would have been of capital importance to distribute this risk homogeneously among t~e five treatment groups. The cardiovascular abnormalities found in the UGDP patients are listed in tab. 7. But the standards laid down by the protocol according to which these clinical situations were apprdsed and c!assified were such as to allow the twelve groups of experimenters considerable elasticity in evaluation. Suffice it to say that the anamnestic element of use of digitalis compounds was considered the index for congestive heart failure, and electrocardiographic changes (Q-waves, ST-wave alterations, complete a-v block, left bundle branch block, and ventricular taehycardia) were considered together which may be found in clinical conditions of widely differing prognostic sig~J~cance. On the other J~and, some conditions of considerable cliniced relevance were completely forgotten by those w~ho elaborated the protocol of the investigation, such as pulmonary diseases (asthma, emphysema, etc.), liver d~sorders (cirrhosis, hepatitis, etc.), cardiac valve defects and some electrocardiographic changes such as non ventricuIar arrhythmias, hypertrophy and overloadL,qg. Finally, one of the selection criteria was a prognosis of a,~ least five years survival but obviously this is an extreme!y elastic and failacious criterion. There can be no doubt for instance that some of the coded electrocardiographic signs, such as left bun.d!le branch block, were scarcely compatible with a prognosis of five years survival. Thus, there is the possibility that the cardiovascular risk factors, though cEstzibuted among the Eve groups with sv~cient 'qualitative' homogeneity, were less so from the 'quantitative' po~t of view, with resulting cF.fferences between the groups which were apt to create a flaw in the e:vperiment. 2) T,be experimentaI p~'otoco~ did no~. take into accoun~ exogeno:~ /~.ctors iiable to /avouz athe~osderosis. There is certainly a possibility that 405
external factors causing cardiovascular damage via a cumulative action may quite accidentally have influenced one group more than another. The experimental protocol did not take these factors into account at all. The UGDP workers did not trouble to provide for the homogeneous distribution of some cardiovascular risk factors among which smoking and profession. As for this last element it is indeed surprising that the experimenters did not consider it in view of the varying prevaIence of atherosclerosis in urban and rural populations and even within the same professional categories according to the type of work the invidual performs.
3) Patients were assigned to the different treatnzent groups regardIess o/ their clinical requirements and o/ their sensi~ivky to the drug applied. The diabetics included in the study differed as to: -
-
age: there were young ones and old ones (tab. 3i);
severity of the disorder: there were cases of severe diabetes and others of chemical diabetes (tabs 32 and 33); -
-
body weight: (tab. 34). -
-
there were over weight subjects and lean ones
These patients showing varying clinicaI features were quite arbitrarily submitted to five different therapeutic schedules which are usually employed for precise indications and in keeping with the severity of the disease. Thus the program gave rise to fictitious situations wtaich were often very far removed from clinical reality: 'abusive' insulin treatment and tolbutamide in subjects for whom it would have been logical and mandatory to limit prescriptions to diet only, This decision must necessarily have introduced errors that no physician would have committed. It is in fact not unlikely that tl~is experimental schedule resulted in five years of a treatment which was either out of proportion or inadequate to the patient's needs. Nor wotdd it be surprising if in some cases this 'abusive' therapy had given rise to counterregulatory reactions. The UGDP workers do in fact admh a: that there have been 'episodes of ascertained or suspected hypoglycemia' (6 in the TOLB group, 26 in the tSTD group and 49 in the IVAR group) but only in the IVAR group dosage was modified. They further admit that in 32 subjects of the PLBO group and 23 of the TOLB group the blood glucose level rose and this was 'associated with other signs and symptoms'. It is equally possible that some of the 204 diabetics who received tolbutamide were not sensitive to this agent and therefore were submitted for years to ineffective treatment. It is in fact welt known that no~ every case of non ketosis-prone maturity-onset diabetes is sensitive to tolbutamide.
4) Th~'ee o/the groups were treated ,,~i:h fixed doses o~ fheiY respective agents. Insulin, sulfonytureas, and biguanides are known to require adaptation of dosage to the metabolic needs of the patient. Totbutamide was given at a standard dose of t.5 g/die. Apart. from other considerations, this is certainly a large dose if we keep in mind that the duration of diabetes in these patients did not exceed one yea> tn many cases therefore this dose mast have been excessive. 406
age at entw
T a b ! e 31 -
mate
female
all patients
20-24
0.3
0.8
0.7
25-29
1.4
2,2
1.9
30-34
t.4
3,0
2.5
35-39
6.i
8.2
7.6
40-44
t0.2
13.4
12.5
45-49
i4.0
12,3
12.8
50-54
15.3
t6.0
15.8
55-59
19.0
15,4
16.5
60-64
t4.6
14.6
i4.6
65-69
10.9
9.3
9.7
70.-74
6.i
3,3
4.1
75-79
0.7
!,6
1.4
no. of patients
294
733
1,027
mean
54.t
52.1
52,7
:~ SD
10.5
11.4
11.2
Percentage Nstribution of patients o{ UGDP study by age at entry and sex v6
I- fasting blood glucose (mg/i00 ml) < 90
99 t
all
TOLB
ISTD
IVAR
PHEN
10.3
11.5
11.8
8.8
10.5
17.6
24.9
29.2
25.5
23.0
19.7
24.0
203
I
90-109
26.6
1t0-I29
18.7
2L1
182
130-179
25.1
28.4
23.9
25.6
22.6
25.t
19.7
22.6
21.5
22.7
19.1
21.1
203
204
209
203
204
1,023
141.0
1.46.6
142.3
144.2
I41.7
]43.2
k180 ~o~ of patients mean
i
Percentage distribution o{ patients by baseline ~as~ing blood g].~acose value i~ UGDP s~udy 7~.
Tahge 3 2 -
407
sum GTT (rag/100 mI)
% frequency in 1,004 patients
275-
349
0.1
350-
424
0.6
425- 499
5.6
500- 574
18.8
575- 649
14.0
650- 724
12.2
725- 799
&t
800-
874
8.1
875- 949
6.3
950- 1,024
5.6
1,025 - t,099
4.7
1,100 - t,174
4.2
1,!75 - 1,249
4.0
1,250 - t,324
2.2
1,325 - 1,399
1.8
1,400 - 1,474
1.2
1,475 - 1,549
0.9
1,550 - t,624
0.7
1,625- 1,699
0.2
t,700 - 1,77.4
0.2
1,775- 1,849
0.4
t,850- 1,924
0.I
1,925 - 1,999 2,000 - 2,074
0.1
Table 33 - Distribution of patients by baseline sum GTT (mean -2: SD = 797.9 -± 283.4) in UGDP stud3,7L
The same may be said{ for the t S T D group trea~ed w i t h a fixed a m o u n t of insulin (10 to 16 U according to b o d y surface) irrespecthTe of their metabolic condition (hypoglycemia or hy-pergiycemia). N o careful, conscientious physician w o u l d d r e a m of doing this in. his daily practice.
5) The evd~ation o/ rnetaboSe control was inadequate. Q u a r t e r l y foIlow-ups w i t h fasting blood glucose assay were considered su~cien~ to 408
relative body weight
% frequency in 1,027 patients
0.60 - 0.69
0.i
0.70- 0,79
02
0.80- 0.89
2,0
0,90 - 0.99
4.7
1.00- 1.09
12,8
t.10 1,19
15.7
1.20 - t 29
16.9
130- 1,39
13,5
1.40 t.49
1t.1
-
-
1.50 - 1.59
8.2
1.60 - 1.69
5.6
1.70- 1.79
3.7
1.80- 1.89
1,8
1.90- t ,99
1.2
ZOO- 2.09
0.8
2,10-2.19
0~7
2.20 - 2.29
0.6
2,30 - Z39
0,2
2.40 - 2.49
0.1
2.50 -
2,59
2,60 - 2,69
0.t
Percentage distribution of patients by baseline relative body weight (mean SD = t.33 ± 027) in UGDP study 7L
Table 34 -
asses the degree of mecaboiic control. This is a very debatable procedure since postprandial glycemia would have yielded much more informative data. t But the choice of this criterion becomes imaeed amazing if we consider the me~hod adopted for evaluating t,he degree of metabolic control a&deved by the rations drugs. W h a t was u s e d in. the U G D P protoco! (see tab, 8) was the baseline bIood g/ucose l e v d of a short G T T initiated 30 rain after administration of the chosen drag. This means that metabolic control was assessed on the basis of blood glucose 1eve1 in the fasting subject and again 30 rain after insulin or tolbutamide or pherfformin or placebo, tt is indeed dif~cu!t to teIl the meaning of these values. 0~
6) Dietary treatmen~ o~ overweigh~ subjects was not programmed. A large proportion of the patients (56%) were overweight. The experimental protocol speaks of the introduction of a diet 'apt to reduce and maintain' body weight within-i 15% of ideal weight. In other words, the UGDP workers were only concerned with correcting major obesity so that many of the subjects included in the study remained overweight. 7) Patienf compliance with dietary prescriptions was not evaluated. The criteria on which the study of patient adherence to the therapeutic schedule was based are debatable. First of all, they completely neglect compliance with dietary norms. Besides, the three categories into which patients are subdivided according to adherence (tab. 9) do not stand up to closer scrutiny: indeed, only in the 3rd category (high), which inc!udes 61.6% of the population studied, was compIiance such as to justify their evaIuation in the study. In our opinion adherence in the remaining two categories was so low that almost 4098 of the population should actuatly have been excluded from evaluation. ERRORS IN THE PERFORMANCE OF THE STUDY
Several errors in the performance of the research have been listed in the large number of critical reviews. We only mention t~celve because they seem to us sufficiently relevant to in~uence the outcome of the investigation. 1) Careful scrutiny of tabs 31 and 32 shows that 63 patients included in the study were certainly not diabetic, i.e. with a sum GTT
4 t~0
-
regular assays of fasting blood glncose levels were done only in 434 of the 823 subjects (fig. 1); -
-
short GTTs were performed regularly in only 396 of the 823 subjects (fig. 2); -
-
- - changes in body weight were regulady checked only in 510 of the £23 subjects (fig. 3). But the conclusions of the UGDP study were drawn from the whole population and not from the data concerning the subjects for whom this very important information was available. We must ask ourselves whether this 40% of the population which had not been so weI1 studied may not hide the explanation of the higher coronary mortality found in the TOLB group. 5) The criterion of a 5-year life expectancy was not always fulfilled; suffice it to point out that the popnIation included subjects above the age of 70 with signs of coronary disease. 6) The diabetics included in the study came from different sources (private practice, hospitalized subjects, outpatients). 7) In some of the Clinical Centers there was a prevalence of women and of white subjects. Myocardial infarction is known to be more frequent in diabetic women and it is equally known that the black race is comparatively immune from infarction. This excess of white women was noted particuIarly in the two Clinical Centers where the higher mortaIity from cardiovascular causes was observed in the TOLB group. 8) Causes of death were partly diagnosed on a dinicat basis and par@ from post-•orterv findings. This is particularly relevant for deaths due to myocardial infarction since in diabetics this is very frequently painless so that it can be diagnosed only at post-mortem. It should be added that in some groups post-morte;e~ examination was performed more frequently than in others: as regards deaths presumably due to cardiovascular causes, 50o26 of patients dying in the TOLB group and onIy 20% of those in the PLBO group had post-mortem examination. 9) At the end of the study (1968) the code according to which electrocardiographic changes were evaluated was substituted by another one. According to the criteria used in the original experimental protocol (Minnesota code), electrocardiographic changes were present in 25% of the TOLB patients and 15% of the PLBO subiects. The cardiovascular risk was therefore higher at baseline in the TOLB group. According to the criteria used subsequently (modified Minnesota code), these percentages had been reduced to 4 and 3 ~ respectively. The authors never gave a satisfactory explanation of this belated and relevant change in their protocol. 10) Cardiovascular risk does not appear to have been homogeneously eaIst_,bnted in the various experimentaI groups. Tab. 35 shows the cardiovascular abnormalities present in the UGDP patients. Taken separately, they were unevenly distributed in the five treatment groups; thus, there was a larger number of hypertensive subiects in the PLBO as compared to the 411
cardiovascular risk factors
PLBO
TOLB
ISTD
IVAR
PtKEN
all
;6.8
30.2
30.9
28.1
28.4
30.9
4.5
7.6
5.8
5.0
4.6
5.5
5.0
7.0
7.7
3.5
7.4
6.1
significant ECG abnormaIity
3.0
4.0
5.3
4.0
5.0
4.3
cholesterol> 300 mg/100 mi
8.6
15.1
16.4
3.4
11.1
13.0
one or more of above factors
.7.3
47.9
50.2
1.5
43.5
46.1
significant ECG abnormality or history of angina pectoris
7.1
9.5
!2.0
7.0
1!.9
9.5
hypertension history of digitalis use
history of pectoris
Table 35 -
angina
Cardiovascular risk factors (%) at entry by treatment groups in UGDP study :~.
P H E N group and a larger number of cases of hypercholesterolemia in the TOLB as compared to the PLBO group. According to the researchers the sum of risk factors was fairly homogeneously distributed among the five groups but we do not exactly know the specific influence of each of the risk factor on the process of atherogenesis: does hypertension carry a greater risk of atherogenesis than hypercholesterolemia, or vice versa? 11) The research on pherdormin (which was started 18 months later than the rest) was limited to 6 Clinical Centers and the comparison has enormously complicated the statisticaI problem. In fact, in the Non-Phenformin Clinics, where the T O L B group was more thoroughly studied, excess mortality was found in this group w191e this excess was not observed in the Phenform~l Clinics. There are therefore valid reasons to suspect that if the TOLB group subjects had been evenly distributed throughout the 12 Clinical Centers, the final data would have been different. 12) In the authors' intentions the research should have been doubleblind. But obviously in the insulin-treated groups both physicians and patients knew that hormone treatment was being given and at least t'he physicians must have been aware of the different dosages applied in the IVAR and ISTD groups. As for oral treatment, the fact that in groups 7, 9, 10, 11, and 12 of tab. 1 placebo was administered in the shape of capsules made it possible to recognize totbutamide which was presented in tablets. 412
ERRORS OF DATA PROCESSING 1) For the purpose of statisficaI processhqg of the data the UGDP examined the possibiIity of risk factor combinations; the simultaneous presence of two or more pathological elements in the same patient is in itself a new factor which deserves to be taken into particular consideration. Experts of statistics therefore took into account the so-called 'unions', i.e. the coincidence of three of the highest risk facto*_'s (electrocardiographic abnormalities, use of digitalis, angina pectoris) and of two additional features (hypertension and hypercholesterolemia). On the other hand they did not consider arterial calci~cation which was particularly prevaJ.ent in the TOLB group. Actually there are no standardized statistical principles for this type of calculation. The computer can only facilitate the task by allowing selected combinations of the factors involved to be carefully analyzed. The principal method applied by UGDP was a slightly modified multiple logistic regression mode1. Apart from the complex equations to be used for cala~lations of this type, the greatest shortcoming of the regression model is that the combination of variables is linear. Thus new combinations are introduced which are the resuh of the sum of ea& variable individually but any effect other than the simple sm-n of the individual action of each variable is consistently neglecte& 2) Throughout the UGDP study any information avaiiable was recorded on special cards and automatic~.Iy transcribed on magnetic tape so as to be always readily available for immediate computarized analysis. This is one of the outstanding features of UGDP because it gives an excellent example of data analysis in a long-term dinicaI study. But even in this field, little attention was paid to the description of the methods used. The data obtMned had to be converted into magnetic traits so as to be suitable for computer calculation. However, we are not told what method was used for this conversion and especially whether there was a direct check on the data once they had beef, registered on magnetic tape. 3) For the purpose of statistical evaluation, only fatal cardiovascular accidents were considered whereas there is only scanty reference to nonfat~ vascular complications. But the main declared purpose of the study had beer, to evaluate the suitability of treatment with oral hypo~ycemic agents as far as the prevention of cardiovascular complications was concerned, irrespective of their outcome. The UGDP report explidtly states that while there was no difference between the four treatment groups as regards the frequency of fatal and nonfatal vascular complications, there was significant excess mortality in the tolbutamide group. We must therefore ask ot~rselves whether this was due to chance, to differences in baseline conditions, or to tolbutamide in itself, or to the interference of factors different from those mentioned above. ERRO,R8 IN DATA ~NTERPRETATION Obviously, after having listed dl these criticisms concerning the approach and operation of the study, little room is 1eft for consideration of 413
mistakes that may have been made by the UGDP investigators in interpreting their data. Perhaps the most serious of these mistakes was made in I969 when it was decided to abandon the tolbutamide experiment although not a few doubts must already have arisen among the experimenters. This decision, based on premature wholesale interpretation of the study and/or on hasty evaluation of the data, perhaps did not influence the results of the study since in 1968 the data which appeared in the press in 1970 had already been gathered. However, it led to two very important consequences: 1) first of alI, the experimenters wasted the opportunity of finding out whether the vascular accidents which they thought induced by totbutamide continued to increase progressively. Perusal of fig. 2 shows teat global and cardiovascular mortality followed a different pattern in time among the four treatment groups. The possibility cannot be dismissed that, if the experiment had been carried on, things might have changed in some way or other; 3 years after starting the study mortality was higher in one of the insulin groups, a difference which subsequently disappeared 6~; 2) besides the decision generated a certain amount of psychological obstinacy in the research workers as to the final interpretation of their data. Thus the UGDP shunned the dialectic confrontation which might have been fruitful for everyone. Because research, in whatever field and especially if it is costly and exacting, belongs to all man kind. CONCLUSIONS The problem of the influence of oral hypoglycemic agents on the onset and evolution of atherosderosis in diabetes, which came to the fore drarnaticaliy and unexpectedly after the St. Louis Meeting, has degenerated during the last four years so as to become an intricate controversy which is difficult to sotve. The principal aim of the present review has been to offer a panoramic view of this vexata quaestio, and at the same time to help readers to understand the reasons for its controversial character. First and foremost, there is the intricacy of the basic problem. Atherosclerosis as a disease is somehow connected with the process of aging and/or the cumulative effect of external agents. But, as is true for ali phenomena of aging, it does not start in all subjects at the same age and does not progress at the same rate and to the same degree. External agents may play a more or less decisive role. As for diabetes melIitus, it adds a whole series of further problems to this already intricate question~ Only two things are certain: first~ that diabetes anticipates, aggravates and ex~ends the disease; and secondl that nowadays atheroscterosis is the prindpal cause of death in diabetics. Since there are, at least in theory, four possibilities of treating adultonset diabetes (diet only, insulin, sulfonylureas, bi~,lanides), one must IogicalIy ask oneself whi& of these therapies is most apt to delay or timit vascular degeneration in these patients. The di~cnlty of having ac one's disposal weli-ba!anced groups of diabetics with similar clinical characteristics cannot have escaped the reader. 414
For an ideal experiment of this kind the following would have to be postulated: - -
the largest possible number of patients;
- -
having the same age;
-
-
w
i
t
h
the same type and severity of the disorder;
- - with the same cardiovascular risk (geneticalIy determined, metabolic origin, bIood pressure, etc.); - - with the same risk arising from exogenous atherogenic factors (smoking, stress, social class, profession, etc.); - -
following the same diet;
and in whom by any one of the various types of treatment the best possible metabolic control should be obtained. - -
Even given the availability of a group of patients meeting all these requirements, by submitting them at random to therapies of diverse nature and efficacy one would create anomalous experimental situations which would prove detrimental to the study. One would in fact find oneself with diabetics whose treatment was inadequate to their needs and others in whom the therapy appLed would be insuNcient. On the other hand, in order to use treatment schedules of varying intensity and indication one would be compelled to use groups of patients with different degrees of severity of the disease so as to apply the various treatments appropriately. But this would disturb the homogeneousness of the population examined. These considerations account for the difficulty encountered by all studies aimed at comparing the effect of various treatments on the cardiovascular apparatus in diabetes metlitus. The present review reports the results obtained by different groups of research workers in studies intended to evaluate the cardiovascular effects of the treatment of diabetes with hypoglycemic agents: 9 of these studies provide evidence of a protective action of these drugs or at least exclude negative consequences; 4 suggest that there is the possibility that oral hypo~ycemic agents are responsible for coronary disease. A sMpIe numerical comparison of these studies woutd certainly be naive and based on a faIiacy. In fact, among these 13 studies the one conducted by the UGDP stands out for its ambitious polycentric experimental design, the large number of cases examined, the many famed research workers involved and the large amount of money spent. This is why the UGDP report has been viewed critically by so many diabetologists, and why the analysis of the UGDP data takes up most of the space dedicated to the present review. tt is in fact of paramount importance to know whether these data should be considered reliable or not. If they were established beyond discc~ssion, ali the other studies, whether in accordance or not with the UGDP data, wou!d count for tittle. In the UGDP study cardiovascular mortality was notoriously more prevalent in the tolbutamide-treated subjects, and statistical analysis of the 415
data showed the difference from control subjects to be reasonably signifi. cant. But this higher prevalence might be accounted for in different ways if the following six arguments are kept in mind, 1) The data of the various Clinical Centers show astonishing discrepancies. Thus, the fatal cases reported by 4 Centers (Chicago, St. Louis, San Juan, Seattle) did not include a single tolbutamide-treated patient while in another 3 Clinics (Cincinnati, Minneapolis, Boston) they amounted to 24-32% of all fatalities in the T O L B group (tab. 36). Further, 2 Centers (Cincinnati and Minneapolis) accounted for almost 5 0 % of the deaths due to cardiovascular events in the TOLB group. This arouses the suspicion that the experimental model was not suf/iciently rigid to avoid the unbalanced distribution of cardiovascular risk factors in the various groups. I f in fact we could neglect the data of these 2 Clinics, cardiovascular mortality in the TOLB group woutd not differ from that of the other groups.
PLBO
TOLB
ISTD
IVAR
no. of deaths
0.0 30,4 5.3 t3,6 13,6 8.7 15.4 6.7 9.1 t0.0 0.0 0.0
4.5 31.8 5.6 33.3 10.0 18.2 18.2 29.4 0.0 0.0 0.0 0.0
0.0 t6,7 0.0 20.8 9.5 13.0 0.0 12,5 8,3 8.3 7,7 9,1
0,0 23.8 10.0 8.3 I3.6 12.5 0.0 6.7 9.1 0.0 7.7 0.0
23 4 18 10 12 4 9 3 2 2 t
10,2
14,7
95
8.8
89
0.0 8,7 0.0 9.1 13.6 4.3 0.0 6.7 9.1 0.0 0.0 0.0
4,5 31.8 5.6 25.0 10.0 13.6 18,2 23.5 0.0 0.0 0.0 0.0
0.0 i6,7 0.0 8,3 0.0 8,7 0,0 6,3 8.3 8,3 7,7 9,t
0.0 19.0 5.0 8.3 0.0 12.5 0,0
1 17 2 12
0,0 0.0 0.0
1 1
4.9
12,7
6,2
5.9
61
dl causes (.%) 1. 2. 3, 4. 5. 6. 7. 8. 9. 10. 11. 12.
Baltimore Cincinnati Cleveland Minneapolis New York Williamson Birmingham Boston Chicago St. Louis San Juan Seattle
all clinics
1
cardiovascular cause (%) 1. 2. 3. 4. 5. 6. 2. 8. 9. t0. tl. 12.
Baltimore Cincinnati Cleveland MinneapoEs New York Wiltiamson Birmingham Boston Chicago St. Louis San Juan Seattle
all clinics
Table" 3 6 -
4t6
Causes of death in the Clinic Centers of UGDP study 7%
6.7 9.1
I
59
2 I
37 1
2) If tolbutamide were really harmful to the cardiovascular apparatus, it would provoke not only severe and fatal compiications but also minor clinical incidents. There should therefore have been an increase in nonfatal cardiovascular episodes in the TOLB group. But this was not the case. On the contrary, if all fatal and nonfatal cardiovascular events are added together, their prevalence in the TOLB group is lower than that of the PLBO group. Stringent logic would therefore lead us to hold tolbutamide responsible not for an increase in coronary disease but for a more severe prognosis of episodes of infarction. However, this appears to apply only to the 2 Centers (Gncinnati and Minneapolis) where half the carddovascular fatalities were registered. There are therefore valid reasons to maintain that in these 2 Clinics the death rate among patients with infarction was higher than in the other experimental Centers. As a resuh, we are left with the suspicion that treatment of infarction, in these 2 Centers was somehow different from that of the others. 3) Next, the fact that post-~nortem was carried out in about 50% of the fatalities of the TOLB group and in about 20% of the PLBO group, This difference is partly clue to the fact that the majority of deaths in the TOLB group occurred in hospitalized patients (70%) while in the PLBO group this proportion was lower (50%). tn view of the targe proportion of diabetics in whom myocardial infarction is asymptomatie, it is possible that the difference in cardiovascular mortality is due to the different frequency with which post-v~orge,~ was performed, 4) It is indeed amazing that among the 2i deaths in the PLBO group not a single one was attributable to myocardial infarction, which is known to be the most frequent cause of death in diabetics and is found in 40% or more of all the published series. The fact that, for reasons which escape statistical anaIysis but which can easiIy be guessed, this totally unlikely situation has occurred in the PLBO group is certainly to a large extent responsible for the difference between the TOLB and PLBO groups as regards death from cardiovascular causes. Keen ,4 called this absence of coronary death in the PLBO group 'the thirteenth chime which casts doubt on all the rest'. 5) Then there is the incomprehensib!e finding shown in fig. 3 concerning body weight in the experimental groups: only the PLBO group showed a certain weight reduction. (2-3%) persisting throughout the experiment. In the other 3 groups (data concerning body weight have not been made known for the PHEN group) a slight weight loss (about 1%) was soon fotlowed by a return, to £qe initial values and subsequently by a further increase of the average so that at the end of the experiment body weight was some 2% above the starting values. This can only be accotmted for by the neglect of dietetic norms by subjects in the TOLB, IVAR, a~d ISTD groups who must have increased their food intake. We cannot help asking ourselves why this did not occur in the PLBO group, The di~culty of carrying out a trial of this kind in double-blind 417
conditions has already been stressed. Could it be that the experimental protocoI led the patients of the PLBO group or their physicians to lay greater ,stress on diet? Or should we come to the conclusion that insulin or tolbutamide brought about an increase in appetite and hence in food intake? AtI this leads us to think that only the PLBO group had satisfactory compliance with dietary prescriptions. We should like to consider this the most precious tipshot of the whole research in that it underscores the importance of diet in the prevention of cardiovascular complications of diabetes. 6) Finally it should be pointed out that, apart from the failure to normalize body weight, not even metabolic control was satisfactory. As shown in fig. 1, fasting blood glucose rose rather quickly to values which were even higher than the starting ones, except in the IVAR group, thus failing to attain the prime objective that every physician aims at in the treatment of his diabetic patients. Indeed, the fact that in a study on the effectiveness of hypoglycemic therapies in preventing the cardiovascular risk involved in diabetes mdlitus, adequate care was not taken to ensure the foremost resuh of the treatment applied, i.e. correction and maintenance of metabolic balance in at least one of its most obvious aspects, viz. control of hyperglycemia, casts serious doubt, if not discredit, on the global evaluation of the study and therefore on the interpretation of its findings. These six considerations together with the large number of shortcomings in the approach and performance of the study cast heavy doubts on the validity of its results and on the justification of the conclusions drawn from them. They are like six red flashlights on the dashboard which arouse the suspicion that the research has not been planned and/or carried out correctly. Indeed, it has been most appropriately pointed out that 'computers and statistics furnish answers that are infallible and indisputable only when the original data are infallible and indisputable" 50 If we consider the enormous financial and organizing effort that was made to obtain results which carry so little conviction, we cannot help remembering the devastating judgement pronounced by DOLGERon this study: 'an American iqasco'. The reasons for this failure can be defined only if and when the UGDP research workers will abandon their present rigid 'and adamant defence and wilt cooperate in trying to find the errors which at present are not sufficiently clearly documented in the four UGDP publications. And this will eertaiMy be a use6al effort because clinical pharmacologT is in need of longitudinal multicenter studies like that of the UGDP provided these studies are reliable and tmthfial. Physicians may with a dear conscience continue to prescribe oraI hypogtycemic agents, though they should always remember that diet, carefully worked out, patiently explained and scrupulously observed remains the most important prophylaxis of diabetic complications. The danger does not arise from these drags but from. their use in the absence of precise d{etary control. 418
REFERENCES 1) AMERICANDIAbetes AssociaTion: Labeling of Oral tIypoglycemic Drugs - Diabetes 21, 833, 1972. 2) Assoc~t~TtoNDES DIg~TOLO¢LTESU~ LaNGUZF~AN~AlSZ: 7'~ propos de t'UGDP - Diab~te {Le Raincy) 19, t20, 1971. 3) AUSTt~.ALIANDRUG EVALUATION COMMITTEE: OraI }-{ypoglycaemic Agents in the Treatment of Diabetes Mellitus - New- Ethieals 9, 197!. 4) BaLOn~MOSM. C., GLEASONR. E., B~.aDZ~E~:R. F , MA~Le A.: Long-Term Tolbutamide Therapy in Diabetes: a Controlled Study of the Frequency of Cardiovascular Disease and Other Findings - In: BUTTEr.5"IeLnW. J. H., VAN Wzs'rEzlNa W. (Eds): ToIbutamide... after Ten Years. Excerpta Medica FouMation, Amsterdam, 1967; p. 270. 5) BALODIMOSM, C., ?V~ARBLEA., RIPPEY J. H , LEGGM. A., KuwaBa~a T., GLEASON R. E.: Pathological FindLngs after Long-Term Sulphonylurea Therapy - Diabetes t7, 503, 1968. 6) BEm4~-mlMR.: Les antidiab&iques oraux jouent-its un r61e duns Ia mortalit6 cardiovasculaire du diab~te? Apropos du rapport de I'U,G.D.P. et des controverses qu'il a suscitdes. Tab!e ronde - Presse me'd. 79, 2437, 1971, 7) BLoonwo.~r~ J. M. B. Jr., HA~rwI G. J.: Histopathologie Lesions Associated with SuIfonyluma Administration - Diabetes 10, 90, 1961. 8) BLt~eN~a5 H. T., MoRris A., Gocm:~B~aa S.: A Study of Ia:sions of the Intramural Coronary Artery Branches in Diabetes Mellitus - Arch. Path. 70, 27, i960, 9) BOSTON COLLA~ORATIWDF.UG SUZW'XLL~WCEPROGRAM: Coffee Drinking and Acute Myocardial Infarction - Lancet 2, 1278, 1972. 10) BoYsz D., BHATIa S. K., HADVeN D. R., Mo~'raOM~a~ D. A. D., We~.w~ J. A.: Ischaemic Heart-Disease in Diabetics. A Prospective Stu@ - Lancet 1, 338, 1972. t l ) BRaDteY R. F., FOiS~AM P. H., DOL~ez H., As.~tMo~e J, D,, SEL~Z~t H., Bz.assl S.
12
13 14) I5) t6) 17) 18) !9) 20) 21)
22)
F~5:cs P. W., F~nMa.WR., F~_~:ro~R. I., Fx~nm.~ G. J., Ga:rzs E. W., t:[A~ R. L., I-L~e.wzE. A., JAco~soa W. E., K~ar G., Kaos~c:< A., ~vL,~a~.eA., MI~5~ L,, Moss J. M., McDo.~a~ G. W,, Ovv~-~r~Ix~zxH. J., O ' S u ~ v a ~ J. B., PEEBLEs-MEYEP.SM., SCUV~AC>a~RO, P., S~tUM~ C., SZLV~ A. A., SIs~ C. W., SOEL~X~ J. S., S~P~NS J. W., WELSH G.: Treatment o~ Diabetes. Statement from a Newly Formed Committee - Northw. Mad. (Seattle) 70, 8, 1971. CAN'AA.NIZ.: Some Evidence for the Ianocence of Long-Term Tolbutamlde Treatment on the Frequency of Cardiovascular Disease - In: SaAFR~ E, (Ed.): Impact of Insulin on Metabolic Pathways. International Symposium Commemorating the 50th Arm_iversary of InsuIin. Jerusalem 197t; Israel J. med. Sci. g, 699, I972. Ca~.a~.~ Dn~.~eTic Assoc~.r~o~: Statement Concerning To!butamide - Rx BuItetin, May 1971; p. 65. CAXLSO>~L. A., BG~Tm~xL. E.: isehaemic Heart-Disease in Relation to Fasting Values of Plasma Triglycerides and C~nolestero!. Stockholm Prospective Study - Lancet 1, 865, t972. Calzs:rz6tI S., Pzasso~ G., Scs:~zs:r~,~ B.: A Critical Evaluation of a Diabetes Survey: Ten Years Afterw-ards - 8th bieeting of the European Study Group for Diabetes Epidemiology, Milan, May 3-4, 1973. Co~,~x~r~'ee o~ Sa~e~x" o~ De,~:c-s: Report for 1969 and 1970 - I-Ier Majesty's Stationery Office, London, 1971. CONS~rAMG. R.: Is the Use of SuIphonylurea Derivatives in the Treatment of Diabetes MeI1ims Dangerous? - DiabetoIogia 7, 237, 197i. Co~,~ieLn J.: The Universig; Group Diabetes Program. A Further Statistical Analysis of the Mortality Findings - I. Amer. ned. Ass. 217, 1676, !97t. D e u ~ s c ~ D~A~e~s-Gss~L~sc~.r~T: Kein Risiko bel Tolbutami&Therapie des Diabetes - Dtsch. Arzt 2204, 1970 D z u r s c ~ DIAI~ETEs-GEsELLSC,~.~T; Langzeitbehandlung der Zuckerkrankhdt n i t Tabletten -Dts&. Arzt 640, 1971. Doza~ J. T., Daw~x T. R., Kx~a~z~ W, B., K~CH S. t-L, K~a>m-H. A.: The Relationship of Cigarette Smoking to Coronary Heart Dbease. The Second Report of the Combined Experience of the Albaw (N~ YJ and Framingham (Mass.) Studies - J. Amer. ned. Ass. 190, 886, 1964. Daz_r~yM, io,T*i~Io,.xsYF, J.: Oral I-~i?/poglycaemicAgents and Cardiovascular Deaths in Diabetes - Acta diabet, lat. 9, 645, 1972.
4!9
23) ED*TOR*AL: Are Antidiabetic Drugs Dangerous? - Brit. reed. J. 4, 444, 1970, 24) Enzro~zaL: Status of Problem of Usage of Tolbutamide. Pre!iminary Statements Diabetes 19, 467, 1970. 25) Et~i:rO~IAL: The Tolbntamide Evidence - Lancet t, 171, 197t. 26) EDI?fOICIAL:Use of Tolbutamide and Other Sutphonyinreas in Diabetes Mellitus, in the Light of the University Group Diabetes Program (UGDP) Report -Med. Proc. 17, 265, 1971. 27) Enw~aec~s C. C.: Oral Hypoglycemie Agents. Report of the Food and Drug Administration - Diabetes I9 (Suppl. 2), viii, 1970. 28) FAJANS S. S.: Review of University Group Diabetes Progrmva (UGDP) - Evian Diabetes Symposium, September 3-4, 1971. 29) F~AXXLEZG. R.: PharmacologicaI Fibrinolysis in Occlusive Vascular Disease, and its Relation to Diabetes Mellitus - Postgrad. med. J. 45, 52, 1969. 30) FE~NSTEINA. R.: Clinical Biostatistics. VIII. An Analytic Appraisal of the University Group Diabetes Program (UGDP) Stud?" - t e n . Pharmacol. Ther. 12, 167, 1971, 31) F~Lm J.: Discussion Remark. Panel Discussion on: OraI Substitutes for Insulin: Modern Appraisal of Theb Action and Indications - In: SHA~xr~ E. (Ed.): hnpact of Insulin on Metabolic Pathways. International Symposium Commemorating the 50th Anniversary of Insulin. Jerusa!em 1971; Israel J. reed. Sci g, 708, 1972. 32) FINE5 J.: Vascular Complications in Diabedc Patients in the Jerusalem Area Treated by Diet, Tolbutamide and Insulin .- In: SHa~RI~ E. (Ed.): Impact of Insulin on Metabolic Pathways. International Symposium Commemorating the 50th Anniversary of Insulin. Jerusalem 1971; Israel J. reed. g, 705, 1972. 33) Foo~ ~'~n D~u6 P~MINISTXA'rtON: Final Labeling of Oral Hypoglycemic Drugs - FDA Drug Bull., Magg[o 1970. 34) G.~cIa M. J., Gormo>~T., McNA~aaRAP. M., KANNELW. B.: Morbidity and Mortality in Diabetics in a General Population. Sixteen-Year Follow-up Experience in the Framingham Study" - Diabetes 19 (Suppl. 1), 375, 1970. 35) GEL~ J., E~aR~sm I., D~'CI>~'.T!~E P., R_TCHAXDJ. L.: L'~Iectroeardiogramme dans une popu!ation ~ haul risque - In: Enqu~.te @iddmiologique sur les facteurs de l'athdroseldrose. (Premiers rdsnhats 1963-1964). Bull. Inst. nat. Sant6. et Rech. mdd. 22, 279, 1967. 36) GOL~NE~ M. G,: introductory Remarks. Panel Discussion on: Oral Substitutes for Insu!in: /Vlodern Appraisal of Their Action and indications - In: SHAI:XI~:E. (Ed.): Impact of InstHin on MetaboIic Pathways. International Symposium Commemorating the 50th Anniversary of Insulin. Jerusalem 1971; Israel J. reed. Sci. g, 677, 1972. 37) G o L ~ x M. G., tgx-~a~rTe~umG. L., P~ouT T. E.: E~ects of Hypoglycemic Agents on Vascular Complications in Patients with Adult-Onset Diabetes. III. Ctinical Implica. lions of UGDP Resuhs - J. Amcr. reed. Ass. 21& 1400, t971. 38) Goemo~ T., Ka.,',~,~eLW. B.: Predisposition to A:herosclerosis in the Head, Heart, and Legs. The Framingham Study - J. Amer. reed. Ass. 22t, 661, 1972. 39) Ha~D~N D. R., M o ~ ' ~ o r , , ~ D. A. D., W~aw~ J. A.: Myocardial Infarction in Maturity-Onset Diabetics. A Retrospective Study - Lancet, t, 335, 1972. 40) JA~ee'rT R. J.: Diabetes, Hyperglycaemia and Arterial Disease - In: Lr~NDBmKK., KeEN H. (Eds): Nood Vessd Disease in Diabetes Mellitus. Casa Ed. ~ II Ponte >>, Milano, 1971; Acts diabet, tat. g (Suppl. 1), 7, 1971. 4t) KaPee~:c A.: Tobacco and Diabetic Angiopathy - In: LUN~;eK K., K~N H. (Eds): Blood Vessel Disease in Diabetes Mdlims. Casa Ed. <>,Milano, 1971; Acta diabet, la.< g (Suppl. 1), 429, 1971. 42) K~N H.: Discussion Remark - 30th Annual Meeting of the Amer. Diabetes Ass., St. Louis, 1970. 43) K~E,,;H.: Minimal Diabetes and Artet:iaI Disease: Prevalence and the Et~ec~ of Treatmerit - In: CA~x~e~za~-DY,vAnoS R. A., COnE H. S (Eds): Early Diabetes. Academic Press, New York-London, 1970; p. 437. 44) K~m~ H.: Factors Influencing the Progress of Ad~.erosderosis in the Diabetic - in: I,u~Basa K., K~ex H. (Eds): Blood Vessel Disease in Diabetes MeIiit,as. Casa Ed. <~I1 Ponte ~>, Milano, 1971; Acts diabet, tat. 8 (Suppl. 1), 444, 1971. 45) K~e~ H., JAa~'r'r R. J~: The Effect of Carbohydrate Tolerance on Plasma Lipids and Atherosclerosis in Man - in: 7o~7~s R. J. (Ed.): Atherosderosis. Proceedings of the Second International Symposium on AtheroscIerosi~, Chicago 1969. SprLnge>Verlag, Berlin-Heidelberg-New York, 1970; p. 435. 420
46) KE~N H., JA~ET:t' R. J., C.H~OUVE~aKISC., BOYNS D. R.: The Effect of Treatment of Moderate Hyperglycaemia on the Incidence of Arterid Disease - Postgrad. reed. J. 44, 960, 1968. 47) KEEN H., Ros~. G. A., PY~:E D. A., BoYNs D. R., CHSOUVE~aKIS C., M~s:ea-e S.: Blood Sugar and Arterial Disease - Lancet 2, 505, 1965. 48) Ka~.usE U.: To!butamid und Myocardi~arkt - Pharm. Ztg (Frankfnrt) 15, 523, 1971. 49) Labding of Oral Hypoglycemic Drags - Diabetes 21,833, 1972. 50) LOU~ATIh~SA.: Therapeutic Aspects of Diabetes. Remarks Concerning Some Recent Reports on Therapy of Diabetes - In: SaAva~a E. (Ed.): Impact of In.sulin on Metabolic Pathways. International Symposium Commemorating the 50th Anniversary of InsuHm Jerusalem 1971; Israel J. reed. Sd. 8, 883, 1972. 51) LVBETZKIJ.: R~flexions sur t'angiopathie diab&ique. Rdlations avec Ins traitements hypogIycdmiants - Nouv. Presse todd. 1, 715, 1972. 52) L~DB.~K K., KaaENH. (Eds): Blood Vesset Disease in Diabetes Mdiitus - Casa Ed. ~ tl Ponte ,,, Mdtano, 1971; Acta diabet, int. g (Suppl. 1), 1971. 53) MtCeLE*:R.: The Effect of Diabetes and Insulin on Bioche~NcaI Reactions of the A> terial Wail - In: Lu~sm:< K., KEeN H. (Eds): Blood Vessel Disease in Diabetes Melliras. Casa Ed. ~{I1 Ponte ~>, Milano, 1971; Acta diabet. Int. 8 (Suppl. 1), 68, 1971. 54) MAY~'EE. E., B~.maEs J. M., WF.AwI~.I.A.: Platelet Adhesiveness, Plasma Fibrinogen and Factor VIII Levels in Diabetes MeIlims - Diabetologia 6, 436, 1970. 55) MInLex M.: Factors Influencing the Progress of Atherosclerosis in the Diabetic - In: Lu~B.~K K., KeEN H. (Eds): BIood Vessei Disease in Diabetes MeIlims. Casa Ed. <<11 Ponte >>, .bfitano, 1971; Acta diabet, lat. g (Suppt. 1), 434, 1971. 56) ODEOAARDA. E., SKF.AL~GGS. A., HELLEM A. J.: Increased Activity of 'Anti-Willebrand' Factor in Diabetic PIasma- Thrombos, Diathes. haemorrh. (Stuttg.) 11, 27, 1964. 57) OST~.~EX L. D. jr., F~aNcIs T. Jr., HA'ZNE~N. S., KJELS~ERO M. O., E>S~'E~N F. H.: The RelafionsMp of Cardiovasctdar Disease to Hyperglycaemia - _Ann.intern. Med. 52, 1188, 1965. 58) PaAs~rav~ J.: Long-Term Tolbutamide Treatment after Myocardial Infarction. A Ctinica1 and Biochemiced Study of 178 Patients without Overt Diabetes - Acta me& scan& 307 (Suppl.), 1, 1970. 59) Paas~K~vx j.: Long-Term To!butamide Treatment of S,arvivors from Myocardial Infe¢ction .. In: L b ~ . ~ K K., K~Ex H. (Eds): Blood Vessel Disease in Diabetes Mellhus. Casa Ed. <~II Ponte ~>, NElano, 1971; Acta diabet lat. g (Suppl. 1), 437, t971. 60) P~AS~K~WJ., \graz4~a~ F.: Preventive Totbutan-~.deTreatment and Arterial Disease in MJId HypergIycaemia - Diabetologia 7, 323, 1971. 61) P~Ez~ea E. F.: Factors influencing the Epidemiology of and InterreIations between Atherosderosis and Diabetes Mellitus - In: Lu~*~.~: K., K ~ ¢ H. (Eds): Blood Vessel Disease in Diabetes MeUAtus. Casa Ed. <ee, A. G., L~_~.K. T., ScoTT R. F., Goo>aL~ F., T~OMAS W. A.: Comparison of Adipose Tissue Fat~- Acids a~.d PIasma Lipids in Diabetics from East Africa and the United States with Different Frequendes of Myocardial Infarction - Amer. I. CardioI. !0, 390, t962.
421
70) Statement on Treatment of Diabetes - Diabetes 19, 527, 1970. 71) Statement Regarding the University Group Diabetes Program (UGDP) Study. B2vIA Council on Drugs - Diabetes 19 (Suppt. 2), vi, t970. 72) STouz R. W.: Insulin-StimuIated Lipogenesis in Arterial Tissue in Reladou to Diabetes and Atheroma - Lancet 2, 702, I968. 73) TEUSCHE~A.: Sind ora!e Antidiabeti~a unxvirksam-oder gar gef'Shrtich? Das <~University Group Diabetes Program ~> (U.G.D.P.) als Ausgangspunkt einer Kontroverse tiber die Therapie des Ahersdiabetes - Schweiz. reed. Wscbr. 102, 848, 1972. 74) TEUSCm~ A.: Sind orale Antidlabetika unwirksam - oder gar gef~hrlich? - Tggl. Prax. I4, 177, t973. 75) TZAGOUm'~ISM., ILiaNEW,SON R.: Mortality from Coronary I-Ieart Disease during Phenformin Therapy - Ann. intern. Med. 76, 587, 1972. 76) Ua'Iwi~siT'z GRouP DIm3ETES PaOORAM: A Study of the Effects of Hypogtycemic Agents on Vascular Complications in Patients with Adult-Onset Diabetes. Part I: Design, Methods and Basdine Resuhs - Diabetes 19 (Suppl. 2), 747, 1970. 77) UNXWaSITY GRouP DIAbeTES Pt~oOsa51: A Study of the Effects of HypoNycemic Agents on Vascular Complications h'a Patients with Adult-Onset Diabetes. Part II: Mortality Results - Diabetes 19 (Suppl. 2), 789, 1970. 78) UNIVERSITYGROUP DIABETESPROORaM: Effects of Hypo~ycemie Agents on Vascular Complications in Patients with Adult-Onset Diabetes. IV. A Preliminary Report on PhevXormin Resuhs - J. Amer. reed. Ass. 217, 777, 197t. 79) WEF2CCEZ7. H.: Diabetologists Score FDA Action - Intern. Med. Diag. News 3, no. 24, t, 1970. 80) W~*SEN~ELD S.: Discussion Remark. Panel Discussion on: Orai Substitutes for Insulin: Modern Appraisal of Their Action and Indications - In: SHAF~m E. (Ed.): Impact of Insulin on Metabolic Pathways. InternationaI Symposium Commemorating the 50th Anniversary of Insulin. Jerusalem 1971; Israel J. reed. Sci. 8, 708, 1972. 8I) W~m3o~NT. A., CUMPSTON G. N., CULL~ K. J., CURNO'~VD. H., McCALL M. G., STEa~0USE N. S.: The Prevalence of Coronary Heart Disease and Associated Factors in an Australian RuraI Commnnity - Amer. J. Epidem. gg, 521, t969. 82) WELL~I.SN~K. F., LazARus S. S., VoLK B. W.: The Viscera in Sulfonylurea-Treated Diabetics - Metabolism 12, 959, 1963. 83) Woi~u) H~ALT~ ORGANIZATIO~:EpidemioIogica[ and Vital Statistic Report - World HeMth Organization, GenSve, 1969.
422
INSTRUCTIONS TO AUTHORS ACTA DIABETOLOGICA LATINA pubIishes original papers directly or indirectly ¢oneerned with diabetes metlims.
Mam~scripts: Only orlginal papers written in Eng!ish, and net published or in course of publication elsewhere, will be considered. The Editom reserve the right to make any linguistic corrections to the text and, if necessary, return the paper for rewriting to the author(s). Manuscripts should be submitted in duplicare (with two sets of rab!es and illu~strations or photocopies of the originals), typewritten on one side of the paper only and double spaced with a wide margin. Tit!e page: The first page of each paper should carry the title (main title underIined), the authors' names and the name of the Institute where the authors have conducted tbeir research work. A sho~ fide for the page headings should also be provided. Summary: Each paper must include a summa..~¢ of not more than 250 words. Key.wot&~: A list of 5d5 key-words for indexing purposes should be given by the authors. Chapters': ExperimenraI papers must be divided into c~hapters followtng this order: I) introduction; 2) material ~td mounds; 3) remits; 4) discussion; 5) summary; 6) references, Ciinical papers (case reports) must be divided into d~e ~ollowing chapters: 1) introduction; 2) case report; 3) dismassion; 4) summary; 5) references. Review articles: Review articles may be divided iron chapters at the authors' discretion. Tables and illustration;: Tables shoutd be wped on separam sheets with number and tide. Figures and graphs must be numbered aM accompanied by a clear tegend. The EdimriaI Board reserves the right to redraw graphs, figures and tables. Re~erences: References must be limited to publlcatior~ quoted in the text; they must be listed in alphabeticaI order, according to the surname of the go'st author, and be cited as numbers in the text according to ~ e quotation. For p~tiedieals, the following data must be successively reported: surname and initiaI(s) of the f~st name(s) of the author and all co-authors, bah dtIe of the publication, abbreviation of the perlodicai according to the World Medical Periodicals List, volume, first page and year of publication. For example: BANTING F. G,, BEST C. H.: The Internal Secretion of the Pancreas - J. Lab, din. Med. 7, 251, 1922. Book citations must include also Publislaers, city of publication, year of publication and page number. For e~mple: JOSLIN E. P., ROOT H. P., WHITE P., MARBLE Ax The Treatment of Diabetes Mettims - Lea & Febigcr, Philadelphia, 1959; p. 3i. For ardcles in books: surname and inltlal(s) of the first name(s) of author(s), title of article, Editor(s), fide of hook, edition (other than first), Publishing House, city of publication, year of puMication, page. For example: SAMOLS E., TYLER J. M,, KAJINUMAN H.: Influence of the Sulfonamides on Pancreatic Humoral Secretion and Evidence for an Insutin-Ghicagon Fe~tback System - In: !~ODRtGLrEZ R. R., VALLANCE-OWEN J. (Eds): Diabetes. Proc. 7th Congr. Int. Diabetes Fed,, Buenos Aires 1970. Excerpta Medica Foundation, Amsterdam, 197i; p. 636. Trade T,'avzes: Each pharmaceutical compound must he cited according to its chemical name, and ~ade names must be avoided. Footnotes: They should be avoided; if essentiaI, they should appear at the bottom of the corresponding page and be numbered consecutively. Fvdl address: At the end of each manuscript the exact postal address, eomnlete with postal code, of the senior author must be given. Should requests for reprints be handlred by someone other t h ~ r~e senior author (co-author); this should be indicated aecordingD. Gal&y proofs: The first proofs will be sent by the Publisher to the first named author (unless otherwise indicated), a ~ they must be returned wi~in 10 days from the date of disvatch, with the author's approval. At the expiry of this time limk the Publisher will provide directly for their correction. Correction of the second proofs will be carried out by ~ e internal staff. Re2rim~s: The senior author wit1 receive 50 reprints, free of charge~ In the case of papers with several au(hors, the Publisher will arrange the number o~ reprints directly with ~bem. Publication of the text, rabies, graphs, biocks in. black and white, and coloured plates is free of charge, All papers must be sen~ exctusiveIy to the following address: THE PUBLISHING HOUSE ~ IL PONTE ~ Via M. A. Colonna 27 - 20149 Milano, Its17
P u b b t i c a z i o e e bimesera~e e d i t a a c u r a d e l l a Casa E d i t ~ i c e ~ I L P O N T E >~ S . r , l . - C a s e i ! a P o s ~ d e n . 1071 20100 M i l a n o - D i e e t t o r e R e s p o n s a b i l e : l~roL G , P o z z a - A'atorizzaz{ooe n . 6598 de1 T r i b u n a I e d i M i ! a n o d e I ~0 l u g l i o 1964 - Ofi~cine G r a ~ c h e C h r i s t e n ,* V i a H o r n s , 39 - R o m a - S p e d i z i o n e i n a b b o n a r n e n t o ~os~ale g r , IV~
ACTA DIABETOLOGICA LATINA VOLUME XI
SEPTEMBER-OCTOBER 1974
NUMBER. 5
MISUNDERSTANDING OR ERROR? the UGDP report after four years edited by SERGtO NL&RIGO
co~ec~to~s
LUCIANO ADEZATI, PAOLO BRUNETTI, GUIDO POZZA, LIVIO ROBBA, GIANCARLO URBINATI