The Accuracyof Alkaline Phosphatase Isoenzyme Determination GEOFFREY C. LAMB, MD, GREGORY ROUSH A/ka/ine p h o s p h a t a s e isoenzyme determination (APID) is in c o m m o n u s e despite evidence s u g g e s t / n g that the resu/ts correlate poorly with a c t u a / s i t e s of disease. To assess the predictive v a l u e of this t e s t / n c/in/cal p r a c ~ c e , 99 APIDs p e r f o r m e d on 94 paffents were identified a n d the pat/ents" charts w e r e reviewed. Results of APID were c o m p a r e d with actual p a t i e n t d / a g n o s e s a s determined b y other means. The liver isoenzyme fraction w a s not v e r y a c c u r a t e / n p r e d / c t / n g the p r e s e n c e of//ver d i s e a s e / p o s i t / r e pred/ct/ve v a l u e 68°~). In contrast, the b o n e isoenzlm2e fraction w a s insensitive (56%) but a positive test predicted bone disease well (posi. rive predictive value 93%). The assoc/at/on of elevated transaminases with elevated alkaline pbosphatase On a chemistry prefi/e w a s a s u s e f u / a s A/OlD in identiflH.ng liver disease, suggesting that APID should not be done in this setting. Using this information, APID can be helpful in the assessment of an f l / p a t i e n t with an elevated alkaline pbosphatase. Key words: a/ka//ne pbospbatase isoenzyme determ/nation; isoenzymes, a / k a / / n e p h o s p h a t a s e . J GEN INYEI~N MED 1 9 8 8 ; 3 : 3 2 6 - 3 3 0 .
ELEVATIONof the concentration of alkaline phosphatase in serum m a y be a valuable clue in the e v a l u a tion of patients who h a v e poorly defined disease. However, b e c a u s e alkaline p h o s p h a t a s e is found in a variety of h u m a n tissues, a number of diagnostic possibilities must be considered. Therefore, the possibility of identifying the specific tissue source of the enzyme using alkaline p h o s p h a t a s e isoenzyme determination (APID) L 2 is attractive. Various studies h a v e b e e n d o n e to determine the ability of APID to predict the site of disease, s-s Some s. s suggest poor correlation b e t w e e n the tissue source identified by isoenzyme analysis a n d the actual site of disease as determined by other means, while others e conclude that the test is useful. Unfortunately, most of the r e s e a r c h h a s b e e n descriptive, without formal statistical analysis of results. The test itself is expensive a n d time-consuming. In our hospital, the c h a r g e for APID is $92.00, a n d it requires four hours of a technician's time. Despite this, isoenzyme testing remains in common use. Because of our concerns r e g a r d i n g the clinical usefulness of APID, we did a retrospective study of APID to determine the a c c u r a c y of the test in predicting the site of d i s e a s e in our hospital setting. Receivedfrom the Department of Internal Medicine,Northeastern Ohio UniversitiesCollegeof Medicine,Canton.Ohio (GCL),and Northeastern Ohio UniversitiesCollegeof Medicine.Rootstown. Ohio (GR). Presentedat the MidwesternConferenceof the Societyfor Research and Educationin Primary Care Internal Medicine.November 1985. Address correspondenceand reprint requeststo Dr. Lamb: Division of GeneralInternal Medicine,Milwaukee County MedicalComplex, 8700 West Wisconsin Avenue, Milwaukee. Wl 53226.
3~6
METHODS The study w a s conducted at Aultman Hospitalin Canton, Ohio, a 687-bed community-based facility providing primary, secondary, a n d tertiary care for a population of approximately 620,000. All APIDs performed on inpatient serum specimens b e t w e e n J a n u a r y a n d August 1984 were identified a n d e a c h patient's chart w a s reviewed. Every patient h a d h a d a n elevation of total alkaline p h o s p h a t a s e docum e n t e d prior to performance of APID. The decision to obtain a n APID w a s m a d e at the physician's discretion. Isoenzyme determination w a s performed by a combination of cellulose a c e t a t e electrophoresis, h e a t denaturation, a n d L-phenylalanine inhibition according to the method of Seide a n d Seiffert.7 Electrophoresis w a s done with Beckman m e m b r a n e s a n d r e a g e n t s a n d quantitated using a H e l e n a colorimetric densitometer. Electrophoretic tracings a n d physiochemical results were interpreted by a staff pathologist. Eighty per cent denaturation of e n z y m e at 55" C w a s used as the dividing point b e t w e e n bone a n d liver fractions on initial screening; then electrophoretic mobility, stability at 65 ° C a n d Lp h e n y l a l a n i n e inhibition were c o m p a r e d for consistency. Chart review w a s conducted to document admission diagnoses, past medical history, discharge diagnoses, results of APID, a s p a r t a t e aminotransferase (AST), a l a n i n e a m i n o t r a n s f e r a s e (ALT), serum calcium, radiologic studies, biopsy results, a n d medications. One h u n d r e d thirty-three specimens from 128 patients were identified. Thirty-four patients were excluded b e c a u s e of lack of a d e q u a t e information regarding final diagnosis, APID result, or other diagnostic tests performed. The remaining 94 patients formed the basis for subsequent analysis. Ninety-nine APIDs w e r e performed on these patients. The final diagnoses for e a c h patient w e r e determined from the chart i n d e p e n d e n t of APID result. Liver disease w a s considered present w h e n a test of liver p a r e n c h y m a (liver/spleen scan, CT scan, ultrasonography, endoscopic retrograde cholangiop a n c r e a t o g r a p h y , or biopsy) indicated a d i s e a s e known to elevate alkaline phosphatase, the past medical history r e v e a l e d known liver disease, or the history a n d physical examination were highly indicative of liver disease. Liver d i s e a s e w a s considered
327
JOURNALOFGENERALINTERNALMEDICINE, Volume 3 (July/August), 1988
not p r e s e n t w h e n no studies of liver p a r e n c h y m a w e r e positive, a n d no suggestion of liver d i s e a s e w a s identified b y history or physical examination. Bone d i s e a s e w a s c o n s i d e r e d p r e s e n t if a b o n e scan, a p p r o p r i a t e r a d i o g r a p h s , or a well-docum e n t e d history indicated a disorder such a s Paget's disease, b o n e m e t a s t a s e s , osteomyelitis, or r e c e n t fracture. Bone d i s e a s e w a s ruled out w h e n no radiog r a p h s or b o n e s c a n s w e r e positive, a n d there w a s no e v i d e n c e of b o n e d i s e a s e b y history or examination. The sensitivity, specificity, a n d predictive v a l u e of a n e l e v a t e d liver or b o n e isoenzyrne fraction in predicting d i s e a s e of liver or bone, respectively, w e r e calculated. Isoenzyrne determinations for liver a n d b o n e fractions w e r e c o n s i d e r e d a s s e p a r a t e a n d i n d e p e n d e n t tests. The relationships b e t w e e n results of other liver function tests c o m m o n l y found on a n a u t o m a t e d chemistry profile a n d the liver isoenzyrne fraction of alkaline p h o s p h a t a s e w e r e also a s s e s s e d . The sensitivity, specificity, a n d predictive v a l u e of a n elevation of AST or ALT in predicting the p r e s e n c e of liver d i s e a s e a n d the result of the APID w e r e calculated, as w e r e the sensitivity, specificity, a n d predictive v a l u e of simultaneous elevations of liver AP a n d AST or ALT. To a s s e s s the potential impact of the e x c l u d e d patients, the test characteristics w e r e r e c a l c u l a t e d using e x c l u d e d patients a s false positives a n d a s all true positives. S t a n d a r d error determination for 2 X 2 tables w a s u s e d to estimate c o n f i d e n c e levels. 8
TABLE 1 Results of Clinical Evaluation of Patients Based on Discharge Diagnoses and Criteria Cited in Text
Patients
Tests Performed
Liver disease Biliary disease Metastatic cancer Congestive heart failure with hepatomegaly Hepatitis (infectious and toxic) Cirrhosis Infiltrative
40
43
Bone disease Malignancy Fracture Paget's 0steomyelitis
20
Final Diagnosis
14 8
16 8
6 7 4 3
6 8 4 3 21
11 10 4 1
Bone and liver
11 10 5 1
6
6
Other
28
29
TOTAL
94
99
TABLE ~. Results of Isoenzyme Determinations Liver Bone Intestinal Liver and bone Liver and intestinal Normal TOTAL
57 13 3 2 4 20 99
RESULTS The 94 patients w e r e e v e n l y divided b e t w e e n m e n (48) a n d w o m e n (46). Ages r a n g e d from 18 to 93 years, but it w a s g e n e r a l l y a n older population, m e a n a g e 66. Various d i s e a s e s w e r e r e p r e s e n t e d . Most c o m m o n w e r e liver diseases, including cholelithiasis, h e p a t i c metastases, a n d congestive h e a r t failure with h e p a t i c involvement. Bone abnormalities w e r e p r e d o m i n a t e l y fractures, skeletal m e t a s tases, a n d Paget's d i s e a s e (Table 1). The results of the 99 i s o e n z y m e determinations a r e summarized in Table 2. Fifty-seven elevations w e r e r e p o r t e d a s of liver origin, 13 a s of b o n e origin, a n d t h r e e a s intestinal. In addition, two w e r e reported a s originating in both liver a n d bone, a n d four a s both liver a n d intestinal. Another 20 d e t e r m i n a tions w e r e r e p o r t e d to show n o r m a l total AP a n d fractionation. The criteria for liver d i s e a s e w e r e met b y 46 patients (49 tests) a n d those for b o n e d i s e a s e b y 26 patients (27 tests) (Table 1). Nineteen n e w d i a g n o s e s w e r e m a d e b a s e d o n tests o r d e r e d following completion of the APID. T h e s e included six c a s e s of biliary tract disease, five c a s e s of hepatitis, both toxic a n d infectious, five c a s e s of liver metastases, two c a s e s of congestive
liver disease, a n d o n e fracture. D a t a w e r e not availa b l e as to w h e t h e r or not t h e s e tests would h a v e b e e n p e r f o r m e d without APID. In 16 c a s e s n e w d i a g n o s e s w e r e m a d e b a s e d on tests o r d e r e d prior to the completion of APID. The APID w a s a c c u r a t e in these c a s e s but h a d no impact on diagnostic workups. In 15 c a s e s the APID result w a s consistent with a diagnosis k n o w n prior to the hospital admission. In eight c a s e s the APID e r r o n e o u s l y indicated that o n e tissue w a s involved w h e n actually a n o t h e r was, potentially hindering a c c u r a t e diagnosis. In s e v e n of the eight, APID r e v e a l e d a n i n c r e a s e d liver fraction in the f a c e of b o n e d i s e a s e a n d n o r m a l tests of liver p a r e n c h y m a . In the eighth, a n e l e v a t e d intestinal fraction w a s indicated in the p r e s e n c e of a c u t e cholecystitis. Tables 3 a n d 4 show sensitivity, specificity, a n d predictive value. The n u m b e r of liver i s o e n z y m e results includes those r e p o r t e d a s liver a n d bone, liver a n d intestine, a n d liver alone. Bone i s o e n z y m e resuits w e r e t a b u l a t e d similarly. In o n e case, in which the report indicated elevations of AP of both liver a n d b o n e origin, it w a s n e c e s s a r y to e x c l u d e the
Sensitivity
20 29
49
42 ?
49
Liver isoenzyme elevation + Liver isoenzyme elevation --
29 ~=0.81 7+29
42+20
42
29 20 + 29
TABLE 4
(±0.07)
0.68 ( ± 0 . 0 6 )
0.59 ( ± 0 . 0 7 )
0.52 ( ± 0 . 1 )
13+71
71
0.85 ( ± 0 . 0 4 )
14 ~ = 0.93 ( ± 0 . 0 7 ) 14+ 1
71 1 + 71 = 0.99 ( + --0 . 0 1 )
14 14 + 13
27 - - = 27 °/0 99
Bone isoenzyme elevation --
Bone isoenzyme elevation +
27
13
14
Jr
1
72
71
--
B Bone Disease
84
15
(±0.06) (±0.07) (±0.06) (+0.06)
9~
o.57 (+o.o6)
0.45 ( ± 0 . 0 7 ) 0.73 ( ± 0 . 0 6 ) 0.63 ( ± 0 . 0 8 )
49
49
36
62
0.43 0.75 0.75 0.43
36
27
AST or ALT elevation --
27
36
AST or ALT elevation --
63
13
22
AST or ALT elevation +
9
26
AST or ALT elevation +
35
+
B Liver Disease
Jr
A Uver Isoenzyme Elevation
63
35
49
31
49
AST" or ALT + liver isoenzyme elevation --
0.69 ( ± 0 . 0 9 ) 0.57 ( ± 0 . 0 6 )
o.84 (±o.os)
0.37 (-I-0.07)
41
18
AST or ALT + liver isoenzyme elevation +
8
C Liver Disease
Relationships Between Other Enzymes of Liver Origin Found on Chemistry Profiles and the Liver Isoenzyme of Alkaline Phosphatase (Standard Errors in Parentheses)
Negative predictive value
9~
36
62
42 42 + 7 = 0.86 ( ± 0 . 0 5 )
49 - - = 50% 98
--
A Hepatic Disease 4"
Positive predictive value
Specificity
Sensitivity
Prevalence
Positive predictive value Negative predictive value
spedfic~y
TABLE 3
Sensitivities, Specificities. and Predictive Values of Elevated Liver Isoenzyme and Elevated Bone lsoenzyme (Standard Errors in Parentheses)
I
72
26
EJ
=o
0o
JOURNALOFGENERALINTERNALMEDICINE.Volume 3 (July/August), 1988
liver fraction from the analysis b e c a u s e it w a s impossible to determine from the chart whether liver d i s e a s e w a s present. This r e d u c e d the total number of APIDs from 99 to 98 w h e n calculating the results for hepatic disease. I n a d e q u a t e n u m b e r s p r e v e n t e d analysis of the s e v e n intestinal fractions a n d rarer tumor isoenzyrnes such a s the R e g a n fraction. Twenty of the APIDs r e v e a l e d normal total values a n d fractionations. All of these w e r e found in patients w h o s e initial alkaline p h o s p h a t a s e values on the a u t o m a t e d multichannel chemistry analyzer w e r e greater than 105 m U / m l a n d less than 145 m U / m l (normal r a n g e 30 to 105 mU/rnl). Of the APIDs results excluded b e c a u s e of inadeq u a t e data, 31 suggested liver a n d four suggested b o n e origin. Assuming all of these to b e false positives, thus presuming the hypothetical worst case, the positive predictive value of the liver isoenzyme d e c r e a s e d to 0.45 (±0.05) a n d the negative predictive v a l u e to 0.72 (±0.07). For the b o n e isoenzyme the positive predictive value d e c r e a s e d to 0.74 (±0.1) a n d the negative predictive v a l u e w a s 0.62 (±0.04). Alternatively, if the "best c a s e " w e r e true a n d all of the excluded results w e r e considered true positives, the liver isoenzyme positive a n d negative predictive values would b e c o m e 0.78 (±0.04) a n d 0.82 ( ± 0.06), respectively. Corresponding predictive values for b o n e isoenzyrne would c h a n g e to 0.95 (0.05) a n d 0.89 (_-+0.03).
DISCUSSION The purpose of APID is to guide the choice of s u b s e q u e n t diagnostic tests in the setting of a patient with poorly defined d i s e a s e a n d e l e v a t e d alkaline phosphatase. To demonstrate that APID is useful for this purpose, it must b e shown that the results of the test a r e accurate, that it provides information not a l r e a d y a v a i l a b l e through the history, physical examination, a n d other simple tests, a n d that the final diagnosis c a n b e obtained more efficiently a n d cost effectively if the APID is used. Our results s h o w e d that the APID w a s most helpful w h e n a n elevation of b o n e isoenzyrne w a s identified. In the population studied, the p r e v a l e n c e or prior probability of b o n e d i s e a s e w a s 27% a n d that of liver d i s e a s e w a s 50%. Therefore, with a positive predictive value of 93%, the finding of a n elevation of b o n e isoenzyme w a s v a l u a b l e in defining the differential diagnosis. A normal b o n e isoenzyme fraction w a s also s o m e w h a t helpful, with a negative predictive value of 85%. On the other hand, the identification of a n elevation of liver isoenzyrne w a s of v e r y little use, with a positive predictive v a l u e of only 68%. A normal liver isoenzyme fraction w a s more informative, with a negative predictive value of 81%. The intestinal isoenzyme fraction w a s not e v a l u a t e d in this study, but it h a s b e e n shown that 20% of the
3Z9
population c a n h a v e a n elevation of intestinal isoenzyme a s a normal variant, limiting its value a s a marker of disease.1 The low positive predictive value of a n elevation of the liver isoenzyrne raises the question whether equivalent information could b e obtained from the serum t r a n s a m i n a s e s routinely determined a s part of the a u t o m a t e d chemistry profile. We found that a n elevation of serum t r a n s a m i n a s e s w a s not a very sensitive marker of e l e v a t e d liver alkaline phosphatase, but the t r a n s a m i n a s e s w e r e just a s reliable a s the liver alkaline p h o s p h a t a s e fraction in predicting clinical liver disease. This suggests that if AST, ALT, or another chemical indicator of liver function such a s gamma-glutamyl transferase (GGT) is abnormal, APID should not b e performed unless other clinical evidence suggests the possibility of concomitant b o n e disease. A large proportion of APIDs s h o w e d no abnormality, suggesting that the original elevations of alkaline p h o s p h a t a s e in these c a s e s w e r e false positives. Alkaline p h o s p h a t a s e m e a s u r e m e n t s c a n b e affected b y postural changes, food consumption, a n d a variety of medications. 2 All of the a b n o r m a l APID results w e r e found for patients-who h a d h a d elevations of AP of less than 50% in their original samples, suggesting that patients with such mild elevations should h a v e the AP determinations rep e a t e d w h e n they a r e supine a n d fasting before the expensive a n d time-consuming APID is performed. Clearly, in a retrospective study of this sort there a r e problems that c a n l e a d to bias in the results. The major difficulty is in the quality of the "gold standard." The selection of tests a n d the decision to pursue specific d i a g n o s e s w e r e entirely at the discretion of the individual physician. As a result, the p r e s e n c e of a d i s e a s e process w a s generally well documented, but the possibility of the other d i s e a s e s w a s often not ruled out thoroughly. Since the APID w a s part of the evaluation selected b y the physician, it is also likely that the results of the APID themselves influenced the nature of the workup, leading to a more intense investigation of a n y d i s e a s e they suggested. The resulting bias would l e a d to a greater correlation b e t w e e n test outcome a n d p r e s e n c e of disease. A second problem is the lack of a uniform stand a r d in t h e decision to order a n APID. Again, this w a s d e p e n d e n t upon the practice patterns of the physicians in the community a n d varied b e t w e e n individuals. It could affect the p r e v a l e n c e of d i s e a s e in the population studied a n d the gener~]~ability of the results. However, it is r e a s o n a b l e to a s s u m e that the practice of physicians in this community hospital parallels that found in community hospitals elsewhere, although it m a y not b e possible to extend this assumption to other types of institutions.
330
Lamb, Roush, ALKALINE PHOSPHATASEISOENZYMEDETERMINATION
The large number of cases excluded from the final analysis could also h a v e led to bias. However, recalculation of the d a t a using the best c a s e / w o r s t c a s e scenarios indicated that this h a d little effect on the results. To summarize, these results help to clarify the role that APID c a n play in the evaluation of a patient with poorly defined d i s e a s e who h a s a n e l e v a t e d alkaline phosphatase. The test should be performed only after determining that results of other tests of liver function a r e normal. If bone d i s e a s e is a question in the face of a n e l e v a t e d ALT, AST, or GGT, a n APID m a y be helpful, but in terms of cost a n d information to be g a i n e d it should be w e i g h e d a g a i n s t a bone scan. If the APID reveals elevations of liver or intestinal enzymes they a r e probably best ignored a n d further workup pursued as though APID h a d not b e e n done. An elevation of the bone isoenzyme fraction is a strong indication of a bone disorder, a n d subsequent evaluation should focus on the skeleton, including bone s c a n or appropriate radiographs.
The authors thank Dr, DonaldLaskowskiand DoloresWilkins for their assistancein the Aultman ClinicalLaboratory: Dr. Andre Ognibeneand Dr, FrederickWhittier for ongoingcriticism and review of the manuscript; and Judy Theis for typing it.
REFERENCES 1. Posen S, Doherty E. The measurement of serum alkaline phosphatase in clinical medicine. Adv Clin Chem 1981;22:163-245. 2. McComb R, Bowers G. Posen S, eds. Alkaline phosphatase. New York: Plenum Press, 1979. 3. Winkelman J, Nadler S, Demetriou J. Pileggi V. The clinical usefulness of alkaline phosphatase isoenzyme determinations. Am J Clin Patho] 1972;57:625-34. 4. Fishman W. Perspectives on alkaline phosphatase isoenzymes. Am J Med 1974;56:617-50. 5. Ewen L. Separation of alkaline phosphatase isoenzymes and evaluation of the clinical usefulness of this determination. Am J Clin Patho] 1973;61 : 142-54. 6. Gorman L, Statland B. Clinical usefulness of alkaline phosphatase isoenzyme determinations. Clin Biochem 1977; 1O: 171-4. 7. Seide WH, Seiffert UB. Quantitative alkaline phosphatase isoenzyme determination by electrophoresis on cellulose. Clin Chem 1977;23:28-34. 8. Rosner B. Fundamentals of biostatistics. Boston: Duxberry Press, 1982; 155-8.