J Cancer Res Clin Oncol (1994) 120:553-557
Cancer esearch Clinical 9 9 Springer-Verlag 1994
Analysis of acute myeloid leukemia cells by flow cytometry, introducing a new light-scattering classification Naoki Harada, Seiichi Okamura, Akira Kubota, Kazuya Shimoda, Wataru Ikematsu, Seiji Kondo, Mine Harada, Yoshiyuki Niho The First Department of Internal Medicine, Faculty of Medicine, Kyushu University, Fukuoka, Japan Received: 3 February 1994/Accepted: 14 March 1994
Abstract. A combined flow-cytometric evaluation of light scattering and the immunophenotype of acute myeloid leukemia (AML) cells from 71 newly diagnosed consecutive patients was conducted. Light-scattering characteristic of AML cells examined by flow cytometry and multiple surface markers were also analyzed using the same samples, to enable a comparison with the French-American-British (FAB) classification. Our AML cases could be classified into three light-scattering classification (LSC) types according to their physical properties on flow cytometry. These were type A, where forward light scattering (FSC) of the leukemic cell population was larger than that of lymphocytes, while side light scattering (SSC) was the same or larger than that of lymphocytes but smaller than that of monocytes; type B, where FSC of the leukemic cell population was larger than that of lymphocytes and SSC spread toward that of monocytes; and type C, where both FSC and SSC of the leukemic cell population spread beyond those of monocytes. Although a clear relationship between the FAB classification and LSC classification by the light-scattering profile of AML was not established, we observed the following findings. The majority of cases were classified as type A (58%), while type B comprised 25% and type C comprised 17%. While CD7 expression on AML cells is considered to be an immature characteristic, CD7 was expressed more frequently among LSC type A cases. Furthermore, all but one of the FAB M 1 cases were classified as type A. On the other hand, CD7 was not expressed on type C leukemic cells. The percentage of cases in which more than 60% of leukemic cells possessed another immature surface antigen, CD 346, was 13/18 (72%) among FAB M1 cases, much higher than among FAB M2 (35%) or FAB M4 (27%) cases. A negative correlation was observed Partly supported by grants in aid from the Ministry of Education, Science and Culture of Japan (03670325, 04247102, 04454572 and 05670916) and from the Fukuoka Anti-cancer Society Abbreviations: AML, acute myeloid leukemia; LSC, light-scattering classification; FAB, French-American-British classification; CD, cluster of differentiation Correspondence to: S. Okamura, The First Department of Internal Medicine, Faculty of Medicine, Kyushu University Hospital, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812, Japan.
between mature antigen CD33 and CD34 among the FAB M2 cases. The frequency of CD7 expression was 25% among the total cases, and CD7-positive cases were frequent among FAB M1 and M2, but not among FAB M3 cases. These findings concerning LSC and immunophenotyping indicate that the scattergram pattern analysis may contribute towards more precise immunophenotyping, in that it reflects the maturation stage of each AML case.
Key words: Acute myeloid leukemia - Light-scattering classification - Immunophenotyping
Introduction A popular classification of acute leukemia, based on morphology and cytochemistry, is the French American-British (FAB) classification and this has been widely used since 1976 (Bennett et al. 1976). However, several problems have been observed. One of these was the serious problem of inter-observer variability, e. g. only 70% of cases were consistently classified, even among trained hematologists (Argyle et al. 1989). Using flow cytometry to analyze leukemic cell morphology more objectively, we examined the physical properties of leukemic ceils in addition to using conventional surface phenotype analysis. Those properties were forward light scattering, which indicates cell size, and side light scattering, which indicates cell granularity (Salzman et al. 1975). In this article, the relationship between our classification by flowcytometry scattergram and the FAB classification was investigated, in addition to the conventional multiple surface phenotype analysis on newly diagnosed AML cells.
Materials and methods Patients and preparation of leukemic cells. From March 1989 to March
1992, peripheral blood and/or bone marrow aspirates from 71 patieuts with AML, aged between 14 and 83 years, were presented to our institution for surface marker analysis. The diagnosis of AML was established according to the revised French-American-British (FAB) criteria (Ben-
554 nett et al. 1985). The FAB subclasses among our 71 AML cases were as follows: M1, 18; M2, 18; M3, 8; M4, 15; M5, 8; M6, 2; and M7, 2. Mononuclear cells, obtained following density-gradient centrifugation using lymphocyte separation medium (LSM: Litton Bionetics Inc., Charleston, S.C., USA), were washed twice with Dulbecco's Ca2+/Mg2+free phosphate-buffered saline (PBS), and were then resuspended in RPMI-1640 medium containing 10% fetal calf serum (FCS). The cell concentration was adjusted to lxl07/ml with the same buffer for flowcytometric analysis.
Classification by scattergram. We classified AML into three types by using the scattergram pattern of the leukemic cell population seen on a flow cytometer (FACScan, Becton Dickinson, San Jose, Calif., USA) (Fig. 1A, B, C). These three types were type A, where forward light scattering (FSC) of the leukemic cell population was larger than that of lymphocytes, while side light scattering (SSC) was the same or larger than that of lymphocytes but smaller than that of monocytes; type B, where FSC of the leukemic cell population was larger than that of lymphor
9
!
1A
A 1
1B
7
/
oi 03
i
t oi
7
,5R7%~:
J
/ q
]
,
4
,
:
9
~
'
.'
'
4
'
,,2r, ~'
i
ported previously (Kondo et al. 1992; Shimoda et al. 1992). Briefly, cells were incubated for 20 rain at 4 ~ C with fluorescein-dye-labelled murine monoclonal antibodies (mAb) against the following antigens: CD2(Leu5), CD3(Leu4), CD4(Leu3), CD5(Leul), CD7(Leu9), CD10(CALLA), CD19(Leul2) and CD20(Leul6) (Becton Dickinson, Mountain View, Calif., USA) as lymphoid markers; CD13(My7), CD14(My4) and CD33(My9) (Coulter Immunology, Hialeah, Fla., USA) as myeloid markers; and CD34(HPCA-1), HLA-DR and CD56(Leu19) (Becton Dickinson), after being preincubated with heatinactivated human immunoglobulin (Sigma, St. Louis, Mo., USA) in order to block Fc binding. Background staining was detected using fluorescein-conjugated non-specific mouse IgG. For the detection of CD34, antibody reactivity with test cells was determined by indirect immunofluorescence. After Fc binding had been blocked, cells were incubated for 20 rain at 4 ~ C with the murine monoclonal Ab (IgG), washed twice, and further incubated with fluorescein-conjugated goat anit-(mouse IgG) (Becton Dickinson). After two additional washing steps, fluorescein-coated cells were detected on a flow cytometer. When we analyzed the data, the leukemic cells were gated on a light scattergram and the normal lymphocyte population was excluded. We also used Paint-a-gate software (Becton Dickinson) to determine clearly the leukemic cell population and the normal lymphocyte population. A positive reaction was defined as being when 20% or more of the gated cells were more fluorescent than the control.
Statistical analysis. The correlation between the percentage of cells pos-
1c
1
Surface phenotype analysis. Immunophenotyping of AML has been re-
FSC
FSC 4I
l
?fb
cytes, while SSC spread toward that of monocytes; and type C, where both FSC and SSC of the leukemic cell population spread beyond those of monocytes. We here refer to this classification as the light-scattering classification (LSC).
i
i1
itive for the surface markers of gated populations was evaluated using Spearman's rank correlation coefficient analysis.
1D'
1
o7 ~4
o9 i
if
!
l d
)4~
-~::b
1
g :c"~-.;e
!
Results
LSC and immunophenotyping
1
I n
FSC
FSC
Fig. 1 A-D. Light-scatter classification (LSC) of acute myeloid leukemia (AML) cells on flow cytometry. Scattergram pattern of AML cells was categorized into three types by flow cytometry (A-C). These were, type A, where forward light scatter (FSC) of the leukemic cell population was larger than that of lymphocytes, while side light scatter (SSC) was the same or larger than that of lymphocytes but smaller than that of monocytes; type B, where FSC of the leukemic cell population was larger than that of lymphocytes, while SSC spread toward that of monocytes; and type C, where both FSC and SSC of the leukemic cell population spread beyond those of monocytes. We named this classification LSC. A scattergram of normal peripheral blood cells is shown for reference (D)
T h e L S C and i m m u n o p h e n o t y p e o f A M L are g i v e n in Table 1. T h e m a j o r i t y o f cases w e r e classified as type A (58%), w h i l e type B c o m p r i s e d 2 5 % and type C 17%, W h i l e 75% of C D 3 4 - p o s i t i v e cases w e r e classified as L S C type A, C D 1 4 p o s i t i v e cases w e r e infrequent a m o n g the L S C type A cases. It is n o t e w o r t h y that C D 7 positivity was significantly higher a m o n g L S C type A than a m o n g L S C type B cases, and was significantly higher a m o n g L S C type B than a m o n g L S C type C cases (P = 0.024, M a n t e l extension test) (Fig. 2).
LSC and FAB classification The relationship b e t w e e n F A B classification and L S C is s h o w n in Table 2. All the F A B M1 cases e x c e p t for one fell
Table 1. Surface marker frequency according to light-scattering classification (LSC) type LSC type
A B
C Total
Marker frequency (no. positive/no, tested) CD13
CD14
CD33
CD34
T marker b
B marker c
HLA-DR
CD7
CD4
31/41 14/18 10/12 55/71
8/41 5/18 6/12 19/71
38/41 17/18 11/12 66/71
30/40 10/18 4/11 44/69
14/41 5/18 2/12 21/71
3/41 1/18 0/12 4/71
36/41 15/18 6/12 57/71
13/40 4/18 0/12 17/70
10/41 6/16 4/11 20/68
a According to light-scattering properties. See Fig. 1 b Either CD2-, CD3-, CD5- or CD7-positive c Either CD l 0-, CD 19- or CD20-positive
555 %
into the L S C type A group, while none fell into the LSC type C group. Furthermore, all four cases of FAB M6 or M7 were also type A cases. No FAB M3 cases were found among the LSC type A cases. On the other hand, most of the LSC type B cases were classified as FAB M2, M3 or M4, while LSC type C cases fell into the FAB M3, M4 or M5 groups.
100
801 CO
Immunophenotyping and FAB classification Table 3 summarizes the surface marker frequency according to FAB classification. Positivity of CD34 indicates immature cells (Civin et al. 1984) and the frequency of CD34 positivity among the total cases examined was 64%. FAB M3 showed a low positivity o f 29%. There were no significant differences in CD34 positivity among FAB M1, M2 or M4. However, cases possessing a high percentage (60% or more) of CD34positive leukemic cells were significantly more numerous among the FAB M1 subclass (72%) than among the M2 or M4 subclass (33% and 27% respectively) (P < 0.05) (Fig. 3). None of the FAB M3 cases possessed the H L A - D R antigen on their leukemic cells (Table 3). The m y e l o i d antigens CD13 and CD33 were expressed at a high frequency on leukemic cells from patients with A M L , regardless of the FAB
% 100 O
80
I'-- 6 0 ,
s O
,,
40. @
9
:i.- .i,"
0,
ot
""
A
9
9
'IP 'llO
o4oI 20
9
,, 9
'" , 9
9
o M1
9
,,,
|
-at 2
3
FAB
9
9
9
9
oo
9
9
4
m . . . . . . 5 6 7
Subclass
Fig. 3. CD34 positivity according to FAB classification. Cases that possessed more than 60% CD34-positive leukemic cells were much more frequent among M1 cases than among the other subclasses
subclass. CD14- or CD4-positive cases were often observed among FAB M4 and M5 cases. One of the T-cell-related antigens, either CD2, CD5 or CDT, was expressed on 30% of all A M L cases, as shown in Table 3. CD7 was expressed more frequently (25%) than the rest of the T-cell-related antigens. CD7 seemed to be expressed more frequently among M1 cases than among the other subclasses. CD3 expression was not detected on A M L leukemic cells. The expression rate of the B-cell-related antigens, CD10, CD19 and CD20, is shown in Table 3. These antigens were expressed less frequently than the T-cell-related antigens on A M L leukemic cells. Only four cases were positive with CD19 (6%), while two cases were positive with CD10. CD20 was negative in all our cases. CD56, which is known to be a neural cell adhesion molecule, was also examined on leukemic cells in about half our cases. None of our M1 cases expressed this antigen (Table 3).
B
"
C
Table 2. Relationship between FAB classification and LSC type a
A B C Total
ee
~
Relationship among surface antigens
;
Fig. 2. CD7 positivity according to LSC type. CD7 positivity was significantly higher among LSC type A than among LSC type B cases, and significantly higher among LSC type B than among LSC type C cases (P = 0.024, Mantel extension test)
LSC type
60
~
FAB classification (no. of each subclass)
As shown in Fig. 4, only a weak, negative correlation between the percentage of CD33 positivity and the percentage of CD34 positivity was observed within the whole A M L population (r = - 0 . 2 7 , P < 0.05). However, this negative correlation could be clearly observed among FAB M2 cases (r = - 0 . 6 3 , P < 0.05). The relationship between CD14 and CD4 antigens, both of which are frequently expressed in FAB M4 and M5 cases (Table 3), is shown in Fig. 5, but no significant correlation was observed.
Total
M1
M2
M3
M4
M5
M6
M7
17 1 0 18
11 6 1 18
0 3 5 8
5 7 3 15
4 1 3 8
2 0 0 2
2 0 0 2
Discussion
41 18 12 71
According to light-scattering properties on flow cytometry. See Fig. 1
We examined the leukemic cells of A M L using flow cytometry. The advantages of this method were (a) many cells could be analyzed at once, (b) the cell population of interest could be separately analyzed, and (c) it was thought to be more objective. In this series, we examined a relatively enriched leu-
556 Table 3. Surface marker frequency according to FAB classification FAB
Marker frequency (no. positive/no, tested)
M1 M2 M3 M4 M5 M6 M7 Total
CD13
CD14
CD33
CD34
HLA-DR
CD56
CD2
CD5
CD7
CD3
CD19
CD10
CD4
14/18 16/18 5/8 10/15 7/8 1/2 2/2 55/71
0/18 3/18 2/8 9/15 4/8 1/2 0/2 19/71
18/18 15/18 8/8 14/15 8/8 1/2 2/2 66/71
13/18 12/17 2/7 11/15 3/8 1/2 2/2 44/69
15/18 17/18 0/8 13/15 8/8 2/2 2/2 57/71
0/10 3/12 1/3 1/8 2/5 2/2 0/2 9/42
2/18 0/18 2/8 2/15 0/8 0/2 0/2 6/71
0/14 0/14 0/7 0/14 0/8 0/2 1/2 1/61
7/18 5/17 0/8 1/15 2/8 1/2 1/2 t7/69
0/18 0/18 0/8 0/15 0/8 0/2 0/2 0/71
3/18 0/18 0/8 1/15 0/8 0/2 0/2 4/71
1/18 0/17 0/6 1/15 0/8 0/2 0/2 2/68
2/18 3/18 1/7 7/13 5/8 2/2 0/2 20/68
Each of the above values refers to the number of positive cases/number of cases tested. A positive reaction was defined as being when 20% or more of the gated leukemic cells were more fluorescent than the control
M1
TOTAL CASES % 100 . . . . . . . . . . . . . . . . . . . . . 9
9
ee
9
8
401 9
9.
9
9
0
9
.9.
100
9
60
o 9
"
20
M2
%
. . . . . . . . . . . . . . . .
80
9
9
eli 9
O
80
gee
I co60
% 100 9 ' 9 n . . . . . . . . . . . . . . . .
a.
20
0 40
9
:..
9
40
2O
q
4O
0
60
80
100 %
20
. . . . . .
0
J
20
. . . . . . .
40
CD33 %
~0
o ~
9
60
. . . .
80
100%
20
40
CD33
M3
100 . . . . . . . . . . . . . . . . . . . .
8O
~o-)6O C3 O 40
%
M4 % 100 . . . . . . . . . . . . . . . . .
80
100
100%
80
M5 .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
80
00
~c% 6C C3
8
20
60
CD33
4(
4C
2(
2C 9o
9
0
.
20
9
.
.
.
40
.
".
60
CD33
.
80
"."
.
100 %
.
0
.
.
.
.
20
.
.
.
.
q
40
.
.
60
CD33
.
9. ~
80
.
9
.
100 %
0
20
40
60
80
I
100%
CD33
Fig. 4. Relationship between CD34 and CD33 antigen expression. Among the total number of cases, only a weak, negative correlation was observed (r = -0.27, P < 0.05), however a relatively strong negative correlation was observed among the FAB M2 cases (r = -0.63, P < 0.05)
kemic cell population following density gradient (LSM) separation to remove red blood cells and mature granulocytes and a proper window setting to separate normal lymphocytes. The usefulness of a flow-cytometric scattergram to classify A M L has been examined. Terstappen et al. (1991) showed a relationship between FAB classification and scattergram patterns and categorized their cases into nine patterns. However, their classification is complicated since they require orthogonal light-scattering signals transformed according to a polynomial function. On the other hand, we
have here introduced a simple classification of light scattering of A M L cells, LSC. Our A M L cases could be classified into three types according to their physical properties on flow cytometry (Fig. 1), these being forward light scattering, which indicates cell size, and side light scattering indicating cell density or granularity (Salzman et al. 1975) on the base of normal lymphocytes and monocytes. Since scattergrams can sometimes be changeable according to the setting of flow cytometry and the sample conditions, we used normal lymphocytes and monocytes as the standard to establish LSC,
557
%
~ O - -
.
.
.
.
.
,
. ,
.
.
.
.
.
.
.
.
.
.
.
low side light scattering. Furthermore, CD7-positive cases, which might also reflect cell immaturity, were frequent among type A cases, but not among type C cases. We conclude that a scattergram of A M L cells, not only aids immunophenotyping but is also helpful for clarifying the properties o f A M L cells.
.
80 60 a
9
040
I
9
2ol, ~,o 9
0
0
Acknowledgements. We thank Drs. K. Akazawa and M. Sakamoto, of
9
9
9
DO 20
40
60
80
I
100%
CD4 Fig. 5. CD14 and CD4 antigen expression. The relationship between CD 14 and CD4 was analyzed, but a significant correlation was not demonstrated
and then compared the scattergram of our samples with this standard (Fig. 1). We also used fresh samples, since the freezing of the leukemic cells could influence the scattergram or affect the immunophenotyping. With particular regard to the surface phenotype CD34, which is known to be a hematopoietic progenitor cell surface antigen (Civin et al. 1984), the frequency among the total cases examined was 66%, as presviously reported (Geller et al. 1990). There was no significant difference in the frequency of CD34-positive cases among the FAB subclasses M1, M2 and M4. However, the percentage of cases in which more than 60% o f leukemic cells possessed the CD34 antigen was 13/18 (72%) among the FAB M1 cases, which was much higher than among the M2 (35%) or M4 (27%) cases (P < 0.05). Borowitz et al. (1989) reported that CD33 was observed at a significantly higher frequency among CD34negative groups than among CD34-positive groups. A significant negative correlation between the percentage of CD33 positivity and the percentage of CD34 positivity was observed among our FAB M 2 cases, as shown in Fig. 4. The frequency of CD7-positive A M L cases was 25% among the total cases, similar to the results reported by Zutter et al. (1990). According to our observations, CD7-posirive cases were frequent among FAB M1 and M2 cases, but not among FAB M3 cases. This phenomenon m a y be related to the cell immaturity of CD7-positive leukemia, as has been suggested by some investigators (Lo Coco et al. 1989; Zutter et al. 1990, Osada et al. 1990; Kondo et al. 1992). CD4, which is weakly expressed on normal monocytes (Stewart et al. 1986), was expressed on more than half of the M4 and M5 cases. We thought that both CD14 and CD4 were useful for characterizing M4 and M5 cases, and so we further analyzed the relationship of the expression ratios between these two antigens. However, no significant correlation was observed (Fig. 5). F r o m our observations, it seemed noteworthy that most of the FAB M1 cases were CD34-positive and, in particular, the percentage o f CD34-positive cells was high and could be categorized as LSC type A: low forward light scattering with
the Department of Medical Informatics, Faculty of Medicine, Kyushu University, for the stastistical analysis. The English used in this manuscript was corrected by Miss K. Miller (Royal English Language Centre, Fukuoka, Japan).
References Argyle JC, Benjamin DR, Lampkin B, Hammond D (1989) Acute nonlymphocytic leukemias of childhood. Inter-observer variability and problems in the use of the FAB classification. Cancer 63:295-301 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DAG, Gralnick HR, Sultan C (1976) Proposals for the classification of the acute leukaemias. Br J Haematol 33:451-458 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DAG, Gralnick FIR, Sultan C (1985) Proposed revised criteria for the classification of acute myeloid leukemia. A report of the French-AmericanBritish Cooperative Group. Ann Intern Med 103:626-629 Borowitz MJ, Gockerman JP, Moore JO, Civin CI, Page SO, Robertson J, Binger SH (1989) Clinicopathologic and cytogenic features of CD34 (Myl0) positive acute nonlymphocytic leukemia. Am J Clin Patho191:265-270 Civin CI, Strauss LC, Brovall C, Fackler MJ, Schwartz JF, Shaper JH (1984) A hematopoietic progenitor cell surface antigen defined by a monoclonal antibody raised against KG-la cells. J Immunol 133:157-165 Geller RB, Zahurak M, Hurwitz CA, Burke PJ, Karp JE, Piantodosi S, Civin CI (1990) Prognostic importance of immunophenotyping in adults with acute myelocytic leukemia: the significance of the stemcell gtycoprotein CD34 (My 10). Br J Haemato176:340-347 Kondo S, Okamura S, Harada M, Gondo H, Hayashi S, Shibuya T, Okamura T, Yamaga S, Asano Y, Omori F, Taniguchi S, Akashi S, Tanaka T, Niho Y (1992) CD7-positive acute myeloid leukemia: further evidence of cellular immaturity. J Clin Exp Oncol 118:386-388 Lo Coco F, De Rossi G, Pasqualetti D, Lopez M, Diverio D, Latagliata R, Fenu S, Mandelli F (1989) CD7 positive acute myeloid leukaemia: a subtype associated with cell immaturity. Br J Haematol 73:480-485 Osada H, Emi N, Ueda R, Seto M, Koike K, Suchi T, Kojima S, Obata Y, Takahashi T (1990) Genuine CD7 expression in acute leukemia and lymphoblastic lymphoma. Leuk Res 14:869-877 Salzman GC, Crowell JM, Martin JC, Trujillo TT, Romero A, Mullaney PF, LaBauve PM (1975) Cell classification by laser light scattering: identification and separation of unstained leukocytes. Acta Cytol 19:374-377 Shimoda K, Okamura S, Harada N, Ikematsu W, Kondo S, Kawasaki C, Tanaka T, Etou T, Akashi K, Okamura T, Shibuya T, Harada M, Niho Y (1992) Granulocyte colony-stimulating factor receptors on human acute leukemia: biphenotypic leukemic cells possess granulocyte colony-stimulating factor receptors. Cancer Res 52:3052-3055 Stewart S, Fujimoto J, Levy RJ (1986) Human T-lymphocytes and monocytes bear the same Leu3 (T4) antigen. J Immunol 136:3773-3776 Terstappen LWMM, Konemann S, Safford M, Eoken MR, Zurlutter K, Buchner Th, Hiddemann W, Wormann B (1991) Flow cytometric characterization of acute myeloid leukemia. Part 1. Significance of light scattering properties. Leukemia 5:315-321 Zutter MM, Martin PJ, Hanke D, Kidd PG (1990) CD7+ acute nonlymphocytic leukemia: evidence for an early multipotential progenitor. Leuk Res 14:23-26
