Skip to main content
Device Global Header

Subscribe
or Renew

Advanced Search

Original Article

Clinical Importance of Myeloid-Antigen Expression in Acute Lymphoblastic Leukemia of Childhood

List of authors.
  • Susan R. Wiersma, M.D.,
  • Jorge Ortega, M.D.,
  • Eugene Sobel, Ph.D.,
  • and Kenneth I. Weinberg, M.D.

Abstract

Background.

Leukemic cells in 15 to 25 percent of patients with acute lymphoblastic leukemia (ALL) express myeloid antigens as well as lymphoid antigens (the latter reflecting B-cell or T-cell lineage). The relations of myeloid-antigen expression to other features of ALL and to prognosis have been controversial.

Methods.

We analyzed clinical and laboratory features present at diagnosis in 236 consecutive cases of ALL in children. Immunophenotyping, including single- and dual-fluorescence analyses, was used to classify leukemic cells as B or T lymphoblasts and also to identify myeloid-antigen expression — the simultaneous expression of lymphoid-associated antigens and at least one of three myeloid-associated antigens (CD33, CD13, and CD14) on cells classified as L1 or L2 according to the French—American—British system.

Results.

Forty-five of 185 patients with B-lineage ALL had myeloid-antigen expression, as did 8 of 41 patients with T-lineage ALL. In 10 patients, the lineage could not be determined. Myeloid-antigen expression was associated with L2 morphology (P<0.05), but it did not correlate with other prognostic features recognized previously. Multivariate analysis showed that myeloid-antigen expression was an important predictor of relapse in childhood ALL and the most significant prognostic factor statistically (P<0.0001). A white-cell count ≥50×109 per liter at diagnosis was also an important and highly significant prognostic feature (P<0.001). After 40 months, the estimated disease-free survival for patients with ALL was 84 percent for those without myeloid-antigen expression and with a low white-cell count, 57 percent for those without myeloid-antigen expression and with a high white-cell count, 47 percent for those with myeloid-antigen expression and a low white-cell count, and 26 percent for those with myeloid-antigen expression and a high white-cell count (P<0.00001).

Conclusions.

Myeloid-antigen expression is an important independent predictor of a poor response to chemotherapy in childhood ALL. (N Engl J Med 1991; 324: 800–8.)

Introduction

ACUTE lymphoblastic leukemia (ALL) is a clonal expansion of cells committed to differentiation as either B or T lymphocytes. These cells retain the differentiation characteristics of the corresponding normal lymphoid precursors.1 The development of lineage-specific monoclonal antibodies has provided a means of identifying the lymphoid lineage and the particular stage of differentiation of the leukemia-cell progenitor.2 3 4 In some cases of childhood ALL, immunophenotyping demonstrates that in addition to surface antigens associated with differentiation as B or T lymphocytes, antigens associated with myeloid differentiation are expressed at the same time.5 6 7 8 9 10 The phenomenon of simultaneous expression of myeloid-associated surface antigens on ALL cells has been cited as an example of lineage infidelity or mixed-lineage expression.11 12 Previous studies of myeloid-antigen expression in ALL have, however, been complicated by the possible nonspecificity of the monoclonal antibodies used and the difficulty of interpreting samples of bone marrow in which normal cells are not completely replaced by leukemic cells.

In addition to the biologic issues raised by the expression of myeloid antigens in ALL, the clinical and prognostic importance of this phenomenon has been controversial. Previous investigators have reported conflicting results with regard to the rates of remission and disease-free survival among patients with ALL and myeloid-antigen expression.4 , 6 , 8 9 10 These inconsistencies may reflect differences in the definitions of myeloid-antigen expression and in immunophenotyping techniques, study populations, and treatments.

We report the results of a prospective study of myeloid-antigen expression in children with newly diagnosed ALL at a single institution. Dual-fluorescence analysis and the use of monoclonal antibodies specific for myeloid antigens allowed a well-defined group of patients with myeloid-antigen—positive ALL to be identified. Their presenting features and clinical outcomes were then compared with those of control patients with myeloid-antigen—negative ALL.

Methods

Patients

From January 1, 1985, through December 31, 1989, 236 children were given diagnoses of ALL at Childrens Hospital Los Angeles. All patients were under 18 years of age at diagnosis. Patients with chronic myelogenous leukemia or leukemias with two different populations of leukemic cells were excluded from the study. At the time of diagnosis, information was obtained about presenting clinical features, morphology, histochemistry, immunophenotype, and cytogenetics. The patients were treated according to the protocols of the Children's Cancer Study Group. The treatments were classified as standard13 or intensive (i.e., the Berlin—Frankfurt—Munich regimen, with a delayed intensification phase,14 , 15 the LSA2L2 regimen,16 the New York regimen,17 , 18 or a regimen containing high doses of methotrexate19).

Morphologic and Cytogenetic Analysis

Fresh bone marrow treated with heparin and occasionally samples of peripheral blood obtained from patients at the time of diagnosis were used in the analysis.

The morphologic diagnosis of ALL was made by examination of smears of peripheral blood and bone marrow (both buffy-coat and direct smears). A minimum of 500 cells were counted. Specimens were routinely stained with Wright—Giemsa stain, periodic acidSchiff reagent, and Sudan black B stain and were stained for myeloperoxidase and alpha-naphthyl acid esterase. The leukemias were subclassified as L1 or L2, according to the French—American—British classification.20 , 21

Cytogenetic analysis was performed after bone marrow cells were cultured in RPMI 1640 (Irvine Scientific, Santa Ana, Calif.) for 24 to 72 hours, harvested, and treated with demecolcine (Colcemid) for 1 hour. The cells were treated with a hypotonic solution of potassium chloride (0.4 percent) and fixed in methanol—acetic acid (3:1). Slides were prepared, and Giemsa banding was performed.

Immunophenotyping

Cell Preparation

Heparin-treated bone marrow samples were diluted 1:1 with Hanks' buffered saline solution (Irvine Scientific), layered onto Ficoll–Hypaque gradients (Pharmacia, Piscataway, N.J.), and centrifuged at 2000 rpm for 30 minutes. Mononuclear cells were collected at the interface of Ficoll–Hypaque and plasma and washed twice with Hanks' buffered saline solution. After the final wash, the cells were resuspended in RPMI 1640 with L-glutamine (2 mM), penicillin (50 U per milliliter), and streptomycin (50 mg per milliliter) added.

Monoclonal Antibodies

Table 1. Table 1. Antigenic Determinants.

The monoclonal-antibody panel was selected to identify lymphoid-specific and myeloid-specific differentiation antigens (Table 1). The use of the panel of antibodies allowed both the lineage and the stage of differentiation of the leukemic cells to be identified.2

Indirect Immunofluorescence

The leukemic cells were incubated with optimal concentrations of each monoclonal antibody (on the basis of the manufacturers' recommendations) for 30 minutes at 4°C. The cells were washed twice in a Sorvall Cell Washer (Ortho Diagnostic Systems, Raritan, N.J.) with Immusal (American Dade, Denville, N.J.) and incubated with 50 ml of a 1:20 dilution of fluorescein isothiocyanate (FITC)—conjugated affinity-purified goat antimouse immunoglobulin (GAM-FITC) (Cappel, Cooper Biomedical, West Chester, Pa.). After the cells were washed twice, they were fixed in 1 percent paraformaldehyde and analyzed with either a Coulter Epics V fluorescence-activated cell sorter or a Becton Dickinson flow cytometer.

Double-Stained Immunofluorescence

The cells were incubated simultaneously with optimal concentrations of an FITC-labeled monoclonal antibody and a phycoerythrin-labeled monoclonal antibody for 30 minutes at 4°C. The cells were washed twice with Immusal in a Sorvall Cell Washer, fixed in 1 percent paraformaldehyde, and analyzed by flow cytometry.

Terminal Deoxynucleotidyl Transferase (TdT)

The cytocentrifuged smears were fixed in absolute methanol for 30 minutes at 4°C, incubated with Coulter Blocking Agent (CBA-1, Coulter, Hialeah, Fla.) for 20 minutes at room temperature, and then incubated with mouse antihuman TdT monoclonal antibody (Coulter) or a nonspecific mouse antibody for 30 minutes at room temperature. The slides were washed twice with phosphate-buffered saline, incubated with GAM-FITC for 30 minutes at room temperature, and rewashed. After drying, the slides were scored with an Olympus fluorescence microscope with respect to intranuclear fluorescence.

Statistical Analysis

Because many comparisons or statistical tests were undertaken, a P value of 0.01 was used as the criterion for statistical significance. Chi-square and Pearson-correlation test statistics were used to evaluate associations or patterns of relation between the expression of lymphoid antigen and that of myeloid antigen. The same techniques were used to investigate associations between clinical features and myeloid-antigen expression. The chi-square test was used to investigate differences between proportions. Univariate and multivariate Cox proportional-hazards methods were used to model the prognostic importance of potential predictors of relapse or death.22 For the multivariate procedure, stepwise inclusion of predictors was used to build models, and a P value of 0.01 was used for entering the predictors into the models. The normal approximation was used to test for the significance of the coefficients in the model. The Kaplan–Meier product-limit method was used to estimate the survival curves for groups of patients,23 and the log-rank test was used to assess differences in survival curves.24 Results are reported as means ±SE. All statistical tests were two-tailed.

Results

Between January 1985 and December 1989, 236 children were given diagnoses of ALL at Childrens Hospital Los Angeles. On the basis of the French—American—British morphologic criteria, the leukemic cells in 162 of the children had the morphologic features of L1 lymphoblasts, whereas 74 had L2 morphology.20 , 21 No patient had L3 morphology (Burkitt's leukemia).

Definition and Incidence of Myeloid-Antigen Expression

Figure 1. Figure 1. Flow-Cytometric Analysis of Immunophenotypes of Patients with ALL and Myeloid-Antigen Expression.

Panel A shows single-fluorescence staining of a bone marrow sample from one patient with ALL (solid line) after staining with antibodies against CD10 and CD13, as compared with an ascites control (dashed line). The abscissa indicates the logarithm of the relative intensity of green fluorescence with the fluoresceinated antibodies indicated. The ordinate indicates the relative frequency of green fluorescence of fluorescein isothiocyanate as an indicator of antibody staining. Because the bone marrow was not fully replaced with leukemic lymphoblasts, single-fluorescence analysis could not differentiate between the simultaneous expression of CD13 on the leukemic cells and the presence of residual normal myeloid cells. In Panel B, dual-fluorescence analysis of cells from the same patient with ALL shows expression of both CD10 and CD13 on the leukemic lymphoblasts (quadrant 2). The upper left quadrant contains cells that reacted with CD13; the lower right quadrant, cells that reacted with CD10; and the upper right quadrant, cells that reacted with both antibodies and hence express both antigens. Cells that reacted with neither antibody are shown in the lower left quadrant. In Panel C, single-fluorescence staining of a bone marrow sample from another patient with ALL in which the marrow was completely replaced by leukemic lymphoblasts shows that 70 percent of the cells express CD19 and 80 percent express CD33, indicating that the CD19+ cells also express CD33.

All bone marrow samples were initially analyzed with a panel of monoclonal antibodies directed against antigens expressed during multiple stages of B-lymphocyte, T-lymphocyte, and myeloid differentiation.2 Myeloid-antigen expression was determined by staining with monoclonal antibodies directed against any of three myeloid-associated antigens — CD33, CD13, and CD14.25 When single-antibody analysis revealed expression of both myeloid and lymphoid antigens, dual-fluorescence analysis was performed. Leukemia in which both lymphoid antigens and at least one myeloid antigen were expressed by more than 30 percent of the cells was classified as myeloid-antigen—positive. The presence of myeloid-antigen expression was thus determined by the direct documentation of the simultaneous expression of lymphoid-associated and myeloid-associated antigens (Fig. 1A and 1B). When cells were not available for dual-fluorescence staining, myeloid-antigen expression was defined as an overlap of at least 30 percent in the expression of lymphoid and myeloid antigens, as detected by single-antibody analysis (Fig. 1C).

Of the 236 patients with ALL, 185 (78 percent) had leukemia of B lineage, and 41 (17 percent) had leukemia of T lineage. In 10 patients (4 percent), the lineage of the leukemic cells could not be determined by immunophenotyping. Expression of both myeloid and lymphoid surface antigens was detected in 53 of the patients with ALL (22 percent). The 53 patients with ALL and myeloid-antigen expression were divided into two groups on the basis of immunophenotype. The first group consisted of 45 patients with leukemia of B lineage who had expression of at least one myeloid surface antigen. The second group consisted of eight patients with leukemia of T lineage who expressed at least one myeloid surface antigen.

Immunophenotype

Figure 2. Figure 2. Distribution of Immunophenotypes in 53 Patients with ALL and Myeloid-Antigen Expression.

The number of patients with ALL who had one, two, or three myeloid surface antigens (CD13, CD14, or CD33) expressed on their leukemia cells is shown for patients with B-lineage ALL of an immature (CD10–, CD20–), intermediate (CD10+, CD20–), or more mature (CD10+, CD20+) stage of differentiation. In eight patients CD20 expression was not tested (NT). Eight patients with T-lineage ALL were also found to express at least one myeloid antigen. No correlation was found between the stage of B-lymphoid differentiation and the number or pattern of the myeloid antigens expressed.

The distribution of immunophenotypes in the 45 patients with myeloid-antigen—positive ALL of B lineage is presented in Figure 2. In addition to expression of B-lineage differentiation antigens, 2 patients had leukemic cells that expressed three myeloid surface antigens (CD33, CD13, and CD14), and 15 patients had expression of two of these antigens (CD33 and CD13 in 8 patients, CD13 and CD14 in 6, and CD33 and CD 14 in 1). The majority of the patients had expression of only one myeloid antigen: CD 14 in 14 patients, CD33 in 7, and CD 13 in 7. No association could be detected between the pattern of lymphoid-antigen expression and the presence or pattern of myeloid-antigen expression. Specifically, the maturational stage of B lymphoblasts did not correlate with the expression of myeloid antigens. Cells with an immature B-lineage phenotype (CD10–, CD20–) were no more likely to express myeloid antigens than those of intermediate stages (CD 10+, CD20–) or more mature stages (CD10+, CD20 + ) of B-lymphocyte differentiation (P>0.2).

In the patients with myeloid-antigen—positive ALL of T lineage, the leukemic phenotypes were more consistent than those in the patients with ALL of B lineage. In contrast to the latter patients, whose phenotypes corresponded to more than one stage of B-lymphocyte differentiation, all eight patients with myeloid-antigen—positive ALL of T lineage had an immature thymocyte phenotype (CD7+, CD2 +/–, CD5+/–, CD1–, CD3–). In five of these eight patients, only the myeloid antigen CD 13 was expressed. One patient each had expression of CD33 or CD 14, and one patient had expression of both CD 13 and CD33.

Cytogenetic Abnormalities

A wide variety of cytogenetic abnormalities were observed in the patients with myeloid-antigen—positive ALL. Thirty-two of the 53 such patients (60 percent) had metaphases that could be evaluated cytogenetically. One patient with ALL of T lineage had a (4; 11 ) translocation, a cytogenetic abnormality previously associated with the presence of an indeterminate immunophenotype.26 , 27 No patient was found to have a (9;22) translocation or a Philadelphia chromosome. Many patients had aneuploidy, including patients with hyperdiploidy, pseudodiploidy, or hypodiploidy — abnormalities commonly seen in ALL. Nine patients had no detectable cytogenetic abnormality.

Presenting Features

Table 2. Table 2. Features of Patients with ALL at the Time of Diagnosis, According to Presence or Absence of Myeloid-Antigen Expression.

Table 2 compares the presenting features (age, sex, white-cell count, hemoglobin level, platelet count, presence of extramedullary disease, and morphologic features) of the patients with ALL and myeloid-antigen expression with those of the patients without myeloid-antigen expression. Except for an increased frequency of L2 lymphoblast morphology, there was no association between these features, either individually or in combination, and the presence or pattern of myeloid surface-antigen expression. L2 morphology was present in 43 percent of the samples from patients with myeloid-antigen—positive ALL — more than that found in the samples from patients without this antigen (P<0.03). With use of previously described methods of risk-group stratification based on white-cell count, age, sex, the presence of extramedullary disease, French—American—British morphologic classification, hemoglobin level, and platelet count at diagnosis, the distribution of risk groups for patients with ALL and myeloid-antigen expression was similar to that for patients with ALL but without myeloid-antigen expression.21 , 28 According to the criteria of the Children's Cancer Study Group, 19 percent of the patients with ALL and myeloid-antigen expression were at low risk of relapse at the time of diagnosis, 38 percent were at intermediate risk, and 43 percent were at high risk. In the group negative for myeloid antigen, 16 percent were at low risk, 45 percent were at intermediate risk, and 39 percent were at high risk of relapse (P>0.6).

Prognostic Importance of Myeloid-Antigen Expression in ALL

Univariate Analysis

Table 3. Table 3. Prognostic Importance of the Presenting Features in All Patients Studied and in Those with ALL of B Lineage, as Determined by Univariate and Multivariate Analysis.*

The patients with ALL and myeloid-antigen expression had a significantly worse prognosis than those without expression of a myeloid antigen (P<0.0001) (Table 3). In the group without myeloid-antigen expression (183 patients), 2 patients died during induction and 1 patient did not have a remission. All 53 patients with myeloid-antigen expression except 1, who died during induction, had a complete remission. Except for the one death during induction, the only event contributing to the poor outcome in this group was relapse of leukemia. There was no difference in outcome among the patients in this group with regard to the number of myeloid antigens expressed. Cytogenetic features associated with a low risk of relapse (e.g., hyperdiploidy) did not correlate with prolonged event-free survival. The only other feature in these patients that was shown to be significant in prognosis on univariate analysis was a high initial white-cell count (≥50×109 per liter) (P<0.001). T-cell immunophenotype (P<0.02) and the presence of extramedullary disease (P<0.04) at diagnosis approached significance. Features that were not predictive of event-free survival in our population were L2 morphology, a hemoglobin level ≥10 g per deciliter, age (divided into three groups: <1 year, 1 to 9 years, and ≥10 years), sex, a platelet count < 100X 109 per liter, and the type of treatment received (intensive or standard).

Figure 3. Figure 3. Kaplan–Meier Curves of Estimated Event-free Survival for Patients with ALL with (Dotted Line) and without (Solid Line) Myeloid-Antigen Expression.

Rates for all patients studied are shown in Panel A, and rates for patients with ALL of B lineage in Panel B.

After 40 months, the estimated event-free survival in the patients with ALL and myeloid-antigen expression was 39± 13 percent, as compared with 78±5 percent in patients without such expression (P<0.0001) (Fig. 3A). Although only 22 percent of the children with ALL had myeloid-antigen expression, 40 percent of all the patients with an adverse event had expression of such antigens.

Multivariate Analysis

Figure 4. Figure 4. Kaplan–Meier Curves of Estimated Event-free Survival of Patients with ALL, as Stratified According to Presence or Absence of Myeloid-Antigen Expression and Initial White-Cell Count.

A denotes no myeloid-antigen expression and a white-cell count <50 × 109 per liter, B no myeloid-antigen expression and a white-cell count ≥50 × 109 per liter, C myeloid-antigen expression and a white-cell count <50×109 per liter, and D myeloid-antigen expression and a white-cell count ≥50× 109 per liter. Log-rank analysis showed statistically significant differences between curves (P<0.00001 ). Pairwise comparison of the curves showed statistically significant differences between A and B (P<0.01), A and C (P = 0.0004), A and D (P<0.00001), and B and D (P = 0.007). The difference between C and D approached significance (P<0.03). B and C were not significantly different (P = 0.40).

Some prognostic features in ALL are interrelated or correlated — e.g., T-lineage phenotype and extramedullary disease. Therefore, a multivariate analysis was performed to determine the features that, when considered collectively, provided the best predictive model of event-free survival or relapse. In this analysis, myeloid-antigen expression was again found to be the most significant predictor of relapse (Table 3). When a P value of 0.01 was used in entering features, myeloid-antigen expression (P<0.0001) and a high initial white-cell count (P<0.001) were jointly significant. The combination of these two features was the most powerful predictor of relapse in childhood ALL. Log-rank analysis of survival curves demonstrated that myeloid-antigen expression and the initial white-cell count were the most significant factors in predicting event-free survival (P<0.00001). The estimated event-free survival rates after 40 months were 84 ±4 percent for patients negative for myeloid antigens and with a low white-cell count, 57±12 percent for those negative for such antigens and with a high white-cell count, 47 ±14 percent for those positive for myeloid antigens and with a low white-cell count, and 26±13 percent for those myeloid-antigen—positive and with a high white-cell count (Fig. 4).

Effect of Therapy

Intensive chemotherapeutic regimens have been used in an effort to improve event-free survival for patients considered to be at a high risk of relapse.17 , 18 Our data revealed that for patients with ALL and myeloid-antigen expression, the use of intensive chemotherapy regimens was ineffective. The event-free survival rate after 40 months for such patients treated with intensive therapy was 40±13 percent. When all patients who received intensive chemotherapy were evaluated separately by multivariate analysis, myeloid-antigen expression was again the most significant predictor of relapse (P<0.001), whereas a high white-cell count approached significance (P<0.02) (data not shown).

Analysis According to Lineage

When the patients with ALL of B lineage were analyzed separately, the association of myeloid-antigen expression with L2 morphology approached significance (P<0.06), and myeloid-antigen expression did not correlate with any other presenting feature. Univariate analysis showed that myeloid-antigen expression was the most significant predictor of relapse (P<0.0001), followed by a high initial white-cell count (P<0.001) (Table 3). After 40 months, the estimated event-free survival rate of patients with myeloid-antigen—positive ALL of B lineage was 43 ± 14 percent, as compared with 80±6 percent for patients with ALL of B lineage who were negative for myeloid antigens (P<0.005) (Fig. 3B). Multivariate analysis showed that the combination of myeloid-antigen expression and a high initial white-cell count was the best predictor of relapse in children with B-lineage ALL, with myeloid-antigen expression being the more important (Table 3). The similarity of these results to those of the analyses of the entire group of patients with ALL is not surprising, since 78 percent of our study group had B-lineage ALL.

Similar analyses were performed for the patients with T-lineage ALL. The same trends were seen, but they were not statistically significant, perhaps because of the small number of patients in this group. Among these patients, the estimated event-free survival rate after 40 months was 23±14 percent for those with myeloid-antigen expression, as compared with 68±9 percent for those without such expression (P<0.13).

Discussion

We have prospectively documented myeloid-antigen expression among a cohort of 236 patients who had recently been given diagnoses of childhood ALL. In 53 patients (22 percent), analysis of the leukemic marrow cells demonstrated simultaneous expression of either B- or T-lymphocyte surface antigens and myeloid surface antigens on the cells. The percentage of patients with newly diagnosed ALL who had mixed-lineage expression was comparable to that reported by others.8 9 10

Dual-fluorescence analysis is the best method for directly demonstrating simultaneous expression of lymphoid and myeloid antigens, especially when marrow samples contain residual normal elements. Normal elements can complicate single-fluorescence analysis, because staining of the leukemic cells cannot always be distinguished from that of normal marrow cells expressing myeloid antigens.

The use of monoclonal antibodies that are not lineage-specific has confounded some other studies.12 For example, CD15, which is recognized by the monoclonal antibody My1, is a hapten that is not unique to myeloid cells.12 , 29 CD11b has also been used in a previous study to identify myeloid-antigen—positive ALL. However, CD11b is a complement receptor found on large granular lymphocytes as well as granulocytes and monocytes.25 To avoid the problem of nonspecificity, in our study the definition of myeloid surface-antigen expression was restricted to CD33, CD 13, and CD 14, three antigens that have not been found on normal T or B lymphocytes.25

Previous studies of childhood ALL have demonstrated the prognostic importance of clinical and laboratory features present at the time of diagnosis.13 , 21 , 28 The factors predicting a poor prognosis have been an initial white-cell count above or equal to 50 × 109 per liter, age over 10 years or under 1 year, immunophenotype of T lineage, presence of bulky extramedullary disease, male sex, a platelet count below 100× 109 per liter, or a hemoglobin level over 10 g per deciliter. Many of these features are interdependent, such as T-lineage immunophenotype with older age, presence of extramedullary disease, a high white-cell count, and normal hemoglobin values. Multivariate analyses have demonstrated the importance of the initial white-cell count, age, T-lineage immunophenotype, and the presence of bulky extramedullary disease as the chief prognostic factors in childhood ALL.21 , 28 , 30 , 31 The predictive value of these presenting features is not absolute, however. For example, some children thought to be at low risk of leukemic relapse have a poor response to therapy.

The present study identifies myeloid-antigen expression as the most statistically significant predictor of a poor outcome in childhood ALL. In both univariate and multivariate analyses, myeloid-antigen expression was more important than the initial white-cell count, age, T-lineage immunophenotype, the presence of extramedullary disease, or any other feature in predicting leukemic relapse. Analyses of the relations between myeloid-antigen expression and other presenting features demonstrated that except for an increased frequency of L2 morphology, myeloid-antigen expression was an independent variable, unrelated to the other prognostic factors. The only statistically significant features (P<0.01) that predicted event-free survival in a multivariate analysis were myeloid-antigen expression and the initial white-cell count. The power of these two factors to predict relapse in childhood ALL is shown by the fact that the prognosis was found to be worst in children with the combination of myeloid-antigen expression and an initial white-cell count greater than or equal to 50 × 109 per liter (Fig. 4). Children whose leukemic cells expressed myeloid antigens and who had an initial white-cell count under 50 × 109 per liter had outcomes similar to those of children with a white-cell count greater than or equal to 50 × 109 per liter but who did not have myeloid-antigen expression.

The identification of risk factors for relapse has been used to stratify children with ALL into different treatment programs. Patients at a high risk of relapse can then receive the most intensive chemotherapy to improve the control of their leukemia. Because the patients in the present study received chemotherapy that was stratified on the basis of previously identified risk factors, we examined whether intensive therapy could overcome the poor prognosis accompanying myeloid-antigen expression in ALL. Despite the administration of intensive chemotherapy regimens, the patients with myeloid-antigen expression still had a worse prognosis than those without such expression. Thus, myeloid-antigen expression identifies a group of patients for whom current intensive chemotherapy regimens may not be effective.

The results of the current study differ from those of other reports. In adults with ALL, myeloid-antigen expression with B-lineage ALL has been associated with a low rate of induction of remission.9 In the current study, the failure rate for induction of remission was no different in the patients with or without myeloid-antigen expression. The discrepancy may be due to differences between children and adults in the biology of ALL. Our results also differ from those reported in children with ALL by Pui et al.10 In their study, no difference in outcome was found in children with ALL with or without myeloid-antigen expression who received either of two intensive-therapy regimens. Nearly 40 percent of the myeloid-antigen—positive patients with ALL in their study were included because they had expression of CD 11b, an antigen found on nonmyeloid cells. CD11b expression was not used as a criterion for inclusion in the current study because of its broad range. In addition, the therapy used in the study by Pui et al. was not stratified according to risk category. The patients were randomly assigned to receive one of two intensive-therapy regimens. The patients in the current study were stratified on the basis of previously identified risk groups.

Improvement in the outcome of myeloid-antigen—positive ALL will require understanding of the biologic basis of myeloid-antigen expression in ALL. It is not known whether the leukemic-cell immunophenotype in this form of the disease corresponds to that of a normal lymphohematopoietic cell or whether it is caused by the aberrant expression of myeloid surface antigens after transformation.11 , 12 Several experiments have suggested that in some cases, the simultaneous expression of lymphoid and myeloid antigens on leukemic cells reflects the transformation of a multipotent progenitor cell. Occasionally, treatment of leukemias has resulted in conversion in vivo from a lymphoid to a myeloid phenotype, or vice versa.32 , 33 Similar results with in vitro treatment with growth factor have been reported.34 Thus, one explanation for the expression of myeloid-associated antigens in ALL is that the leukemias are the progeny of transformed multipotent cells that represent a particularly promiscuous stage of normal differentiation. In this intermediate stage before commitment to either myeloid or lymphoid differentiation, the pattern of surface-antigen expression may be more variable than at later stages.12

The patients with B-lineage ALL and myeloid-antigen expression did not have consistent patterns of myeloid-antigen expression. For example, the expression of CD 14, a monocytic marker, was frequently found without the expression of other myeloid surface antigens. The lack of consistent myeloid-antigen expression suggests that the CD 14 expression represents an aberration, rather than the activation of a particular program for differentiation. It should be noted, however, that recent experiments with virally transformed murine cells suggest that the pathways of B-lymphocyte and monocyte differentiation may be closer than previously appreciated.35 In the present study, there was no evidence that the leukemic cells expressing the phenotype of an immature B-lymphocyte precursor (e.g., DR+, CD19 +, CD24+, CD10+/—, or CD20–) were more likely to demonstrate mixed-lineage expression than cells with the more mature phenotype (DR+, CD19 +, CD24+, CD10+, or CD20 + ).

A wide variety of cytogenetic abnormalities were observed in the ALL cells with myeloid-antigen expression. Previous reports have indicated an association between structural abnormalities of 11q23 and mixed-lineage expression.26 , 27 In the present series, there was only one leukemia with a structural abnormality of 11q23 observed among the patients with ALL and myeloid-antigen expression. No association could be shown between the types of cytogenetic abnormalities in these patients and prognosis.

Mirro et al. have proposed the term "acute mixed-lineage leukemia" to describe both ALL with myeloid-antigen expression and acute nonlymphoblastic leukemia with expression of lymphoid-associated antigens.8 Mixed-lineage leukemia may be a useful term for characterizing the clinically important phenomenon of simultaneous expression of myeloid and lymphoid antigens on leukemic cells. One must be cautious, however, about the use of this term. With the possible exception of the T-lineage ALL expressing CD 13 or CD33, the mixed-lineage expression described in our study is heterogeneous with regard to patterns of surface-antigen expression, morphology, cytogenetics, and clinical presentation. In many defined subtypes of leukemia — e.g., Burkitt's leukemia and lymphoma —correlation between the cellular and molecular biology and the clinical features supports the idea that the subtype is a single entity.36 In contrast, ALL with myeloid-antigen expression probably represents a diverse group of disorders. Elucidation of its biologic features may lead to the development of improved therapeutic approaches for patients with these disorders.

Funding and Disclosures

We are indebted to Geralyn Annett, Judith Brooks, Raymond Chan, and Felix Burotto for expert assistance with the immunophenotyping and flow cytometry; to Lisa Wiersma for data management; to Pei-Jiuan Lee for statistical analyses; to Drs. Carl Lenarsky, Donald Kohn, and Lennie Sender for reviewing the manuscript; and to Dr. Robertson Parkman for his encouragement, helpful discussions, and critical review.

Author Affiliations

From the Divisions of Hematology/Oncology (S.R.W., J.O.) and Research Immunology/Bone Marrow Transplantation (K.I.W.), Childrens Hospital Los Angeles, and the Departments of Pediatrics and Preventive Medicine (E.S.), University of Southern California School of Medicine, Los Angeles. Address reprint requests to Dr. Wiersma at the University of Wisconsin, Department of Pediatrics, 600 Highland Ave., Madison, WI 53792.

References (36)

  1. 1. Greaves MF. . Differentiation-linked leukemogenesis in lymphocytes . Science 1986; 234:697–704.

  2. 2. Foon KA, Todd RF III. . Immunologic classification of leukemia and lymphoma . Blood 1986;68:1–31.

  3. 3. Nadler LM, Korsmeyer SJ. Anderson KC, et al. . B cell origin of non-T cell acute lymphoblastic leukemia: a model for discrete stages of neoplastic and normal pre-B cell differentiation . J Clin Invest 1984; 74:332–40.

  4. 4. Reinherz EL, Kung PC, Goldstein G, Levey RH. Schlossman SF. . Discrete stages of human intrathymic differentiation: analysis of normal thymocytes and leukemic lymphoblasts of T-cell lineage . Proc Natl Acad Sci U S A 1980;77:1588–92.

  5. 5. Smith LJ, Curtis JE, Messner HA, Senn JS, Furthmayr H, McCulloch EA. . Lineage infidelity in acute leukemia . Blood 1983; 61:1138–45.

  6. 6. Pui C-H. Dahl GV, Melvin S. et al. . Acute leukemia with mixed lymphoid and myeloid phenotype . Br J Hematol 1984; 56:121–30.

  7. 7. Neame PB. Soamboonsrup P, Browman G, et al. . Simultaneous or sequential expression of lymphoid and myeloid phenotypes in acute leukemia . Blood 1985;65:142–8.

  8. 8. Mirro J,Zipf TF, Pui C-H, et al. . Acute mixed lineage leukemia: clinicopathologic correlations and prognostic significance . Blood 1985; 66:1115–23.

  9. 9. Sobol RE, Mick R, Royston I, et al. . Clinical importance of myeloid antigen expression in adult acute lymphoblastic leukemia . N Engl J Med 1987; 316:1111–7.

  10. 10. Pui C-H, Behm FG. Singh B, et al. . Myeloid-associated antigen expression lacks prognostic value in childhood acute lymphoblastic leukemia treated with intensive multiagent chemotherapy . Blood 1990; 75:198–202.

  11. 11. McCulloch EA. . Stem cells in normal and leukemic hemopoiesis (Henry Stratton lecture, 1982) . Blood 1983;62:1–13.

  12. 12. Greaves MF, Chan LC, Furley AJW, Watt SM, Molgaard HV. . Lineage promiscuity in hemopoietic differentiation and leukemia . Blood 1986; 67:1–11.

  13. 13. Miller DR, Leikin S, Albo V, Sather H, Karon M, Hammond D. . Prognostic factors and therapy in acute lymphoblastic leukemia of childhood: CCG-141 . Cancer 1983; 51:1041–9.

  14. 14. Riehm H. Gadner H, Henze G, et al. Acute lymphoblastic leukemia: treatment results in three BFM studies (1970–1981). In: Murphy SB, Gilbert JR. eds. Leukemia research: advances in cell biology and treatment. New York: Elsevier, 1983:251–60.

  15. 15. Gaynon P. Steinherz P, Reaman G, et al. . Delayed intensification for children with acute lymphoblastic leukemia with high risk features . Proc Am Soc Clin Oncol 1985; 4:173. abstract.

  16. 16. Duque-Hammershaimb L, Wollner N, Miller DR. . LSA2–L2 protocol treatment of stage IV non-Hodgkin's lymphoma in children with partial and extensive bone marrow involvement . Cancer 1983; 52:39–43.

  17. 17. Steinherz PG, Gaynon P, Miller DR, et al. . Improved disease-free survival of children with acute lymphoblastic leukemia at high risk for early relapse with the New York regimen — a new intensive therapy protocol . J Clin Oncol 1986; 4:744–52.

  18. 18. Gaynon P, Steinherz PG. Bleyer WA, et al. . Intensive therapy for children with acute lymphoblastic leukemia and unfavourable presenting features . Lancet 1988; 2:921–4.

  19. 19. Reaman G, Lange B, Feusner J, et al. . Intensive systemic and intrathecal chemotherapy improves the outcome for infants with acute lymphoblastic leukemia . Proc Am Soc Clin Oncol 1989; 8:211. abstract.

  20. 20. Bennett JM, Catovsky D. Daniel M-T, et al. . Proposals for the classification of the acute leukemias: French-American-British (FAB) Cooperative group . Br J Haematol 1976; 33:451–8.

  21. 21. Bleyer WA, Sather H, Coccia P, Lukens J, Siegel S, Hammond GD. . The staging of childhood acute lymphoblastic leukemia: strategies of the Childrens Cancer Study Group and a three-dimensional technic of multivariate analysis . Med Pediatr Oncol 1986; 14:271–80.

  22. 22. Cox DR. . Regression models and life-tables . J R Stat Soc [B] 1972; 34:187–220.

  23. 23. Kaplan EL, Meier P. . Nonparametric estimation from incomplete observations . J Am Stat Assoc 1958; 53:457–81.

  24. 24. Lawless JF. Statistical models and methods for lifetime data. New York: John Wiley, 1982.

  25. 25. Civin CI. . Human monomyeloid cell membrane antigens . Exp Hematol 1990; 18:461–7.

  26. 26. Arthur DC, Bloomfield CD, Lindquist LL, Nesbit ME. . Translocation 4;11 in acute lymphoblastic leukemia: clinical characteristics and prognostic significance . Blood 1982; 59:96–9.

  27. 27. Mirro J, Kitchingman G, Williams D, et al. . Clinical and laboratory characteristics of acute leukemia with the 4;11 translocation . Blood 1986; 67:689–97.

  28. 28. Robison LL, Sather HN, Coccia PF, Nesbit ME, Hammond GD. . Assessment of the interrelationship of prognostic factors in childhood acute lymphoblastic leukemia . Am J Pediatr Hematol Oncol 1980; 2:5–13.

  29. 29. Stockinger H, Majdic O, Liszka K, et al. . Exposure by desialylation of myeloid antigens on acute lymphoblastic leukemia cells . J Natl Cancer Inst 1984;73:7–11.

  30. 30. Kersey JH, LeBien TW, Hurwitz R, et al. . Childhood leukemia-lymphoma: heterogeneity of phenotypes and prognoses . Am J Clin Pathol 1979; 72:746–52.

  31. 31. Sallan SE, RitzJ, Pesando J, et al. . Cell surface antigens: prognostic implications in childhood acute lymphoblastic leukemia . Blood 1980; 55:395–402.

  32. 32. Hershfield MS, Kurtzberg J, Harden E, Moore JO, Whang-Peng J, Haynes BF. . Conversion of a stem cell leukemia from a T-lymphoid to a myeloid phenotype induced by the adenosine deaminase inhibitor 2'-deoxycoformycin . Proc Natl Acad Sci U S A 1984; 81:253–7.

  33. 33. Stass S, Mirro J, Melvin S, Pui C-H, Murphy SB, Williams D. . Lineage switch in acute leukemia . Blood 1984; 64:701–6.

  34. 34. Griesinger F, Arthur DC, Brunning R, et al. . Mature T-lineage leukemia with growth factor-induced multilineage differentiation . J Exp Med 1989; 169:1101–20.

  35. 35. Hanecak R, Zovich DC, Pattengale PK, Fan H. . Differentiation in vitro of a leukemia virus-induced B-cell lymphoma into macrophages . Mol Cell Biol 1989; 9:2264–8.

  36. 36. Croce CM, Nowell PC. . Molecular basis of human B cell neoplasia . Blood 1985;65:1–7.

Citing Articles (132)

    Letters

    Figures/Media

    1. Table 1. Antigenic Determinants.
      Table 1. Antigenic Determinants.
    2. Figure 1. Flow-Cytometric Analysis of Immunophenotypes of Patients with ALL and Myeloid-Antigen Expression.
      Figure 1. Flow-Cytometric Analysis of Immunophenotypes of Patients with ALL and Myeloid-Antigen Expression.

      Panel A shows single-fluorescence staining of a bone marrow sample from one patient with ALL (solid line) after staining with antibodies against CD10 and CD13, as compared with an ascites control (dashed line). The abscissa indicates the logarithm of the relative intensity of green fluorescence with the fluoresceinated antibodies indicated. The ordinate indicates the relative frequency of green fluorescence of fluorescein isothiocyanate as an indicator of antibody staining. Because the bone marrow was not fully replaced with leukemic lymphoblasts, single-fluorescence analysis could not differentiate between the simultaneous expression of CD13 on the leukemic cells and the presence of residual normal myeloid cells. In Panel B, dual-fluorescence analysis of cells from the same patient with ALL shows expression of both CD10 and CD13 on the leukemic lymphoblasts (quadrant 2). The upper left quadrant contains cells that reacted with CD13; the lower right quadrant, cells that reacted with CD10; and the upper right quadrant, cells that reacted with both antibodies and hence express both antigens. Cells that reacted with neither antibody are shown in the lower left quadrant. In Panel C, single-fluorescence staining of a bone marrow sample from another patient with ALL in which the marrow was completely replaced by leukemic lymphoblasts shows that 70 percent of the cells express CD19 and 80 percent express CD33, indicating that the CD19+ cells also express CD33.

    3. Figure 2. Distribution of Immunophenotypes in 53 Patients with ALL and Myeloid-Antigen Expression.
      Figure 2. Distribution of Immunophenotypes in 53 Patients with ALL and Myeloid-Antigen Expression.

      The number of patients with ALL who had one, two, or three myeloid surface antigens (CD13, CD14, or CD33) expressed on their leukemia cells is shown for patients with B-lineage ALL of an immature (CD10–, CD20–), intermediate (CD10+, CD20–), or more mature (CD10+, CD20+) stage of differentiation. In eight patients CD20 expression was not tested (NT). Eight patients with T-lineage ALL were also found to express at least one myeloid antigen. No correlation was found between the stage of B-lymphoid differentiation and the number or pattern of the myeloid antigens expressed.

    4. Table 2. Features of Patients with ALL at the Time of Diagnosis, According to Presence or Absence of Myeloid-Antigen Expression.
      Table 2. Features of Patients with ALL at the Time of Diagnosis, According to Presence or Absence of Myeloid-Antigen Expression.
    5. Table 3. Prognostic Importance of the Presenting Features in All Patients Studied and in Those with ALL of B Lineage, as Determined by Univariate and Multivariate Analysis.*
      Table 3. Prognostic Importance of the Presenting Features in All Patients Studied and in Those with ALL of B Lineage, as Determined by Univariate and Multivariate Analysis.*
    6. Figure 3. Kaplan–Meier Curves of Estimated Event-free Survival for Patients with ALL with (Dotted Line) and without (Solid Line) Myeloid-Antigen Expression.
      Figure 3. Kaplan–Meier Curves of Estimated Event-free Survival for Patients with ALL with (Dotted Line) and without (Solid Line) Myeloid-Antigen Expression.

      Rates for all patients studied are shown in Panel A, and rates for patients with ALL of B lineage in Panel B.

    7. Figure 4. Kaplan–Meier Curves of Estimated Event-free Survival of Patients with ALL, as Stratified According to Presence or Absence of Myeloid-Antigen Expression and Initial White-Cell Count.
      Figure 4. Kaplan–Meier Curves of Estimated Event-free Survival of Patients with ALL, as Stratified According to Presence or Absence of Myeloid-Antigen Expression and Initial White-Cell Count.

      A denotes no myeloid-antigen expression and a white-cell count <50 × 109 per liter, B no myeloid-antigen expression and a white-cell count ≥50 × 109 per liter, C myeloid-antigen expression and a white-cell count <50×109 per liter, and D myeloid-antigen expression and a white-cell count ≥50× 109 per liter. Log-rank analysis showed statistically significant differences between curves (P<0.00001 ). Pairwise comparison of the curves showed statistically significant differences between A and B (P<0.01), A and C (P = 0.0004), A and D (P<0.00001), and B and D (P = 0.007). The difference between C and D approached significance (P<0.03). B and C were not significantly different (P = 0.40).

      More Research