Increased coexpression of c-KIT and FLT3 receptors on myeloblasts: Independent predictor of poor outcome in pediatric acute myeloid leukemia

Authors

  • Surender Kumar Sharawat,

    1. Department of Medical Oncology, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Ritu Gupta,

    1. Laboratory Oncology Unit, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Vinod Raina,

    1. Department of Medical Oncology, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Lalit Kumar,

    1. Department of Medical Oncology, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Atul Sharma,

    1. Department of Medical Oncology, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Sobuhi Iqbal,

    1. Department of Medical Oncology, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Radhika Bakhshi,

    1. Biomedical Sciences, University of Delhi, New Delhi, India
    Search for more papers by this author
  • Sreenivas Vishnubhatla,

    1. Biostatistics Unit, University of Delhi
    2. All India Institute of Medical Sciences, New Delhi, India
    Search for more papers by this author
  • Sameer Bakhshi

    Corresponding author
    1. Department of Medical Oncology, University of Delhi
    2. Dr. B. R.A. Institute Rotary Cancer Hospital, New Delhi, India
    3. All India Institute of Medical Sciences, New Delhi, India
    • Correspondence to: Sameer Bakhshi, Additional Professor of Pediatric Oncology, Department of Medical Oncology, Dr. B. R.A. Institute Rotary Cancer Hospital, All India Institute of Medical Sciences, New Delhi 110 029, India. E-mail: sambakh@hotmail.com

    Search for more papers by this author

Abstract

Background

Significance of mutations in FLT3 and c-KIT genes in AML has been well established, but role of their coexpression has not been evaluated. The aim of this study was to evaluate clinical significance of FLT3 (CD135) and c-KIT (CD117) coexpression on myeloblasts in AML.

Methods

Using flow-cytometry, we prospectively observed in 115 AML patients that CD135, CD117, and CD135+CD117 coexpression was expressed in 95 (82%), 104 (90%), and 81 (70%) patients respectively. Median expression of CD135, CD117, and their co expression was used as cut off for high and low expression.

Results

FLT3 ITD (internal tandem duplication) was present in 20 (17%) patients. High coexpression did not correlate with FLT3 ITD (P = 0.432) and cytogenetics (P = 0.244). Out of 115 patients, 86 (74.7%) achieved remission. At median followup of 15.5 months, EFS and OS was 29% and 35%, respectively for the entire cohort. Patients with high coexpression of CD135 and CD117 in comparison to those with low coexpression had significantly inferior EFS (20% vs 38% P < 0.001) and OS (27% vs 44% P = 0.001). In step wise Cox regression multivariable analysis, hazard ratio for high hemoglobin, WBC count, and coexpression of CD135 and CD117 was 0.63, 1.73, and 2.46 respectively for EFS, and for OS only CD135+CD117 coexpression emerged as an independent predictor (hazard ratio 2.25).

Conclusions

This is the first study to show that high coexpression of CD135+CD117 is an independent predictor of poor outcome in AML and is easily measurable by routine diagnostic flow-cytometry. © 2013 International Clinical Cytometry Society

The pathogenesis of acute myeloid leukemia (AML) is linked to imbalance between proliferation and apoptosis. Amongst proliferative markers, receptor tyrosine kinases (RTK) have significant contribution in leukemogenesis. The key proliferative RTKs for AML include c-KIT receptor (CD117) and FLT-3 receptor (CD135). CD117 is a diagnostic marker for AML and is expressed in >85% of patients with AML([1]). There is variable data to suggest that CD117 overexpression may or may not be associated with outcome in AML ([2, 3]). Overexpression of CD135 has been associated with poor outcome in AML ([4]).

Since proliferation is one of the mechanisms for AML and both CD117 and CD135 are proliferative markers, higher expression of these markers would imply a higher proliferative potential. Data pertaining to combined expression of both these proliferative markers and its impact on outcome in AML has not been evaluated. We hypothesized that higher coexpression of CD135 and CD117 would imply a higher proliferative potential of myeloblasts, and that it would adversely affect outcome. Thus, we conducted a prospective study to assess impact of coexpression of CD135 and CD117 on outcome in AML patients.

MATERIALS AND METHODS

Patients and Treatment

Newly diagnosed consecutive de novo AML patients ≤60 years of age were enrolled from March 2008–June 2010 prospectively after institute ethical approval and taking informed consent. Acute promyelocytic leukemia (APML) was excluded. Induction included 3+7 regimen (Daunorubicin and cytosine arabinoside) with Daunorubicin at 60 mg/m2 for 3 days; patients who were not in complete remission (CR) post first induction received ADE protocol ([5]). Patients who were not in CR after two inductions were declared refractory. Post remission, patients received three cycles of high dose cytarabine at 18 gm/m2/cycle.

Flow Cytometry

Five ml peripheral blood (if peripheral blast count was more than 30%) or otherwise 5 ml bone marrow was collected using ethylenediaminetetraacetic acid and processed within 24 hours of collection. Mononuclear cells (MNC) were separated using density gradient media. Monoclonal antibodies for CD45, CD135 and CD117 were tagged with fluorochromes fluorescein isothiocyanate, phycoerythrin and phycoerythrin cyanine-5 respectively (BD Biosciences, USA). Multiparameter flow-cytometry analysis was performed using a FACS Calibur flow-cytometer and cell quest pro software (Becton Dickinson, San Jose, CA, USA). Total 20,000 events were acquired and CD135 and CD117 percentage expression seen on gated myeloblasts (dim CD45) (Fig. 1). Unstained cells (CD45 only) were used as negative controls for CD135 and CD117 expression assessment. Greater than 20% myeloblasts expressing the protein was classified as positive expression.

Figure 1.

Flow cytometric based expression of CD135, CD117, and their coexpression. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Statistical Analysis

Outcome analysis was assessed based on event-free-survival (EFS) and overall-survival (OS). EFS were defined as time between diagnosis and first event such as failure to achieve CR, relapse or death. OS was defined as the time between diagnosis and death or last followup. The end point of study was 5 September, 2012. Data was expressed as median (range) and mean ± SD; the differences between values were determined by using Kruskal Wallis test and Student's t test respectively. Median values of quantitative variables were used as cut off point for categorization into high or low expression. Mean values of CD135, CD117, and their coexpression were compared with baseline patient characteristics. Kaplan Meier curves were obtained for survival analysis followed by log rank test. Stepwise multivariable Cox regression method was employed to evaluate independent prognostic factors. Prognostic index (PI) of individual patients was calculated with hazard coefficient and using this index, predicted EFS of patients was calculated. P value ≤ 0.05 was considered significant. All statistical analysis was done using STATA 11.0.

RESULTS

Patient Selection, Baseline Characteristics and Outcome

During the study period total 168 patients were registered in our centre and after exclusion [Biphenotypic leukemia (n = 4), age > 60 years (n = 11), APML (n = 14), therapy not taken (n = 10) and logistical issues (n = 14)], 115 patients were finally recruited in current study. Median age was 16 years (range: 1–59 years); median hemoglobin, white blood count (WBC), and platelet counts were 7.2 g/dl (range: 2.0–14.5 g/ dl), 22,000/mm3 (range: 700–6,85,000/mm3) and 31,500/mm3 (range: 1900-2,72,550 /mm3), respectively. FLT3-internal tandem duplication was present in 20 (17%) patients. Cytogenetics was available for 74% patients (23% good risk, 59% intermediate risk and 18% poor risk). CR rate was 74.7% for entire cohort. At median followup of 15.5 months (range: 0.5–53.4 months), EFS was 28.7% ± 4.28% (CI: 20.64–37.24) and OS 35.42% ± 4.90% (CI: 25.99–44.96).

CD135, CD117, and CD135+CD117 Coexpression and Correlation with Patient Characteristics

CD135, CD117, and their coexpression on myeloblasts was observed in 82%, 90%, and 70% AML patients respectively (Table 1). Median expression of CD135, CD117, and their coexpression was 53.2% (range: 1.21%–99.91%), 74.7% (range: 0.69%–99.48%) and 34.1% (range: 0.1%–97.33%) respectively. Patients > 18 years had a significant association towards higher mean expression of CD135 (62% vs 46%, P = 0.024) and coexpression of CD135 and CD117 (46% vs 35%, P = 0.029); no other baseline patient characteristic including cytogentics and FLT3-ITD expression was associated with CD135, CD117, and their coexpression (Table 2).

Table 1. Expression of CD135, CD117, and CD135+CD117
 ALL patients (N = 115)Pediatric patients (N = 64)Adult patients (N = 51)
VariablesPositiveMedian expressionPositiveMedian expressionPositiveMedian expression
  1. aCoexpression denotes expression of both CD135 and CD117 on the same blast and is different from a patient having blasts which express both CD135 and CD117.
CD13583% (95)53.281% (52)42.284% (43)71.1
CD11790% (104)74.791% (58)69.890% (46)76.3
CD135 + CD117a70% (81)34.170% (45)31.571% (36)44.0
Table 2. Association of CD135+CD117 Coexpression with Baseline Patient Characteristics
VariablesCD135 (mean ± SD)PCD117 (mean ± SD)PCD135+CD117 (mean ± SD)P
Age 0.024 0.192 0.029
Pediatric ( 64)46.1 ± 29.7 62.6 ± 27.3 34.9 ± 24.2 
Adults( 51)61.8 ± 32.3 69.1 ± 25.7 46.0 ± 29.7 
Sex 0.457 0.505 0.908
Male (77)52.8 ± 30.9 66.7 ± 26.1 39.7 ± 26.8 
Female (38)57.5 ± 33.4 63.1 ± 28.1 40.3 ± 28.5 
Hemoglobin (g/ dl) 0.307 0.408 0.746
<7.2 (56)57.5 ± 31.1 63.4 ± 26.9 40.7 ± 26.74 
≥7.2 ( 59)51.4 ± 32.2 67.5 ± 26.6 39.1 ± 27.9 
WBC (/mm3) 0.362 0.893 0.699
<22000 ( 57)51.6 ± 33.0 65.2 ± 27.5 38.9 ± 28.0 
≥22000 ( 58)57.1 ± 30.4 65.8 ± 26.1 40.8 ± 26.6
Platelets (/mm3) 0.159 0.586 0.163
<31,500 (57 )58.8 ± 31.7 66.9 ± 26.0 43.6 ± 27.6 
≥31,500 ( 57)50.4 ± 31.4 64.1 ± 27.8 36.4 ± 26.7 
FLT3 0.273 0.704 0.432
Positive (20)46.7 ± 27.1 65.0 ± 25.0 34.1 ± 21.7 
Negative ( 95)56.0 ± 32.4 65.6 ± 27.2 41.1 ± 28.2 
Cytogenetics 
Good risk (20)49.5 ± 31.60.63261.7 ± 27.10.25734.2 ± 24.70.341
Intermediate risk (50)54.8 ± 30.9 64.8 ± 25.8 40.2 ± 25.8 
Poor risk (15)57.6 ± 30.6 74.9 ± 21.7 48.1 ± 29.3 
AML subtype 
M2 (66)52.5 ± 31.80.56161.8 ± 27.80.13336.5 ±25.70.244
M4 (16)62.1 ± 31.3 76.5 ± 15.4 53.3 ± 27.4 
Others (33)54.4 ± 32.0 67.8 ± 27.7 40.2 ±28.9 

The mean of fluorescent intensity was significantly higher for CD135 and CD117 in patients >18 years of age; no other baseline patient characteristic was associated with their fluorescent intensities (Supporting Information Table 1).

Univariate Analysis of CD135, CD117, and CD135+CD117 Coexpression

Over expression of CD135 and coexpression of CD135 and CD117 was significantly associated with poor CR rate, EFS and OS (Table 3) (Figs. 2, 3). When median fluorescent intensity was used for analysis, it was observed that the CR rate (P = 0.07), EFS (P = 0.04), and OS (P = 0.06) was inferior in the cohort with relatively higher CD135 expression (Supporting Information Table 2). Amongst the baseline patient characteristics, poor EFS had a trend with lower median hemoglobin (20.1% vs 36.3%, P = 0.067) and significant association with higher than median WBC (22.9% vs 34.6%, P = 0.043) (Fig. 2D), whereas poor OS was noted with higher WBC (30.23 % vs 40.84%, P = 0.080) (Fig. 3D).

Figure 2.

Kaplan-Meier EFS curves for (A) CD135, (B) CD117, (C) coexpression of CD135 and CD117, and (D) WBC. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 3.

Kaplan- Meier OS curves for (A) CD135, (B) CD117, (C) coexpression of CD135 and CD117, and (D) WBC. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Table 3. Univariate Survival Analysis for the CD135, CD117, and CD135 + CD117 Coexpression
VariablesCR%PEFSa ± SE (CI)POSa ± SE (CI)P
  1. aSurvival at 54 Months.
For all patients (N = 115)
CD135 (%)<53.289 (51)<0.00135.63 ± 6.60 (23.06–48.39)<0.00141.97 ± 7.44 (27.37–55.89)0.001
 ≥53.261 (35) 21.21 ± 5.43 (11.73–32.55) 27.50 ± 6.44 (15.84–40.47) 
CD117 (%)<74.682 (47)0.07729.82 ± 6.06 (18.61–41.87)0.45839.21 ± 6.98 (25.69–52.46)0.357
 ≥74.668 (39) 27.70 ± 6.03 (16.69–39.84) 32.12 ± 6.71 (19.63–45.29) 
CD135+ CD117 (%)<34.186 (49)0.00637.66 ± 6.60 (24.96–50.31)<0.00143.73 ± 7.29 (29.30–57.28)0.001
 ≥34.164 (37) 19.76 ± 5.33 (10.57–31.03) 26.90 ± 6.54 (15.15–40.11) 
For pediatric patients (N = 64)
CD135 (%)<42.2100 (32)0.00034.38 ± 8.40 (18.70–50.56)0.01743.49 ± 9.24 (25.33–60.36)0.029
 ≥42.266 (21) 26.00 ± 7.90 (12.30–42.06) 29.46 ± 8.68 (14.07–46.69) 
CD117 (%)<69.890 (29)0.09231.25 ± 8.19 (16.38–47.34)0.40038.48 ± 8.79 (21.74–55.01)0.6361
 ≥69.875 (24) 29.25 ± 8.19 (14.65–45.53) 36.62 ± 9.23 (19.35–54.09) 
CD135+ CD117 (%)<31.5100 (32)0.00040.63 ± 8.68 (23.83–56.79)<0.00148.53 ± 9.11 (30.04–64.74)<0.001
 ≥31.566 (21) 19.50 ± 7.13 (7.93–34.85) 23.57 ± 8.31 (9.71–40.83) 
For adult patients (N = 51)
CD135 (%)<71.076 (19)0.11636.67 ± 10.64 (17.05–56.60)0.01240.23 ± 13.25 (15.72–63.88)0.005
 ≥71.056 (14) 16.00 ± 7.33 (5.02–32.54) 23.47 ± 9.36 (8.32–42.95) 
CD117 (%)<76.268 (17)0.38428.00 ± 8.98 (12.42–45.98)0.97839.91 ± 11.02 (19.10–60.08)0.722
 ≥76.260 (15) 26.67 ± 9.17 (11.07–45.22) 30.77 ± 10.28 (12.82–50.85) 
CD135+CD117 (%)<44.068 (17)0.38436.67 ± 10.64 (17.05–56.60)0.03736.13 ± 12.22 (14.20 - 58.81)0.076
 ≥44.060 (15) 16.00 ± 07.33 (05.02–32.54) 25.13 ± 09.92 (08.91-45.43) 

Multivariable Analysis and Prognostic Index

In step-wise multivariable analysis, all baseline patient characteristics as shown in Table 2 and study variables as shown in Table 3 were subjected to cox-proportion hazard regression model. Inferior EFS was significantly associated with low hemoglobin, high WBC and high coexpression of CD135 and 117 whereas poor OS was significantly associated only with high coexpression of CD135 and 117 (Table 4). On the basis of hazard coefficient for each variable independently predictive for EFS in multivariate analysis, following formula for each individual patient was used: PI = (−0.5 × Hemoglobin score) + (0.5 × WBC score) + (0.9 × CD135 + CD117 score) where score is 1 if it is less than median and 2 if more than median; based on this EFS of 4 different PI [0.5, 1.0, 1.5, and 2.0] is shown in Figure. 4A. The difference in means of PI of patients with and without an event was significantly different (1.1 ± 0.53 vs 1.45 ± 0.59, P = 0.003) (Fig. 4B).

Figure 4.

A: Difference in EFS of the 4 cohorts based on prognostic index as created by hazard coefficient of all the significant predictors for EFS in multivariable analysis. B: Difference in prognostic index of the patients based on presence or absence of event. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Table 4. Step Wise Multivariable Analysis with Significant Baseline Characteristics (Multivariate Cox Proportional-Hazards Regression model)
 EFSaOSa
Variables Hazard ratioStandard errorP95% Confidence intervalHazard ratioStandard errorP95% Confidence interval
  1. asurvival at 54 Months. WBC, white blood cells; CR, complete remission; EFS, event free survival; OS, overall survival; SE, standard error; CI 95%, confidence interval.
All patients
Hemoglobin (g/ dl )<7.21.000.1440.0450.4060.989
 ≥7.20.63        
WBC (/mm3)<22,000.01.000.3940.0171.1032.694
 ≥22,000.01.73        
CD135 + CD117 (%)<34.11.000.5760.0001.5413.8671.000.5740.0001.5074.270
 ≥34.12.46    2.25    
Pediatric subgroup
Hemoglobin (g/ dl )<6.71.000.1340.0070.2270.7901.00
 ≥6.70.42    2.25    
CD135 + CD117 (%)<31.51.000.9950.0001.6575.8201.001.050.0011.5866.018
 ≥31.53.1    3.08    
Adult Subgroup
CD135 (%)<71.061.000.9180.0111.2385.1531.001.20.0061.3916.900
 ≥71.062.52    3.09    

Significance of CD135, CD117, and CD135+CD117 Coexpression Separately in Pediatric and Adult AML Patients

In univariate analysis, when data was separately analyzed for pediatric and adult AML patients, the data was same as the entire cohort in pediatric AML patients, whereas in adult AML coexpression of CD135 and CD117 lost its significance for OS (Table 3). Likewise in multivariable analysis, in pediatric AML patients coexpression of CD135 and CD117 emerged as the strongest factor for predicting EFS and OS, whereas in adult patients only CD135 expression emerged as an independent predictor for both EFS and OS (Table 4).

DISCUSSION

In our cohort, >20% expression of CD135 and CD117 in myeloblasts was observed in 82% and 90% patients, respectively. Previous studies have suggested that CD135 is expressed in 62%–90% ([4, 6, 7]), whereas CD117 is expressed in 48%–87% AML patients ([8-11]). This variability is linked to method of detection whether flow-cytometry or immunohistochemistry, and within flow-cytometry as to what cutoff may have been taken as positive expression.

Both CD135 and CD117 belong to RTK class III and share virtually all common characteristics with respect to their structure, mode of action, and signaling pathways. CD135 has been shown in many previous studies that its higher expression predicts poor outcome([4, 12]), while it is not so with CD117 ([3]). We also observed that high expression of CD135 predicted poor EFS and OS, but CD117 was not predictive of either EFS or OS. The two proliferating transcripts have been evaluated at molecular level but their coexpression is not technically feasible in a molecular study ([13]).

Our study is unique in the sense that although many studies have individually evaluated CD135 and CD117, none of the previous studies have evaluated their coexpression in myeloblasts. We tried to address the question if their simultaneous expression was a more powerful prognostic marker than their individual expression. We observed that 70% AML patients had coexpression of CD135 and CD117, and their higher expression was associated with adult patients at initial presentation. This coexpression denotes expression of both CD135 and CD117 on the same blast and is different from a patient having blasts which express both CD135 and CD117.

Interestingly we observed that their coexpression significantly predicted EFS and OS. Further, in multivariable analysis, individual expression of CD135 and CD117 did not predict survival, but their coexpression emerged as the strongest independent factor for predicting EFS and OS with hazard ratio of 2.46 and 2.25 respectively. Infact, for OS the only predictive factor for poor outcome was the coexpression of CD135 and CD117. We established a PI for EFS in our group of patients based on low hemoglobin, high WBC count and higher coexpression of CD135 and CD117.

Our study cohort had a predominance of pediatric patients, and when data was separately analyzed for pediatric and adult patients coexpression of CD135 and CD117 still emerged as the strongest factor for predicting EFS and OS in pediatric patients, but not so in adult AML patients.

Our study has for the first time indicated a possible synergy of higher expression of both RTKs with each other, which may have been responsible for poor survival of AML patients. Many clinical trials have used FLT3 antagonists, and one wonders if c-KIT inhibitors would be additionally useful in patients who coexpress both CD135 and CD117 ([14, 15]).

To conclude, we demonstrated for the first time that coexpression of CD135 and CD117 is observed in 70% of patients with AML with significant association toward increased expression in older age. Further, this coexpression can be easily measured by routine diagnostic flow-cytometry techniques in AML and may serve as a useful independent prognostic marker for predicting outcome in pediatric AML patients. Prospective validation of this easily measured prognostic marker is suggested in other large cohorts, and also to evaluate biologically the synergy between c-KIT and FLT3 in proliferation of myeloblasts.

Acknowledgment

Surender Kumar Sharawat is thankful to Indian Council of Medical Research (ICMR) and All India Institute of Medical Sciences New Delhi, India for research fellowship.

Ancillary