Prognostic impact of absolute lymphocyte counts at the end of remission induction in childhood acute lymphoblastic leukemia
Jeffrey E. Rubnitz MD, PhD,
Department of Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee
Department of Pediatrics, College of Medicine, University of Tennessee Health Science Center, Memphis, Tennessee
Corresponding author: Jeffrey E. Rubnitz, MD, PhD, Department of Oncology, St. Jude Children's Research Hospital, 262 Danny Thomas Place, Mail Stop 260, Memphis, TN 38105-2794; Fax: (901) 521-9005; firstname.lastname@example.org
We thank David Galloway, ELS, for expert editorial review; Tad McKeon for data collection; and Julie Groff for preparing the figures.
Ching-Hon Pui is an American Cancer Society Professor.
Absolute lymphocyte counts (ALC) during treatment have been associated with outcome in children and adults with hematologic malignancies. However, the impact of ALC relative to that of other prognostic factors on the outcome of children with acute lymphoblastic leukemia (ALL) treated in recent trials is unknown.
Outcomes of 399 patients aged ≤18 years with newly diagnosed ALL who were enrolled in the Total Therapy XV study at St. Jude Children's Research Hospital were analyzed according to ALC at the end of remission induction therapy.
An ALC ≥500 cells/μL was significantly more prevalent among patients with B-lineage ALL, in those with favorable presenting features, and in those who achieved negative minimal residual disease (MRD) status on day 43 of treatment. Both overall survival (OS) and event-free survival (EFS) were superior among patients with higher ALC, but only the association with OS was statistically significant in a univariate analysis. In multivariable analyses, ALC was not a significant predictor of outcome after controlling for age, leukocyte count, lineage, risk group, and MRD status at the end of induction (P > .1 for all comparisons). However, among MRD-negative patients, those with low ALC had a 5-year OS rate of 84.2% ± 8.9% versus 97.3% ± 1.0% for patients with higher ALC (P = .036).
In patients with acute leukemia undergoing hematopoietic cell stem transplantation (HSCT), the absolute lymphocyte count (ALC) after transplantation serves as an indicator of the speed and quality of engraftment and hematopoietic cell reconstitution and may also reflect the potential for graft-versus-leukemia activity. Thus, several studies have demonstrated that a higher ALC post-transplantation predicts a superior outcome in patients with either acute myeloid leukemia (AML) or acute lymphoblastic leukemia (ALL). A rapid regeneration of normal hematopoiesis after intensive chemotherapy also should be a favorable prognostic indicator, as this may reflect a deeper leukemia cytoreduction with a consequently higher functional capacity of the hematopoietic microenvironment. Conceivably, immune cells also may exert an antileukemic effect and suppress relapse. However, whether the ALC has prognostic impact in patients receiving chemotherapy alone is less clear. To this end, Behl et al reported that an ALC ≥500 cells/μL after induction chemotherapy was associated with superior overall survival (OS) and event-free survival (EFS) in 103 patients with AML. Investigators of the Children's Oncology Group (COG) examined the effect of ALC in childhood leukemia.[5, 6] In an analysis of 171 patients aged ≤21 years, De Angulo et al observed that ALC during induction therapy was a significant predictor of survival in both ALL and AML. In a subsequent study of 171 pediatric ALL cases, Rabin et al observed that ALC at Day 29 of induction therapy was independently associated with relapse-free survival and OS.
The level of minimal residual disease (MRD) during and after remission induction chemotherapy is the strongest prognostic factor in childhood ALL. MRD is an independent predictor of outcome and can define risk subgroups among clinically or genetically defined ALL subtypes. In the St. Jude Total XV study for children and adolescents with newly diagnosed ALL, we incorporated MRD measurements in risk-classification algorithms and used them to guide therapy, a strategy that resulted in 5-year EFS and OS rates of 85.6% and 93.5%, respectively. The study by Rabin et al indicated that ALC after remission induction therapy predicted outcome even after adjusting for MRD status, suggesting that these were independent prognostic factors. To determine whether ALC held its prognostic significance in the context of MRD-directed therapy, we examined the impact of ALC on outcome in our Total XV cohort.
MATERIALS AND METHODS
From June 2000 to October 2007, 411 patients aged ≤18 years with newly diagnosed ALL were enrolled in the Total Therapy XV study at St. Jude Children's Research Hospital. Treatment regimens, risk classification, and MRD methods have been previously described. ALC values were obtained at the time of remission evaluation, after 6 weeks of remission induction therapy. The study was approved by the institutional review board, and written informed consent or assent, as appropriate, was obtained for all patients.
ALC values were analyzed separately as a categorical variable (<500 cells/μL vs ≥500 cells/μL) and as a continuous variable. Comparisons of ALC, treated as a categorical variable, between groups were performed using the chi-square test. ALC treated as a continuous variable between groups was compared using the Wilcoxon rank-sum test when 2 groups were compared and using the Kruskal-Wallis test when >2 groups were compared simultaneously. The duration of EFS was measured from the date of enrollment to the date of the first treatment failure of any kind (relapse, death, lineage switch, or second malignancy) or to the date of the most recent follow-up. Failure to induce remission was considered an event at time zero. The duration of OS was calculated from the date of enrollment to the date of death from any cause or the date of the most recent follow-up. Data for patients who were alive at the most recent contact date were censored at the time of that contact. The Kaplan-Meier method was used to estimate the probability of OS and EFS, standard errors were determined using the method of Peto et al, and comparisons between groups were made using the log-rank test. The Cox proportional hazards regression model was used to assess each potential prognostic factor in univariate and multivariable analyses, and hazard ratios were estimated with 95% confidence intervals.
All analyses were performed using SAS software (SAS Institute, Cary, NC), Windows version 9.2. All reported P values are 2-sided. The criterion for significance in all analyses was a probability of ≤ .05 (P ≤ .05), and no P value adjustments were made for multiple comparisons.
Absolute Lymphocyte Count at the End of Remission Induction Therapy and Relation With Clinical Features of Acute Lymphoblastic Leukemia
The ALC at the end of remission induction ranged from 120 to 6942 cells/μL (median, 1024 cells/μL). Because the ALC was associated with OS when cutoff points of 500 cells/μL and 600 cells/μL were used but was marginally associated with EFS only when a cutoff point of 500 cells/μL was chosen, we used 500 cells/μL for subsequent analyses. Patients with an ALC ≥500 cells/μL at the end of induction, as indicated in Table 1, were more likely than patients with lower counts to have B-lineage ALL (P < .001), to have low-risk leukemia as defined by age and leukocyte count (P = .001), and to have MRD-negative status (<0.01% leukemic cells among bone marrow mononucleated cells) at the end of induction (P = .005). When ALC was analyzed as a continuous variable (data not shown), higher counts were associated with ages 1 to 9 years (P < .001), B-lineage ALL (P < .001), and low-risk features (P < .001). Thus, higher ALC was characteristic of patients who had favorable prognostic features and excellent initial treatment response.
aP values were calculated using the Pearson chi-square test.
Age at diagnosis, y
Leukocyte count, ×103/μL
Initial risk group
Low ALC at the end of induction therapy may reflect a delayed recovery of bone marrow function. Thus, we compared the leukocyte, platelet, and absolute neutrophil counts between patients with ALC values >500 or <500 cells/μL. Indeed, the median leukocyte count in patients who had an ALC <500 cells/μL was 1800 cells/μL (range, 400-3900 cells/μL) compared with 2900 cells/μL (range, 500-33,300 cells/μL) for patients with higher ALC (P = .003); platelet counts were 135 × 103/μL (range, 17-344 × 103/μL) versus 277 × 103/μL (range, 23-1220 × 103/μL), respectively (P < .001).
Relation Between Absolute Lymphocyte Count and Treatment Outcome
In a univariate analysis, ALC ≥500 cells/μL was significantly associated with a superior OS: the 5-year OS estimate was 95% ± 1.2% versus 82.4% ± 6.8% for patients who had ALC values <500 cells/μL (P = .023) (Fig. 1a); ALC was also significantly related to OS when analyzed as a continuous variable (P = .029). Among the other factors included in the analysis, age (P < .001), MRD status (P < .001), leukemic cell lineage (P = .029), and initial risk group (P = .001) were predictive of OS (Table 2). Age (P = .005), MRD status (P < .001), leukemic cell lineage (P = .002), initial risk group (P < .001), and leukocyte count at diagnosis (P = .040, when analyzed as a continuous variable) also were significantly associated with EFS (Table 2). However, ALC was not significantly associated with EFS: the 5-year EFS estimates for patients with ALC values >500 vs <500 cells/μL were 88% ± 1.8% versus 76.5% ± 7.6%, respectively (P = .053) (Fig. 1b); likewise, ALC lacked prognostic significance if analyzed as a continuous variable (P = .092).
Table 2. Univariate Analysis of Event-Free and Overall Survival
In a multivariable Cox proportional hazards regression model that included age, leukocyte count at diagnosis, leukemic cell lineage, initial risk group, and MRD status at the end of induction, ALC, whether analyzed as a categorical variable (P = .362) (Table 3) or a continuous variable (P = .234; data not shown), was not associated with OS. We also performed a multivariable analysis that included the absolute neutrophil count in the Cox model. However, inclusion of the absolute neutrophil count in the model did not change the results—ALC was not significantly associated with OS. Similarly, ALC was not significantly associated with EFS when analyzed as a categorical variable (P = .930) (Table 3) or a continuous variable (P = .506; data not shown). In the multivariable analyses, MRD retained a significant association with OS and EFS (P < .001); age was associated with OS (P = .005) and initial risk group with EFS (P = .003). To assess whether ALC may provide useful prognostic information in settings in which MRD measurement is not possible, the multivariable analyses were repeated without MRD in the model. In these analyses, ALC was not significantly associated with OS (P = .176) or EFS (P = .534).
Table 3. Multivariable Cox Proportional Regression Model With Absolute Lymphocyte Count As a Categorical Variable
Although the multivariable analyses provided no evidence that ALC was independently associated with outcome, we nevertheless performed subgroup analyses to determine whether ALC values may be prognostically useful for a subset of patients (Table 4). After adjusting for age, ALC was not significantly associated with OS or EFS in MRD-positive, low-risk, or standard/high-risk patients, but ALC was associated with OS among MRD-negative patients (Fig. 2; Table 4). The 5-year estimates of OS were 97.3% ± 1% for MRD-negative patients who had ALC values ≥500 cells/μL and 84.2% ± 8.9% for patients with lower ALC values (P = .036) (Fig. 2). Three hematologic relapses occurred in the low ALC, MRD-negative group, including 2 patients with T-lineage ALL and 1 patient with B-precursor ALL, all of whom had MRD levels >1% at day 15 of remission induction therapy.
Table 4. Outcome According to Absolute Lymphocyte Count, Risk Group, and Minimal Residual Disease
We assessed the prognostic impact of ALC at the end of remission induction in the context of contemporary MRD-directed therapy. We observed that a higher ALC (≥500 cells/μL) was significantly more prevalent among patients with B-lineage ALL than among those with T-lineage ALL, in those with favorable presenting features, and in those who had a good response to remission induction therapy and achieved negative MRD status on day 43. Patients who had higher ALC at the end of induction also had higher leukocyte and platelet counts at the same time point. Both OS and EFS were superior among patients with higher ALC, but only the association with OS was statistically significant in a univariate analysis. ALC was not significantly associated with EFS or OS after adjusting for other risk factors, including age, leukocyte count, lineage, risk group, and MRD status.
A recent COG study indicated that end-of-induction ALC is a significant predictor of outcome in children with ALL even after accounting for MRD. Several factors may explain the differences between the COG report and our findings. First, remission induction therapy on the COG trials consisted of 3 or 4 drugs given over a 4-week period, whereas the Total Therapy XV trial included 8 drugs in a 6-week induction period. Thus, ALC and MRD were evaluated at day 29 in the COG study and at approximately day 43 after more intensive therapy in our study. This difference in timing and intensity of therapy was reflected in higher ALC values in the COG study (median, 2280 cells/μL; range, 192-9890 cells/μL) than in our study (median, 1024 cells/μL; range, 120-6942 cells/μL). The most important difference, however, was the outcome for MRD-positive patients in the 2 studies. In the COG report, relapse-free survival and OS rates for patients with low ALC and positive MRD status were 33% and 41%, respectively, compared with EFS and OS rates of 69% and 77%, respectively for the corresponding group in our study. In the COG study, the ALC identified MRD-positive patients who had dismal or very good OS rates (41% vs 92% for those with low or high ALC, respectively), whereas the ALC was not associated with survival among MRD-positive patients in our study (77% vs 84%).
Our results suggest that the modification of therapy based on MRD, pharmacogenetics, and pharmacodynamics, along with careful supportive care, can abrogate the prognostic effect of both traditional and novel factors. Although age was associated with OS and risk group was predictive of EFS, MRD was the only factor that was independently associated with both OS and EFS. By contrast, multivariable analysis demonstrated that leukemic cell lineage, leukocyte count, and ALC were not independently associated with OS or EFS. This suggests that any negative effect potentially associated with low ALC, particularly among MRD-positive patients, was overcome by intensification of chemotherapy. It is noteworthy that MRD-negative patients who had ALC values <500 cells/μL in our study had significantly worse outcomes than those who had higher counts within this group, a finding that requires further investigation.
It has been proposed that the favorable impact of a higher ALC after chemotherapy or HSCT for hematologic malignancies reflects enhanced immunologic effects of autologous or allogeneic lymphocytes.[2, 6, 8, 9] Antileukemic effects have been ascribed to T-cell subsets[18, 19] and natural killer (NK) cells.[2, 9] It is clear that, among patients undergoing HSCT, allogeneic NK cells play a key role, because donor NK cells may exert potent antileukemic effects when the patient's leukemia cells lack the class I epitope for the donor's inhibitory killer immunoglobulin-like receptors. In addition, the normalization of NK cell receptor expression may be prognostically important in patients receiving chemotherapy. A study of adult patients with AML revealed that the acquisition of an NK cell-triggering receptor (NCR)-dull phenotype was reversible in patients who achieved remission. An association between the NCR-dull phenotype and poor survival in these patients suggests that recovery of normal NCRs may allow the recovery of normal NK cell function, which, in turn, contributes to sustained remission. Although this and other studies suggest an important role for immune surveillance after chemotherapy,[2, 5, 6] the results of the current study indicate that intensive chemotherapy may compensate for the lack of such immune effects.
This work was supported in part by Cancer Center Support (CORE) grant P30 CA021765-30 from the National Institutes of Health and by the American Lebanese Syrian Associated Charities (ALSAC).