Time to Minimal Residual Disease (MRD) Negativity Is Independently Predictive of Outcome in Adults with Acute Lymphoblastic Leukemia (ALL) Receiving Hyper-CVAD

Background: MRD is an established prognostic/predictive factor in ALL.  Achieving MRD negativity (MRD^Neg) early during treatment is associated with superior outcomes with pediatric regimens.  However, little is known about how to use MRD assessments with hyperCVAD (hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone, alternating with methotrexate and cytarabine), one of the most commonly used regimens in adults with ALL. MRD^Neg at 3 months has been shown to be predictive of outcome with hyperCVAD + tyrosine kinase inhibitor (TKI) in Philadelphia chromosome (Ph)+ ALL (Ravandi, Blood, 2013, p. 1214), but this is relatively late in the treatment course.  We hypothesized that achieving MRD^Neg early during treatment would be associated with better outcomes with hyperCVAD.

Methods:  We performed a retrospective analysis of our center's experience since 2005 under an IRB-approved protocol.  We included pts with ALL (excluding Burkitt and mixed-phenotype) age >18 years (yrs) who received hyperCVAD as initial therapy.  MRD was assessed primarily by either multiparameter flow cytometry or BCR-ABL quantitative PCR on bone marrow, though other techniques (e.g., cytogenetics [CG]) were considered if obtained.  Timing and nature of assessments were left to treating physicians.  Pts were not defined as MRD^Neg until all assays performed were unable to detect any disease.  Clinical risk at diagnosis was defined by age (> 35 yrs), white blood cells (WBC; > 30 for B-ALL, > 100 for T-ALL), and CG (Ph+, MLL rearranged, -7, +8, complex, low hypodiploid, and near triploid).  Events included morphologic or MRD recurrence, change in treatment due to inadequate response, death from any cause, or secondary malignancy.  Frequencies of characteristics between groups were compared using a Fisher exact test.  Cox proportional hazards (PH) models were used to investigate associations between variables.  A test of PH was used to assess the impact of time on the association between MRD and outcome, in which MRD was modeled as a time-dependent covariate with left-truncation.  Clinical follow-up was updated as of June 2015.

Results: We identified 142 pts for this analysis: 18% T-ALL, 73% > 35 yrs, 24% had high WBC (6 unknown), and 48% had high-risk CG, 71% of which (34% of total) were Ph+ (11 unknown).  Sixty-five pts (46%) underwent hematopoietic cell transplantation (HCT) in first remission (CR1): 32% with reduced-intensity and 68% with myeloablative conditioning.  All Ph+ pts received TKI with hyperCVAD: 23 (48%) received imatinib and 25 (52%) received dasatinib.

Median time of 1^st MRD assessment (relative to start of treatment) was 37 days, with 27% occurring by 21 days and 85% by 90 days; 42% were MRD^Neg at 1st assessment, 26% became MRD^Neg later, and 32% did not become MRD^Neg during treatment with hyperCVAD.  Incidences of age > 35 yrs (P = 1), high-risk CG (P = 0.08), Ph+ (P = 0.72), and high WBC (P = 0.38) were not significantly different in MRD^Neg pts.  HCT in CR1 was more common in MRD^Neg pts (P = 0.05).  In Cox PH models adjusted for HCT in CR1, CG, and WBC, MRD^Neg pts had significantly better overall survival (OS; hazard ratio [HR] 0.43, 95% confidence interval [CI] 0.23-0.81; P = 0.01; 48 events) and event-free survival (EFS; HR 0.27, 95% CI 0.16-0.46; P < 0.01; 84 events). 

Through an exploratory landmark analysis, the benefit of MRD^Neg on EFS became more pronounced after 8 weeks: pts who were MRD^Neg by 8 weeks had significantly better EFS (HR 0.53, 95% CI 0.32-0.89; P = 0.02) than those who were not.  Further, achieving MRD^Neg earlier during treatment was associated with better EFS. In a test of PH on the time-dependent Cox PH model, time to MRD^Neg had a significant impact on the association between MRD status and EFS (P = 0.02) but not OS (P = 0.19).  This is depicted in Fig 1, where being MRD^Neg earlier after treatment initiation is associated with a smaller beta (i.e., significantly less hazard for an event).   

Conclusions: MRD^Neg is an independent predictor of superior OS and EFS in adults receiving hyperCVAD for ALL.  Further, achieving MRD^Neg earlier during treatment was associated with better EFS.  When considering the time-dependent nature of MRD status relative to EFS, 8 weeks after initiation of hyperCVAD may represent an important prognostic time point.  If confirmed in an independent dataset, this may prove to be a useful surrogate in both routine clinical practice and future clinical trials with this regimen.