Time from Diagnosis to Initiation of Curative Chemotherapy Affects Outcomes in DLBCL - Data from a Large Asian Cohort

Diffuse large B-cell lymphoma (DLBCL) is a high-grade lymphoma that requires treatment. It remains unclear how promptly curative anthracycline-based immunochemotherapy should be started and the prognostic effects time from diagnosis to treatment (TDT) has. We retrospectively analyzed the impact of TDT on progression-free survival (PFS) and overall survival (OS) using patient characteristics available at diagnosis.

From the databases of two large lymphoma treatment centers in Singapore, we included 581 DLBCL patients sequentially diagnosed and treated with R-CHOP (or R-EPOCH) between 2002 and 2014. All transformed lymphomas and retroviral positive cases were excluded. Patients with TDT > 7 weeks were also excluded as only a few (N=17) in our dataset were, and in real world practice would be, treated beyond this time. The effect of TDT on PFS and OS was examined in Cox regression models which included other known prognostic variables of age, extranodal (EN) sites, lactate dehydrogenase (LDH), performance status, stage, international prognostic index (IPI) risk and cell-of-origin (COO) based on Hans' algorithm. TDT was treated as a continuous variable in per week time units. As well as in a binary normal vs. high model, LDH was analyzed as a continuous variable and then categorically following the NCCN-IPI concept of a normalized LDH to provide more clinical relevance.

The median age was 60 years (range, 15-88) and median follow up in surviving patients 47 months. The number of patients with TDT in week 1 = 97, week 2 = 199, week 3 = 126, week 4 = 81, week 5 = 41, week 6 = 28, week 7 = 9. The median TDT was 14 days (range, 4-49) and was significantly longer in patients with normal LDH level, stage 1-2 disease, EN ≤ 1 site, and low IPI. In univariate analysis, longer TDT was associated with poorer PFS and OS in the high LDH and stage 3-4 groups. Cox regression analysis for PFS showed that poorer performance state of ECOG ≥ 2 (P=0.029), stage 3-4 disease (P<0.001), increased LDH 1-2 fold (P=0.01) and ≥ 2-fold (P <0.001), and longer TDT (P=0.008; HR 1.144, 95% CI 1.036 - 1.263) were predictive. For OS, stage 3-4 disease (P=0.019), increase in LDH 1-2 fold (P=0.005) and ≥ 2-fold (P <0.001), and longer TDT (P=0.01, HR 1.154; 95% CI 1.035 - 1.287) were significantly associated. Considering the statistically significant interactions between LDH and disease stage and TDT, we performed Cox regression separately for patients with high (>normal) and normal LDH, and for advanced stage (3-4) and early stage (1-2) disease. We found that TDT remained a significant predictor of PFS and OS in patients with advanced stage but not early stage disease and in high LDH but not normal level LDH patients.

The analysis demonstrated that delays in starting curative immunochemotherapy have detrimental effects on PFS and OS in patients with high LDH and advanced stage disease. The benefit of starting treatment earlier in DLBCL patients with higher tumor bulk, while intuitive, has not been previously demonstrated. Our findings suggest that treatment delays, whenever possible, should be minimized.