Background: Retrospective studies suggest that change in ctDNA levels after 1-2 cycles of induction therapy in patients (pts) with previously untreated DLBCL is associated with event-free survival and overall survival (OS), using a log-fold change (LFC) cutoff in ctDNA levels of 2.0 (Kurtz DM, et al. J Clin Oncol 2018). The prognostic value of ctDNA has not been prospectively validated; this is a prespecified exploratory analysis of the prognostic value of ctDNA in the POLARIX study (NCT03274492). Methods: Pts in POLARIX had previously untreated DLBCL and were randomized to receive Pola-R-CHP or R-CHOP (Tilly H, et al. N Engl J Med 2022). Plasma ctDNA levels were measured at baseline and Cycle 2 Day 1 (C2D1) using the AVENIO NHL CAPP-Seq assay and are reported as mean mutant molecules per mL (MMPM; Herrera AF, et al. Blood Adv 2022). Plasma-depleted whole blood at baseline was used as a source of germline DNA to filter non-tumor-specific variants. Change in ctDNA levels was characterized as the log10 ratio between baseline and C2D1 MMPM (LFC); a LFC cutoff of 2.0 was used in this analysis. ctDNA clearance was determined with a detection cutoff of p=0.005 (Scherer F, et al. Sci Transl Med 2016). Univariate and multivariate Cox regression were used to identify relationships between ctDNA and progression-free survival (PFS) and OS. Hazard ratios (HR) are reported with 95% confidence intervals. Results: Results for paired ctDNA samples at baseline and C2D1 were available for 319/440 (72.5%; Pola-R-CHP) and 299/439 (68.1%; R-CHOP) pts. A median 135 (range: 1-644) variants were detected per pt. Baseline ctDNA levels were not statistically different between the two arms (median MMPM: Pola-R-CHP, 356; R-CHOP, 237; p=0.2); high baseline MMPM was associated with high baseline IPI score (p<0.001), ABC DLBCL (p<0.001), ECOG performance status >1 (p<0.001), and presence of bulky disease (>7.5cm; p<0.001). In both arms, pts with baseline ctDNA levels above the median MMPM had shorter PFS (Pola-R-CHP: HR, 1.96 [1.21-3.18]; R-CHOP: HR, 1.43 [0.92-2.20]) and OS (Pola-R-CHP: HR, 2.17 [1.07-4.37]; R-CHOP: HR, 2.10 [1.05-4.21]) than pts with baseline ctDNA levels below the median MMPM. Change in ctDNA levels from baseline to C2 as a continuous fold-change value was associated with PFS (HR, 0.75 [0.65-0.87]) and OS (HR, 0.78 [0.63-0.96]) in both arms. Pts with greater reductions in ctDNA levels after one treatment cycle had longer PFS and OS than pts with lower ctDNA reductions after one cycle. LFC ≥2.0 or undetectable ctDNA at C2D1 were achieved by 66.8% (213/319) of Pola-R-CHP-treated pts and 65.6% (196/299) of R-CHOP-treated pts. Stratifying pts at this LFC threshold was strongly prognostic of survival outcomes by arm (LFC <2.0 vs LFC ≥2.0, PFS HR: Pola-R-CHP 1.86 [1.18-2.92], R-CHOP 2.21 [1.43-3.41], OS HR: Pola-R-CHP 1.48 [0.77-2.81], R-CHOP 2.67 [1.34-5.30]). To define an optimal risk cutoff for pts receiving Pola-R-CHP, pts were grouped into training and validation cohorts. A LFC threshold of 2.5 (LFC ≥2.5 Pola-R-CHP: 54.2% [173/319]; R-CHOP: 54.8% [164/299]) was better able to identify pts at risk of poorer outcomes (LFC <2.5 vs LFC ≥2.5, PFS HR 2.89 [1.78-4.69], 24-month estimates 65.7% [58.3-74.0] vs 87.0% [82.0-92.2]; OS HR 1.87 [0.98-3.58], 24-month estimates 86.2% [80.8-92.0] vs 91.8% [87.7-96.0]; Figure A). In the Pola-R-CHP arm, pts with ctDNA clearance at C2D1 had superior outcomes vs pts with detectable ctDNA at C2D1 (not cleared vs cleared, PFS HR 2.98 [1.53-5.80]; OS HR 2.74 [1.07-7.02]). LFC 2.5 and ctDNA clearance were evaluated in separate multivariate analyses; it was found that both LFC 2.5 and ctDNA clearance are prognostic for PFS and OS in Pola-R-CHP treated pts, independent of key baseline risk factors (Figure B). LFC 2.5 and ctDNA clearance were also prognostic in R-CHOP-treated pts. Lastly, the top five commonly mutated genes identified in baseline ctDNA were HIST1H variants, TP53, PIM1, MYD88, and BCL2; analysis of clonal change on- and post-treatment is ongoing. Conclusions: Based on this prespecified exploratory analysis from the POLARIX study, ctDNA analysis has prognostic value for pts with previously untreated DLBCL. Pts who did not achieve ≥2.5 LFC and/or did not have ctDNA clearance following one cycle of Pola-R-CHP had inferior outcomes than those who did. Early changes in ctDNA levels may be of use in risk-adapted trial designs to identify pts in need of alternative treatment. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal
Read full abstract