Despite evolution of therapeutic strategies, there is no universal standard of care for relapsed or refractory (RR) mature T and NK-cell neoplasms (TNKL). Most patients receive multiple lines of therapy and cycle through many available options. 1-3 There is no data to inform optimal therapy sequence, however emerging data suggest that exposure to epigenetic modifiers (EM) can sensitize tumors to other therapies. 4-6 Here we report results of comparative analyses assessing survival outcomes based on therapy sequence using a global RR TNKL patient cohort with data from 15 centers across 6 continents. We conducted a retrospective target-trial 7,8 cohort study using an updated global patient cohort. 9 Patients were excluded from analyses if they had anaplastic large cell lymphoma (as EM is infrequently used 10,11), did not receive first line cytotoxic chemotherapy (CC), or no documented second line (2L) start date. Cohort assignment was based on 2L therapy received: EM (e.g. histone deacetylase or DNA methyltransferase inhibitors), small molecule inhibitors (SI; broad or selective), or CC. Antibody-drug conjugates were included in CC. Outcomes were overall survival (OS; time from 2L start to death) and real-world progression-free-survival-2 (rwPFS2; time from 2L start to fourth line start or death 12 to assess if 2L modifies the effect of third line [3L]). Planned subgroups included histologic subtype and those who received allogeneic hematopoietic transplant (HSCT) after 2L. For sequence analyses, included patients must have received 3L. Outcomes were assessed between all prespecified 2L to 3L sequences using a single comparator, and pairwise. Kaplan Meier curves were used, and adjusted hazard ratios (aHR) were estimated using Cox regression with previously described a priori covariates. 9 Maximal, minimal, and no covariate Cox models were compared using likelihood ratio tests and Akaike's information criterion. Of the 925 RR patients in the global cohort, 472 were included who received EM (n=89), SI (n=46), or CC (n=337) at 2L. The most common SI included immunomodulatory imide drug (26.1%), PI3K inhibitor (21.7%), investigational pathway inhibitor (19.6%) and JAK inhibitor (17.4%). Within EM, SI, and CC, 42.7%, 37%, and 47.2% of patients were primary refractory to first line, respectively. The most common histologies were PTCL-NOS (51.7%, 41.3%, and 57.3%) and AITL (43.8%, 54.3%, and 26.4%) in EM, SI, and CC, respectively. Most patients had a Prognostic Index for T-cell Lymphoma (PIT) score at diagnosis of 1 (27%, 34.8%, and 27%) or 2 (28.1%, 26.1%, and 25.8%) in EM, SI, and CC, respectively. In EM, 34.8% received first-line autologous HSCT, versus 28.2% in SI, and 15.4% in CC. Following 2L, 12.4%, 17.4%, and 9.8% received allogeneic HSCT in EM, SI, and CC, respectively. The Cox model containing histology, PIT, and primary refractory status was used as the final model (vs maximal model: p=0.88). Compared to CC at 2L, EM did not affect OS (aHR 0.88, 95% CI 0.63-1.24; p=0.46) or rwPFS2 (aHR 0.84, 95% CI 0.61-1.16; p=0.28), however SI improved OS (aHR 0.61, 95% CI 0.37-0.97; p=0.038) and rwPFS2 (aHR 0.61, 95% CI 0.38-0.90; p=0.038). In the allogeneic HSCT subgroup, there was no difference in OS (EM vs CC: aHR 0.32, 95% CI 0.04-2.40; p=0.27; SI vs CC: aHR 0.51, 95% CI 0.06-4.69; p=0.55) or rwPFS2 (EM vs CC: aHR 0.30, 95% CI 0.05-1.96; p=0.21; SI vs CC: aHR 0.64, 95% CI 0.05-3.90; p=0.46). There were no differences in PTCL-NOS patients in OS (EM vs CC aHR 0.80, 95% CI 0.51-1.25; p=0.33; SI vs CC: aHR 1.00, 95% CI 0.54-1.84; p=0.99) or rwPFS2 (EM vs CC: aHR 0.91, 95% CI 0.61-1.38; p=0.66; SI vs CC: aHR 0.90, 95% CI 0.49-0.97; p=0.16). In AITL, SI improved OS (aHR 0.34, 95% CI 0.16-0.76; p=0.009) and rwPFS2 (aHR 0.36, 95% CI 0.17-0.77; p=0.008) versus CC, but EM showed no difference in OS (aHR 0.80, 95% CI 0.43-1.46; p=0.46) or rwPFS2 (aHR 0.56, 95% CI 0.31-1.02; p=0.059). There was no difference in rwPFS2 or OS across sequences overall (Figure 1) or between pairwise sequence comparisons (Figure 2). Compared to CC, SI at 2L improved OS and rwPFS2 in AITL patients. Therapy sequence did not affect OS or rwPFS2. Advanced approaches such as machine learning and dynamic treatment regimes have been initiated to fully elucidate the effect of therapy sequence. Our study highlights several equally effective therapy sequences to treat RR TNKL allowing therapy to be individualized based on patient and disease characteristics, and drug access at a given time.
Search from 250 M+ research papersSearch from 250 M+ research papers
Open Access
Just Published
Apr 29, 2024
Paul Sackstein + 5
