Abstract
Background: Peripheral T-cell lymphomas (PTCL) display heterogeneous responses to immune checkpoint inhibitors, including rare durable responses. The basis for such variability remains unclear. Previously, we identified elevated expression of immune checkpoint genes (e.g., CD274 [PD-L1]) in ~50% of PTCL-NOS, associated with an M2 macrophage-like gene expression signature (Sugio et al., Blood Adv 2018). We hypothesized that a subset of relapsed/refractory (r/r) PTCL patients may benefit from PD-1 blockade, with molecular profiles helping to identify them using tumor and liquid biopsies.
Methods: We conducted a Phase II investigator-initiated trial of Nivolumab monotherapy (240mg every 2 weeks) in r/r PTCL patients who failed ≥2 prior therapies (WJHS-NHL02, UMIN000034499). The primary endpoint was overall response rate (ORR); secondary endpoints included CR rate, PFS, and OS. Molecular profiling included tumor RNA-seq, whole exome sequencing, and plasma cfDNA analyses (CAPP-Seq for SNVs/indels [Newman et al., Nat Med 2016], CANARy for CNAs [Chabon et al., Nature 2020], EPIC-Seq for inferred expression [Esfahani et al., NBT 2022], SABER/QUARTZ for TCR analysis [Shukla et al., ASH2022, Sugio et al., ASH2024]). We also performed Imaging Mass Cytometry (IMC) to characterize spatial proteomic profiles of one patient with an exceptionally long response using Hyperion.
Results: Of 20 enrolled patients, 19 received Nivolumab. Median age was 68.5 years; 21% were female. Histologies included PTCL-NOS (n=4), ENKTL (n=6), AITL (n=4), ALCL (n=5; 20% ALK+), and EATL (n=1). Grade 3/4 SAEs occurred in 9 patients, including infectious disease (21%). One patient developed breast cancer and discontinued treatment despite long-term lymphoma response.
The ORR was 10.3% (2/19: 1 PTCL-NOS, 1 AITL) with no CRs, and one patient had SD. Three patients were not evaluable for response due to insufficient imaging follow-up. Three patients achieving PR/SD had durable responses >2 years. All of 4 evaluable ENKTL cases had PD. The median OS and PFS were 5.7 and 2.0 months, respectively. Responders showed enrichment for specific molecular features, including PD-L1/2 mutations, higher total mutation burden (TMB), and distinct T-cell receptor (TCR) dynamics. One PTCL-NOS patient with early PR had a CD274 intron 5 mutation and immune-inflammatory pathway alterations; IMC revealed a focal infiltration of PD-L1+ macrophage and CD8+ T cells with a well-demarcated boundary. A second patient with EATL experiencing >2 years of durable SD had mutations in CD274 3'UTR, PDCD2LG1 (PD-L2), and PTPRD. Another patient achieving PR had typical AITL-associated mutations.
Responders also had significantly greater diversity of TCR in cfDNA (cfTCR). Specifically, among 13 non-ENKTL patients with dominant tumor TCR clonotypes in tumor tissues, 4 lacked detectable monoclonal expansions of these clonotypes in cfTCR, of whom 3 (75%) achieved >2 years PFS. The patients experiencing Nivolumab-induced early ctDNA reductions had higher levels of diversity in cfTCRs, which were non-tumor derived.
All patients with >2 years of PFS showed either complete clearance of ctDNA MRD or a >90% reduction. One patient who discontinued Nivolumab due to breast cancer maintained her long-term PR with sustained absence of ctDNA MRD. Among the 3 patients who were unevaluable radiographically, two showed notable ctDNA reductions (85% and 25%) at the time of discontinuation, whereas all other PD cases showed increased ctDNA levels. ctDNA responders had significantly higher TMB (p = 0.02) and ctDNA levels (p = 0.04) at baseline.
RNA-seq of diagnostic tumor tissue showed higher expression of macrophage-related genes in PR patients. EPIC-Seq of pre-treatment cfDNA revealed elevated expression of macrophage, B-cell, and dendritic cell-related genes, as well as PD-L1. These findings suggest enhanced turnover of both tumor and non–tumor immune cells in patients with PR.
Conclusions: Although the pre-specified efficacy threshold (ORR 30%) was not met, Nivolumab monotherapy induced durable disease controls in 3 (15.8%) patients with r/r PTCL, including 2 (10%) patients with PRs. Durable disease controls were associated with PD-L1/2 and PTPRD mutations, higher TMB, and increased cfTCR diversity. We also observed augmented turnover of both tumor cells and non-tumor immune cells in responders. cfDNA profiling may help identify candidates for PD-1 blockade in PTCL and should be integrated into future trials.
