Longitudinal Monitoring of Circulating Tumor DNA to Predict Survival in Newly Diagnosed Peripheral T-Cell Lymphoma Patients

Introduction Peripheral T-cell lymphomas (PTCLs) are a heterogeneous group of Non-Hodgkin lymphomas characterized by poor prognosis and represent 10% to 15% of all lymphoid malignancies. Given the complex biology and pathogenesis of PTCLs, there is a need to find new biomarkers that can possibly predict prognosis. In recent years, liquid biopsy has emerged as a novel molecular tool to genotype circulating tumor DNA (ctDNA), however, little is known about its role in PTCLs. In this study, we aimed at understanding the relevance of liquid biopsy in PTCL genomics and the potential role of ctDNA in predicting survival and monitoring minimal residual disease (MRD). Methods Formalin-fixed, paraffin-embedded (FFPE) tissue and plasma samples of 51 newly diagnosed patients with different PTCL histologies [nodal T-follicular helper lymphoma (nTFHL; n=18); Peripheral T-cell lymphoma not otherwise specified (PTCL-NOS; n=21); Anaplastic Lymphoma Kinase-negative Anaplastic large-cell lymphoma (ALKneg ALCL;n=12)], were analyzed. Patients were part of the PTCL13 Fondazione Italiana Linfomi prospective study aimed at assessing the role of romidepsin addition to anthracycline-based chemotherapy followed by stem cell transplantation (SCT) (NCT02223208, Chiappella et al, Leukemia 2023). The FFPE samples were collected at baseline, while based on the availability, the plasma samples were collected at baseline, cycle 3, cycle 6, SCT and relapse. We extracted DNA from FFPE tissue and cell-free DNA (cfDNA) from plasma to prepare libraries. The libraries were then sequenced using a targeted panel of 60 genes (Roche) to identify somatic mutations. Sequencing was performed on NextSeq 550 (Illumina). Variants were called and annotated with a bioinformatic pipeline. Results Somatic mutations were detected in 43 out of 51 (84.3%) FFPE samples and 38 out of 51 (74.5%) plasma samples at baseline. Liquid biopsies were able to identify tumor mutations with a sensitivity of 68.3% [95% CI: 59.69% - 75.74%]. TET2 was found to be the most frequently mutated gene (35%) in plasma, followed by IDH2 (20%), RHOA (18%), DNMT3A (16%) and TP53 (14%). Mutations in TET2 were found to be significantly associated with nTFHL histology (11/18, p=0.008) and TP53 mutations were found exclusively in PTCL-NOS patients (7/21, p=0.003). Interestingly, patients with TET2/IDH2 had a better progression free survival (PFS) probability as compared to patients with only TET2 mutations (p=0.015). Circulating tumor DNA levels, calculated as human Genome Equivalents (hGE)/mL, associated with high International Prognostic Index (IPI) score (>2, p<0.0001), high Prognostic Index for T-cell lymphoma (PIT) score (≥2, p=0.02) and increased lactate dehydrogenase (LDH) levels (greater than upper normal limit, p=0.0003). By receiver operating characteristic (ROC) curve, a cut-off value of ctDNA was identified at baseline (1650 hGE/mL) which was able to predict survival (PFS p=0.0022; Overall Survival (OS) p=0.0004). In patients with a complete clinical response, ctDNA levels significantly decreased after end-of-treatment (EOT) with respect to baseline (p=0.0004), however, for patients with a partial clinical response/progressive disease, no significant differences were observed between baseline and EOT ctDNA levels. Notably, irrespective of the clinical response status, the presence of detectable ctDNA (hGE/mL>0) at EOT was able to predict survival (PFS p=0.05; OS p=0.009). At the time point of relapse, the genetic landscape of PTCLs was characterized by the emergence of mutations in CHEK2, FYN and STAT5B genes. Conclusions Considering the high sensitivity of detection, liquid biopsy could be used as a reliable tool to identify somatic mutations in PTCLs. Plasma DNA genotyping unveiled the complex genomics of PTCLs, with the presence of TET2/IDH2 co-mutations in plasma conferring a better prognosis. Baseline ctDNA level was representative of disease burden, but more importantly, it was able to predict long term survival, indicating its potential role as a novel prognostic biomarker to tailor future treatments. Moreover, irrespective of clinical response status, the association of detectable ctDNA at EOT with survival suggests that it could also be used to monitor the presence of MRD. The change in the mutational profile of PTCLs at relapse likely suggests a genetic evolution during disease progression.