Mutation Profiling, Efficacy Monitoring, and Prognosis Evaluation Using Circulating Tumor <scp>DNA</scp> in Peripheral <scp>T</scp> ‐Cell Lymphoma: A Prospective, Multicenter, Observational Study

Peripheral T-cell lymphoma (PTCL) represents a diverse and aggressive category of hematologic malignancies that significantly affects patient outcome [1]. PTCL is characterized by multiple subtypes, including NK/T-cell lymphoma (NKTCL), angioimmunoblastic T-cell lymphoma (AITL), anaplastic large cell lymphoma (ALCL), and PTCL, not otherwise specified (PTCL-NOS), and is associated with considerable morbidity and mortality, primarily due to poor response to conventional therapies [2-4]. Recent advancements in molecular diagnostics have highlighted circulating tumor DNA (ctDNA) as a valuable biomarker for various cancers, providing real-time insights into tumor dynamics and treatment responses [5-7]. Previous studies have established the role of ctDNA in assessing tumor burden, prognostication, and monitoring treatment efficacy across several hematological malignancies [8-10]. Moreover, identifying specific ctDNA alterations associated with treatment response may enhance prognostic evaluation and support therapeutic decision-making [11, 12]. Despite its promising potential, significant gaps remain regarding the clinical applicability of ctDNA in PTCL. In particular, comprehensive and prospective studies examining the association between ctDNA dynamics and treatment outcomes across different PTCL subtypes are lacking. To address these gaps, we conducted a prospective, multicenter, observational cohort design using next-generation sequencing (NGS) to analyze tumor DNA (tDNA) and plasma-derived ctDNA mutation profiles and to evaluate the prognostic significance of ctDNA dynamics in 104 patients with PTCL (Tables S1–S3). The study was approved by the Ethics Committee at the participating centers in accordance with the Declaration of Helsinki. Figure 1A,B illustrate the study design flowchart. Thirty-three patients provided baseline, post-four-cycle, and end-staging ctDNA samples, along with paired baseline tDNA samples for concordance, dynamic monitoring, and prognostic analyses. Thirty patients provided baseline and post-four-cycle ctDNA samples with paired baseline tDNA for concordance and dynamic monitoring analysis. Three patients provided baseline and end-staging ctDNA samples with paired baseline tDNA for concordance, dynamic monitoring analysis, and prognostic analyses. Thirty-eight patients contributed baseline tDNA and paired ctDNA samples for concordance analysis only (Figure 1B). Baseline tDNA detection identified 161 mutated genes, comprising 529 mutation events. The most frequently mutated genes were TET2 (51.0%), followed by RHOA (34.6%), DNMT3A (24.0%), and both STAT3 and TP53 (13.5%) (Figure 1C), similar to the spectrum of ctDNA mutations (Figure 1D). The percentage of cases detectable via ctDNA in each subtype of PTCL is shown in Table S4. Among the 398 concordant mutations detected in both tDNA and ctDNA (Figure 1E), variant allele frequency (VAF) measurements showed a high correlation (R = 0.530, p < 0.001) (Figure 1F). The median VAF was significantly lower in ctDNA than in tDNA (6.1% vs. 11.4%, p < 0.001, Figure 1G). We used tissue-based genetic testing as the gold standard and found that ctDNA sensitivity was 75.2% (398/529) (Figure 1H). Sensitivity varied based on pathological subtypes and was highest in patients with AITL & nodal T-follicular helper cell lymphoma (82.1%, Figure S1), followed by those with ALCL (75.6%, Figure S2), PTCL-NOS (74.2%, Figure S3), and NKTCL (37.1%, Figure S4). Subsequently, we analyzed the relationship between mutations detected in tDNA and ctDNA and clinical parameters. The number of mutations detected in tDNA and ctDNA correlated with patient age (p = 0.001, p = 0.002), lactate dehydrogenase (LDH) levels (p = 0.005, p < 0.001) and pathological subtypes (p = 0.001, p < 0.001) (Figure 2A,B), but not with Ann Arbor stages (p = 0.835, p = 0.245) (Figure S5A,B). Additonally, ctDNA concentration was associated with LDH levels and pathological subtypes (p < 0.001, p = 0.010), tended to be correlated with age (p = 0.105), and was not associated to Ann Arbor stages (p = 0.741) (Figures 2C and S5C,D), suggesting that ctDNA concentration may serve as a potential indicator of tumor burden. We analyzed dynamic changes in ctDNA VAFs and concentrations to monitor tumor burden. Patients were categorized into two groups: the response group, including those with complete response (CR) and partial response, and the non-response group, including those with stable and progressive disease. After four treatment cycles, 58 responding patients experienced a significant reduction in ctDNA VAFs and concentrations (median VAF, 3.5% vs. 0%, p < 0.001; median concentration, 2.0 log10 hGE/mL vs. 0 log10 hGE/mL, p < 0.001). Conversely, persistent VAFs and ctDNA concentrations were observed in the non-response group (median VAF, 1.1% vs. 1.4%, p = 0.666; median concentration, 3.2 log10 hGE/mL vs. 2.0 log10 hGE/mL, p = 0.059) (Figure 2D,E). At the end of treatment, post-treatment ctDNA VAFs and concentrations were lower in responding patients than in non-responders (median VAF, 3.3% vs. 0%, p < 0.001; median concentration, 2.1 log10 hGE/mL vs. 0 log10 hGE/mL, p < 0.001). Furthermore, non-responders exhibited persistent VAFs and concentrations (median VAF, 1.8% vs. 0.8%, p = 0.661; median concentration, 2.2 log10 hGE/mL vs. 2.5 log10 hGE/mL, p = 0.361) (Figure 2F,G). Following induction therapy, 28 patients exhibited a decrease in mean ctDNA VAF, and 8 patients showed no decrease. Notably, changes in mean ctDNA VAF (calculated as the post-treatment mean VAF compared to pre-treatment mean VAF) were not linked to progression-free survival (PFS) (p = 0.942) (Figure 3A). Nevertheless, patients with decreased mean ctDNA concentration experienced better PFS (p = 0.036) (Figure 3B). Additionally, both mean ctDNA VAF at end-staging ≥ 1% and ctDNA concentration at end-staging ≥ 2 log10 hGE/mL predicted worse PFS (p = 0.012, p = 0.009, respectively) (Figure 3C,D). However, no significant prognostic impact of baseline ctDNA status on PFS was observed when considering mean ctDNA VAF (p = 0.241) (Figure S6A) or mean ctDNA concentration (p = 0.106) (Figure S6B), indicating that pre-treatment ctDNA status is not a reliable biomarker for prognostic stratification. To assess the prognostic differences between mean ctDNA VAF and mean ctDNA concentration change from post- to pre-treatment, we compared the discrimination ability of the prognostic model using the concordance index (C-index). The model based on ctDNA concentration at end-staging showed the best discrimination for PFS with a C-index of 0.72 (95% confidence interval [CI]: 0.64–0.80), outperforming models based on mean VAF at end-staging, ctDNA concentration change, and mean VAF change (Figure 3E). The area under the curve (AUC) for predicting 1-year PFS was higher for ctDNA concentration at end-staging and ctDNA concentration change than for mean ctDNA VAF change and mean VAF at end-staging (AUC: 0.77, 0.77, 0.67, and 0.73, respectively), indicating greater accuracy (Figure 3F). Notably, we observed a high sensitivity of 75.2% for concordant mutations between tDNA and ctDNA, especially in patients with AITL, reaching 82.1%. These results align with those of Kim et al. in their PTCL study [13], and findings from other hematological malignancies [10, 14], indicating that plasma ctDNA can effectively reflect the variations detected at the tissue level. We observed heterogeneity in ctDNA detectability across PTCL subtypes, and hypothesized that these differences may be attributed to several factors, including tumor biology, proliferation rate, disease localization, and number of patients. Consistent with previous research [12], our findings revealed significant associations between ctDNA levels and clinical parameters, including age and LDH, highlighting the clinical relevance of ctDNA in evaluating disease burden for patients with PTCL. We found that patients with AITL harbored mutations in RHOA, TET2, DNMT3A, and IDH2 detectable via ctDNA analysis for AITL diagnosis, which aligns with previous research results [15-17]. Previous research highlights the utility of ctDNA in various malignancies as a non-invasive tool for tracking tumor dynamics and treatment efficacy [18, 19]. Our prospective, multicenter, observational study of Chinese patients with PTCL provides the first evidence of the utility of ctDNA in tracking tumor dynamics and treatment efficacy, filling a critical knowledge gap in this aggressive disease category. We found that changes in pre- and post-treatment ctDNA were closely related to the therapeutic efficacy of the patients, indicating that patients with rapid clearance of ctDNA achieved higher CR rates, which is consistent with the findings of Qi et al. [20] Notably, elevated ctDNA concentration, higher mean VAF, or higher ctDNA concentration at end-staging were associated with a poor prognosis. When analyzing ctDNA concentration, NGS is more cost-effective than traditional polymerase chain reaction because it can detect multiple variants simultaneously. Our findings are significant for clinical practice and highlight the benefits of ctDNA in monitoring treatment response and prognosis. Furthermore, standardized protocols and collaboration between clinicians and laboratory specialists are needed to facilitate the implementation of ctDNA monitoring in clinical settings. This study has certain limitations. First, the relatively small sample size limits generalizability of our findings across the broader PTCL patient population, such as patients with NKTCL, and the reliability of prognostic sensitivity requires further validation. Second, the lack of long-term follow-up data limits our understanding of the prognostic value of ctDNA in predicting long-term outcomes. Third, differences in treatment regimens across various PTCL subtypes may affect the performance of ctDNA in predicting efficacy and prognosis. Future research should include larger multicenter cohorts by enrolling patients receiving uniform treatment protocols with long-term follow-up to validate these findings and strengthen statistical power. In conclusion, this study emphasizes the critical role of ctDNA in predicting treatment response and prognosis among patients with PTCL, suggesting that ctDNA analysis may serve as a valuable tool in clinical practice to improve patient management. M.J., Z.J., and J.J. designed the study and approved the final manuscript. Z.D., S.Y., T.H., D.K., C.Q., H.H., W.L., J.H., Z.L., Q.W., and B.O. collected the clinical sample and data. C.S. and Q.J. performed the NGS platform. Z.D., S.Y., and Q.J. did the statistical analysis, wrote, and edited the manuscript. All authors contributed to the article and approved the submitted version. The authors declare that no generative AI tools were used in any stage of the research reported in this paper. The authors declare no conflicts of interest. The data used in the present study are available from the corresponding author on reasonable request. Table S1: Centers recruited. Table S2: Baseline characteristics of 104 patients with PTCL. Table S3: Targeted 521-gene sequencing panel. Table S4: The percentage of cases detectable via ctDNA in each subtype of PTCL. Figure S1: Heatmap showing mutation details for enrolled patients with AITL and nTFHL based on tDNA (A) and ctDNA (B). AITL, angioimmunoblastic T-cell lymphoma; nodal T-follicular helper cell lymphoma (nTFHL); tDNA, tumor DNA; ctDNA, circulating tumor DNA. Figure S2: Heatmap showing mutation details for enrolled patients with ALCL based on tDNA (A) and ctDNA (B). ALCL, anaplastic large cell lymphoma; tDNA, tumor DNA; ctDNA, circulating tumor DNA. Figure S3: Heatmap showing mutation details for enrolled patients with PTCL-NOS based on tDNA (A) and ctDNA (B). PTCL-NOS, peripheral T-cell lymphoma, not otherwise specified; tDNA, tumor DNA; ctDNA, circulating tumor DNA. Figure S4: Heatmap showing the mutation details for enrolled patients with NKTCL based on tDNA (A) and ctDNA (B). NKTCL, NK/T-cell lymphoma; tDNA, tumor DNA; ctDNA, circulating tumor DNA. Figure S5: Correlation analysis between mutation characteristics, Ann Arbor stage, and age. Mutation characteristics include the number of mutations in tDNA (A), the number of mutations in ctDNA (B), and ctDNA concentration (C, D). tDNA, tumor DNA; ctDNA, circulating tumor DNA. Figure S6: Progression-free survival analysis based on ctDNA assessments. Patients were stratified according to mean VAF at baseline (A) and ctDNA concentration at baseline (B). ctDNA, circulating tumor DNA; VAF, variant allele frequency; PFS, progression-free survival. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.