Prognostic Impact of Pre-Treatment Cell-Free DNA Concentration in Newly Diagnosed Peripheral T-Cell Lymphomas

Peripheral T-cell lymphomas (PTCLs) are a heterogeneous group of non-Hodgkin lymphomas with poor prognosis. Although some risk factors associated with worse survival are known, there are no reliable predictive biomarkers of response. Considering the prognostic relevance of cell-free DNA (cfDNA) in B-cell lymphomas, in this work we aimed to explore its role in 75 patients (pts) enrolled in the PTCL13 prospective study (NCT02223208), including newly diagnosed Peripheral T-cell Lymphoma, Not Otherwise specified (PTCL-NOS, n=30), nodular T Follicular Helper Lymphoma (nTFHL, n=26) and Anaplastic Lymphoma Kinase negative Anaplastic Large Cell Lymphoma (ALK-ALCL, n=19).

Cell-free DNA was extracted using the Maxwell® RSC LV ccfDNA Kit (Promega) from plasma samples. Pre-treatment cfDNA concentration had a median value of 19.1 ng/mL of plasma (range: 2.27-1082 ng/mL) and was significantly increased in the presence of high-risk clinical parameters, including disease-related risk factors (increased lactate dehydrogenase levels, p<0.0001; presence of extranodal disease p=0.04), prognostic scores used in PTCLs (International Prognostic Index >2, p<0.0001; Prognostic Index for T-cell lymphoma >1, p=0.0006), and patient condition (poor performance status, p=0.0003), compared with cfDNA concentration in low-risk categories. On the other hand, no association was detected between age, Ann Arbor stage, bone marrow involvement and cfDNA levels. In addition, higher baseline cfDNA concentration was found in pts who progressed compared with those who achieved a response at the end of 6 chemotherapy cycles (p= 0.0303).

Importantly, pre-treatment cfDNA level had a negative impact on both progression-free survival (PFS, Univariate Cox linear model, p=0.0001) and overall survival (OS, Univariate Cox nonlinear model, p=0.0105). Multivariate analysis confirmed a significant influence of cfDNA on PFS [Multivariate Cox Model, HR 1.34, 95% Confidence Interval (CI) 1.14-1.58, p= 0.0004 and HR 1.28, 95% CI 1.11-1.48, p= 0.0006] greater than that of IPI and PIT scores, LDH, performance status, and extranodal disease.

Furthermore, to identify a specific pre-treatment cfDNA concentration associated with a worse prognosis, an analysis by receiver operating characteristic (ROC) curve was performed by dividing the values according to the death event. In doing so, a cut-off of 42.78 ng/mL was identified (Youden's J statistics) above which PFS and OS are significantly reduced (log-rank test p= 0.0177 and < 0.0001, Figure 1 and 2 respectively).

Among the 21 pts with cfDNA>42.78 ng/mL, 8 were PTCL-NOS, 4 nTFHL and 9 ALK- ALCL, with nTFHL being significantly less represented than ALK-ALCL (p=0.043). Multivariate analysis confirmed a significant effect on survival when cfDNA above or below 42.78 ng/mL were tested with IPI, LDH and ECOG (HR 2.76, 95% CI 1.24-6.16, p= 0.0128 for PFS; HR 9.45, 95% CI 3.10-28.81, p=0.0001 for OS). When tested together with PIT, ECOG and extranodal disease, both cfDNA and PIT score were associated with OS (HR 9.30, 95% CI 3.22-26.87, p<0.0001 and HR 0.34, 95% CI 0.12-0.94, p=0.037 respectively), while no significant correlations with PFS were found (cfDNA≥ vs <42.78 ng/mL, HR 1.83, 95% CI 0.86-3.91, p= 0.1178).

Moreover, sequencing of cfDNA and diagnostic biopsy-extracted DNA was performed in an exploratory cohort of 25 pts using a target custom panel of 60 genes. We identified lymphoma mutations in 18 plasma samples (number of mutations in pts with at least a mutation: median=2, range 1-5) and in 20 biopsy samples (median=2, range 1-6). Mutations identified in biopsy were confirmed in coupled plasma samples in 74.4% cases.

In conclusion, our data demonstrate that pre-treatment cfDNA concentration has a strong prognostic impact in PTCLs and represents a superior predictor of shorter survival, compared to canonical parameters. Despite the fact that sequencing is feasible as demonstrated in our preliminary analysis, and is crucial to identify molecular alterations, its applicability in routine clinical practice is limited. On the other hand, cfDNA concentration is an inexpensive biomarker that can quickly provide prognostic information in the real-world scenario. Its influence on survival in the context of a prospective study where PTCLs were homogeneously treated identifies cfDNA as an easily assessable biomarker that could become part of those to be measured in newly diagnosed PTCLs.