Clinical Implications of Ctdna-Based Minimal Residual Disease Detection in Newly Diagnosed Diffuse Large B-Cell Lymphoma

Background: Pretreatment circulating tumor DNA (Pre-ctDNA) has been reported as a reliable surrogate for molecular subtyping and prognostic predicting for diffuse large B cell lymphoma (DLBCL). The ctDNA-based minimal residual disease (MRD) status at the end of treatment (MRD^end) has been demonstrated as even more strong prognostic marker in both clinical trial (Leo Meriranta, Blood) and the real-world (Brian J Sworder, ASH 2023) population for DLBCL patients (pts). However, no data has been shown (i) which clinical and biological risk factors can predict MRD^end status, (ii) which baseline gene alterations can not be cleared, or new gene alterations can be obtained leading to the disease progression or relapse, and (iii) why ctDNA can not be cleared in a subset of population even with the same genetic subtype.

Methods: We enrolled 164 DLBCL pts from the First Affiliated Hospital of Nanjing Medical University undergoing first line (1L) therapy for ctDNA profiling at 2 pre-defined milestones ( matched baseline and end of treatment (EOT) plasma samples) using a 475 gene lymphoma-specific sequencing panel. By the last visit in July 2024, the median follow-up duration was 24.4 (range, 15.3−44.6) months. All pts received R-CHOP-like regimens.

Results: Among the 164 pts, the median age was 57 years, 57.3% had stage III−IV disease and 42.1% had international prognostic index (IPI) scores 3−5. Among the 164 pts, 46 pts (28.1%) were defined as MRD^end positivity (MRD^end+, 23 pts were complete response (CR), 3 pts were partial response (PR) and 20 pts were progressive disease (PD) / stable disease (SD) by EOT PET) with extremely poor outcomes both in progression free survival (PFS) and overall survival (OS) (P<0.001). The other 118 pts were defined as MRD^end negativity (MRD^end-,106 pts were CR, 7 were PR and 5 pts were PD/SD by EOT PET). The MRD^end- ratio in different genetype was 78.4% (29/37) for MCD genetype, 77.8% (7/9) for EZB genetype, 75.0% (21/28) for BN2 genetype, 73.3% (11/15) for ST2 genetype, 68.8% (33/48) for other genetype and 60.9% (14/23) for TP53 disruption.

Five risk factors including high pre-ctDNA burden, TP53^mut, P2RY8^mut, TET2^mut and CD58^wt were identified as independent predictors for MRD^end+ by multivariable risk analysis. Patients can be divided into 3 risk group (score of 0−1; 2−3; 4−5) according to these 5 risk factors with the ratio of MRD^end+ was 10.8%, 43.6% and 100%, respectively.

The EOT top gene alterations (GAs) for MRD^end+ pts were TP53^mut (34.8%), BCL-6^mut (17.4%), PIM1^mut (15.2%), IRF4^mut (15.2%) and KMT2D^mut (15.2%) which were significantly different from baseline plasma mutation profiling. Firstly, we found that most EOT GAs for the MRD^end+ pts were the predominant gene at baseline. Secondly, we found that TP53^mut was the first GA which can not be cleared by the 1L induction therapy. Among the 38 pts of TP53^mut, 18 pts were MRD^end+ (47.4%); among these 18 MRD^end+ pts, TP53^mut was not cleared in 16 pts (88.9%). Thirdly, among the 46 MRD^end+ pts, the EOT ctDNA burden for PET-CR group was the lowest compared with PET-PR and PET-PD/SD groups. Furthermore, we found that the ratio of subsequent PD was extremely high in MRD^end+ pts among the pts who achieved PR or CR by EOT PET scan. Lastly, we found that the EOT ctDNA burden was the key factor to predict weather pts will experience further PD for MRD^end+ pts.

As regard to the divergent transcriptional signatures between MRD^end+ and MRD^end- pts for each genetype, our findings showed distinct gene expression patterns associated with EOT MRD status. Utilizing the Gene Set Enrichment Analysis, we identified that cell cycle, neutrophil extracellular trap formation and oxidative phosphorylation were particularly active in MRD^end+ TP53 disruption and other genetype cases. In contrast, MRD^end+ MCD genetype cases showed downregulated activity in IL-17 signaling pathway and Th17 cell differentiation, whereas MRD^end+ BN2 genetype cases were linked to disrupted Th1 and Th2 cell differentiation and T cell receptor signaling pathway.

Conclusions: These data demonstrate that: (i) EOT MRD status is a strong prognostic predictor for DLBCL after 1L therapy; (ii) High pre-ctDNA burden were the independent predictor for MRD^end+ status; (iii)TP53^mut was the first GA which can not be cleared by the 1L induction therapy; (iv) Divergent transcriptional signatures might be the underline mechanism why a subset of pts cannot achieve MRD clearance among the same genetic subtype.