Response prediction and mutation profiling by circulating tumor DNA (ctDNA) of Relapsed/Refractory Diffuse Large B-cell Lymphoma (R/R DLBCL) patients receiving Glofitamab

Diffuse Large B-cell Lymphoma (DLBCL) represents the most common subtype of aggressive Non-Hodking Lymphoma (NHL). Bispecific antibodies (BsAbs) such as Glofitamab, have expanded therapeutic options for relapsed/refractory (R/R) DLBCL, including patients failing CAR-T. Despite high response rates, resistance and relapse remain major challenges, and current prognostic tools fail to accurately stratify patients and capture disease heterogeneity. In this context, circulating tumor DNA (ctDNA) analysis offers a minimally invasive approach for dynamic monitoring and genomic profiling. We analyzed ctDNA by CAPP-Seq in 48 R/R LBCL patients treated with Glofitamab, at baseline, cycle 3 (C3), end of treatment, or progression. Baseline ctDNA correlated with high IPI and bulky disease but was not predictive of outcome. In contrast, ctDNA at C3 strongly predicted response: all ctDNA+ patients progressed, whereas ctDNA- patients had a significantly improved survival (p<.0001). Early ctDNA outperformed interim PET in outcome prediction and represented a valid marker of minimal residual disease (MRD). Molecular profiling revealed frequent mutations in TP53 (46%), KMT2D (38%), and other canonical DLBCL genes. Surprisingly, alterations in the tumor suppressor TP53 were not associated with adverse prognosis. We hypothesize that a TP53–KMT2D co-mutation might be involved in mitigating the poor outcomes of this high-risk group. Mutations in CCND3 and BCL7A were associated with inferior survival, independently of IPI. We integrated genomic data and tested the performance of three DLBCL molecular classifiers: NMF, LymphGen and DLBClass. NMF represented the most suitable option for our targeted panel and exploratory approach, enabling the classification of each patient into a group. The molecular subtypes were associated with different outcomes: C4 and C5 were linked to dismal survival (mPFS 2.1 and 3.8 months), while C1 and C3 showed favorable prognosis. Incorporation of MRD with significant molecular features improved outcome prediction and showed MRD- patients in C1–C3 had superior outcomes compared with MRD- patients in high-risk C4– C5. This exploratory analysis highlights how ctDNA outperformed PET in early MRD detection during Glofitamab treatment and provided genomic insights into resistance. These findings underscore the potential of combining ctDNA profiling with current imaging and prognostic systems to refine response prediction and guide therapeutic decision-making in the era of bispecific antibodies.