Exploring improved prognostication using PET/CT radiomics in relapsed/refractory, TP53-aberrant diffuse large B-cell lymphoma (DLBCL).

Abstract

7070 Background: Patients (pts) with TP53-aberrant DLBCL have inferior progression-free survival (PFS) and overall survival (OS) compared to pts with wild type disease, and outcomes are worse for those with R/R disease. Improved prognostics are needed to identify those at risk for inferior outcomes to guide enhanced surveillance and early initiation of subsequent therapy. Methods: Retrospective data from pts with R/R, TP53-aberrant DLBCL treated at the University of Colorado system with 2 distinct PET/CT scans at time of R/R diagnosis and post-treatment between 2016 and 2025 were included in this analysis. PET/CT measurements were conducted by two nuclear medicine physicians blinded to outcomes and included SUVmax, total metabolic tumor volume (TMTV) by 41% SUVmax method, and total lesional glycolysis (TLG). Receiver operator curves (ROC) were constructed to determine the optimal cutoff points for PFS at 1 year using Youden’s J statistic. Survival analysis was conducted using Kaplan-Meier and Cox analyses. Results: The 22 pts had a median age of 66.9 years (44-83), median follow-up of 18.7 months (mos.), and a median OS of 28.7 mos. (95% CI 19.0 – 53.8) from R/R diagnosis. 17/22 (77%) pts had primary refractory disease, 7/22 (32%) had bulky disease, 13/22 were GCB, and 8/22 (36%) had double-hit lymphoma. 17/22 pts received CAR-T therapy as next line of therapy, and OS was not associated with therapy type or line by Cox regression. ROC analysis for PET/CT interval changes from R/R diagnosis to post-treatment identified cutoffs of 86.7% for ∆SUVmax, 93.43% for ∆TMTV, and 97.98% for ∆TLG. The known prognostic ∆SUVmax >70% was associated with better OS with hazard ratio (HR) of 0.307 (p=0.025). Each radiomic metric was independently associated with superior PFS (Table 1). R-IPI >3 at progression was not associated with differences in PFS or OS (p=0.751 and p=0.655, respectively). Pts with 2 or more high-risk PET/CT radiomic features (any combination of 2+ makers less than the cutoffs) was associated with inferior PFS (Table 1), OS (HR=0.265, p=0.009), and had a median PFS of 3.5 months and OS of 19.0 months. Conclusions: PET/CT radiomic metrics were independently associated with PFS and OS in R/R, TP53-aberrant DLBCL pts. Pts with 2 or more high-risk PET/CT features were at high risk for progression and inferior OS compared to those with ≤ 1, and this 2-component model significantly outperformed R-IPI for predicting PFS and OS in this exploratory analysis, which warrants validation in future study. Median PFS Median OS HR for PFS (95% CI, p-value) ∆ SUVmax ≥ 86.7% NR 50.1 months 0.259 (0.078-0.861, p=0.0164) ∆ TMTV ≥ 93.43% NR 46.3 months 0.294 (0.085-1.03, p=0.0223) ∆ TLG ≥ 97.98% NR 46.4 months 0.223 (0.066-0.758, p=0.0071) ≥2 high-risk PET/CT features 3.5 months 19.0 months 0.181 (0.055-0.591, p=0.0012) High-risk IPI (3 or higher) 36.7 months 28.6 months 1.207 (0.341-4.279, p=0.751)