PET imaging biomarkers predict survival in peripheral T-cell lymphoma: A systematic review and meta-analysis.

Abstract Background Peripheral T-cell lymphomas (PTCLs) are aggressive and heterogeneous lymphoid malignancies with poor prognoses and limited response durability using CHOP-based regimens. Recent studies suggest that PET/CT-derived biomarkers—including total metabolic tumor volume (TMTV), total lesion glycolysis (TLG), and Deauville scores (DS) at interim and end-of-treatment (EOT)—may hold prognostic value, yet they are not uniformly incorporated into clinical decision-making. As frontline regimens like BV-CHP become more widely adopted, identifying reliable predictors of outcomes is critical. We hypothesized that PET-based biomarkers—TMTV, TLG, and DS—are significantly associated with overall survival (OS) and progression-free survival (PFS) in patients with nodal PTCL treated with CHOP-like or BV-CHP regimens. Methods This systematic meta-analysis (PROSPERO CRD420251041744) identified 196 articles via systematic search of PubMed and Cochrane searches. Inclusion criteria were treatment-naive adults with nodal PTCL treated with CHOP-like BV-CHP regimens studies reporting hazard ratios (HRs) for OS or PFS associated with baseline TMTV, TLG, or DS. Exclusion criteria included case reports, pediatric studies, non-English publications, and studies involving other lymphoma subtypes, ENKTL, cutaneous T-cell lymphomas, and other mature lymphoid neoplasms. Ten studies met the criteria. Five reviewers screened and extracted data. PET variables were dichotomized as study specific threshold of high vs. low (TMTV,) and DS 4–5 vs. 1–3 for interim and EOT comparisons. We conducted Meta-analyses using RevMan 7.2.0 software, inverse-variance methods and random-effects models. Heterogeneity was assessed via I² and τ² (REML). Results Out of 10 eligible studies 9 were retrospective, and ECHELON-2 (E-2) study was prospective with prespecified assessment of PET response using DS. CHOP-like regimens were used in 1075 patients (including 226 from the matched E-2 cohort), while 226 were treated with BV-CHP in E2. The pooled median age was 58.5 years, predominantly male (61.8%) with advanced-stage disease (88.6% Stage III/IV). Patient distribution between low-risk (IPI 0-2) and high-risk (IPI 3-5) groups was balanced. Follow-up duration ranged from 17 to 65.8 months; the estimated pooled median follow-up was 38 months. Baseline TMTV was evaluated in seven studies (n = 680) and was significantly associated with inferior PFS (pooled HR 2.78 [95% CI: 1.24–6.24]; p = 0.01; I² = 88%). Baseline TMTV also correlated with OS across five studies (HR 2.43 [95% CI: 1.49–3.97]; p = 0.0004; I²=83 %). TLG, reported in three studies (n = 254), was associated with worse OS (HR 3.07 [95% CI: 1.29–7.34]; p = 0.01; I² = 73%). Higher baseline SUV was variably associated with OS and PFS across studies, but without consistent statistical significance. DS high at interim PET (6 studies; n = 698) was strongly associated with PFS (HR 3.40 [95% CI: 2.54–4.54]; p < 0.00001; I² = 13%) and OS (HR 3.62 [95% CI: [2.05- 6.38]; p < 0.00001; I²= 65%). DS high at end-of-treatment (EOT) was even more predictive of OS (pooled HR 5.89 [95% CI: 2.15-16.14]; p < 0.0006; I² = 66%) and PFS (HR 4.29 [95% CI: 1.96–9.39]; p < 0.0003; I²=61 %). DS 1–3 was consistently associated with durable remissions, while DS 4–5 often identified primary refractory . Conclusions This meta-analysis confirms that PET-based imaging biomarkers—particularly TMTV, TLG, and DS are significantly associated with OS and PFS in patients with nodal PTCL treated with CHOP or BV-CHP. DS 4–5 at EOT showed the strongest predictive value across cohorts. These findings support incorporating PET biomarkers into clinical risk models and future trials. Limitations include inter-study heterogeneity, mostly retrospective designs, variability in PET methods and cutoff definitions. Prospective validation of optimal cutoffs, and integration with molecular classifiers are needed to refine risk-adapted therapy in PTCL.