5067 Background: 177Lu-PSMA radioligand therapy is now a standard treatment for patients (pts) with mCRPC. Dual-tracer FDG/PSMA PET imaging mismatch (i.e. FDG+ and PSMA- lesions) is currently the main selection criterion used to predict 177Lu-PSMA response. However, this approach is time-consuming, resource-intensive and may not fully capture tumor biological aggressiveness. More accessible and biologically informative biomarkers are therefore needed. Methods: This retrospective, single-center study included pts diagnosed with mCRPC who underwent FDG and PSMA PET for 177Lu-PSMA screening and circulating tumor DNA next-generation sequencing (ctDNA) as part of the STING program (NCT04932525) at Gustave Roussy. A tumor suppressor gene alterations (TSalt) signature was defined by ≥ 2 alterations among TP53, PTEN and/or RB1 . Univariate logistic and linear regression models were used along with area under the curve (AUC) to assess associations between genomic and mismatch as well as PET-derived parameter. Progression-free survival (PFS, PCWG3 criteria) and overall survival (OS) were analyzed using univariate Cox models. Results: Among the 110 pts with available ctDNA and dual-tracer imaging, 18 (16.4%) exhibited mismatch. TSalt+ tumors were identified in 24 pts (21.8%) and were enriched in the mismatch group (50.0 vs 16.3%, p = 0.004) and in case of liver metastases (33.3% vs 10.5%, p = 0.011). TSalt was associated with mismatch (p = 0.003, AUC = 0.625). TSalt predicted higher total lesion glycolysis (TLG; p < 0.001, AUC = 0.720), increased ratio TLG/total lesion activity (TLA; p = 0.008, AUC = 0.657), larger FDG+ tumor volume (p = 0.008, AUC = 0.720), and lower PSMA+ tumor volume (p < 0.001, AUC = 0.725). TSalt+ pts had significantly shorter OS compared with TSalt- pts (median 5.7 vs 16.5 months; HR 3.31, 95% CI 1.93–5.69, p < 0.001). Similarly, pts with mismatch had inferior OS (median 4.7 vs 16.3 months; HR 3.17, 95% CI 1.69–5.95, p < 0.001). Among the 79 pts treated with 177Lu-PSMA, TSalt+ status (n = 14, 17.7%) was associated with shorter median PFS (3.5 vs 5.1 months; HR 2.37, 95% CI 1.29–4.34, p = 0.004) and OS (5.5 vs 15.3 months; HR 4.19, 95% CI 2.17–8.10, p < 0.001). Patients with mismatch (n = 3, 3.8%) also experienced inferior outcomes, with shorter PFS (2.5 vs 4.9 months; HR 3.80, 95% CI 1.14–12.7, p = 0.02) and OS (5.6 vs 15.1 months; HR 4.97, 95% CI 1.48–16.7, p = 0.004). Conclusions: The TSalt genomic signature was associated with FDG/PSMA PET mismatch, adverse metabolic imaging features, and poorer survival in patients with mCRPC, including those treated with 177Lu-PSMA. These findings suggest that ctDNA genomic profiling may help identify molecular imaging mismatch, associated with biologically aggressive disease, such as lineage plasticity phenotypes. Prospective studies are warranted to validate its clinical utility.
Lazerges et al. (Wed,) studied this question.