Exploring the prognostic value of the TRAQinform Profile applied to end-of-treatment PSMA PET/CT in patients with mCRPC and treated with 177Lu-PSMA radioligand therapy: A retrospective, single-center analysis.

person Vishnu Murthy UCLA, David Geffen School of Medicine, Los Angeles, CA info_outline Vishnu Murthy, Ojaswita Lokre, Timothy G. Perk, Pan Thin, Kathleen Nguyen, Lucia Chen, Andrei Gafita, Johannes Czernin, Jeremie Calais

Authors person Vishnu Murthy UCLA, David Geffen School of Medicine, Los Angeles, CA info_outline Vishnu Murthy, Ojaswita Lokre, Timothy G. Perk, Pan Thin, Kathleen Nguyen, Lucia Chen, Andrei Gafita, Johannes Czernin, Jeremie Calais Organizations UCLA, David Geffen School of Medicine, Los Angeles, CA, AIQ Solutions, Madison, WI, UCLA, Los Angeles, CA, University of California, Los Angeles, Los Angeles, CA, Department of Medicine Statistics Core, David Geffen School of Medicine at UCLA, Los Angeles, CA, David Geffen School of Medicine at UCLA, Los Angeles, CA Abstract Disclosures Research Funding No funding sources reported Background: To explore the prognostic value of the TRAQinform Profile trained on mid-treatment prostate-specific membrane antigen (PSMA) PET/CT (PSMA-PET) when applied to end-of-treatment PSMA-PET (ePET) in patients with metastatic castration-resistant prostate cancer (mCRPC) treated with 177 Lu-PSMA Radioligand Therapy (PSMA-RLT). Methods: This was a single-center retrospective study. mCRPC patients who underwent PSMA-RLT with available baseline PSMA-PET (bPET) and ePET within 6 mo of the last PSMA-RLT cycle were eligible. Overall survival (OS) from ePET was collected. TRAQinform IQ technology (AIQ Solutions, Madison, WI) was used to conduct lesion region of interest (ROI)-based analyses at bPET and ePET, and changes in SUV max , SUV mean , volume (cm 3 ), and SUV total (SUV mean x volume) across all lesion ROI from bPET to ePET were extracted. Lesion ROI were matched across time points and categorized into new, increasing, stable, decreasing, and disappeared based on changes in SUV total . Associations between % new, % increasing, % stable, % decreasing, and % disappeared lesions and OS were evaluated using univariate Cox regression models. This study applied TRAQinform Profile (a random forest model), trained on 185 external PSMA-PET, to generate TRAQinform Profile scores for early identification of optimal vs. suboptimal responders to PSMA-RLT. Associations between TRAQinform Profile scores and OS were evaluated by Kaplan-Meier analysis. Results: Twenty mCRPC patients were included. Thirteen of 20 patients (65%) had died at the last follow-up. The median survival time from ePET was 14.5 mo (IQR, 10.2–44.3). The median number of lesion ROI identified on bPET and ePET were 13.5 (IQR: 5.8-59.5) and 33 (IQR: 2.8-86.3) respectively. The median % changes in SUV max , SUV mean , volume, and SUV total from bPET to ePET were -33.9%, -20.7%, 4.3%, and -13.9% respectively. An increase in percent new and percent new or increasing lesions was associated with a higher risk of death (HR = 1.03; p = 0.002, HR = 1.03; p = 0.003 respectively), while an increase in percent disappeared or decreasing lesions was associated with a lower risk of death (HR = 0.97; p = 0.006). The median TRAQinform Profile score was 0.27, and patients with a TRAQinform Profile score above the median had shorter OS compared with patients with a TRAQinform Profile score below the median (median OS, 10.9 mo [IQR, 5.0-13.3] vs. median OS, 44.3 mo [IQR, 14.5-44.3]; p = 0.02). Conclusions: In this retrospective study of 20 mCRPC patients treated with PSMA-RLT, the TRAQinform Profile score was prognostic for OS. Percent new, percent new or increasing, and percent disappeared or decreasing lesions were significantly associated with OS. Validation in larger, prospective multicentric clinical trials is warranted.