Prognostic value of artificial intelligence (AI) –driven tumor estimation of PSMA PET total tumor burden (TTB) in newly diagnosed high-volume metastatic hormone-sensitive prostate cancer (mHSPC).

Abstract

47 Background: Prostate cancer represents a significant economic burden for health systems; the identification of risk factors that allows the identification of subgroups of patients that require closer follow-up or more intense therapeutic regimens has a direct impact on the prognosis and expenses for the patient, there is currently great interest in patients with high-volume mHSPC, this group is defined according to the number of lesions and their location, this could represent a challenge in decision making due to the variety of presentations regarding tumor burden, currently with the development of AI algorithms it is possible to estimate the PSMA TTB by identifying patients who are candidates for specific therapies without exposing them to comorbidities associated to aggressive regimens. Methods: Baseline PSMA PET scans of patients with histopathological corroborated diagnosis of prostate cancer and considered as high volume according to 68Ga PSMA PET/CT imaging in the period from October 2017 to June 2020 were retrospectively analyzed using an automated algorithm to estimate PSMA TTB, the median follow-up was of 44 +/- 9.7 months. These patients were divided into two groups according to the optimal cutoff values of PSMA TTB and estimate progression-free survival after initial therapy with Kaplan-Meier curves and correlation index between tumor burden and prostatic specific antigen. Results: 53 patients with a mean age of 71 years (+/- 5.1) were included, 17 with Gleason 7 (4+3/3+4), 18; G8 (4+4), 15; G9 (5+4/4+5) and 3; G10 (5+5), ISUP group 3 (n=17), group 4 (n=18) group 5 (n=18), the mean PSA value at diagnosis was 62.1 ng/dL (range 25.2- 201.5). PSMA TTB mean was 100.1cm3 (range 21cm3- 518.8 cm3). Log-rank test revealed that PSMA TTB lower than 45 cm3 (n=21) presented higher time of progression-free survival in comparison with patients with PSMA TTB higher than 45 cm3 (n=32) (31.1 mo vs 47.8 mo, p <0.003). TTB and PSA demonstrated a significant correlation (Pearson r= 0.526, p< 0.0001). There was no difference between groups of low and high PSMA TTB and ISUP groups or Gleason score. Sites of visceral disease were lung (n=11), (pleura=9) liver (n=9), adrenal, (n=5) brain, (n=2), and sites of extra spinal and pelvic disease were ribs (n=24) femur (n=16), skull (n=8), scapula (n=7), humerus (n=7). All patients had one or more lesions at previously described sites. Conclusions: PSMA TTB has great value in predicting progression-free survival of patients with newly diagnosed high-volume metastatic hormone-sensitive prostate cancer and has a strong correlation with PSA, strengthening the role of next-generation molecular imaging with PSMA PET as an invaluable tool in initial diagnosis and potentially in follow-up by providing information on the biological behavior of the disease in a systemic way.