Purpose: Prostate-Specific Membrane Antigen (PSMA) PET/CT imaging has high sensitivity for monitoring soft tissue and bone metastases in metastatic prostate cancer (mPC) patients. Here, we extended our Quantitative Total Bone Imaging (QTBI) tool, designed to quantify bone lesion changes, to include soft tissue lesion response quantification using â?¶â?¸Ga-PSMA PET/CT images and validated its output against physician assessment.
Methods: â?¶â?¸Ga-PSMA PET/CT images of 16 mPC patients were gathered retrospectively for preliminary QTBI(PSMA) development. Lesions were identified using an SUV>2.5 g/ml threshold. Physiologic activity such as salivary glands and the renal system was removed manually. Baseline CTs were registered to follow-up CTs using articulated registration, which first divides the CT into bone segments before performing piecewise-rigid registration. Each registration was applied to neighboring soft tissue lesions. Registered lesions were considered matched from baseline to follow-up if their contours had non-zero overlap. Patient-level changes in SUVmax, SUVtotal, and lesion number were compared to physician assessment.
Results: Across baseline and follow-up, QTBI(PSMA) identified 72 and 101 soft tissue and 116 and 81 bone lesions, respectively. 61 lesions were matched by articulated registration. Individual lesion-level response showed 12/16 patients contained at least one increasing/new lesion and at least one decreasing/disappearing lesion. In patients identified by a physician as responding (N=11), progressing (N=4), or stable (N=1), patient-level changes were (median[range]) -47%[-100,+112], 45%[-11,+324], and -6% for SUVmax, -76%[-100,+49], 84%[-76,+7500], and +34% for SUVtotal, and -58%[-100,+43], 60%[-44,+960], and -16% for lesion number, respectively. In 2/16 patients, QTBI(PSMA) patient-level changes were discordant with physician assessment; however, in these patients, physician assessment was influenced by changes in a single lesion.
Conclusion: Semi-automated QTBI(PSMA) successfully assessed soft tissue and bone lesions and agreed with physician-based assessment in 14/16 of patients and identified response heterogeneity in 12/16 patients. Future QTBI(PSMA) analysis will include automated removal of physiologic activity and improved disease detection.
Funding Support, Disclosures, and Conflict of Interest: University of Wisconsin Carbone Cancer Center Support Grant P30 CA014520. Robert Jeraj, Guy Starbuck, and Glenn Liu are co-founders of AIQ Solutions.