Serial PSMA PET for Therapy Monitoring in Clinically Significant Prostate Cancer

Background and Rationale Prostate cancer is the fastest-growing male malignancy in China, with an average annual increase of approximately 12.6% over the past decade. The concept of clinically significant prostate cancer (csPCa) distinguishes patients who require active treatment from those with indolent disease. For this study, csPCa is defined as Gleason score ≥7 (grade group ≥2). PSMA PET has emerged as the most sensitive imaging modality for detecting prostate cancer lesions, outperforming conventional imaging (CT, bone scan). However, its application has been primarily limited to single time-point assessments for initial staging or biochemical recurrence. Standardized response criteria such as RECIP 1.0 (Response Evaluation Criteria in PSMA PET/CT) and PPP (PSMA PET Progression) criteria have been validated in the context of radiopharmaceutical therapy and show robust prognostic value (HR for mortality 3.48; 95% CI: 2.64-4.59) , but their role in patients treated with androgen signaling inhibitors is less clear due to potential "PSMA flare" or treatment-induced heterogeneous expression. Prospective data on serial PSMA PET for therapy monitoring in csPCa are urgently needed. Study Design and Technical Phases This is a prospective, multicenter, observational cohort study. Eligible patients are newly diagnosed csPCa with Gleason score ≥7, scheduled for curative-intent or systemic therapy (radiotherapy, ADT, chemotherapy, or combination). PSMA PET Acquisition Protocol Baseline scan: Performed prior to any treatment. Radiotracer: 68Ga-PSMA-11 or 18F-DCFPyL, dose according to institutional guidelines, acquisition starting 60 minutes post-injection. Follow-up scan: Triggered either by: PSA recurrence: PSA rise ≥2 ng/mL above nadir after radiotherapy, or biochemical progression per PCWG3 criteria; OR Fixed time window: 12-24 months after treatment completion (for patients without biochemical recurrence). Consistency: Same radiotracer and scanner type will be used whenever possible. Images centrally collected and analyzed. Quantitative PSMA PET Parameters (Primary) SUVmax change: Maximum standardized uptake value measured in the most intense PSMA-avid lesion. Up to five lesions per patient may be recorded; the highest SUVmax is used for analysis. Absolute and relative (percentage) change from baseline to follow-up will be calculated. PSMA-avid lesion count change: Total number of PSMA-avid lesions per patient, including primary tumor, nodal metastases, and bone metastases. Counting rules: Primary tumor: counted as 1 lesion if present and PSMA-avid. Lymph nodes: each distinct node with SUVmax \> liver background and typical morphology counted separately; contiguous nodal clusters counted as 1. Bone metastases: each discrete focus counted separately; diffuse marrow involvement counted as 1 "marrow" lesion. Excluded: sites of physiological uptake (e.g., salivary glands, bowel, renal collecting system). Two independent nuclear medicine physicians will perform counts; discrepancies resolved by consensus. Composite Reference Standard for Treatment Response Treatment response (ground truth) will be categorized as: complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD) based on an adjudicated composite reference standard: PSA kinetics: PSA50, PSA90, PSA nadir, time to PSA progression per PCWG3. Conventional imaging: CT and bone scan, assessed per RECIST 1.1 (for soft tissue) and PCWG3 (for bone) by two independent radiologists blinded to PSMA PET results; disagreements resolved by a third reader. Clinical outcomes: Need for treatment switch, occurrence of symptomatic progression, or death. The composite standard will be determined by a clinical adjudication committee blinded to PSMA PET results. Statistical Considerations (Detailed) Sample Size Justification Target enrollment: 110 patients (allowing 10% loss to follow-up, aiming for 100 evaluable). Assumptions based on csPCa (Gleason ≥7) population: Prognostic analysis: Expected 2-year progression rate 30% → \~30 events. With 100 patients, 80% power to detect a hazard ratio of 2.0 between PSMA PET-defined responders and non-responders (two-sided alpha 0.05, log-rank test). This detectable HR is consistent with prior RECIP 1.0 validation studies (HR 3.2). Concordance analysis: For Cohen's kappa, 100 patients yield a 95% CI half-width of ±0.10 when kappa ≈ 0.70, sufficient for agreement assessment. Subgroup analyses: Descriptive only; no formal power calculation. Statistical Analysis Plan All analyses will be performed using R (version ≥4.2). The final SAP will be finalized before database lock. Primary analysis: Linear mixed-effects models (LMM) for log-transformed SUVmax and lesion count, with fixed effects for time point (baseline, follow-up), treatment type, and random intercept per patient. Multinomial logistic regression to assess association between ΔSUVmax/Δlesion count and the composite reference standard response categories, adjusting for baseline disease volume. Survival analysis: Progression-free survival (PFS): time from treatment initiation to radiographic progression (per composite standard), PSA progression (PCWG3), or death. Time to castration resistance (for ADT patients). Kaplan-Meier curves and log-rank tests comparing PFS between PSMA PET response groups (e.g., responders vs. non-responders, defined by ≥30% decrease in SUVmax or lesion count). Cox proportional hazards models for baseline and follow-up parameters, adjusting for clinical covariates (PSA, Gleason, T stage, treatment type). Agreement analysis: Cohen's kappa with quadratic weights for agreement between PSMA PET response categories (binary or ordinal) and PSA response categories (PSA50, PSA90, PSA progression). Percent agreement and 95% CI reported. Clinical management impact: Proportion of patients with treatment change following serial PSMA PET, with exact binomial 95% CI. Multiple testing: Benjamini-Hochberg procedure for secondary endpoints (FDR 5%). No adjustment for exploratory subgroup analyses. Data Management and Quality Assurance Data will be stored centrally in a REDCap database with 3-step authentication. Data entry every 3-6 months. PSMA PET images centrally reviewed by two nuclear medicine physicians blinded to clinical outcomes, using standardized software (MIM Software, Hermes, or institutional platform). Inter-reader agreement for lesion counts and SUVmax measurement will be assessed using intraclass correlation coefficients (ICC).