Development and validation of a nomogram for predicting the likelihood of metastasis in prostate cancer patients undergoing Ga-68 PSMA PET/CT due to biochemical recurrence

Abstract

Objective

To develop a nomogram based on commonly used clinical data for predicting the likelihood of metastasis in gallium-68 prostate-specific membrane antigen PET/computed tomography (Ga-68 PSMA PET/CT) scans of prostate cancer patients with confirmed biochemical recurrence (BCR).

Methods

One-hundred thirty-five ( n = 135) patients who underwent Ga-68 PSMA PET/CT due to BCR were included in the study. Predictors of metastasis in Ga-68 PSMA PET/CT were determined with multivariable logistic regression analysis. Coefficients derived from the regression model were used to develop a prediction nomogram. The performance of the prediction model was evaluated with receiver operating characteristic analysis. Internal validation was performed with 50 bootstrap resamples, and the nomogram’s clinical benefit was assessed with decision curve analysis.

Results

Multivariable logistic regression analysis revealed that ISUP group, prostate-specific antigen (PSA) before PET and PSA doubling time were independent predictors of metastasis in Ga-68 PSMA PET/CT. A prediction nomogram was developed according to this model [the area under curve: 0.866; 95% confidence interval (CI), 0.788–0.944]. The best cutoff value of the nomogram-derived likelihood for predicting metastasis was 60%, with a bootstrap-corrected accuracy of 78.8%. An online version of the nomogram was implemented on pro-gram.nzm.co (https://pro-gram.nzm.co).

Conclusion

The proposed nomogram provides a practical approach for predicting the likelihood of imaging-based metastasis according to Ga-68 PSMA PET/CT in patients with BCR, with results ≥60% being the most accurate cutoff for referring patients to Ga-68 PSMA PET/CT. If validated in a larger cohort, this tool can serve as a guide for the appropriate use of Ga-68 PSMA PET/CT.