Accuracy of Prostate Specific Antigen Density in Predicting Prostate Cancer- A single-centre UK experience

Abstract

Introduction and objectives: PSA has been a useful tool in identifying prostate cancer since its discovery. However relying on this alone can lead to unnecessary over investigation as it can be falsely raised in other conditions. PSA density is a possible enhanced marker for prostate biopsy indication, however it is not standard practice or recommended in national guidelines. Previous studies have shown that PSA density is most useful when the PSA is in the range of 4-10. We aim to study the effectiveness of PSA density at detecting prostate cancer within that range as well as outside that range. Methods: Retrospective single centre study at Peterborough City Hospital including 500 patients who had a MRI prior to prostate biopsy between July 2017 to July 2018. Patients undergoing repeat biopsy already on a cancer pathway were excluded. PSA density was calculated by dividing PSA from the prostate volume recorded on the MRI. A cut of value of PSA density was chosen at 0.15 and 0.10. Results: Data from 500 patients with a mean age of 65 and mean PSA of 10.7 (PSA range 0.32 to 99). 251 (50.2%) patients had cancer on histology with 152 having clinically significant cancer. Mean PSA density was 0.11ng/ml/cm3 with benign histology, 0.36ng/ml/cm3 for all cancer and 0.48ng/ml/cm3 for clinically significant cancer. Overall the mean PSA density in prostate cancer was found to be 0.15. Sensitivity of PSAD at detecting all prostate cancer at cutoff of 0.10 with PSA 4-10 was 76% and 88% for clinically significant prostate cancer (Gleason 7>). The negative predictive value was 94% for clinically significant prostate cancer. With PSA of 10 and PSA density of 0.10 the sensitivity was 93% (all cancer) and 96.6% (Gleason 7>) with a negative predictive value of 93.7% for clinically significant prostate cancer. At a cut off PSA density of 0.15 sensitivity and negative predictive value was 39% and 65% (all prostate cancer) and 50% and 87% (Gleason 7 and above) respectively with PSA of 4-10. ROC curves were created for each PSA range comparing cancer cases versus benign to determine area under curve values. No statistically significant difference was found for PSA range 10ng/ml for all cancer vs benign. However, there was a statistically significant difference when comparing AUC for PSAD in the PSA range of 4-10ng/ml for all cancer vs benign. (p < 0.01). Conclusions: This study is useful as a high powered study with a large study population which shows a direct correlation of PSAD to Gleason scoring. PSAD cut off value of 0.10ng/ml/cm3 has a higher sensitivity and negative predictive value especially when compared against clinically significant prostate cancer. When the PSAD cut off value is increased, sensitivity decreases, however specificity improves. PSAD can be used to avoid unnecessary biopsies however also runs the risk of missing clinically significant cancers.