There is currently sparse data on PSA characteristics, mpMRI findings, and subsequent csPCa detection over time for patients who have a reassuring mpMRI and omit initial biopsy. In our large academic cohort utilizing PHI and mpMRI, we report a 2.4% subsequent csPCa diagnosis rate in this population with median follow up of 23.1 months. This demonstrates a high negative predictive value for initial mpMRI combined with baseline clinical characteristics in selecting patients to forgo biopsy, as well as the safety of serial serum biomarker monitoring with subsequent for-cause mpMRI and prostate biopsy. Importantly, all men with subsequent csPCa detection had upgrade to a PIRADS 4 or 5 lesion on follow-up mpMRI.
Historically, the overall NPV of mpMRI ranges from 80–96%, although rates vary depending on study population and criteria for csPCa and negative mpMRI, and most studies focus on patients who were biopsied after initial mpMRI.(2, 9, 18, 19) Among patients who received initial biopsy, one systematic review and metanalysis of NPV of mpMRI for csPCa reported a pool NPV for biopsy naïve men at 90.8% and 86.8% for ≥ GG2 with negative mpMRI defined as PIRADS 1–2 and PIRADS 1–3, respectively.(20) Included studies had variations in sampling technique, with certain studies performing saturation or transperineal mapping biopsy, MR scoring system, and practice setting, which may all affect overall performance of mpMRI and NPV for ≥ GG2 PCa. Panebianco and colleagues studied the overall effect of negative MRI (defined as PIRADS 1–2) on subsequent csPCa both in patients with and without initial biopsy. Among the patients who omitted initial biopsy, the subsequent csPCa detection rate was 9.1% (24/264) with median 38 month follow up.(10) In addition, patients with initial PIRADS 3 on mpMRI remain controversial, as csPCa detection rates range from 12% in the PRECISION trial to 23.9% in Zhang and colleagues.(6, 21) At our institution, 23% of patients with PIRADS 3 were spared from initial biopsy with consideration of PSAD and PHI, which compares favorably with projections from use of a dedicated nomogram.(13) Our csPCa diagnosis free survival was 97.6% for patients with PIRADS 1–3 who were clinically deemed to be safe to omit initial biopsy. Patients with initial PIRADS 3 had a higher subsequent csPCa detection rate compared to patients with initial PIRADS 1 or 2, suggesting that patients with initial PIRADS 3 lesions should be monitored more closely.
Regarding utility of repeat mpMRI evaluation in patients with initial reassuring mpMRI and negative biopsy, Kinniard and colleagues reported subsequent csPCa detection rate of 32% (17/53) among men with PIRADS ≥ 3 on subsequent mpMRI, whereas none of the patients with subsequent PIRADS 1 or 2 lesions were found to have csPCa (0/20).(22) In our population who avoided initial biopsy, subsequent mpMRI was helpful in making biopsy decision, as all subsequent csPCa detection was associated with PIRADS 4 or 5 lesion. Among 11 patients who were biopsied with PIRADS 1–3 on subsequent mpMRI, no patients were found to have csPCa.
The negative predictive value of PHI has been established. Tosoian and colleagues reported on cohort of 345 men, and found that only three men with PHI < 27 had ≥ GG2 PCa, and no patients with PIRADS 1–3 on mpMRI and PHI < 27 had subsequent detection of prostate cancer.(17) 80.6% and 74% of patients in our initial and follow up cohort received PHI testing, respectively. Overall, our initial tiered screening with PHI prior to mpMRI may increase the incidence of csPCa among patients who received mpMRI. The median PHI among patients with subsequent csPCa was 51 (IQR 41.2, 59.3), and PHI remained a significant factor for prompting both repeat mpMRI and subsequent biopsy. In addition, both PSA and PHI velocity were significantly associated with repeat mpMRI and biopsy, with a PSAV threshold of 0.46 ng/mL/year providing highest discrimination for repeat mpMRI and biopsy. There was no clear PHIV threshold triggering re-evaluation, likely due to limited power and other confounding clinical variables, and defining an ideal PSAV and PHIV threshold will require external validation. Overall, the subsequent use of PHI after not undergoing initial biopsy appears to be a reasonable strategy for monitoring for csPCa.
The limitation of our study is its retrospective nature. There is inherent selection bias in terms of the factors that we identified to risk stratify patients who were deemed safe to omit biopsy. We are unable to ascertain the “true negative” in our cohort because mpMRI and biopsy were only performed if there was subsequent suspicion of csPCa, and we do not have biopsies or whole mount samples for patients who continued to have stable serum biomarkers. Further follow-up is required to determine whether any of these patients develop csPCa, although the follow-up data from Panebianco and colleagues suggest that there may be no significant difference in csPCa diagnosis free rates between 2 years of follow up and 4 years of follow up.(10) However, patients with persistently elevated PSA ≥ 50% from baseline were appropriately further worked up with PHI, mpMRI and biopsy, and patients had appropriate clinical follow up. There was no central review of mpMRI, although many of our radiologists are GU specialized and high-volume readers. At our predominantly academic practice, this may limit generalizability to community practice as mpMRI performance varies widely by reader experience.
Overall, we demonstrate that patients with initial PIRADS 1–3 selected to not undergo biopsy are safe to be observed with serial PSA or PHI and for-cause mpMRI longitudinally. Our data suggests that it is safe to continue observation if subsequent mpMRI shows PIRADS 1–3.