Unified model involving genomics, magnetic resonance imaging and prostate‐specific antigen density outperforms individual co‐variables at predicting biopsy upgrading in patients on active surveillance for low risk prostate cancer

Abstract Background Active surveillance (AS) is the reference standard treatment for the management of low risk prostate cancer (PCa). Accurate assessment of tumor aggressiveness guides recruitment to AS programs to avoid conservative treatment of intermediate and higher risk patients. Nevertheless, underestimating the disease risk may occur in some patients recruited, with biopsy upgrading and the concomitant potential for delayed treatment. Aim To evaluate the accuracy of mpMRI and GPS for the prediction of biopsy upgrading during active surveillance (AS) management of prostate cancer (PCa). Method A retrospective analysis was performed on 144 patients recruited to AS from October 2013 to December 2020. Median follow was 4.8 (IQR 3.6, 6.3) years. Upgrading was defined as upgrading to biopsy grade group ≥2 on follow up biopsies. Cox proportional hazard regression was used to investigate the effect of PSA density (PSAD), baseline Prostate Imaging‐Reporting and Data System (PI‐RADS) v2.1 score and GPS on upgrading. Time‐to‐event outcome, defined as upgrading, was estimated using the Kaplan–Meier method with log‐rank test. Results Overall rate of upgrading was 31.9% (n = 46). PSAD was higher in the patients who were upgraded (0.12 vs. 0.08 ng/ml2, p = .005), while no significant difference was present for median GPS in the overall cohort (overall median GPS 21; 22 upgrading vs. 20 no upgrading, p = .2044). On univariable cox proportional hazard regression analysis, the factors associated with increased risk of biopsy upgrading were PSA (HR = 1.30, CI 1.16–1.47, p = <.0001), PSAD (HR = 1.08, CI 1.05–1.12, p = <.0001) and higher PI‐RADS score (HR = 3.51, CI 1.56–7.91, p = .0024). On multivariable cox proportional hazard regression analysis, only PSAD (HR = 1.10, CI 1.06–1.14, p = <.001) and high PI‐RADS score (HR = 4.11, CI 1.79–9.44, p = .0009) were associated with upgrading. A cox regression model combining these three clinical features (PSAD ≥0.15 ng/ml2 at baseline, PI‐RADS Score and GPS) yielded a concordance index of 0.71 for the prediction of upgrading. Conclusion In this study PSAD has higher accuracy over baseline PI‐RADS score and GPS score for the prediction of PCa upgrading during AS. However, combined use of PSAD, GPS and PI‐RADS Score yielded the highest predictive ability with a concordance index of 0.71.


| INTRODUCTION
Active surveillance (AS) is the reference standard treatment for the management of low risk prostate cancer (PCa). Generally, AS programs require serial digital rectal examination (DRE) and prostate specific antigen (PSA) measurement, annual multi-parametric Magnetic Resonance Imaging (mpMRI) of the prostate and repeat biopsies 2-3 years after diagnosis. 1 This approach defers definitive treatment, providing significant quality of life benefits for those low risk patients recruited. Furthermore, disease progression and metastatic rate are extremely low. 2 Despite promising evidence of safe long term inclusion for AS patients, up to 43.6% still require definitive treatment in 5 years, based on a review of over 10 000 men in AS programs across 12 countries. 3 A more accurate assessment of tumor aggressiveness can also guide recruitment to AS programs to avoid conservative treatment of intermediate and higher risk patients. There are commercially available genomic tests that predict treatment outcome in prostate cancer. 4 A 17 gene RNA expression assay known as the Oncotype Genomic prostate score (GPS) (Exact Science Corp.) is one such test that has been validated in a biopsy setting to predict treatment outcomes following radical prostatectomy (RP), 5 as well as contributing to the decision for AS recruitment in newly diagnosed PCa patients. 6,7 One major concern for AS management lies in underdiagnosis of clinically significant prostate cancer (csPCa), with subsequent underestimation of disease risk. 8 The term "clinically significant prostate cancer" (csPCa) has gained increasing acceptance with the widespread adoption of pre-biopsy mpMRI, and refers to the detection of intermediate and high risk prostate cancer that would necessarily require radical treatment, rather than the detection of low risk prostate cancer which has AS as the standard of care. The ability of mpMRI to detect suspicious lesions with a greater chance of harboring csPCa, has improved clinical staging as well as the accuracy of repeat prostate biopsies via targeted biopsy (TB). 9 mpMRI and TB show promise in the AS setting, however its routine use in AS has not yet been established. 10 In this study, we sought to evaluate the individual roles of GPS and mpMRI in predicting upgrading during AS, as well as exploring the use of a unified model including both these co-variables and PSAD.

| Study population
An institutional review board approved retrospective analysis was performed on 144 patients who were diagnosed with National Comprehensive Cancer Network (NCCN) low risk, and low volume intermediate risk PCa and followed by the AS protocol from October 2013 to December 2020 (see Figure 1, flow chart). All patients had available GPS data, baseline mpMRI data and at least two follow up biopsies which may have included a confirmatory biopsy. Median follow up was approximately 4.8 (IQR 3.6, 6.3) years. Patients were diagnosed with systematic biopsy with or without TB, and second biopsy was defined as subsequent biopsy 12 months after the initial diagnostic biopsy. Repeat biopsies prior to 12 months from diagnostic biopsy were regarded as "confirmatory" (17/144 patients [11.8%]).

| Variables and outcome
Baseline clinical and pathological data included age, race, BMI, PSAD, biopsy Gleason score, PI-RADS Score, GPS and GPS risk group. For PSAD measurement, the volume data were defined by baseline mpMRI measurement. Patients were stratified by PI-RADS score (1, 2, and 3 grouped as low risk vs. 4 and 5 grouped as high risk) and GPS defined as low risk <18, intermediate risk 18-30 or high risk 31-100.
Upgrading was defined as upgrading to grade group ≥2 or higher on follow up biopsies. Upgrading was the primary outcome of the study and resulted in patients being offered definitive therapy, via RP or external beam radiation.

| RESULTS
The overall rate of upgrading was 31.9% (n = 46). Median time to upgrading between diagnostic and upgrade biopsy was 29.6 months (IQR 16.7, 48.7). Baseline characteristics are summarized in Table 1.
The only statistically significant difference between non upgraded and upgraded groups was PSAD (0.08 ng/ml 2 vs. 0.12 ng/ml 2 in no vs.

| DISCUSSION
The results of this study show, with nearly 5 years follow up, PSAD has more value than both mpMRI alone and GPS alone as a tool predicting upgrading on biopsy. With specific reference to PSAD, the results correlate well with previous studies in the literature, which have also reported a PSAD greater than 0.15 is associated with upgrading. 12,13 The results of this study support the utility of mpMRI in AS, not only because high PI-RADS score was a significant predictor of upgrading, but also by measurement of MRI prostate volume to calculate PSAD which is more accurate than TRUS volume measurements, 14     patients, showed baseline MRI prior to confirmatory biopsy, reduced subsequent conversion from AS to treatment by 50% after only 2 years follow up. 22 As with prediagnostic use of MRI, the ultimate goal would be MRI having sufficient accuracy in diagnosis to replace serial biopsy in AS follow up. However, many of these studies are performed in tertiary referral centers, and the reality for most practicing Urologists is that day to day MRIs in AS programs may be excellent but not yet of sufficient quality to replace histopathological examination. Until such time when that is the case, other parameters highlighted in this study will remain valuable tools. 23 Exploring radiogenomic associations may also support the growing utility of mpMRI in the AS setting. In this cohort there was no association between GPS and either low or high risk PI-RADS lesions on baseline MRI (see Table 2. with most studies in that confirmatory biopsy was not done in every patient, and, in the absence of any changes in disease parameters, repeat biopsy is not done until 3 years after diagnosis. All patients diagnosed at this institution were diagnosed with mpMRI fusion biopsy unless their mpMRI was without lesions, and patients received confirmatory biopsy only if they were diagnosed at an outside institution without targeted biopsy and their follow up mpMRI showed lesions suggestive of csPCa. This practice aims to reduce the number of biopsies. Finally, all mpMRI studies were reported prospectively as part of routine clinical workflow at different institutions and only a portion of studies was re-reported (n = 75) by one reader unaware of PSA, PSAD, GPS and biopsy findings, which means differences in MRI acquisition protocol and MRI scanners were present during the study period. On the other hand, this study has a reasonable size and is a single surgeon series, so treatment decision-making did not vary within the cohort.

| CONCLUSION
PI-RADS score of the baseline mpMRI and GPS were shown to behave some potential for predicting upgrading at nearly 5 years follow up in this single center tertiary referral center AS cohort, however their predictive ability was inferior to PSAD. mpMRI and GPS remain useful tools in AS, however, their optimal use is yet to be determined.
Moreover, with the development of scoring systems such as "PRE-CISE", the utility of MRI in AS may be expanding. Nevertheless, the combined use of PSAD, GPS and mpMRI yields the highest predictive ability for identifying pathologic upgrading at follow up biopsy in patients undergoing AS in this study, and on that basis, the use of all three parameters when recruiting patients with newly diagnosed PCa to AS programs is recommended.