A 4K score/MRI‐based nomogram for predicting prostate cancer, clinically significant prostate cancer, and unfavorable prostate cancer

Abstract Background The detection of prostate cancer requires histological confirmation in biopsy core. Currently, number of unnecessary prostate biopsies are being performed in the United States. This is due to the absence of appropriate biopsy decision‐making protocol. Aim To develop and validate a 4K score/multiparametric magnetic resonance imaging (mpMRI)‐based nomogram to predict prostate cancer (PCa), clinically significant prostate cancer (csPCa), and unfavorable prostate cancer (uPCa). Methods and Results Retrospective, single‐center study evaluating a cohort of 574 men with 4K score test >7% or suspicious digital rectal examination (DRE) or Prostate Imaging Reporting and Data System (PI‐RADS) scores 3, 4, or 5 on mpMRI that underwent systematic and/or mpMRI/ultrasound fusion–targeted prostate biopsy between 2016 and 2020. External cohort included 622 men. csPCa and uPCa were defined as Gleason score ≥3 + 4 and ≥4 + 3 on biopsy, respectively. Multivariable logistic regression analysis was performed to build nomogram for predicting PCa, csPCa, and uPCa. Validation was performed by plotting the area under the curve (AUC) and comparing nomogram‐predicted probabilities with actual rates of PCa, csPCa, and uPCa probabilities in the external cohort. 4K score, a PI‐RADS ≥4, prostate volume and prior negative biopsy were significant predictors of PCa, csPCa, and uPCa. AUCs were 0.84, 0.88, and 0.86 for the prediction of PCa, csPCa, and uPCa, respectively. The predicted and actual rates of PCa, csPCa, and uPCa showed agreement across all percentage probability ranges in the validation cohort. Using the prediction model at threshold of 30, 30% of overall biopsies, 41% of benign biopsies, and 19% of diagnosed indolent PCa could be avoided, while missing 9% of csPCa. Conclusion This novel nomogram would reduce unnecessary prostate biopsies and decrease detection of clinically insignificant PCa.


| BACKGROUND
Prostate cancer is the most common cancer in men in the United States, accounting for an estimated 19% of all newly diagnosed cancers in 2018. 1 Prostate-specific antigen (PSA) is the only molecular marker routinely used for detection of prostate cancer (PCa), and screening with PSA has been shown to reduce prostate cancer mortality. 2 However, numerous studies have demonstrated the diagnostic limitations of PSA resulting in overdiagnosis of indolent cancers, frequent unnecessary prostate biopsies, the results of which are benign, and overtreatment with significant morbidity. 3 Alternative strategies for the detection of significant PCa are needed to avoid potentially morbid, invasive procedures in men who are unlikely to be diagnosed with prostate cancer.
4K score test (OPKO Diagnostics, Woburn, MA) is a serum biomarker-based test on a four-kallikrein panel including kallikreinrelated peptidase 2 (hK2), intact PSA, free PSA, and total PSA, as well as incorporating clinical information such as biopsy history and DRE findings. The 4K score test has repeatedly been shown to predict prostate cancer (PCa) biopsy outcome in men with elevated PSAs while also significantly reducing the number of unnecessary biopsies. [4][5][6] Furthermore, multiparametric magnetic resonance imaging (mpMRI) in combination with targeted biopsies has emerged as an effective tool to detect clinically significant disease. [7][8][9] Studies combining biomarkers and MRI-targeted biopsies for prostate cancer detection have also been proven to decrease the number of biopsies and avoid overdetection of indolent cancer to an even greater extent. 10 Although the mpMRI and the 4K score test are both used in American clinical practice for the evaluation of prostate cancer, there are no reports on the impact of using these tests in combination. We hypothesize that both of these tests will provide independent and complementary value, and when combined, will improve the detection of clinically significant disease compared to either test alone. 11 The aim of this study was to develop a nomogram using 4K score test and mpMRI to detect clinically significant prostate cancer (csPCa) in men with an elevated PSA and/or abnormal DRE.

| Outcomes definition and statistical analysis
The outcome for predicting PCa was defined as a Gleason score of ≥3 + 3 on biopsy, and men with this outcome were considered cases.
Men with negative biopsy were considered controls. The outcome for predicting csPCa was defined as a Gleason score of ≥3 + 4 on biopsy; men with this outcome were considered cases, and the remaining men with negative biopsy and men with Gleason score 3 + 3 on biopsy were considered controls. The outcome for predicting uPCa was defined as a Gleason score of ≥4 + 3 on biopsy; men with this outcome were considered cases, while the remaining men with negative biopsy, Gleason score 3 + 3, 3 + 4, were considered controls.
Descriptive statistics for the two groups were calculated. Continuous variables were reported as median and interquartile range (IQR) and compared using a Mann-Whitney U test. Categorical variables were reported as rates and were tested with a chi-square test, as appropriate. The prediction model included the following variables: age, family history of prostate cancer, history of negative prior biopsy, 4K score test, DRE findings, mpMRI prostate volume, and highest PI-RADS score. We are aware that 4K score test incorporates clinical parameters like age, family history, DRE, and prior biopsy history. Therefore, we calculated matrix of correlation coefficients between 4K score test and these predictors. We also ran variance inflation factor analysis (the inflation in the variances of the parameter estimates due to collinearities among predictors) to evaluate the potential presence of substantial multicollinearity between these predictors in our model. Nomogram validation was performed in the external cohort in two stages. First, receiver-operating characteristics (ROCs) were plotted for presence of PCa, csPCa, and uPCa using the same variables that were used to build the nomogram. Second, calibration graphs were plotted by grouping sorted nomogram-predicted probabilities from the development cohort into deciles and then comparing the mean prediction of each group with the observed proportion of men from the validation cohort with PCa or csPCa or uPCa. Using nomogram-derived probability cut-offs, we calculated the number of biopsies that could be avoided without missing PCa, csPCa, or uPCa in the validation cohort.
Decision curve analysis (DCA) was performed to evaluate the performance of the prediction models in the validation cohort. Statistical analyses were performed using STATA 12 (StataCorp LP, College Station, TX, USA) and SAS 9.4 (SAS Institute, Cary, NC, USA). All tests were two-tailed with a significance level of P < .05.

| RESULTS
Of the total 574 men in the development cohort, 232 (40%) were diagnosed with PCa, while the remaining 342 men (60%) were not 3.2 | Nomogram to estimate risk of PCa, csPCa, and uPCa Figure 1A-C illustrate the nomogram developed for prediction of PCa, csPCa, and unfavorable PCa, respectively, in the development cohort.
MRI PI-RADS score, 4K score test, prostate volume, and history of prior negative biopsy were significantly contributing to the total score that eventually determined nomogram probability of PCa, csPCa, and unfavorable PCa.

| Nomogram validation
Area under receiver operating curves (AUCs) for predicting PCa, csPCa, and unfavorable PCa in the validation cohort were 0.84. 0.88, and 0.86, respectively ( Figure 2). 4K score test and mpMRI PI-RADS score showed significant contribution for building AUCs. We evaluated the nomogram's calibration by comparing predicted and actual probabilities of PCa, csPCa, and unfavorable PCa in the validation cohort. There was agreement between the predicted and actual rate of probabilities (0%-100%) for PCa, csPCa, and unfavorable PCa as seen by points on the diagonal line in Figure 3A-C, respectively.
DCA showed superior clinical risk prediction for 5%-95% of nomogram-derived probabilities for predicting PCa, csPCa, and uPCa than relying on 4K score test alone or PI-RADS score alone.
Using our model in the validation cohort, 10% of biopsies could be avoided without missing uPCa and with missing 1% csPCa, avoiding 17% of benign biopsies and avoiding 4% of clinically indolent PCa ( Figure 5)  Studies have shown that the 4K score can improve the diagnostic discrimination of csPCa, reducing the number of required prostate biopsies. It has been suggested it could play an important clinical role in the decision-making process prior to proceeding with initial prostate biopsy in men with an elevated PSA level or abnormal digital rectal examination (DRE) or after a prior negative biopsy and persistently abnormal PSA levels. 15 The recent prospective US validation study showed that 4K score test can predict csPCa with AUC 0.82 and with excellent calibration. 11 In the French arm of the ERSPC (European Randomized Study of Prostate Cancer Screening), AUC for detecting csPCa increased from 0.77 for a basic model (age, total PSA, and DRE) to 0.87 after adding four-kallikrein panel. 16 Our study confirms significant role of 4K score test in predicting csPCa. Additionally, multiple studies have shown that mpMRI helps in identifying a higher proportion of csPCa when compared to transrectal ultrasound-guided prostate biopsies alone.
In the large multicenter, paired-cohort study, PROMIS, (comparing the diagnostic accuracy of mpMRI and TRUS-biopsy against template prostate mapping biopsy), results show that mpMRI and targeted biopsies detected up to 18% more cases of csPCa compared with TRUS-biopsy for all, while avoiding diagnosis of nonsignificant PCa by 5%. 8  History of negative prostate biopsy in the past lowers the chances of finding PCa in the forthcoming prostate biopsy. 18 Finding the cancer at earlier stage is equally important for oncological efficacy as well as addressing quality-of-life issues after the surgery. 19 We found significance of prior negative biopsy for predicting PCa, csPCa, and uPCa.
Our study has some limitations. First, all biopsies were performed by a single experienced, high-volume expert, which could affect  could have a significant impact on patient morbidity and social economic costs within prostate cancer diagnostics.

ACKNOWLEDGMENT
We thank Ms Sima Rabinowitz for editorial revision.

CONFLICT OF INTEREST
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

ETHICAL STATEMENT
The study was conducted at the Icahn School of Medicine at Mount Sinai Health System (ISMMS) after approval from the Institutional Review Board (GCO 19-1711). Informed patient consent waived for this study.