CUAJ – Original Research Griffiths et al ECE vs. PNI in predicting pT3 disease in radical prostatectomy

Introduction: Risk assessment for non-organ-confined prostate cancer (PCa) is important in the surgical planning for radical prostatectomy (RP). Perineural invasion (PNI) on prostate biopsy has been associated with adverse pathological outcomes at prostatectomy. Similarly, the identification of suspected extracapsular extension (ECE) on multiparametric magnetic resonance imaging (mpMRI) has been shown to predict non-organ-confined disease. However, no prior study has compared these factors in predicting adverse pathology at prostatectomy. We evaluated mpMRI ECE and prostate biopsy PNI on multivariable analysis to determine their ability to predict pathological stage at time of RP. Methods: We retrospectively investigated the prostatectomy database at our institution to identify men who underwent prostate biopsy with pre-biopsy mpMRI and subsequent RP from 2013–2017. Multivariable regression analysis was performed to compare the association of mpMRI ECE (mECE) and PNI on prostate biopsy on the likelihood of finding pT3 disease on pathology post-prostatectomy. Results: Of a total 454 RP between 2013 and 2017, 191 patients met our inclusion criteria. Stage pT2 and pT3+ were found in 120 (62.8%) and 71 (37.2%) patients, respectively. Patients with mECE had 4.84 cumulative odds of worse pathological stage on RP (p=0.045) compared to PNI on biopsy, which showed cumulative odds of 2.25 (p=0.048). When controlling only for those CUAJ – Original Research Griffiths et al ECE vs. PNI in predicting pT3 disease in radical prostatectomy 2 © 2021 Canadian Urological Association patients without PNI, mECE was still found to be a significant predictor of pT3 disease at RP (p=0.030). However, in patients without mECE, PNI was not significant (p=0.062). Conclusions: While mECE and biopsy PNI were both associated with worse pathological stage on RP, mECE had significantly higher cumulative odds compared to PNI. The significant predictive ability of mECE adds further clinical value to the use of mpMRI in PCa management. While validation in a larger cohort is required, these factors have important clinical implications with regards to early diagnosis of advanced disease and surgical planning. Introduction In patients with localized prostate cancer, nerve-sparing surgery offers improved post-operative functional outcomes and is considered the standard approach to radical prostatectomy (RP).1,2 Implementing a nerve-sparing approach, however, is not appropriate for every case and the decision to nerve spare is influenced by the estimated probability of extraprostatic extension (EPE) on prostatectomy.3,4 To optimize surgical and treatment planning, risk assessment for features of non-organ-confined disease is performed through nomogram assessment of preoperative clinical features, such as preoperative PSA, biopsy Gleason score, and clinical stage.5-9 However, preoperative estimates are often inaccurate, and in patients where the tumor extends outside of the prostate, the risk of positive surgical margin and adverse pathology may be particularly high.10 In an effort to improve existing predictive risk models, additional preoperative parameters have been studied for their utility. Perineural invasion is defined as the tracking of tumor cells along or around nerve fibers; in prostate cancer this extension is commonly reported on prostate biopsy pathology.11,12 While not traditionally utilized in nomogram assessment, perineural invasion (PNI) on prostate biopsy (PBx) is associated with EPE at RP as well as margin positivity and biochemical recurrence.13-16 Multi-parametric magnetic resonance imaging (mpMRI) is a PCa diagnostic tool that is increasingly popular and is being utilized to direct prostate biopsy and surgical planning.17 Preoperative mpMRI is useful for assessing the presence of significant cancer, predicting organ confined disease, and assessing seminal vesicle invasion.17 While having a low to moderate sensitivity, mpMRI has achieved high specificity and positive predictive values of up to 90% in detecting extracapsular extension and seminal vesicle invasion.18-22 Interestingly, while both PNI on PBx and mpMRI extracapsular extension (mECE) have been shown as independent predictors of adverse outcomes and non-organ-confined disease, no prior study has compared these factors together. To query whether PNI is a significant prognostic variable in the era of mpMRI, we evaluated and compared the ability of mECE and PNI on prostate biopsy to predict advanced pathological stage at time of RP. CUAJ – Original Research Griffiths et al ECE vs. PNI in predicting pT3 disease in radical prostatectomy 3 © 2021 Canadian Urological Association Methods Patient population We retrospectively investigated the prostatectomy database at our institution to identify men who underwent radical prostatectomy from July 2013 through March 2017. Patients were included if they received an mpMRI and subsequent prostate biopsy before radical prostatectomy at our institution. Patients were excluded if they had missing information, incomplete mpMRI, if biopsy or mpMRI were conducted at an outside institution, or history of alternative or previous treatment for prostate cancer such as radiation therapy, cryoablation, or any neoadjuvant therapy. A subgroup analysis was performed to ascertain if mECE or PNI would reach significance in prediction of pT3 when excluding all mECE patients and then all PNI patients from the original cohort. All staging was assigned according to the 2017 American Joint Committee on Cancer (AJCC) staging criteria.23 Protocol for mpMRI Patients were referred for mpMRI in the setting of elevated PSA or PSA density. Urologists at our institution routinely obtained mpMRI prior to biopsy to utilize a fusion approach and for potential surgical planning. All patients underwent mpMRI (3T Verio, Siemens, Germany ®) of the prostate obtained with a 16-channel cardiac coil (Sense, Invivo ®) on the anterior pelvis and an endorectal coil (BPX-30, Medrad, Pittsburgh, PA ®) with its balloon filled with PFC-770 (3M, St. Paul, MN ®). Sequences obtained included tri-planar T2-weighted, diffusion-weighted imaging (DWI) (b-values 0, 500, 1000, and 1500) and separate b-2000 and dynamic contrast enhanced (DCE) sequences. All mpMRI were read by experienced genitourinary radiologists, who assigned risk scores using the PI-RADS v1 and PI-RADS v2 scoring system. If multiple suspicious lesions were present, the highest PI-RADS score reported was used. Extra-capsular extension on MRI was stratified based on radiology report into mECE present, mECE suspicious, or mECE absent. mECE ‘present’ was assigned if ECE was clearly seen, as described by the radiologist, with gross extension of disease. mECE ‘suspicious’ was assigned to reports including descriptions of “abutment”, capsular “bulging”, or “microscopic invasion cannot be ruled out”. mECE ‘absent’ was assigned when there was clearly no evidence of ECE on MRI. Protocol for prostate biopsy All biopsies were performed via transrectal technique by experienced urologists at our institution. MRI-Targeted images were processed on a Dynacad workstation (Invivo, Gainesville, FL®). MR/TRUS image fusion was performed using UroNav® software in conjunction with an IU-22 (Philips Health Care, Best Netherlands) end-fire ultrasound probe. During biopsies of the lesions, one core was obtained in the axial and sagittal planes for a total of two cores per lesion followed by 12-core systematic biopsy as per our institution’s protocol for MRI fusion biopsy. Patients without a targetable lesion identified on MRI received a standard CUAJ – Original Research Griffiths et al ECE vs. PNI in predicting pT3 disease in radical prostatectomy 4 © 2021 Canadian Urological Association 12-core transrectal ultrasound (TRUS) biopsy whereby 2 cores, medial and lateral, are taken from each sextant region in a standard fashion. All biopsy pathology slides were reviewed by our institution’s experienced genitourinary pathologist. Statistical methods Univariable analysis was performed to determine if there were differences in age, PSA at MRI, prostate volume, PIRADS score, and ECE between pT2 and pT3+. A Mann-Whitney U test was run for non-normally distributed continuous variables and chi square was run for nominal parameters. A cumulative odds ordinal logistic regression with proportional odds was run to determine the effect of 6 variables on pathological cancer staging after radical prostatectomy. These variables included age, PNI on PBx, mECE, PI-RADS, PSA, and MRI prostate volume. A binomial logistic regression analysis was performed on all the variables to ascertain the likelihood of subjects having positive margins. PI-RADS, PSA, age, and prostate volume were analyzed as continuous variables. Proportional odds were assessed by a full likelihood ratio test comparing the fitted model to a model with varying location parameters. The model fit was tested through a deviance goodness-of-fit test. Collinearity of the independent variables was assessed and ruled out through inspection of correlation coefficients and tolerance/VIF values. SPSS 24.0.0 for Windows (SPSS Inc., Chicago, IL, USA) was used for all statistical analysis with p < 0.05 considered statistically significant. Results During the study period, 454 patients underwent RP at our intuition. 191 patients met our inclusion criteria with baseline characteristics shown on Table 1. Stage pT2 and pT3 were found in 120 (62.8%) and 71 (37.2%) patients after prostatectomy. Biopsy pathology classified 28 (14.7%), 81 (42.4%), and 82 (42.9%) patients into Gleason grade group 1, 2, and 3-5 respectively. Prostate needle biopsy PNI was positive in 55 (22.8%) and absent in 136 (71.2%) of patients. On mpMRI, mECE was found in 12 (6.3%), suspicious mECE in 32 (16.8%), and mECE absent in 147 (77%) patients (Table 1). Multiparametric MRI findings for mECE and suspicious mECE were compared to final pathology to d