Study designs

This retrospective study was approved by the ethic committee of Third Affiliated Hospital of Sun Yat-Sen University and that of Third Affiliated Hospital of Sun Yat-Sen University, Yuedong hospital, with informed consents from all patients involved. All aspects of this research were conducted in accordance with the Declaration of Helsinki. Biopsy-naïve patients who received PI-RADS compliant mpMRI, TR-CDFI with a subjective blood-flow rating scale [6] and other routine examinations prior to ultrasound-guided targeted (TBx) or systematic prostate biopsies (SBx) in two centers were preliminarily included. Then, we further selected those who received surgical interventions (transurethral enucleation of prostate or radical prostatectomy, based on the results of biopsies) and had histo-pathological analyses on their removed prostate tissues, in order to achieve a more accurate pathological classification.

Patients

752 biopsy-naïve ethnic Chinese patients from the Third Affiliated Hospital of Sun Yat-Sen University (n = 502) and Third Affiliated Hospital of Sun Yat-Sen University, Yuedong hospital (n = 250) who were suspected with PCA and received ultrasound-guided SBx or TBx as well as subsequent surgical intervention, under the timeframe between Jan. 2015 and Feb. 2022, were included. Suspicion of PCA was defined as possessing at least one of the three features: (1) elevated serum PSA level (threshold varied from 2.5ng/mL to 4ng/mL among different urologists in actual clinical practices, and is in accordance to the standards in an existed literature [7]). (2) suspicious rectal examination (cT ≥ 2). (3) family history of PCA. Their detailed baseline characteristics are described in Table 1.

Table 1 Detailed baseline characteristics of patients included (n = 752) Assessment of TR-CDFI

The patient was placed in the left lateral decubitus position and the radiologist performed a digital-rectal examination. All patients were then examined using an endo-cavity 5- and 7-MHz probe (Aloka pro-sound 3500, Japan). Radiologists with more than 5 years of experience on prostate ultrasonography first performed the B-mode gray-scale examination, followed by color Doppler sonography. The volumes of prostate were first estimated. Subsequently, all TR-CDFI images were re-assessed and scored by Dr. Lu and Dr. Liu, two urologists with 15 years’ worth experience in prostate ultrasonography, in reference to a modified subjective blood-flow rating scale (mSBRS) based on the work of Mitterberger et.al [6]: score 1: definitely benign, minimal enhancement (only capsular and periurethral flow); score 2: probably benign, mild enhancement (symmetric radial flow from capsular branches); score 3: indeterminate, mildly increased enhancement (asymmetric/increased flow in the prostate); score 4: probably malignant, moderately increased enhancement (asymmetric/increased flow in the prostate); score 5: definitely malignant, substantially increased enhancement (asymmetric/increased flow in the prostate). We considered those TR-CDFI findings with mSBRS score ≥ 4 to be of PCA suspicion.

Assessment of multi-parametric MRI

Patients of both centers received mpMRI scan with the same field strength (3.0T) and vendor (Siemens). The scan included the sequences of triplanar T2-weighted, dynamic contrast-enhanced, diffusion-weighted imaging, and MR spectroscopy, abiding existing protocols [8]. Despite initial scoring and analysis by multiple experienced practitioners, all mpMRI images were re-evaluated and scored by biopsy performer Dr. Lu and Dr. Liu (both with 10 years of experience in prostate MRI), in compliance with Prostate Imaging Reporting and Data System (PI-RADS) Version 2.1 [9]. MRI outcomes with PI-RADS score ≥ 3 were designated suspicious of PCA.

Biopsy protocol and histopathological analysis

All biopsy procedures were performed utilizing a real-time dedicated ultrasound platform (Aloka pro-sound 3500). We integrated systematic biopsy (12 cores for each patient: apex, mid-gland and base, lateral and medial aspects of each sextant, on right and left lobes) and targeted biopsy (2–4 cores per target, up to 4 targets for each patient) for the process. Every systematic core and target set was collected in separated containers, for accurate localization of pathological findings. Biopsy samples were analyzed by two experienced pathologists from two centers, and reports included histological definitions according to the International Society of Urological Pathology (ISUP), Gleason score, percentage, and total length of core involvement by PCA for each core. We defined clinically insignificant PCA (cisPCA) as ISUP grade group 1 (Gleason score 3 + 3), disregarding numbers or percentage of cores involved. Conversely, clinically significant PCA (csPCA) was defined as tumors graded ISUP group 2 or higher (Gleason score ≥ 3 + 4). Highest ISUP (Gleason score) was used to grade the tumor should multiple cores presented with different gradings.

PCA risk analysis and diagnostic pathways

Following the concept of risk-adapted MRI pathways put forth in recent literatures [10, 11], we constructed a dedicative nomogram via multivariate logistic regression analysis. A patient’s risk of PCA over 0.5 determined through nomogram was considered as “High risk”, as risk below 0.5 was classified into the “Low risk” subgroup. Afterwards, 4 diagnostic pathways were retrospectively designated and reviewed: (1) Biopsy all pathway (Reference pathway): Biopsy for all patients in the study; (2) MRI-directed pathway: Biopsy for patients with positive MRI findings (PI-RADS ≥ 3), and patients with negative MRI findings (PI-RADS < 3) would not undergo further diagnostic tests; (3) MRI-directed TR-CDFI pathway: Biopsy for patients with strong positive MRI findings (PI-RADS ≥ 4) or patients with positive MRI findings (PI-RADS = 3) and positive TR-CDFI findings (mSBRS ≥ 4), as those with negative MRI findings (PI-RADS < 3) or positive MRI findings (PI-RADS = 3) but negative TR-CDFI findings (mSBRS < 4) would not receive subsequent diagnostic tests. (4) Risk-based MRI-directed TR-CDFI pathway: Inspired the recently proposed risk-based prostate biopsy strategy [10], patients would be first assessed with the nomogram mentioned above and classified into high and low risk group. Then, based on the risk status, which already incorporated mSBRS scoring of TR-CDFI, combined with the patients’ MRI findings, candidates for biopsies would be selected. The details of all 4 diagnostic sequences were illustrated in Fig. 1.

Fig. 1

Flow chart of diagnostic pathways for men suspicious of PCA. Four diagnostic pathways (biopsy for all, MRI-directed, MRI-directed TR-CDFI and risk-based MRI-directed TR-CDFI pathways) are presented graphically, with respective outcomes listed in result boxes

Outcome measurements and statistical analysis

Respective calculations on medians with interquartile ranges, frequencies, and proportions were made for continuous and categorical variables. We chose overall PCA detection rate, csPCA detection rate, cisPCA detection rate as well as biopsy avoidance rate and missed csPCA detection rate as the main outcomes of this study. Univariate logistic regression analysis was adopted to identify independent risk factor for the subsequent multivariate logistic regression analysis (enter mode was adopted), upon which a predictive nomogram would be produced. To better compare the performances of all diagnostic pathways, we utilized decision curve analysis [12], which focused on the net clinical benefits a patient could receive under different threshold probability levels, thus revealing the actual clinical impacts of certain procedures. Net benefit in our study could be calculated as “true positive rate – (false positive rate x weighting factor)”, where weighting factor equals “Threshold probability/1—threshold probability”. P-values < 0.05 were considered of statistical significance. All statistical analyses were conducted using R version 4.03, with rmda package.