Patient characteristics
A total of 379 consecutive patients suspected of prostate cancer were referred for a prostate biopsy. 32 patients were excluded. 347 patients were enrolled into the study and evaluated with CETRUS followed by collection of prostate biopsy samples. No adverse events related to the contrast agent were observed in any of the 347 patients. The mean patient age was 68.5 years (range: 53-79), mean prostate volume was 52.8 mL3 (range: 18-136), and median the PSA level was 8.9 ng/mL (range: 1.9-382). The number of men with a PSA level < 4, 4-10, 10-20 and >20 were 28, 160, 116, and 43, respectively. Prostate cancer was found in a total of 164 of 347 (47.3%) patients. The proportion of patients with Gleason scores ≤ 6, =7 and ≥8 were 22.6% (37), 43.3% (76), and 31.1% (51), respectively. Among the 164 patients, 91 patients underwent a laparoscopic radical prostatectomy. 17 patients were excluded. Finally, 74 patients with localized prostate cancer were enrolled and followed up as shown in Figure 1. Of the patients treated with a radical prostatectomy, the proportion of patients with Gleason scores ≤ 6, =7 and ≥8 were 24.7% (19), 41.6% (32), and 33.7% (26), respectively.
CETRUS is a useful tool for reducing the performance of unnecessary biopsies
The percentage of prostate cancer patients in CETRUS score groups 1-5 were 0% (0/42), 11.5% (6/52), 32.5% (40/123), 87.1% (61/70), and 95.0% (57/60), respectively. A significant positive correlation (r = 0.69, p < 0.001) was found between CETRUS scores and prostate cancer incidence. When cut off at 4 (≤ 3 as BPH and ≥ 4 as prostate cancer), the CETRUS score had an accuracy of 83.3% (289/347) in correctly diagnosing prostate cancer, with a sensitivity and specificity of 90.8% (118/130) and 78.8% (171/217), respectively. ROC analysis showed the AUC of the CETRUS score was 0.89 (95% CI: 0.85-0.92), indicating that patient CETRUS scores can be used to differentiate prostate cancer from benign prostatic hyperplasia (BPH) (Additional Table 1, Fig. 2). Moreover, when cut off at 2 (1 = BPH and ≥ 2 as prostate cancer), CETRUS scores demonstrated a specificity of 100% for differentiating prostate cancer from BPH, indicating that application of CETRUS using a cut-off score of 2 might be an effective tool for reducing the performance of unnecessary prostate biopsies. In this study, performing biopsy for cases with a CETRUS score ≥ 2 could have reduced the number of biopsies by 12.1% (42/347) without missing any cancer diagnoses and spared 23.0% (42/183) of men from undergoing an unnecessary biopsy (Table 1).
CETRUS can predict biochemical recurrence following a radical prostatectomy in patients with localized prostate cancer
The clinicopathologic characteristics of the 77 patients diagnosed with localized prostate cancer treated with a radical prostatectomy are summarized in Table 2. CETRUS score results were categorized in low score (≤ 3) and high score (> 3) groups. No significant correlation was found between CETRUS score and patient age, tumor stage, Gleason score, PSA level, BMI, or ECOG PS (p > 0.05, Table 2).
The median age of the 77 patients was 65.1 years (range: 49-74 years), and the median follow-up time after surgery was 30 months (range: 8-56 months). Biochemical recurrence was observed in 22% (17/77) of patients during the follow-up period. The 3-year biochemical recurrence-free survival rates were 86% (95% CI: 73%-93%) for patients with low CETRUS scores and 59% (95% CI: 53%-67%) for patients with high CETRUS scores (Fig. 3). Univariate Cox regression analysis revealed that CETRUS score, clinical stage, Gleason score, PSA level, and ECOG PS were significantly correlated with biochemical recurrence-free survival (p = 0.015, 0.042, 0.011, 0.037 and 0.047, respectively, Table 3), while other clinicopathologic variables, including age and BMI, were not (p = 0.618, 0.205, respectively, Table 3). Using multivariate analysis to further examine the parameters identified as significant by univariate analysis, we determined that patient CETRUS score at the time of diagnosis was an independent predictor of biochemical recurrence (HR: 7.02; 95% CI: 2.00-24.69; p = 0.002; Table 3).
To develop a more accurate prognostic tool, we used Cox proportional hazards regression to construct a prognostic model combining CETRUS scores with other clinicopathologic risk factors. A time-dependent ROC curve was used to compare the predictive accuracy of the combined model with the predictive accuracy of CETRUS scores alone or individual clinicopathologic factors alone. As shown in Figure 4, the model combining patient CETRUS scores, Gleason scores, tumor stage, and PSA levels (AUC at 3 years: 0.886; 95% CI: 0.754-1.000) had a better prognostic value than relying on CETRUS scores alone (AUC at 3 years: 0.696; 95% CI: 0.520-0.890; p=0.006), Gleason scores alone (AUC at 3 years: 0.679; 95% CI: 0.554-0.811; p=0.018), tumor stage alone (AUC at 3 years: 0.620; 95% CI: 0.500-0.748; p<0.001), or PSA levels alone (AUC at 3 years: 0.611; 95% CI: 0.500-0.792; p=0.004). The Memorial Sloan Kettering Cancer Center (MSKCC) nomogram is currently the most commonly used predictor of biochemical recurrence and survival for prostate cancer. Therefore, we compared pre-radical prostatectomy CETRUS scores (0.696) and MSKCC nomograms (0.709) for 3-year AUC and found there was no significant difference between the two methods (Fig.5). Therefore, CETRUS scores may add prognostic value to the clinicopathologic risk factors currently used in the context of localized prostate cancer.