Abstract
Purpose
To evaluate the predictive performance of the prostate health index (PHI) and PHI density (PHID), for clinically significant prostate cancer (csPCa) in patients with a PI-RADS score ≤3.
Materials and Methods
Patients tested for total prostate-specific antigen (tPSA, ≤100 ng/mL), free PSA (fPSA), and p2PSA at Peking University First Hospital were prospectively enrolled. Possible predictive factors of csPCa were analyzed using the receiver operating characteristic (ROC) curve. Results were expressed as area under the curve (AUC) with 95% confidence intervals (CI). The cutoff values of PHI and PHID were determined.
Results
We enrolled 222 patients in this study. The prevalence of csPCa in the PI-RADS ≤3 subgroup (n=89) was 22.47% (20/89). Age, tPSA, F/T, prostate volume, PSA density, PHI, PHID, and PI-RADS score were significantly associated with csPCa. PHID (AUC: 0.829 [95% CI: 0.717–0.941]) was the best predictor of csPCa. PHID >0.956 was set as the threshold of suspicious csPCa with a sensitivity of 85.00% and a specificity of 73.91%, avoiding 94.44% of unnecessary biopsies but missing 15.00% csPCa. A threshold of PHI ≥52.83 showed the same sensitivity but a rather lower specificity of 65.22% that avoided 93.75% of unnecessary biopsies.
Conclusions
PHI and PHID have the best predictive performance of csPCa in patients with PI-RADS score ≤3. A threshold value of PHID ≥0.956 may be used as the criterion for biopsy in these patients.
Keywords: Biomarker, Clinical laboratory techniques, Diagnostic imaging, Multiparametric magnetic resonance imaging, Prostate cancer
INTRODUCTION
Prostate cancer (PCa) is the most common cancer among men [1]. Prostate-specific antigen (PSA) has been widely used for PCa screening. However, given the rather low specificity of PSA, overdiagnosis and overtreatment have become issues of concern. Furthermore, other conditions such as benign prostate hyperplasia (BPH) or urological procedures may also lead to a rise in the PSA level [2]. Other blood or urine based biomarkers, such as prostate-specific membrane antigen and prostate cancer antigen 3 (PCA3), as well as multivariate models have been discovered and developed [3,4]. Discovered in 2000, a precursor form of PSA called [-2] proPSA (also known as p2PSA), is known to be associated with PCa and can help distinguish PCa from BPH because of its high specificity for PCa [5,6]. Prostate health index (PHI), a mathematical combination of serum concentrations of p2PSA, free PSA (fPSA), and total PSA (tPSA) was developed by Beckman Coulter. It has a better PCa prediction performance, especially for patients with PSA level 2–10 ng/mL.
Multiparametric magnetic resonance imaging (mpMRI) is a useful and precise method to evaluate probable prostatic lesions, and it provides orientations for further workups such as prostate biopsy [7,8]. Lesions discovered on mpMRI have been assessed using the Prostate Imaging Reporting and Data System (PI-RADS) version 2.1 [9] and categorized based on their likelihood for PCa. Per clinical experience, a few borderline lesions are often scored 3 (named “equivocal lesions”), indicating an undetermined probability of PCa. Nevertheless, some non-suspicious lesions (i.e., PI-RADS score 1 or 2) are finally diagnosed as cancers with time. Therefore, a deeper classification of non-suspicious or equivocal lesions may further lower the proportion of false negative PCa. The present study aimed to evaluate the predictive performance of PHI and PHI density (PHID) for PCa in patients with PI-RADS scores ≤3 and to determine whether PHI and PHID can help guide the decision to perform a biopsy for these patients.
MATERIALS AND METHODS
1. Study population
Patients who were tested for tPSA, fPSA, and p2PSA at Peking University First Hospital between July 2017 and October 2021 were prospectively enrolled with their information recorded. Overall, 399 consecutive patients were enrolled. All the patients were Asian men. The exclusion criteria were tPSA >100 ng/mL, absence of biopsy, tumor other than prostate adenocarcinoma on biopsy, values exceeding the maximum limit of detection, mpMRI with b-value <1,400 s/mm2 that cannot be used for PI-RADS v2.1 scoring, and insufficient clinical data. Ultimately, 222 patients were included (Fig. 1). This study protocol was reviewed and approved by the Ethics Committee of Peking University First Hospital (protocol code 2016-1275). The need for informed consent was waived by the Ethics Committee of Peking University First Hospital. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Fig. 1. The inclusion and exclusion flow chart of the study cohort. tPSA, total prostate-specific antigen; fPSA, free PSA; PHI, prostate health index; GS, Gleason score; mpMRI, multiparametric magnetic resonance imaging; PI-RADS, Prostate Imaging Reporting and Data System.
2. Clinical outcome
All patient data were collected (clinical and pathological characteristics). Prior to biopsy, age, and body mass index information was measured and/or recorded for each patient. Serum tPSA, fPSA, and p2PSA levels were also tested. The free-to-total ratio of PSA (F/T) was calculated by dividing tPSA by fPSA. The prostate volume (PV) was calculated using the formula width×length×height×0.52 with the parameters measured on mpMRI. PSAD was calculated by dividing tPSA by PV. PHI was calculated using the following formula:
PHID was calculated by dividing PHI by PV.
3. The mpMRI technique
Baseline prostate mpMRI was performed on a 3-T scanner (Discovery® HD 750; GE Healthcare) using a phased-array coil. The protocol used in this study included b-values of 0, 800, and 1,400 s/mm2 that met the recommended minimal requirements for PI-RADS v2.1. Details of the mpMRI parameters are presented in Supplementary Table 1.
4. Image interpretation
The images were prospectively interpreted by a collaboration group of experienced radiologists and urologists. PI-RADS v2.1 structured criteria developed by the European Society of Urogenital Radiology [9] was applied and scores were given to each patients.
5. Biopsy procedure
All included patients underwent prostate biopsy. A total of 218 patients received transrectal ultrasound-guided systematic biopsy and 149 of them also underwent a concomitant cognitive mpMRI-targeted biopsy, i.e., extra cores were obtained from suspicious regions found on mpMRI, and these cores were specifically labeled. Four patients received only targeted biopsy. A group of experienced genitourinary pathologists evaluated all the specimens to determine the diagnosis and Gleason score in positive cases. Clinically significant prostate cancer (csPCa) was defined as having a Gleason score 3+4, 4+3, 8, 9, or 10 (the Epstein criteria [10]). PCa, including csPCa and insignificant PCa, was diagnosed if malignancy was detected in any biopsy core.
6. Statistical analysis
Statistical analysis was performed using Python 3.7.6 with statsmodels 0.11.0, scipy 1.4.1, and sklearn 0.22.1 unless specified otherwise. Statistical significance was set at p<0.05, and all analyses were two-sided.
The primary endpoint of the study was the identification of csPCa on biopsy. The association between the possible risk factors and the presence of csPCa or PCa was investigated using univariate analyses. Continuous variables were compared using Student’s t-test and the Mann–Whitney U test, and categorical variables were compared using the Fisher’s exact test or chi-square test, as deemed appropriate. Multivariate logistic regression analysis was conducted to identify independent predictors of csPCa, as well as build combined models of carefully selected individual factors. Independent predictors and combined models were analyzed using the receiver operating characteristics (ROC) (through the estimation of the area under the curve, AUC) to evaluate their diagnostic values for csPCa. The DeLong test was used to compare AUCs. The one with largest Youden index was selected as the most appropriate cutoff value. The decision curves were plotted using R 4.2.3 after the output of predictive probabilities of factors/models calculated using Python 3.7.6 with statsmodels 0.11.0.
RESULTS
In all 222 patients, 100 (45.05%) had gray-zone PSA level (2–10 ng/mL). Among them, 44 (44.00%) had csPCa and 56 (56.00%) had PCa. The mean age (p=0.001), PSAD (p<0.001), PHI (p<0.001), PHID (p<0.001), and PI-RADS v2.1 score (p<0.001) of patients with csPCa were significantly higher than those of patients without csPCa (Table 1). A significantly lower PV (p<0.001) and F/T (p=0.006) were associated with csPCa (Table 1). In patients with PSA level 10–20 ng/mL (n=72, Supplementary Table 2), above 20 ng/mL (n=50, Supplementary Table 3) or the whole cohort (Supplementary Table 4), higher PSAD, PHI and PHID as well as lower F/T and PV were associated with csPCa, showed a similar pattern of those with gray-zone PSA level (the currently recommended PSA level for PHI testing).
Table 1. Patient characteristics of subgroups.
| Variable | PSA 2–10 ng/mL | PI-RADS score 1–3 | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total | csPCa | PCa | Total | csPCa | PCa | ||||||||||
| YES | NOa | p-value | YES | NO | p-value | YES | NOa | p-value | YES | NO | p-value | ||||
| Patients | 100 (45.05) | 44 (44.00) | 56 (56.00) | - | 57 (57.00) | 43 (43.00) | - | 89 (40.09) | 20 (22.47) | 69 (77.53) | - | 31 (34.83) | 58 (65.17) | - | |
| Age (y) | 65.09±7.35 | 67.73±6.93 | 63.02±7.00 | 0.001 | 67.18±6.48 | 62.33±7.52 | 0.001 | 64.63±6.97 | 68.45±6.49 | 63.52±6.71 | 0.005 | 66.71±6.90 | 63.52±6.76 | 0.040 | |
| BMI (kg/m2) | 24.46±3.10 | 24.95±3.23 | 24.08±2.93 | 0.165 | 24.61±3.13 | 24.27±3.04 | 0.598 | 24.60±2.87 | 25.39±2.93 | 24.37±2.81 | 0.165 | 24.92±2.72 | 24.43±2.94 | 0.452 | |
| tPSA (ng/mL) | 6.71 (5.17–8.29) | 6.75 (5.14–8.56) | 6.65 (5.36–8.07) | 0.265 | 6.67 (5.18–8.43) | 6.81 (5.05–7.95) | 0.308 | 9.77 (6.62–15.28) | 10.13 (6.66–15.35) | 9.75 (6.62–15.16) | 0.283 | 10.10 (6.19–15.09) | 9.76 (7.12–15.33) | 0.360 | |
| F/T | 0.14 (0.09–0.20) | 0.12 (0.08–0.17) | 0.16 (0.11–0.21) | 0.006 | 0.13 (0.09–0.18) | 0.15 (0.11–0.20) | 0.068 | 0.14 (0.09–0.21) | 0.09 (0.08–0.15) | 0.15 (0.10–0.21) | 0.007 | 0.10 (0.08–0.19) | 0.15 (0.11–0.20) | 0.027 | |
| PV (mL) | 44.27 (29.41–63.27) | 36.80 (26.90–47.81) | 50.51 (37.50–72.50) | <0.001 | 38.00 (28.00–53.00) | 48.93 (38.75–69.78) | 0.006 | 58.16 (41.50–80.36) | 38.28 (24.63–50.55) | 65.50 (48.78–87.05) | <0.001 | 46.22 (27.13–58.72) | 66.88 (49.47–94.31) | <0.001 | |
| PSAD (ng/mL2) | 0.14 (0.09–0.23) | 0.19 (0.11–0.29) | 0.11 (0.09–0.17) | <0.001 | 0.17 (0.11–0.26) | 0.11 (0.09–0.15) | 0.002 | 0.17 (0.11–0.29) | 0.42 (0.18–1.10) | 0.15 (0.10–0.24) | <0.001 | 0.31 (0.16–1.10) | 0.15 (0.10–0.20) | <0.001 | |
| PHI | 53.04 (33.36–72.68) | 67.75 (57.81–89.23) | 37.09 (29.78–56.08) | <0.001 | 65.53 (50.94–86.93) | 34.83 (29.61–46.18) | <0.001 | 47.47 (30.59–72.36) | 75.33 (54.04–92.70) | 43.02 (29.79–61.57) | 0.001 | 65.17 (36.73–84.71) | 43.57 (30.08–62.24) | 0.010 | |
| PHID | 1.12 (0.68–2.06) | 1.94 (1.15–3.23) | 0.77 (0.44–1.22) | <0.001 | 1.78 (0.96–2.95) | 0.75 (0.44–1.15) | <0.001 | 0.74 (0.44–1.26) | 1.88 (1.11–3.73) | 0.65 (0.42–1.01) | <0.001 | 1.24 (0.67–3.01) | 0.66 (0.43–0.92) | <0.001 | |
| PI-RADS v2 score | <0.001 | <0.001 | 0.006 | 0.002 | |||||||||||
| 1 | 2 (2.00) | 0 (0.00) | 2 (3.57) | 0 (0.00) | 2 (4.65) | 7 (7.87) | 0 (0.00) | 7 (10.14) | 1 (3.23) | 6 (10.34) | |||||
| 2 | 17 (17.00) | 2 (4.55) | 15 (26.79) | 3 (5.26) | 14 (32.56) | 29 (32.58) | 2 (10.00) | 27 (39.13) | 4 (12.90) | 25 (43.10) | |||||
| 3 | 26 (26.00) | 7 (15.91) | 19 (33.93) | 12 (21.05) | 14 (32.56) | 53 (59.55) | 18 (90.00) | 35 (50.72) | 26 (83.87) | 27 (46.55) | |||||
| 4 | 40 (40.00) | 24 (54.55) | 16 (28.57) | 29 (50.88) | 11 (25.58) | - | - | - | - | - | |||||
| 5 | 15 (15.00) | 11 (25.00) | 4 (7.14) | 13 (22.81) | 2 (4.65) | - | - | - | - | - | |||||
| Gleason Grade Group | - | - | - | - | |||||||||||
| Not cancer | 43 (43.00) | - | 43 (76.79) | - | 43 (100.00) | 58 (65.17) | - | 58 (84.06) | - | 58 (100.00) | |||||
| 1 | 13 (13.00) | - | 13 (23.21) | 13 (22.81) | - | 11 (12.36) | - | 11 (15.94) | 11 (35.48) | - | |||||
| 2 | 21 (21.00) | 21 (47.73) | - | 21 (36.84) | - | 8 (8.99) | 8 (40.00) | - | 8 (25.81) | - | |||||
| 3 | 11 (11.00) | 11 (25.00) | - | 11 (19.30) | - | 5 (5.62) | 5 (25.00) | - | 5 (16.13) | - | |||||
| 4 | 9 (9.00) | 9 (20.45) | - | 9 (15.79) | - | 4 (4.49) | 4 (20.00) | - | 4 (12.90) | - | |||||
| 5 | 3 (3.00) | 3 (6.82) | - | 3 (5.26) | - | 3 (3.37) | 3 (15.00) | - | 3 (9.68) | - | |||||
Values are presented as number (%), mean±standard deviation, or median (interquartile range).
csPCa, clinically significant prostate cancer; PCa, prostate cancer; BMI, body mass index; tPSA, total prostate-specific antigen (PSA); F/T, free-to-total PSA ratio; PV, prostate volume; PSAD, PSA density; PHI, prostate health index; PHID, PHI density; PI-RADS, Prostate Imaging Reporting and Data System.
a:Patients who were cancer-free or having insignificant cancer (i.e., positive cores ≥1 and Gleason score ≤6) were included in csPCa-negative subgroup.
Notably, in patients whose lesions were equivocal or negative on mpMRI (i.e. PI-RADS score ≤3, n=89), 20 patients (22.47%) had csPCa with a significantly higher age (p=0.005), PSAD (p<0.001), PHI (p=0.001) and PHID (p<0.001), meanwhile their F/T (p=0.007) and PV (p<0.001) were significantly lower (compared to those who did not have csPCa) (Table 1).
The multivariate logistic regression model for csPCa was established as a function of age, tPSA, F/T, PV, PHI, and PI-RADS score (Supplementary Table 5). All variables but tPSA and F/T remained significantly associated, suggesting that these variables could be independent predictors for the diagnosis of csPCa on biopsy in all patients. For the subgroups, in the multivariate logistic regression model analysis for csPCa in association with age, F/T, PV, PHI, and the PI-RADS score, only F/T was not significantly associated. PHI and PHID significantly increased the predictive performance of multivariate combined model (age+tPSA+ F/T+PV+PHI+PI-RADS score) in all patients (PHI, p=0.025; PHID, p=0.045) but it showed no significant difference in the PI-RADS ≤3 subgroup (PHI, p=0.165; PHID, p=0.493) and the gray-zone PSA subgroup (PHI, p=0.236; PHID, p=0.144) (combined model: age+F/T+PV+PHI+PI-RADS score), as shown in Supplementary Table 6. PHID showed the best diagnostic accuracy (Fig. 2, Table 2) in the whole cohort (AUC=0.891, 95% confidence interval [CI]: 0.849–0.933, p<0.05). This was also the case in the PI-RADS ≤3 subgroup (AUC=0.829, 95% CI: 0.717–0.941, p<0.05) and the gray-zone PSA subgroup (AUC=0.830, 95% CI: 0.751–0.908, p<0.05). While adding PHI or PHID to each individual factors (Table 2), it significantly increased the AUC of ROC curve of each factors in all patients, the gray-zone PSA subgroup, and the PI-RADS ≤3 subgroup (except age+PHI, p=0.102; BMI+PHI, p=0.054; PSAD+PHI, p=0.662; PSAD+PHID, p=0.076 in the PI-RADS ≤3 subgroup).
Fig. 2. Univariate ROC curve for csPCa prediction. (A) The whole cohort (n=222). (B) PI-RADS ≤3 subgroup (n=89). (C) Gray-zone PSA subgroup (n=100). ROC, receiver operating characteristic; csPCa, clinically significant prostate cancer; AUC, area under the curve; BMI, body mass index; tPSA, total prostate-specific antigen; F/T, free-to-total PSA ratio; PV, prostate volume; PSAD, PSA density; PHI, prostate health index; PHID, PHI density; PI-RADS, Prostate Imaging Reporting and Data System.
Table 2. AUC of ROC curve for csPCa prediction.
| Variable | All (n=222) | PI-RADS ≤3 (n=89) | Gray-zone PSA (n=100) | |
|---|---|---|---|---|
| Age | Standalone | 0.690 (0.621–0.759) | 0.699 (0.563–0.835) | 0.688 (0.582–0.795) |
| Age+PHI | 0.894 (0.852–0.936) | 0.809 (0.697–0.920) | 0.833 (0.753–0.914) | |
| p-value | <0.001 | 0.102 | 0.006 | |
| Age+PHID | 0.911 (0.873–0.949) | 0.822 (0.707–0.938) | 0.877 (0.809–0.944) | |
| p-value | <0.001 | 0.007 | <0.001 | |
| BMI | Standalone | 0.560 (0.484–0.635) | 0.607 (0.465–0.749) | 0.576 (0.462–0.690) |
| BMI+PHI | 0.873 (0.828–0.918) | 0.754 (0.626–0.883) | 0.832 (0.750–0.914) | |
| p-value | <0.001 | 0.054 | <0.001 | |
| BMI+PHID | 0.898 (0.857–0.939) | 0.830 (0.721–0.940) | 0.846 (0.770–0.922) | |
| p-value | <0.001 | 0.001 | <0.001 | |
| tPSA | Standalone | 0.649 (0.577–0.720) | 0.543 (0.395–0.690) | 0.537 (0.422–0.652) |
| tPSA+PHI | 0.875 (0.830–0.921) | 0.758 (0.625–0.891) | 0.829 (0.746–0.912) | |
| p-value | <0.001 | 0.010 | <0.001 | |
| tPSA+PHID | 0.892 (0.849–0.935) | 0.825 (0.710–0.940) | 0.834 (0.755–0.913) | |
| p-value | <0.001 | 0.001 | <0.001 | |
| F/T | Standalone | 0.294 (0.225–0.363) | 0.317 (0.169–0.466) | 0.355 (0.247–0.464) |
| F/T+PHI | 0.878 (0.832–0.923) | 0.759 (0.612–0.907) | 0.838 (0.756–0.921) | |
| p-value | <0.001 | 0.002 | <0.001 | |
| F/T+PHID | 0.888 (0.844–0.932) | 0.812 (0.685–0.940) | 0.830 (0.752–0.909) | |
| p-value | <0.001 | <0.001 | <0.001 | |
| PV | Standalone | 0.286 (0.217–0.356) | 0.216 (0.105–0.326) | 0.307 (0.203–0.412) |
| PV+PHI | 0.903 (0.863–0.942) | 0.856 (0.758–0.953) | 0.841 (0.765–0.917) | |
| p-value | <0.001 | <0.001 | <0.001 | |
| PV+PHID | 0.891 (0.848–0.933) | 0.845 (0.743–0.946) | 0.834 (0.755–0.913) | |
| p-value | <0.001 | <0.001 | <0.001 | |
| PSAD | Standalone | 0.776 (0.712–0.839) | 0.750 (0.619–0.881) | 0.700 (0.595–0.805) |
| PSAD+PHI | 0.876 (0.831–0.921) | 0.778 (0.647–0.908) | 0.827 (0.743–0.911) | |
| p-value | 0.001 | 0.662 | 0.034 | |
| PSAD+PHID | 0.890 (0.848–0.933) | 0.840 (0.744–0.936) | 0.828 (0.748–0.907) | |
| p-value | <0.001 | 0.076 | 0.002 | |
| PI-RADS | Standalone | 0.831 (0.781–0.882) | 0.701 (0.616–0.787) | 0.753 (0.663–0.842) |
| PI-RADS+PHI | 0.903 (0.864–0.942) | 0.827 (0.719–0.934) | 0.853 (0.775–0.932) | |
| p-value | <0.001 | 0.002 | 0.006 | |
| PI-RADS+PHID | 0.908 (0.870–0.945) | 0.864 (0.772–0.955) | 0.861 (0.791–0.930) | |
| p-value | <0.001 | <0.001 | 0.001 | |
| PHI | 0.872 (0.826 – 0.918) | 0.748 (0.610–0.886) | 0.821 (0.736–0.907) | |
| PHID | 0.891 (0.849 – 0.933) | 0.829 (0.717–0.941) | 0.830 (0.751–0.908) | |
Values are presented as AUC (95% confidence interval).
AUC, area under the curve; ROC, receiver operating characteristic; csPCa, clinically significant prostate cancer; BMI, body mass index; tPSA, total prostate-specific antigen (PSA); F/T, free-to-total PSA ratio; PV, prostate volume; PSAD, PSA density; PHI, prostate health index; PHID, PHI density; PI-RADS, Prostate Imaging Reporting and Data System.
The detailed diagnostic performance of PHI and PHID was analyzed, as shown in Table 3. Cutoff values of PHI and PHID for csPCa were determined using the Youden index. For patients with a non-suspicious or equivocal mpMRI result, PHI ≥51.84 or PHID ≥0.956 (sensitivity=85.00%) was the best criterion for csPCa prediction. The missing rate using this criterion was 15.00%. However, PHID presented with a higher specificity (73.91%) and a higher negative predictive value (NPV) of 94.44% than PHI (specificity=65.22% and NPV=93.75%), and this could reduce the rate of unnecessary biopsies. For the gray-zone PSA subgroup, PHI ≥48.68 had a higher sensitivity of 88.64% and a higher NPV of 88.64% than the threshold PHID ≥1.42 (sensitivity=68.18% and NPV=77.05%). Notably, the criteria for csPCa prediction in all patients and patients with non-positive mpMRI results were associated with a significantly lower percentage of tumorous tissue in biopsy cores (Table 3). The decision curve analysis (DCA) showed PHI (except the PI-RADS ≤3 subgroup) and PHID had more benefit than other factors (Supplementary Fig. 1).
Table 3. Cutoff values of PHI and PHID for predicting csPCa.
| Group | Prevalence of csPCa (%) | Variate | Cutoff value (≥) | Sensitivity (%) | Specificity (%) | Youden index | PPV (%) | NPV (%) | False negative rate (missed cases, %) | False positive rate (misdiagnosed cases, %) | Max% in core, median (IQR) | p-value | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Above cutoff | Below cutoff | ||||||||||||
| All (n=222) | 54.05 | PHI | 60.37 | 89.17 | 70.59 | 0.5975 | 78.10 | 84.71 | 10.83 | 29.41 | 67.00 (66.50–90.00) | 32.50 (10.00–80.00) | <0.001 |
| PHID | 1.45 | 83.33 | 83.33 | 0.6667 | 85.47 | 80.95 | 16.67 | 16.67 | 67.00 (67.00–90.00) | 50.00 (27.50–80.00) | <0.001 | ||
| PI-RADS ≤3 (n=89) | 22.47 | PHI | 51.84 | 85.00 | 65.22 | 0.5022 | 41.46 | 93.75 | 15.00 | 34.78 | 60.00 (34.00–80.00) | 25.00 (7.50–32.50) | 0.001 |
| PHID | 0.956 | 85.00 | 73.91 | 0.5891 | 48.57 | 94.44 | 15.00 | 26.09 | 55.00 (29.50–80.00) | 30.00 (7.50–32.50) | 0.003 | ||
| Gray-zone PSA (n=100) | 44.00 | PHI | 48.68 | 88.64 | 69.64 | 0.5828 | 69.64 | 88.64 | 11.36 | 30.36 | 67.00 (50.00–90.00) | 50.00 (20.00–75.00) | 0.102 |
| PHID | 1.42 | 68.18 | 83.93 | 0.5211 | 76.92 | 77.05 | 31.82 | 16.07 | 67.00 (60.00–85.00) | 55.00 (31.00–80.00) | 0.068 | ||
PHI, prostate health index; PHID, PHI density; csPCa, clinically significant prostate cancer; PPV, positive predictive value; NPV, negative predictive value; IQR, interquartile range; PI-RADS, Prostate Imaging Reporting and Data System; PSA, prostate-specific antigen.
DISCUSSION
Prostate biopsy is the only confirmative procedure (to diagnose PCa) recommended for patients who have elevated PSA level and/or suspicious digital rectal examination (DRE) and/or abnormal imaging. Given the aggressiveness and potential risks of biopsy (e.g., hemorrhage and infection), several triage strategies to avoid unnecessary biopsies have been studied. Attempts have been made using PSA related-indexes (F/T and PSAD), PI-RADS scores, and other biomarkers such as p2PSA and PCA3 [11]. To date, no consensus has been reached.
Approved by the FDA in 2012 for PCa screening in men older than 50 years with normal DRE, and PSA levels between 4 and 10 ng/mL, PHI was shown to be the best predictor when compared with tPSA, PSAD, and PV [12]. A meta-analysis study done after showed that PHI had a higher diagnostic performance for PCa than %fPSA in patients with PSA levels between 2 and 10 ng/mL [13]. PHI also showed a better predictive performance of csPCa in patients with PSA level >10 ng/mL [6] and patients with prior negative biopsy results [14]. A PHI above the range 31–34 is considered to be associated with csPCa with a sensitivity of 80%–85% and a specificity of 38%–46% [15]. In a prospective Asian cohort study, a threshold PHI ≥36.96 with a sensitivity of 90% and a specificity of 52.1% had better performance than tPSA, to predict csPCa detection on biopsy. Combining PHI and PV, PHID was also reported to have a better performance [16,17]. A better predictive performance of PHI and PHID (than the other potential predictors for csPCa) was also shown in the present study.
Considering its higher diagnostic performance on PCa, when compared with tPSA, transrectal ultrasound and DRE, mpMRI has been used in the diagnostic workup of PCa and its role in pre-biopsy triage has been proven [18]. Studies have shown the synergistic effect of both mpMRI and PHI for csPCa detection [19]. PI-RADS, a classification system associating imaging results and likelihood of PCa, is scored from 1 to 5. Scores 1 and 2 are considered to be non-suspicious of cancer with a NPV of approximately 90%, but there is still a considerable chance of missing PCa cases. Scores 4 and 5 are considered to be suspicious for cancer (40% to 80% of patients with these scores are definitely diagnosed with csPCa) [20]. PI-RADS score 3 lesions are considered equivocal for PCa [21]. Whether this type of lesion needs further diagnostic workup remains controversial. Studies have shown that the prevalence of csPCa in PI-RADS score 3 lesions ranged from 10% to 30% (mostly around 20%) [20,22]. If prostate biopsies were performed in patients with PI-RADS scores of 3, there would be approximately 80% unnecessary aggressive procedures, and these procedures put the patients at risk. On the other hand, leaving them for observation may not be an acceptable choice because of the high missing rate (20%). Some studies showed that a lesion with a PI-RADS score of 3 was associated with a low likelihood of csPCa [23], and suggested that such lesions with a prior negative biopsy should not undergo a second biopsy [24]. Recently, the image texture analysis of T2WI and ADC maps of such lesions [25], as well as further classification of these patients was considered to reduce the missing rate. A meta-analysis showed that a threshold PSAD value ≥0.15 ng/mL2 in these cases is an indication for csPCa confirmation by combined systematic plus targeted biopsy [26]. Nomograms combining PI-RADS score, age, and PSAD were also developed for csPCa prediction in patients with non-suspicious or equivocal mpMRI result (i.e., PI-RADS score ≤3) [27]. The predictive performance of the combination of other biomarkers has also been studied [28]. PHI and PHID were also attempted to further classify patients with positive mpMRI results [29]. However, PHI and PHID have never been used for csPCa prediction in patients with PI-RADS scores ≤3 that were difficult to make decisions of further workup. In the present study, a threshold of PHI ≥51.84 or PHID ≥0.956 was discovered based on the Youden index with the same sensitivity of 85.00% for csPCa prediction. Regarding the higher AUC of the ROC curve (PHID: 0.829 [95% CI: 0.717–0.941] vs. PHI: 0.748 [95% CI: 0.610–0.886], p<0.05) and relatively higher NPV of the cutoff value (PHID: 94.44% vs. PHI: 93.75%) than that of PHI, and an acceptable missing rate (i.e., false negative rate) of 15.00%, PHID was considered to be the best predictor of csPCa in mpMRI non-suspicious or equivocal patients with the highest benefit in DCA. Admittedly, the sensitivity (85.00%) of PHID threshold in patients with PI-RADS score 1 to 3 did not exceed PHI in those with gray-zone PSA level (sensitivity 88.64%; Table 3), but PHID did play an important role of csPCa prediction in patients with seemingly negative result on mpMRI, helping decision-making regarding biopsy in clinical setting.
There are several limitations to the present study. First, this single-centered small-sized cohort of Asian men was recruited from a tertiary medical center that may not be a good representation of the whole population. The cutoff value of PHI varied between Asian and European population [30] as reported. Second, the outcome of csPCa or PCa was based on systematic (69/222, 31.1%), targeted (4/222, 1.8%), and combined biopsies (149/222, 67.1%). As PHI and other biomarkers, and biochemical indexes were obtained for all the lesions, all positive cores were included (irrespective of the biopsy method). Biopsy-naïve (207/222, 93.2%) and repeated-biopsy (15/222, 6.8%) patients were also even-powered. Hence, further studies with larger populations and a PI-RADS v2.1 scoring system may be needed to confirm the improved performance of PHI and PHID for predicting csPCa.
CONCLUSIONS
In conclusion, PHI and PHID are useful predictors of csPCa. Not only can both serve as a triage tool for the whole population (as earlier reported) but they can also decrease the number of missed cases among patients with non-suspicious or equivocal MRI result (i.e., PI-RADS score ≤3). For this group of patients, a PHID ≥0.956 is associated with a higher likelihood of csPCa and is an indication for biopsy.
ACKNOWLEDGMENTS
The authors acknowledge the entire staff of the Department of Urology, Peking University First Hospital.
Footnotes
CONFLICTS OF INTEREST: The authors have nothing to disclose.
FUNDING: This study was supported by Beckman Coulter.
- Research conception and design: Yuanchong Chen and Gang Song.
- Data acquisition: Dong Xu, Mingjian Ruan, and Haixia Li.
- Statistical analysis: Yuanchong Chen, Guiting Lin, and Gang Song.
- Data analysis and interpretation: Yuanchong Chen and Gang Song.
- Drafting of the manuscript: Yuanchong Chen.
- Critical revision of the manuscript: Mingjian Ruan and Gang Song.
- Obtaining funding: Haixia Li.
- Administrative, technical, or material support: Haixia Li and Gang Song.
- Supervision: Gang Song.
- Approval of the final manuscript: all authors.
SUPPLEMENTARY MATERIALS
Supplementary materials can be found via https://doi.org/10.4111/icu.20230060.
Imaging protocols at 3.0 T
Patient characteristics of PSA 10-20 ng/mL subgroup (n=72)
Patient characteristics of PSA >20 ng/mL subgroup (n=50)
Patient characteristics of the whole cohort (n=222)
Multivariate analysis for pathological outcome (csPCa)
AUC of ROC curve of combined models for csPCa prediction
Decision curve for csPCa prediction. (A) The whole cohort (n=222). (B) PI-RADS ≤3 subgroup (n=89). (C) Gray-zone PSA subgroup (n=100). csPCa, clinically significant prostate cancer; BMI, body mass index; tPSA, total prostate-specific antigen; F/T, free-to-total PSA ratio; PV, prostate volume; PSAD, PSA density; PHI, prostate health index; PHID, PHI density; PI-RADS, Prostate Imaging Reporting and Data System.
References
- 1.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 2020;70:7–30. doi: 10.3322/caac.21590. [DOI] [PubMed] [Google Scholar]
- 2.Park SC, Shin YS, Zhang LT, Kim DS, Kim SZ, Park NC, et al. Prospective investigation of change in the prostate-specific antigens after various urologic procedures. Clin Interv Aging. 2015;10:1213–1218. doi: 10.2147/CIA.S84570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hendriks RJ, van Oort IM, Schalken JA. Blood-based and urinary prostate cancer biomarkers: a review and comparison of novel biomarkers for detection and treatment decisions. Prostate Cancer Prostatic Dis. 2017;20:12–19. doi: 10.1038/pcan.2016.59. [DOI] [PubMed] [Google Scholar]
- 4.Ferro M, De Cobelli O, Lucarelli G, Porreca A, Busetto GM, Cantiello F, et al. Beyond PSA: the role of prostate health index (phi) Int J Mol Sci. 2020;21:1184. doi: 10.3390/ijms21041184. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Guazzoni G, Nava L, Lazzeri M, Scattoni V, Lughezzani G, Maccagnano C, et al. Prostate-specific antigen (PSA) isoform p2PSA significantly improves the prediction of prostate cancer at initial extended prostate biopsies in patients with total PSA between 2.0 and 10 ng/ml: results of a prospective study in a clinical setting. Eur Urol. 2011;60:214–222. doi: 10.1016/j.eururo.2011.03.052. [DOI] [PubMed] [Google Scholar]
- 6.Na R, Ye D, Qi J, Liu F, Helfand BT, Brendler CB, et al. Prostate health index significantly reduced unnecessary prostate biopsies in patients with PSA 2-10 ng/mL and PSA >10 ng/mL: results from a Multicenter Study in China. Prostate. 2017;77:1221–1229. doi: 10.1002/pros.23382. [DOI] [PubMed] [Google Scholar]
- 7.Shukla-Dave A, Hricak H. Role of MRI in prostate cancer detection. NMR Biomed. 2014;27:16–24. doi: 10.1002/nbm.2934. [DOI] [PubMed] [Google Scholar]
- 8.NCCN. Prostate Cancer (Version 2.2020) [Internet] Plymouth Meeting: NCCN; 2020. Available from: https://www.nccn.org/professionals/physician_gls/pdf/prostate.pdf . [Google Scholar]
- 9.Turkbey B, Rosenkrantz AB, Haider MA, Padhani AR, Villeirs G, Macura KJ, et al. Prostate Imaging Reporting and Data System version 2.1: 2019 update of Prostate Imaging Reporting and Data System version 2. Eur Urol. 2019;76:340–351. doi: 10.1016/j.eururo.2019.02.033. [DOI] [PubMed] [Google Scholar]
- 10.Epstein JI, Walsh PC, Carmichael M, Brendler CB. Pathologic and clinical findings to predict tumor extent of nonpalpable (stage T1c) prostate cancer. JAMA. 1994;271:368–374. [PubMed] [Google Scholar]
- 11.EAU. EAU-EANM-ESUR-ESTRO-SIOG guidelines on prostate cancer 2020 [Internet] Arnhem: EAU; 2020. Available from: https://uroweb.org/guideline/prostate-cancer/ [Google Scholar]
- 12.Catalona WJ, Partin AW, Sanda MG, Wei JT, Klee GG, Bangma CH, et al. A multicenter study of [-2]pro-prostate specific antigen combined with prostate specific antigen and free prostate specific antigen for prostate cancer detection in the 2.0 to 10.0 ng/ml prostate specific antigen range. J Urol. 2011;185:1650–1655. doi: 10.1016/j.juro.2010.12.032. Erratum in: J Urol 2011;186:354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Bruzzese D, Mazzarella C, Ferro M, Perdonà S, Chiodini P, Perruolo G, et al. Prostate health index vs percent free prostate-specific antigen for prostate cancer detection in men with “gray” prostate-specific antigen levels at first biopsy: systematic review and meta-analysis. Transl Res. 2014;164:444–451. doi: 10.1016/j.trsl.2014.06.006. [DOI] [PubMed] [Google Scholar]
- 14.Liu AQ, Remmers S, Lau SY, Yip SY, Leung CH, Mak CW, et al. Initial Prostate Health Index (phi) and phi density predicts future risk of clinically significant prostate cancer in men with initial negative prostate biopsy: a 6-year follow-up study. Prostate Cancer Prostatic Dis. 2022;25:684–689. doi: 10.1038/s41391-021-00444-y. [DOI] [PubMed] [Google Scholar]
- 15.Loeb S, Sanda MG, Broyles DL, Shin SS, Bangma CH, Wei JT, et al. The prostate health index selectively identifies clinically significant prostate cancer. J Urol. 2015;193:1163–1169. doi: 10.1016/j.juro.2014.10.121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chiu ST, Cheng YT, Pu YS, Lu YC, Hong JH, Chung SD, et al. Prostate health index density outperforms prostate health index in clinically significant prostate cancer detection. Front Oncol. 2021;11:772182. doi: 10.3389/fonc.2021.772182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Peters R, Stephan C, Jung K, Lein M, Friedersdorff F, Maxeiner A. Comparison of PHI and PHI density for prostate cancer detection in a large retrospective caucasian cohort. Urol Int. 2022;106:878–883. doi: 10.1159/000517891. [DOI] [PubMed] [Google Scholar]
- 18.Schoots IG, Padhani AR, Rouvière O, Barentsz JO, Richenberg J. Analysis of magnetic resonance imaging-directed biopsy strategies for changing the paradigm of prostate cancer diagnosis. Eur Urol Oncol. 2020;3:32–41. doi: 10.1016/j.euo.2019.10.001. [DOI] [PubMed] [Google Scholar]
- 19.Hsieh PF, Li WJ, Lin WC, Chang H, Chang CH, Huang CP, et al. Combining prostate health index and multiparametric magnetic resonance imaging in the diagnosis of clinically significant prostate cancer in an Asian population. World J Urol. 2020;38:1207–1214. doi: 10.1007/s00345-019-02889-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Israël B, Leest MV, Sedelaar M, Padhani AR, Zámecnik P, Barentsz JO. Multiparametric magnetic resonance imaging for the detection of clinically significant prostate cancer: what urologists need to know. Part 2: interpretation. Eur Urol. 2020;77:469–480. doi: 10.1016/j.eururo.2019.10.024. [DOI] [PubMed] [Google Scholar]
- 21.Woo S, Suh CH, Kim SY, Cho JY, Kim SH. Diagnostic performance of prostate imaging reporting and data system version 2 for detection of prostate cancer: a systematic review and diagnostic meta-analysis. Eur Urol. 2017;72:177–188. doi: 10.1016/j.eururo.2017.01.042. [DOI] [PubMed] [Google Scholar]
- 22.UpToDate. The role of magnetic resonance imaging in prostate cancer 2020 [Internet] Waltham (MA): UpToDate; 2020. Available from: https://www.uptodate.com/contents/the-role-of-magnetic-resonance-imaging-in-prostate-cancer . [Google Scholar]
- 23.Liddell H, Jyoti R, Haxhimolla HZ. mp-MRI prostate characterised PI-RADS 3 lesions are associated with a low risk of clinically significant prostate cancer - a retrospective review of 92 biopsied PI-RADS 3 lesions. Curr Urol. 2015;8:96–100. doi: 10.1159/000365697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Sathianathen NJ, Konety BR, Soubra A, Metzger GJ, Spilseth B, Murugan P, et al. Which scores need a core? An evaluation of MR-targeted biopsy yield by PI-RADS score across different biopsy indications. Prostate Cancer Prostatic Dis. 2018;21:573–578. doi: 10.1038/s41391-018-0065-6. [DOI] [PubMed] [Google Scholar]
- 25.Giambelluca D, Cannella R, Vernuccio F, Comelli A, Pavone A, Salvaggio L, et al. PI-RADS 3 Lesions: role of prostate MRI texture analysis in the identification of prostate cancer. Curr Probl Diagn Radiol. 2021;50:175–185. doi: 10.1067/j.cpradiol.2019.10.009. [DOI] [PubMed] [Google Scholar]
- 26.Maggi M, Panebianco V, Mosca A, Salciccia S, Gentilucci A, Di Pierro G, et al. Prostate imaging reporting and data system 3 category cases at multiparametric magnetic resonance for prostate cancer: a systematic review and meta-analysis. Eur Urol Focus. 2020;6:463–478. doi: 10.1016/j.euf.2019.06.014. [DOI] [PubMed] [Google Scholar]
- 27.Zhang Y, Zeng N, Zhang F, Huang Y, Tian Y. How to make clinical decisions to avoid unnecessary prostate screening in biopsy-naïve men with PI-RADs v2 score ≤ 3? Int J Clin Oncol. 2020;25:175–186. doi: 10.1007/s10147-019-01524-9. [DOI] [PubMed] [Google Scholar]
- 28.Rajwa P, Huebner NA, Hostermann DI, Grossmann NC, Schuettfort VM, Korn S, et al. Evaluation of the predictive role of blood-based biomarkers in the context of suspicious prostate MRI in patients undergoing prostate biopsy. J Pers Med. 2021;11:1231. doi: 10.3390/jpm11111231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hsieh PF, Li TR, Lin WC, Chang H, Huang CP, Chang CH, et al. Combining prostate health index and multiparametric magnetic resonance imaging in estimating the histological diameter of prostate cancer. BMC Urol. 2021;21:161. doi: 10.1186/s12894-021-00928-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Chiu PK, Ng CF, Semjonow A, Zhu Y, Vincendeau S, Houlgatte A, et al. A multicentre evaluation of the role of the Prostate Health Index (PHI) in regions with differing prevalence of prostate cancer: adjustment of PHI reference ranges is needed for European and Asian settings. Eur Urol. 2019;75:558–561. doi: 10.1016/j.eururo.2018.10.047. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Imaging protocols at 3.0 T
Patient characteristics of PSA 10-20 ng/mL subgroup (n=72)
Patient characteristics of PSA >20 ng/mL subgroup (n=50)
Patient characteristics of the whole cohort (n=222)
Multivariate analysis for pathological outcome (csPCa)
AUC of ROC curve of combined models for csPCa prediction
Decision curve for csPCa prediction. (A) The whole cohort (n=222). (B) PI-RADS ≤3 subgroup (n=89). (C) Gray-zone PSA subgroup (n=100). csPCa, clinically significant prostate cancer; BMI, body mass index; tPSA, total prostate-specific antigen; F/T, free-to-total PSA ratio; PV, prostate volume; PSAD, PSA density; PHI, prostate health index; PHID, PHI density; PI-RADS, Prostate Imaging Reporting and Data System.