Free PSA performs better than total PSA in predicting prostate volume in Chinese men with PSA levels of 2.5–9.9 ng ml⁻¹

Abstract

This study investigated whether free prostate-specific antigen (fPSA) performs better than total PSA (tPSA) in predicting prostate volume (PV) in Chinese men with different PSA levels. A total of 5463 men with PSA levels of <10 ng ml⁻¹ and without prostate cancer diagnosis were included in this study. Patients were classified into four groups: PSA <2.5 ng ml⁻¹, 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹. Pearson/Spearman’s correlation coefficient (r) and receiver operating characteristic (ROC) curves were used to evaluate the ability of tPSA and fPSA to predict PV. The correlation coefficient between tPSA and PV in the PSA <2.5 ng ml⁻¹ cohort (r = 0.422; P < 0.001) was markedly higher than those of the cohorts with PSA levels of 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ (r = 0.114, 0.167, and 0.264, respectively; all P ≤ 0.001), while fPSA levels did not differ significantly among different PSA groups. Area under ROC curve (AUC) analyses revealed that the performance of fPSA in predicting PV ≥40 ml (AUC: 0.694, 0.714, and 0.727) was better than that of tPSA (AUC = 0.545, 0.561, and 0.611) in men with PSA levels of 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹, respectively, but not at PSA levels of <2.5 ng ml⁻¹ (AUC: 0.713 vs 0.720). These findings suggest that the relationship between tPSA and PV may vary with PSA level and that fPSA is more powerful at predicting PV only in the ‘‘gray zone’’ (PSA levels of 2.5–9.9 ng ml⁻¹), but its performance was similar to that of tPSA at PSA levels of <2.5 ng ml⁻¹.

INTRODUCTION

Prostate volume (PV) is a key factor in the initiation of treatment for benign prostatic hyperplasia (BPH)1 because it strongly predicts outcomes, such as acute urinary retention and the need for surgery,2,3 as well as response to treatment.2–4 Although transrectal ultrasound (TRUS) is the gold standard for assessing PV, it is not practical to routinely use it for screening and follow-up since it may not be available in many primary care settings and there are often long waiting periods. Moreover, patients may feel uncomfortable with the procedure and the associated cost may be too high. Thus, there is a need for simple, accurate, and inexpensive methods of assessing PV to guide diagnostic and treatment decisions, especially in outpatient settings. It has been shown that serum prostate-specific antigen (PSA) closely correlates with PV and may serve as a surrogate index for predicting PV.5,6 Recent studies show that free PSA (fPSA) is more powerful at predicting PV than total PSA (tPSA) in patients without prostate cancer.7–9 However, almost all available data on the relationship between fPSA and PV in BPH patients are based on TRUS-guided biopsies. Although these studies showed that the chance of prostate cancer was decreased to the maximal extent clinically, they were limited to a small number of cases, which may limit their applicability to the general population.10,11 Moreover, most data on the relationship between tPSA/fPSA and PV are based on populations in Western countries. It is possible that the relationship between tPSA/fPSA and PV differs between races or between developed and developing countries. Furthermore, in the Chinese population, the correlation between fPSA and PV was significantly higher in the 0–4 ng ml⁻¹ cohort than that in the 4–10 ng ml⁻¹ cohort, suggesting that the ability of fPSA to predict PV may depend on PSA level.8 However, that study did not use receiver operating characteristic (ROC) curve analysis to assess the ability of tPSA and fPSA to predict PV in different PSA-stratified cohorts. Usually, when the level of tPSA is less than 4 ng ml⁻¹, and in the absence of abnormal digital rectal examination (DRE) findings or abnormal hypoechoic lesions based on TRUS, prostate cancer is not suspected and a prostate biopsy is not carried out; such cases are defined as clinical BPH. In this study, we enrolled participants with pathologically-proven BPH, as well as those with “clinical BPH”, and assessed the relationship between age, tPSA, and fPSA and PV in Chinese men with different PSA levels.

PATIENTS AND METHODS

The data on patients who visited the Department of Urology of the Guangzhou First People’s Hospital (Guangzhou, China) from December 1999 to December 2016 and complained of lower urinary tract symptoms (LUTS) were retrospectively retrieved from the patient registry. A total of 8746 men who underwent tPSA testing and transrectal prostate volume measurement but were not diagnosed with prostate cancer were identified. Biopsies were taken from patients who met one or more of the following criteria: (i) tPSA level of >4 ng ml⁻¹, and the ratio of fPSA to tPSA and magnetic resonance imaging (MRI) were also considered, (ii) suspicion of prostate cancer based on DRE, and (iii) abnormal hypoechoic lesions based on TRUS. This is a retrospective cohort study that was performed using a clinical database. The study was deemed exempt, and informed consent was waived by the Guangzhou First People’s Hospital for there was no active enrollment or active follow-up of study subjects, and no data were collected directly from individuals. All procedures performed were in accordance with the ethical standards of the 1964 Helsinki declaration.

The aim of our study was to determine the relationship between age, tPSA, and fPSA and PV in BPH patients. To minimize the probability of including patients with occult prostate cancer, cases with tPSA levels of >10 ng ml⁻¹ were excluded. Because BPH mainly occurs in middle-aged and elderly patients, people aged <40 years were excluded. In addition, those who had other diseases, were undergoing treatment, or had undergone surgical procedures that may have affected PV or PSA level were excluded. These included a history of acute prostatitis or urinary retention in the previous month of hospital visit, treatment with 5-alpha reductase, or prostatectomy. Patients with missing data, including age, tPSA, fPSA, or PV, were also excluded. A total of 5463 men met the inclusion criteria and were included in the study.

The following parameters were collected in all patients: age, tPSA, fPSA, and PV. DRE was performed by a qualified urologist to exclude malignancy or phlogosis. The serum levels of tPSA and fPSA were determined using the chemiluminescent microparticle immunoassay (CMIA) before prostatic manipulations such as DRE, TRUS, and biopsy.

Patients were divided into the PSA <2.5 ng ml⁻¹, 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ groups on the basis of PSA level. The relationships between age, tPSA, and fPSA and PV were analyzed using the Pearson/Spearman’s correlation coefficient. ROC curve analysis was used to analyze and compare the ability of serum tPSA and fPSA to predict PV ≥40 ml. Curves were constructed for PV cutoff values of 30 ml, 40 ml, and 50 ml against continuous tPSA and fPSA values in different PSA strata. Statistical analyses were performed using SPSS (version 18.0, SPSS Inc., Chicago, IL, USA). All statistical tests were two-sided, with P < 0.05 considered statistically significant.

RESULTS

The demographic features of the study cohort are shown in Table 1. Of the 5463 subjects, 2908 (53.2%) had a PSA level of <2.5 ng ml⁻¹, 805 (14.7%) had a PSA level of 2.5–3.9 ng ml⁻¹, and 1750 (32.0%) had a PSA level of 4.0–9.9 ng ml⁻¹.

Table 1.

Characteristics of the study cohort (n=5463)

Characteristic	PSA level (ng ml−1)
	0–9.9	<2.5	2.5–3.9	4.0–9.9	2.5–9.9
Participants (n)	5463	2908	805	1750	2555
Age (year), mean±s.d.	69.76±10.28	67.74±10.79	71.35±9.34	72.36±9.06	72.04±9.16
Total PSA (ng ml−1), mean±s.d.	3.15±2.64	1.12±0.64	3.21±0.44	6.51±1.69	5.47±2.09
Free PSA (ng ml−1), mean±s.d.	0.66±0.59	0.30±0.19	0.69±0.28	1.25±0.66	1.07±0.63
Prostate volume (ml), mean±s.d.	45.05±29.37	32.56±18.44	49.55±25.64	63.74±34.76	59.27±32.83

PSA: prostate-specific antigen; s.d.: standard deviation

In the whole patient population, PV significantly correlated with age, tPSA, and fPSA (r = 0.204, 0.598, and 0.646, respectively; all P < 0.001; Table 2). The correlation coefficient between tPSA and PV in men with PSA levels of <2.5 ng ml⁻¹ (r = 0.422; P < 0.001) was markedly stronger than those of men with PSA levels of 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ (r = 0.114, 0.167, and 0.264, respectively; all P ≤ 0.001), although the correlation coefficient of fPSA and PV did not differ significantly (r = 0.433 vs r = 0.403, 0.425, and 0.462, respectively; all P < 0.001), as shown in Table 2. Figure 1 shows the ROC curves for the prediction of PV ≥40 ml using tPSA and fPSA in the whole patient population and in various PSA-stratified cohorts. In the whole population and the cohort with PSA levels <2.5 ng ml⁻¹, the predictive value of fPSA was similar to that of tPSA (Figure 1a and 1b). In the 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ cohorts, fPSA was superior to tPSA in predicting PV (Figure 1c–1e).

Table 2.

Correlations between age, total prostate-specific antigen, free prostate-specific antigen, and prostate volume at different prostate-specific antigen levels

PSA level (ng ml−1)	Correlation
	Age-PV	P	tPSA-PV	P	fPSA-PV	P
0–9.9	0.204	<0.001	0.598	<0.001	0.646	<0.001
<2.5	0.136	<0.001	0.422	<0.001	0.433	<0.001
2.5–3.9	0.093	0.008	0.114	0.001	0.403	<0.001
4.0–9.9	0.077	0.001	0.167	<0.001	0.425	<0.001
2.5–9.9	0.086	<0.001	0.264	<0.001	0.462	<0.001

tPSA: total prostate-specific antigen; fPSA: free prostate-specific antigen; PV: prostate volume; PSA: prostate-specific antigen

Figure 1.

ROC curves for tPSA and fPSA to predict prostate volume ≥40 ml in different PSA groups: (a) 0–9.9 ng ml⁻¹, (b) <2.5 ng ml⁻¹, (c) 2.5–3.9 ng ml⁻¹, (d) 4.0–9.9 ng ml⁻¹, and (e) 2.5–9.9 ng ml⁻¹. ROC: receiver operating characteristic; AUC: area under ROC curve; tPSA: total prostate-specific antigen; fPSA: free prostate-specific antigen; PSA: prostate-specific antigen.

In the groups with PSA levels of 2.5–9.9 ng ml⁻¹, 2.5–3.9 ng ml⁻¹, and 4.0–9.9 ng ml⁻¹, the area under ROC curve (AUC) for fPSA as a continuous variable in predicting PV was larger than those for tPSA at the three PV cutoff points (30 ml, 40 ml, and 50 ml; Table 3). Thus, fPSA performed significantly better than tPSA in predicting PV thresholds.

Table 3.

Area under curve estimates for receiver operating characteristic curves predicting prostate volume of 30 ml, 40 ml, and 50 ml stratified according to prostate-specific antigen groups

AUC	PSA level (ng ml−1), PV=30 ml			PSA level (ng ml−1), PV=40 ml			PSA level (ng ml−1), PV=50 ml
	2.5–3.9	4.0–9.9	2.5–9.9	2.5–3.9	4.0–9.9	2.5–9.9	2.5–3.9	4.0–9.9	2.5–9.9
tPSA	0.571	0.525	0.588	0.545	0.561	0.611	0.556	0.583	0.636
fPSA	0.736	0.732	0.741	0.694	0.714	0.727	0.685	0.709	0.727

tPSA: total prostate-specific antigen; fPSA: free prostate-specific antigen; PV: prostate volume; PSA: prostate-specific antigen; AUC: area under ROC curve

The cutoff fPSA value for predicting PV of 30 ml, 40 ml, and 50 ml was determined as 0.418 ng ml⁻¹ (specificity: 76.6%, sensitivity: 73.3%), 0.505 ng ml⁻¹ (specificity: 75.3%, sensitivity: 76.3%), and 0.585 ng ml⁻¹ (specificity: 74.1%, sensitivity: 79.7%), respectively (Table 4).

Table 4.

Optimal serum free prostate‐specific antigen cutoff values to predict prostate volume (30 ml, 40 ml, and 50 ml) according to the receiver operating characteristic curves

Diagnostic test characteristic	PV=30 ml	PV=40 ml	PV=50 ml
fPSA level (ng ml-1)	0.418	0.505	0.585
Sensitivity (%)	73.3	76.3	79.7
Specificity (%)	76.6	75.3	74.1

fPSA: free prostate-specific antigen; PV: prostate volume

DISCUSSION

This study examined the relationship between patients’ age, tPSA, and fPSA and PV in a large PSA-stratified cohort of Chinese men. Notably, we found that the correlation coefficient between tPSA and PV in men with PSA levels of <2.5 ng ml⁻¹ (r = 0.422) was significantly higher than that of men with PSA levels of 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ (r = 0.114, 0.167, and 0.264, respectively). Furthermore, the AUC values for fPSA at predicting PV of ≥40 ml in men with PSA levels of 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ (AUC: 0.694, 0.714, and 0.727, respectively) were better than those of tPSA (AUC: 0.545, 0.561, and 0.611, respectively), but not in men with PSA level of <2.5 ng ml⁻¹ (AUC: 0.720 vs 0.713, respectively).

fPSA has been previously reported to perform significantly better than tPSA in predicting PV.^8,9,12 A study involving a European population found no differences in the efficacy of tPSA versus fPSA in predicting PV.13 A study by Kayikci et al.7 showed that the overall effect of fPSA in predicting PV was comparable to that of tPSA and superior to tPSA in patients with a PV of <40 ml. Mao et al.8 analyzed the correlation between fPSA and PV in PSA-stratified cohorts and found that the Pearson’s correlation coefficient in the cohort with PSA levels of 0–4 ng ml⁻¹ (r = 0.561; P < 0.001) was much higher than that in the cohort with PSA levels of 4–10 ng ml⁻¹ (r = 0.398; P < 0.001). In the present study, the correlation coefficient of fPSA and PV did not differ among PSA groups.

There are several potential reasons for the discrepancies between the present results and previous studies. First, the tPSA or fPSA-PV relationship might differ by race.14,15 Yu et al.16 found that enlarged prostate tissues from Chinese men have a higher glandular density than those of American men, while American men have a higher percentage of stromal tissues compared with Chinese men. We have previously showed that the percent fPSA does not improve the effectiveness of prostate cancer detection in Chinese men with PSA levels of 2.5–20.0 ng ml⁻¹.17

Second, the population in our study differs from previous studies on fPSA and PV because it involves biopsy-proven benign prostatic hyperplasia and “clinical BPH” cases with tPSA levels of <4 ng ml⁻¹, without abnormal findings from DRE or TRUS.13,14 Although the incidence of occult prostate cancer is possible, its chance is low, and therefore, it has a limited impact on the conclusion. Current data in the literature suggest that prostate cancer characteristics differ between Western and Asian countries and the incidence of prostate cancer in Chinese men with PSA level of 2.5–10.0 ng ml⁻¹ and 10.1–20.0 ng ml⁻¹ was much lower in Western men.18–21 It should be noted that the biopsy protocol used in our study is similar to the one used in the European randomized study for the screening of prostate cancer.

Third, compared with the level of tPSA, the level of fPSA is more dependent on the volume of benign prostate tissue. Although both PSA and PV increase with age, the rate of increase per decade is higher for PSA (35.9%) than for PV (12.4%).5 PV increase is mainly attributed to the enlargement of transitional zone nodules.22 In addition to PV, other factors may cause an increase in PSA, including tumors, prostatitis, or urinary retention. Probably, the lower the PSA, the less likely it is to be affected by other factors. Differences have also been noted in the cellular composition of BPH. Asian men have a higher glandular component and lower stromal component than white or black American men.15 Furthermore, PSA isoforms are differentially expressed in peripheral and transitional zone tissue.23 Given the lower fPSA to tPSA in patients with prostate cancer than in patients with BPH, it is believed that the serum levels of fPSA may be more dependent on the amount of prostate transitional zone or benign tissues. Thus, fPSA level may indicate PV more accurately than tPSA. However, this possibility should be confirmed using large population-based studies and further basic research on the pathophysiology of fPSA.

There are a number of limitations to this study, including its retrospective and single-center study design. Furthermore, although occult prostate cancer may occur, this may be partly mitigated by the study’s large cohort.

CONCLUSIONS

We conclude that the relationship between tPSA and PV might differ on the basis of varying PSA levels. The correlation coefficient of tPSA and PV in the cohort with PSA levels of <2.5 ng ml⁻¹ (r = 0.422) was significantly higher than that of cohorts with PSA levels of 2.5–3.9 ng ml⁻¹, 4.0–9.9 ng ml⁻¹, and 2.5–9.9 ng ml⁻¹ (r = 0.114, 0.167, and 0264, respectively). The level of fPSA was more powerful at predicting PV only in the ‘‘gray zone’’ (PSA level of 2.5–9.9 ng ml⁻¹), but its performance was similar to that of tPSA at PSA level of <2.5 ng ml⁻¹. The level of fPSA may predict PV, thereby guiding treatment decisions and longitudinal follow-up in Chinese men with BPH. However, this warrants further study for verification of the results.

AUTHOR CONTRIBUTIONS

PT conceived the study and reviewed the manuscript. MPH contributed to data acquisition, data analyses, manuscript writing, and revision. CSK read and revised the manuscript. XHW, WD, JGF, and THH assisted with data acquisition, data interpretation, and manuscript preparation. HC and KJX supervised the project and participated in its coordination. All authors read and approved the final manuscript.

COMPETING INTERESTS

All authors declare no competing interests.