Use of low free to total PSA ratio in prostate cancer screening: detection rates, clinical and pathological findings in Brazilian men with serum PSA levels <4.0 ng/mL
Eliney Ferreira Faria, Department of Surgery/Urology, Cancer Hospital Barretos, Alameda Nicaragua 252, Barretos, SP, Brazil. e-mail: firstname.lastname@example.org
Study Type – Diagnostic (inception cohort)
Level of Evidence 2b
What's known on the subject? and What does the study add?
In spite of its low specificity, PSA is the most widely used screening test for prostate cancer (PCa), and is considered the main cause of the stage migration recently observed. The ratio of free to total PSA (%fPSA) has been shown to increase PSA accuracy in cancer detection; however, few screening studies have systematically evaluated its role in cancer detection rates in men with PSA levels <4.0 ng/mL and normal DRE.
The present study supports a possible role of %fPSA as an adjunct to screening in men with total PSA 2.5–4.0 ng/mL and normal DRE, with a marked increase in cancer detection rates in a large Brazilian PCa screening study. We believe that %fPSA maybe a useful refinement to biopsy indications in men with low PSA levels.
- • To evaluate the role of the free to total prostate-specific antigen ratio (%fPSA) in identifying prostate cancer (PCa) in men with a prostate-specific antigen (PSA) level of 2.5–3.9 ng/mL and a normal digital rectal examination (DRE).
PATIENTS AND METHODS
- • A prospective PCa screening study was conducted, which included 17 571 men aged ≥45 years, across six Brazilian states, where men were recalled for further evaluation in the case of either a suspicious DRE and/or PSA ≥4.0 ng/mL, or PSA 2.5–3.9 ng/mL and %fPSA ≤15.
- • We evaluated the impact of a %fPSA ≤15 on cancer detection rates and the clinical and pathological stage of tumours in men with a normal DRE and PSA 2.5–3.9 ng/mL.
- • When suspicious DRE and/or PSA ≥4.0 ng/mL were considered as criteria to prompt further evaluation, the cancer detection rate was 3.1%. When %fPSA ≤15 in men with total PSA levels of 2.5–3.9 ng/mL were considered as criteria, the PCa detection rate increased to 3.7%. Considering %fPSA ≤15 in men with PSA 2.5–3.9 ng/mL and normal DRE, the positive predictive value of biopsy was 31.1%.
- • Clinical stage was more favourable among men with PSA 2.5–3.9 ng/mL, normal DRE, and %fPSA ≤15 compared with men with normal DRE and PSA ≥4.0 ng/mL (P= 0.02).
- • Among those who underwent radical prostatectomy, pathological stage and the proportion of insignificant tumours were similar between men with PSA 2.5–3.9 ng/mL, normal DRE findings and %fPSA ≤15, and men with PSA ≥4.0 ng/mL.
- • The use of %fPSA ≤15 as a biopsy indication in men with normal DRE and PSA 2.5–4.0 ng/mL in a PCa screening programme, increased cancer detection rates. Tumours in this subset of patients had similar pathological characteristics.
- • Using %fPSA ≤15 to indicate biopsy in men with PSA 2.5–3.9 ng/mL is a useful adjunct to PCa screening.
free to total PSA ratio
European Randomized Study on Screening of Prostate Cancer
positive predictive value
mobile cancer prevention unit
In 1986, PSA testing was approved for the monitoring of men with prostate cancer (PCa), and was recognized as a cancer detection test in 1994 by the US Food and Drug Administration. In spite of its low specificity, PSA is the most widely used screening test for PCa, and is considered to be the main cause of the stage migration and the decreased cancer specific mortality rates observed in countries where screening is carried out . The problems of overdetection and overtreatment are real but not yet well defined. The European Randomized Study on Screening of Prostate Cancer (ERSPC) trial has suggested that 1410 men had to be screened and 48 men had to be treated to avoid one death from PCa ; however, the Gottenborg branch of the ERSPC (with a younger population) reported that 293 men had to be screened and 12 men had to be treated to avoid one death . The ideal PSA threshold is still being debated. Important results from the PSA studies developed by Catalona et al. in the 1990s  have suggested that 2.5 ng/mL should be the PSA threshold, as a substantial proportion of men with PSA ≥2.5 ng/mL would convert to PSA ≥4.0 ng/mL in the subsequent years. In addition, positive predictive values (PPVs) of prostate biopsies were similar for men with PSA 2.5–4.0 ng/mL and men with PSA 4.0–10.0 ng/mL; a substantial number of PCa cases have recently been diagnosed in men within the so-called ‘normal’ range (PSA <4.0 ng/mL) . In fact, most surgical series report that these men usually have more organ-confined tumours, and superior recurrence-free and mortality rates than men diagnosed with PSA ≥4.0 ng/mL. However, some argue that lowering the PSA threshold may lead to overdiagnosis, increased risks associated with biopsies, and unacceptable cost increases to the health system, with questionable benefits. Furthermore, it would expose more men to other negative effects of screening, such as psychological distress . In attempts to improve PSA performance as an indication, many derivatives have been proposed, such as PSA density, age-specific PSA ranges, PSA velocity, free to total PSA ratio (%fPSA), complexed PSA, and other PSA isoforms, e.g. proPSA [5–9].
The free to total PSA ratio has been consistent in improving the accuracy of cancer detection when PSA levels are mildly elevated and may also help to identify more aggressive cancers [4,5]. Preliminary reports also suggest that %fPSA can provide increased specificity in cancer detection when PSA levels are <4.0 ng/mL . The ideal value of the %fPSA is still debatable; we chose 15 as a trigger value for biopsy in an attempt to optimize accuracy, while reducing unnecessary biopsies. In the present study, we explored the role of %fPSA in identifying PCa in men with PSA 2.5–3.9 ng/mL and DRE not suspicious for cancer in a large cohort of men undergoing PCa screening in Brazil .
PATIENTS AND METHODS
From January 2004 to December 2007, 17 571 men aged ≥45 years voluntarily underwent PCa screening at a mobile cancer prevention unit (MCPU). This programme included 231 municipalities across six Brazilian states. Each MCPU had a general physician, trained in performing PCa screening, and nursing personnel. After signing a consent form, participants responded to a standardized epidemiological questionnaire. We did not categorize men into different races, since it is very difficult to perform racial analyses in the Brazilian population in which racial miscegenation has occurred widely.
A blood sample was collected for total and free PSA measurement, and DRE was performed [10,11]. The same central laboratory processed all serum samples, using Hybritech® PSA assays (Beckman Coulter Inc., Fullerton, CA, USA) for total PSA and %fPSA. Men were recalled for further evaluation where there was a suspicious DRE and/or serum PSA ≥4.0 ng/mL. From November 2004, we started recalling for additional evaluation men with total PSA 2.5–3.9 ng/mL and %fPSA ≤15.
All prostate biopsies were provided through the public health system at our institution, which was the reference centre for the diagnosis and treatment of patients in the screening study. Biopsy protocol consisted of 12 sextant biopsy cores of the peripheral zone with two additional cores from the transition zone, and occasional sampling of suspicious areas on ultrasonography (e.g. hypoechoic areas in the peripheral zone).
Positive biopsies were graded using the Gleason score system. MRI of the abdomen and pelvis and bone scans were performed in all patients with cancer and staging was determined according to the TNM system. In men who underwent surgery, we used Epstein's criteria to define clinically insignificant cancers .
We evaluated the impact of %fPSA on cancer detection rates, histological characteristics and clinical stage of tumours in men with PSA 2.5–3.9 ng/mL, normal DRE findings and %fPSA ≤15. Analyses were performed using the software Statistical Package for Social Sciences, version 13.0, (SPSS®, Chicago, IL, USA).
The ethics committee at each participating hospital approved the trial.
Mean (sd; range) age at the first screening visit was 61.2 (9; 45–98) years, with 84.8% of men aged 50–74 years. LUTS were absent in 90.1% of the population. Overall, 71.4% of the screened men had never been submitted to a previous DRE, and 70.9% had never undergone PSA tests. Of the 17 571 men screened, 1647 underwent TRUS-guided biopsy, and PCa was diagnosed in 652 men. When only suspicious DRE and/or PSA ≥4.0 ng/mL were considered as criteria to prompt further evaluation, the cumulative cancer detection rate was 3.1%. With the inclusion of %fPSA ≤15 in men with total PSA levels of 2.5–3.9 ng/mL as criteria for further evaluation, the cumulative PCa detection rate increased to 3.7%. Of the men who underwent biopsy, when the indication for biopsy was PSA ≥4.0 ng/mL the PPV of the biopsy was 44.1%. For men who underwent biopsy because of abnormal DRE findings with normal PSA levels, the PPV was 23.5%. The PPV was greater for men with both abnormal DRE findings and PSA ≥4.0 ng/mL (70.9%). When we considered %fPSA ≤15 in men with PSA 2.5– 4.0 ng/mL and normal DRE findings, the PPV of the biopsy was 31.1% (Table 1). This PPV was somewhat lower than that found for men with PSA 4.0–10.0 ng/mL (37.6%).
Table 1. Biopsy PPVs according to total PSA, %fPSA and DRE
|PSA ≥4.0 ng/mL, %fPSA >15%, normal DRE||211/547||38.6|
|PSA ≥4.0 ng/mL, %fPSA ≤15%, normal DRE||188/357||52.7|
|PSA ≥2.5 and <4.0 ng/mL, %fPSA ≤15%, normal DRE||94/302||31.1|
|Suspicious DRE, PSA <4.0 ng/mL, %fPSA >15%||68/301||22.6|
|Suspicious DRE, PSA ≥4.0 ng/mL, %fPSA >15%||46/72||63.9|
|Suspicious DRE, PSA <4.0 ng/mL, %fPSA ≤15%||8/23||34.8|
|Suspicious DRE, PSA ≥4.0 ng/mL, %fPSA ≤15%||37/45||82.2|
|All biopsy parameters||652/1647||39.6|
The mean (sd) %fPSA was 12 (8), with a median (range) of 10 (1–68) in those with cancer, and was 15 (9), with a median (range) of 13 (1–96) in those with negative biopsy findings (P < 0.001, CI 95% [−3.86; −2.14], Student's t-test).
Of the 17 571 men initially screened, 2841 (16.1%) were recalled for further evaluation, of whom 484 (17.0%) presented with the biopsy criteria of PSA 2.5–3.9 ng/mL, normal DRE and %fPSA ≤15. Of these, 381 (78.7%) were reassessed and 302 (79.3%) underwent biopsy. Of the 302 men, 94 had PCa (31.1%). All of them had clinical stage T1cN0M0 disease. With regard to biopsy Gleason score, 72 men (76.6%) had Gleason score ≤6, 19 (20.2%) had Gleason score 7 and three (3.2%) had Gleason score ≥8 disease.
In the subset of men undergoing biopsy with PSA 2.5–3.9 ng/mL, normal DRE and %fPSA ≤15, we analysed the biopsy PPV for %fPSA ≤10 and the biopsy PPV for %fPSA 11–15. For %fPSA 11–15 we performed biopsies in 155 men and found 41 cases of PCa (PPV = 26.5%). Using a biopsy threshold of %fPSA of ≤10, we performed biopsies in 147 men and found 53 cases of PCa (PPV = 36.1%).
Clinical stage was more favourable among men undergoing biopsy with PSA 2.5–3.9 ng/mL, normal DRE findings and %fPSA ≤15, compared with men biopsied with normal DRE and PSA ≥4.0 ng/mL (P= 0.02; Table 2). There was also a trend towards lower biopsy Gleason scores among men with PSA 2.5–3.9 ng/mL, normal DRE findings and %fPSA ≤15 (P= 0.05; Table 2).
Table 2. Epidemiological, clinical and pathological characteristics of the men screened for PCa at the MCPU, according to serum PSA level
|Mean (sd) age, years||63.6 (6.9)||68.0 (8.3)||0.8|
|Final clinical stage, n (%)|| || || |
| I||94 (100)||362 (90.9)||0.02|
| II||0 (0.0)||3 (0.8)|
| III||0 (0.0)||9 (2.3)|
| IV||0 (0.0)||24 (6.0)|
|Biopsy Gleason score, n (%)|| || || |
| 6||72 (76.6)||257 (64.4)||0.05|
| 7||19 (20.2)||111 (27.8)|
| 8–10||3 (3.2)||31 (7.8)|
|Pathological stage, n (%)|| || || |
| pT2a||8 (17.5)||20 (27.7)||0.83|
| pT2b||10 (21.7)||15 (20.8)|
| pT2c||14 (30.4)||18 (25.0)|
| pT3a||14 (30.4)||18 (25.0)|
| pT3b||0 (0.0)||1 (1.5)|
Among those men who underwent radical prostatectomy, 33 presented with PSA 2.5–3.9 ng/mL and %fPSA ≤15, and 69 had PSA ≥4.0 ng/mL, all with a normal DRE. Among these, pathological findings were similar (Table 3). The proportion of insignificant tumours in men with PSA 2.5– 3.9 ng/mL and %fPSA ≤15 was 22.0%, compared with 23.4% among men with PSA ≥4.0 ng/mL [data not shown].
Table 3. Comparison of pathological findings in men with PSA of 2.5–3.9 ng/mL with normal DRE and men with a PSA of 4.0–10 ng/mL treated with radical prostatectomy (N= 102)
|Radical prostatectomy|| N= 33|| N= 69|| |
|Gleason score (specimen), n (%)|| || || |
| 2–6||22 (66.6)||39 (56.5)||0.40|
| 7||12 (33.4)||27 (39.1)|
| 8–10||0 (0.0)||3 (4.4)|
|Pathological stage, n (%)|| || || |
| pT2||28 (82.4)||52 (75.4)||0.29|
| pT3||6 (17.6)||17 (24.6)|
| pT4||0 (0.0)||0 (0.0)|
In spite of controversy about the effectiveness of PCa screening, some guidelines advocate lowering PSA thresholds . Few studies have systematically evaluated cancer detection rates in men with PSA levels <4.0 ng/mL and normal DRE [6,14–16]. Catalona et al.  were pioneers in performing biopsy in men with PSA 2.6–4.0 ng/mL because they had observed that more than a third of the men with such PSA levels would develop PSA >4.0 ng/mL after 4 years of screening. In their pivotal study, Catalona et al.  reported a PPV of 22% for men with non-suspicious DRE and PSA 2.6–4.0 ng/mL. Haese et al.  performed biopsies in 219 men with PSA 2.0–4.0 ng/mL, and reported a PPV of 18.7%. In the Prostate Cancer Prevention Trial  prostate biopsies were systematically performed in all men at the end of the study, and for men with PSA 2.1–4.0 ng/mL the biopsy PPV was 26.4%. Thus, the main argument against performing biopsies in men with a normal DRE and PSA 2.5–4.0 ng/mL is the relatively low PPV and the increase in the number of negative biopsies, which would be hard to justify in a population screening programme, in which most men are asymptomatic. In our study, we found more men with no symptoms, because the presence or not of prostatic symptoms was ascertained through simple questions (with yes or no answers), and we believe that affirmative responses were received mostly from men with moderate and severe symptoms.
The %fPSA has been established as a useful refinement to the indications for biopsy in men with normal DRE and PSA <10.0 ng/mL, although the ideal threshold is still debated. In a study of men with serum PSA 4.0–10.0 ng/mL, Partin et al. suggested using a %fPSA threshold of ≤15, which would detect all advanced, non-organ-confined and large volume tumours, while avoiding 80% of biopsies in men with insignificant disease. Subsequently, using a %fPSA threshold of ≤0.27, in biopsies of men with total PSA 2.6–4.0 ng/mL, Catalona et al.  were able to increase the PPV to 24%, while avoiding 18% of benign biopsies.
In the present study, the decision to perform a biopsy in men with PSA 2.5–3.9 ng/mL, normal DRE and %fPSA ≤15 resulted in a PPV of 31.1% in this subset of men. By indicating biopsy in this subset of men, we increased biopsy recommendation by 17%, while increasing cancer detection rates by 19.3% (from 3.1 to 3.7%). Although the PPV of the prostate biopsy was somewhat lower in men with PSA 2.5–3.9 ng/mL and %fPSA ≤15 than in men with PSA ≥4.0 ng/mL, overall the lower the %fPSA is, the higher the PPV will be. For instance, using a %fPSA ≤10, would have increased the biopsy PPV to 36.5% (data not shown). In fact, most screening studies published to date do not use this criterion, but in the present study its biopsy PPV was higher than that of an abnormal DRE (31.1% vs 23.5%).
Overall, 94 (14.4%) of the cancers detected in the present study would not have been diagnosed using only abnormal DRE and/or PSA ≥4.0 ng/mL as indications for biopsy. The cancers detected with lower PSA levels and %fPSA ≤15 were of lower clinical stage, with a trend towards lower Gleason scores; however, in those men who underwent radical prostatectomy, such cancers were not different from those of the men of PSA ≥4.0–10.0 ng/mL in terms of pathological stage and grade or the proportion of insignificant cancers. There was a proportional increase in the diagnosis of Gleason 7 cancers after surgery in comparison with pre-treatment biopsy Gleason score (65% increase in men with total PSA levels of 2.6–3.9 ng/mL and 40% increase for men with a PSA level of 4.0–10).
In spite of the fact that our biopsy PPV using a %fPSA ≤15 as a biopsy criterion in men with normal DRE and a total PSA 2.5–3.9 ng/mL was greater than that observed in the published literature, our findings have to be viewed in perspective, as we did not perform biopsies in all men with normal DRE and PSA 2.5 and 3.9 ng/mL. This is the main limitation of the present study and may have resulted in ascertainment bias. We acknowledge that the PPV found in the present study for men with PSA 2.5–3.9 ng/mL and %fPSA ≤15 was somewhat (but not much) higher that that observed in the Prostate Cancer Prevention Trial, in which all men underwent biopsy at the end of the study, for total PSA levels of 2.1 and 4.0 (31.1% vs 26.4%). However, we chose the threshold of %fPSA ≤15 based on the published literature regarding %fPSA in this total PSA range, in which this threshold was found to be the most useful in maximizing cancer detection rates while avoiding unnecessary biopsies.
In conclusion, in a large Brazilian PCa screening study, by adding the criterion of %fPSA ≤15 as a biopsy indication in men with normal DRE and PSA 2.5–3.9 ng/mL, we increased cancer detection rates. In this subset of patients, similar pathological characteristics were found in all tumours. We believe that %fPSA maybe a useful refinement to biopsy indications in screening men with PSA 2.5–3.9 ng/mL.
CONFLICT OF INTEREST