Abstract
Background: To investigate the cumulative probability of developing prostate cancer according to prostate-specific antigen (PSA) velocity (PSAV) from first-to second-round PSA-based population screening in men with low baseline serum PSA levels. Patients and Methods: A total of 11,913 men aged between 54 and 69 years with baseline PSA levels of ≤2.0 ng/ml at the first population screening and who underwent population screening at least twice, were enrolled. The cumulative probability of developing prostate cancer according to age, baseline PSA and PSAV was investigated. The clinicopathological features of screen-detected cancer were also investigated. Results: Out of the 11,913 men, 110 (0.92%) were pathologically diagnosed with prostate cancer during the observation period. The cumulative probability of developing prostate cancer in all participants after 5 and 10 years was 0.64% and 1.79%, respectively. Univariate and multivariate analyses determined that baseline PSA levels and PSAVs were significant predictors of developing cancer and the hazard ratio increased with increasing baseline PSA levels and PSAVs. The optimal PSAV cut-off levels for prostate cancer development were 0.069, 0.106 and 0.285 for the baseline PSA ranges of 0.0-1.0, 1.1-1.5 and 1.6-2.0 ng/ml, respectively. There were no significant differences in baseline PSA levels and PSAVs according to the clinical characteristics of the screen-detected prostate cancer patients. Conclusion: The present study demonstrated that serum PSA levels at second round screening were a strong predictor of cancer development in men with baseline PSA levels ≤2.0 ng/ml at the first population screening.
The European randomized study of screening for prostate cancer (ERSPC) (1) and the Göteborg prospective randomized study (2) demonstrated that prostate-specific antigen (PSA)-based screening improved the prostate cancer-specific survival rate of men aged 50-69 years. The American Urological Association recently recommended shared decision-making for men aged 55-69 years who were considering PSA-based screening (3). As previously reported, the high rate of PSA testing among middle-aged men may have partially contributed to the continuous decrease in the prostate cancer mortality rate in various countries (4).
In a screening program, it is necessary to establish an optimal screening system that maximizes the benefits of PSA-based screening in terms of mortality reduction and cost-effectiveness while minimizing the drawbacks of screening, such as the likelihood of over-detection. To achieve this, setting individualized screening protocols, including screening intervals, cut-offs for biopsy indication and an upper limit of age for screening may reduce the cost of screening as well as the likelihood of overdetection and false-positive PSA test results, while reducing mortality. The Kanazawa population-based screening cohort is a database that links individual screening results with the medical information of screen-detected cancer patients. Our database may contribute baseline information regarding how to establish individualized screening systems in the future.
We have previously reported the cumulative probability of prostate cancer development in men with serum PSA levels of ≤2.0 ng/ml during their first population screening and demonstrated that the risk of developing cancer significantly increased with higher baseline PSA ranges (5). In the present study, we investigated the clinical significance of serum PSA levels at the second-round screening in men with baseline PSA levels of ≤2.0 ng/ml.
Patients and Methods
Study population. Since 2000, PSA-based annual population screening for prostate cancer has been provided for men aged 54-69 years in Kanazawa City, Japan (6, 7). Serum total PSA (tPSA) levels were measured in all participants using a Tosoh II PA kit (Tosoh, Tokyo, Japan) as the primary screening modality. Participants with serum tPSA levels of ≤2.0 ng/ml did not receive additional examinations and were recommended to undergo annual screening. Serum free PSA (fPSA) levels were measured in participants with tPSA levels of 2.1–10.0 ng/ml using an Immulyze Free PSA kit (Nippon DPC Co. Ltd., Chiba, Japan). From 2000 to 2002, participants with serum tPSA levels of >2.0 ng/ml were recommended to undergo further examination by urologists at primary medical care clinics. After 2003, participants with serum tPSA of >10.0 ng/ml and those with a f/t PSA ratio ≤0.22 within the PSA reflex range of 2.1-10.0 ng/ml were recommended to undergo closer examination (6, 7). Since 2012, the age range of participants has expanded to 54-75 years.
PSA levels were measured again in all men who underwent a closer examination, including both digital rectal examination (DRE) and transrectal ultrasonography (TRUS). Systematic TRUS-guided prostate biopsy (6-12 cores) was recommended in men with any abnormal findings on rechecking PSA, DRE or TRUS. If individuals refused to undergo prostate biopsy or were not recommended to undergo biopsy by urologists, they were followed-up using annual PSA tests during subsequent population-based screening.
In patients diagnosed with prostate cancer, local pathologists reported pathological tumor grading. In addition, clinical staging was determined according to the Union for International Cancer Control tumor node metastasis classification published in 1997 (8) based on DRE, TRUS, computed tomography, magnetic resonance imaging and bone scan results at each urological department. Medical information such as PSA levels at the primary screening, biopsy results and clinicopathological findings were reported to the office of Kanazawa Medical Association.
In the Kanazawa population-based cohort, 22,252 men participated in the 14-year screening program between 2000 and 2013, and 18,464 (83.0%) had serum PSA levels of ≤2.0 ng/ml at the first screening. Of these, 11,913 (64.5%) participants aged 54-69 years who were annually screened at least twice during the observation period were enrolled in the present study. PSA velocity (PSAV) was calculated as follows: (PSA [second round] − PSA [first round])/the number of years between PSA tests. The cumulative risk of prostate cancer development according to age, baseline serum PSA levels at first screening and PSAV was then assessed. The time to prostate cancer detection was calculated from the date of the first population screening and participants without evidence of cancer were censored at the date of the last population screening.
We also investigated the clinicopathological features of screen-detected prostate cancer stratified into favorable or unfavorable cancer according to the definition of active surveillance studies (9, 10). We considered prostate cancer with clinical stage T1cN0M0, PSA of ≤10 ng/ml and a Gleason score of ≤6 as clinically favorable cancer. This definition was partially consistent with the patients who would not benefit from radical prostatectomy according to a European randomized trial (11).
Statistical analysis. This retrospective study followed the principles of the Declaration of Helsinki. Comparisons between two groups were performed using the Mann–Whitney U tests or Fisher's exact tests. The cumulative rate of prostate cancer detection was analyzed using univariate and multivariate Cox proportional hazards regression models according to age, baseline serum PSA levels and PSAVs. Patient age at first screening was divided into three categories: 54-59, 60-64 and 65-69 years. Baseline serum PSA readings were divided into categories of 0-1.0, 1.1-1.5 and 1.6-2.0 ng/ml. The risk associated with PSAV was analyzed as quartiles calculated according to the distribution among the patient cohort. The PSAV cut-off for the detection of prostate cancer was investigated by analyzing the receiver operating characteristic (ROC) curves. The probability of prostate cancer detection according to PSAV was examined using the Kaplan–Meier analysis and the significance of the differences was analyzed using log-rank tests. All statistical assessments were performed and the figures were prepared using commercially available software (SPSS Statistics: IBM; Armonk, NY, USA and Prism: GraphPad Software; San Diego, CA, USA). In all analyses, p<0.05 indicated statistical significance.
Results
The population screening results of men eligible for this study are shown in Table I. Out of the 11,913 men enrolled in this study, the number of participants with baseline serum PSA levels of 0.0-1.0, 1.1-1.5 and 1.6-2.0 ng/ml at the first screening were 8,086 (67.9%), 2,523 (21.2%) and 1,304 (10.9%), respectively. The serum PSA levels of 1,728 men (14.5%) subsequently increased to >2.0 ng/ml during the observation period and 1,184 (9.94%) underwent closer examination in a urology department. Out of the 1,728 men whose serum PSA level increased to >2.0 ng/ml, 409 (23.7%) underwent prostate biopsy and 110 (0.92% of all participants and 6.37% the individuals whose serum PSA increased to >2.0 ng/ml) were diagnosed pathologically with prostate cancer during the observation period. The PSAV values between the first and second round screening according to baseline serum PSA range are presented in Table I.
The correlation between the first and second round screening results of prostate cancer patients and participants without cancer is shown in Table II. In terms of prostate cancer detection, there were no statistically significant differences in the age at first population screening between prostate cancer patients and participants without prostate cancer. Baseline serum PSA levels at first screening, serum PSA levels at second-round screening, the duration between the first and second screening and PSAV from the first to second round screening were significantly higher in the prostate cancer patients than in participants without cancer. The observation period was longer in the prostate cancer patients; however, the number of population screenings was comparable between the two groups.
The cumulative probability of prostate cancer detection in all participants after 5 and 10 years was 0.64% (95% confidence interval (CI)=0.46-0.82) and 1.79% (95% CI=1.37-2.22), respectively. Table III shows the results of univariate and multivariate Cox proportional hazards regression analysis for time to prostate cancer detection based on age, baseline serum PSA levels and PSAVs from the first to the second-round screening. Most of the participants were within the baseline serum PSA range of 0.0–1.0 ng/ml (67.9%) and had a PSAV of ≤0.10 ng/ml/year (70.2%). According to univariate and multivariate analyses, baseline serum PSA levels and PSAVs were significant predictors of prostate cancer development and the hazard ratio increased with increasing baseline serum PSA levels and PSAVs.
The optimal PSAV determined from the ROC curves and the corresponding areas under the ROC curves on the basis of baseline serum PSA levels are shown in Table IV. The optimal PSAV cut-off values for prostate cancer development were 0.069, 0.106 and 0.285 for the baseline PSA ranges of 0.0-1.0, 1.1-1.5 and 1.6-2.0 ng/ml, respectively.
Kaplan–Meier curves of the cumulative probability of prostate cancer development according to PSAV are shown in Figure 1. The probability of developing prostate cancer in men with PSAVs of ≤0.10, 0.101-0.289, 0.290-0.499 and ≥0.50 ng/ml/year was 0.15%, 0.63%, 2.41% and 6.18% after 5 years, and 0.67%, 1.71%, 10.59% and 8.65% after 10 years, respectively (Figure 1A). There were statistically significant differences in the probability of prostate cancer development between individuals with PSAVs of ≤0.10, 0.101-0.289 and 0.290-0.499. However, no statistically significant difference between men with PSAVs of 0.290-0.499 and ≥0.50 was observed. On the basis of these findings and the PSAV cut-offs determined from the ROC curves, Kaplan–Meier analyses was performed on three stratified groups: PSAVs of ≤0.10, 0.101-0.289 and ≥0.29 (Figure 1B–D). In men with baseline PSA levels of 0.0-1.0 ng/ml and 1.1-1.5 ng/ml the probability of developing prostate cancer was significantly lower in individuals with a PSAV of ≤0.10 than in those with PSAV of 0.101-0.289 ng/ml/year. However, there was no significant difference between subjects with PSAVs of 0.101-0.289 and ≥0.29 ng/ml/year (Figures 1B and C). In individuals with baseline PSA levels of 1.6-2.0 ng/ml the probability of developing prostate cancer was significantly higher in those with a PSAV of ≥0.29 compared with those with a PSAV of 0.101-0.289 ng/ml/year. However, there was no significant difference between individuals with PSAVs of 0.101-0.289 and ≤0.10 ng/ml/year (Figure 1D).
The clinical characteristics of the prostate cancer patients in the present study are shown in Table V. Among the 110 patients, 4 (3.6%) had PSA levels of ≥10.1 ng/ml, 2 (1.8%) had locally advanced cancer and 2 (1.8%) had metastatic disease. There was no statistically significant correlation between PSAV and Gleason score in biopsy specimens (Gleason score ≤6 vs. ≥7, p=0.0593; Gleason score ≤7 vs. ≥8, p=0.1258; Mann–Whitney U-test). A total of 37 (33.6%) patients had favorable clinicopathological features that made them suitable according to the above-mentioned definition in the present study. Table VI shows the population screening findings of the 108 patients defined as favorable or unfavorable cancer patients in the present study. There were no significant differences in baseline PSA levels at first screening, PSAV and the duration from screening to cancer detection between the favorable and unfavorable cancer patients (Table VI).
Discussion
PSA-based population screening was found to benefit prostate cancer-specific survival in previous reliable randomized studies (1, 2). However, a standard screening system that maximizes mortality reduction and cost-effectiveness while minimizing the drawbacks of screening such as over-detection, subsequent over-treatment and adverse effects on quality of life has not yet been established. To establish an optimal early detection program, the PSA test should be investigated thoroughly in terms of its potential as an objective risk factor, the setting of optimal cut-offs for biopsy indication and screening intervals, as well as the usefulness of individual PSA kinetics for evaluating tumor aggressiveness. The Kanazawa population-based screening cohort is now prospectively establishing a database that includes individual medical information, screening results, subsequent biopsy results and the clinicopathological features of screening-detected prostate cancer. In the present study, we investigated PSAV and the risk of developing prostate cancer in men with low baseline PSA levels at the first population screening and then explored individual PSA kinetics as a risk factor using the Kanazawa population-based screening cohort.
Several studies, including one from our laboratory, demonstrated the cumulative probability of developing prostate cancer in subsequent screening in men with baseline serum PSA levels below the cut-off value and revealed that the risk of cancer development increased significantly with higher baseline PSA ranges (5, 12-15). In individuals with a serum PSA of <1.0 ng/ml, the risk of being diagnosed with cancer in the next 4-7 years after initial screening is very low (<1.0%). Based on these results, individualized screening intervals for men with baseline serum PSA levels of ≤1.0 ng/ml are recommended in current clinical guidelines for prostate cancer (8 years in the European Association of Urology guidelines, 3 years in the Japanese Urological Association guidelines) (16, 17). However, the participants enrolled in these previous studies did not undergo annual screening, while individual PSA kinetic factors including PSAV were not monitored during the 2- to 4-year screening intervals. The present study revealed the natural history of participants screened annually in the baseline PSA reflex range of 0.0-2.0 ng/ml during a median observation period of 4.61 years (1,683 days).
The current study demonstrated that both baseline serum PSA levels and PSAVs from the first to second round screening were independent risk factors for prostate cancer development. Only a small number of studies have assessed the distribution of PSAV in a population-based setting in men with low baseline serum PSA levels. Loeb et al. demonstrated that PSAV was significantly higher in prostate cancer patients than control individuals with PSA levels of <3 ng/ml. The mean PSAV was 0.02 ng/ml/year in control individuals with PSA levels of <3 ng/ml in the Baltimore Longitudinal Study of Aging (BLSA) (18). In the present study, the mean PSAV was 0.07 ng/ml/year in participants with baseline serum PSA levels of ≤2.0 ng/ml and PSAV was 0.100 ng/mlL/year in approximately 70% of the participants. It was notable that PSAV continuously increased with tPSA in the BLSA study (18), which is consistent with our results that the PSAV cut-offs determined from the ROC curves increased in a baseline serum PSA level-dependent manner (0.069 ng/ml/year in individuals with a PSA of 0.0-1.0 ng/ml, 0.106 ng/ml/year with a PSA of 1.1-1.5 ng/ml and 0.285 ng/ml/year in PSA of 1.6-2.0 ng/ml). These findings suggest that the PSAV cut-off levels that were previously associated with cancer detection in prostate biopsies (0.4 ng/ml/year (19)) may not be useful when screening participants with baseline PSA levels of ≤2.0 ng/ml.
In contrast, unfavorable cancer clinicopathological features were not related to PSAV in prostate cancer patients in the present study. Moreover, considering the very low probability of developing prostate cancer in men with baseline serum PSA levels of ≤2.0 ng/ml, particularly those of ≤1.0 ng/ml, a high PSAV alone is not an indication for prostate biopsy. These observations are consistent with a previous study in which the incidence of prostate cancer in men with a low PSA and a high PSAV was very low (20). The results of the current study may support the authors' recommendation that men with a PSA level below the biopsy threshold but a high PSAV should not undergo immediate prostate biopsy but should undergo PSA testing at shorter intervals (21). In the clinical setting, physicians could define individual screening intervals according to the serum PSA values obtained during a second round screening in men with baseline serum PSA levels of ≤2.0 ng/ml based on our present results. Men with baseline serum PSA levels of ≤1.0 ng/ml and 1.1-2.0 ng/ml should undergo a second round PSA testing after 2 and 1 years, respectively. If PSAVs calculated from two serum PSA values are below the cut-off levels, they should undergo their next PSA test with a longer interval due to the very low cumulative risk for developing prostate cancer demonstrated in the present study. The regression model used to determine screening intervals for participants with low baseline serum PSA levels should be established after additional observations, which will define optimal individual screening intervals. However, serum PSA levels at the second-round screening may simply indicate the individual interval from second to third round PSA screening.
There is an increasing awareness of the potential for over-diagnosis and over-treatment for “insignificant” or “minimal” prostate cancer detected using PSA-based population screening, particularly with the low serum PSA cut-off levels in the screening system. When the clinicopathological features of prostate cancer in men with baseline serum PSA levels of ≤2.0 ng/ml in the present study cohort are considered, a small number of patients had advanced stage prostate cancer, including clinical stage T3 or higher (1.8%) and serum PSA levels of ≥10.1 ng/ml at diagnosis (3.6%). However, a relatively high percentage of the patients (64.5%) had unfavorable clinicopathological features, which excluded the criteria in similar active surveillance studies (9, 10). These findings suggest that continuous population-based screening in men with baseline PSA levels below biopsy thresholds is clinically significant.
The present study has several limitations based on the retrospective nature of the analysis in the screening cohort. First, the cancer detection rates may be underestimated because not all subjects who required closer examination underwent prostate biopsy. In the present study, the low biopsy rate (23.7%; 409 biopsies from 1728 cases eligible for biopsy according to the protocol) could be a significant bias. The lack of central pathologists and revisions to the Gleason grading system (the 2005 updated International Society of Urological Pathology guidelines) (23) during the study period also caused potential bias. Second, there was a possible bias for the time from the first to the second round screening because of the irregular rescreening intervals. Third, the increase in non-cancer-related PSA levels due to inflammation (23) could not be excluded in the present study because PSAV was calculated from two serum PSA measurements. Therefore, further prospective large-scale studies using fixed rescreening intervals are needed. However, in the present study, we successfully demonstrated the usefulness of PSAV from the first to second round screening as a predictive factor for prostate cancer development in men with baseline serum PSA levels of ≤2.0 ng/ml. This contributes to the establishment of individualized and natural history-adjusted screening systems. Serum PSA levels at second round screening may help determine the individual intervals to the next PSA test in men with baseline serum PSA levels below the threshold for biopsy. The goal of the Kanazawa population-based screening cohort is to facilitate the establishment of future individualized and natural history-adjusted screening systems.
Acknowledgements
We thank Dr. Yasuo Takeda and Dr. Atsushi Hashiba of the Kanazawa Medical Association for providing data from the Kanazawa population-based screening cohort.
Footnotes
-
Conflicts of Interest
There are no conflicts of interest to declare.
- Received July 11, 2014.
- Revision received July 29, 2014.
- Accepted July 30, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved