Detection rate of prostate cancer using prostate specific antigen in patients presenting with lower urinary tract symptoms: A retrospective studyPR Chavan1, SV Chavan1, NR Chavan1, VD Trivedi2
1 Department of Biochemistry, Lokmanya Tilak Municipal Medical College and Lokmanya Tilak Municipal General Hospital, Sion, Mumbai-400 022, India
2 Department of Urology, Lokmanya Tilak Municipal Medical College and Lokmanya Tilak Municipal General Hospital, Sion, Mumbai-400 022, India
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0022-3859.43548
Source of Support: None, Conflict of Interest: None
Background: Need for undertaking prostate biopsies for detection of prostate cancer is often decided on the basis of serum levels of prostate specific antigen (PSA). Aim: To evaluate the case detection rate of prostate cancer among patients presenting with lower urinary tract symptoms (LUTS) on the basis of PSA levels and to assess the scope of prostate biopsy in these patients. Setting and Design: A retrospective study from a tertiary care center. Materials and Methods: The clinical and histopathological data of 922 patients presenting with LUTS in the last five years was obtained from the medical record section. They had been screened for prostate cancer using PSA and /or digital rectal examination examination followed by confirmation with prostate biopsy. Statistical Analysis Used: Detection rate and receiver operating characteristic curve were performed using SPSS 16 and Medcalc softwares. Results: The detection rate of prostate cancer according to the PSA levels was 0.6%, 2.3%, 2.5%, 34.1% and 54.9% in the PSA range of 0-4, 4-10, 10-20, 20-50 and >50 ng/ml, respectively. Maximum prostate cancer cases were detected beyond a PSA value of 20 ng/ml whereas no significant difference in the detection rate was observed in the PSA range of 0-4, 4-10 and 10-20 ng/ml. Conclusion: A low detection rate of prostate cancer observed in the PSA range of 4-20 ng/ml in LUTS patients indicates the need for use of higher cutoff values of PSA in such cases. Therefore we recommend a cutoff of 20 ng/ml of PSA for evaluation of detection rate of prostate cancer among patients presenting with LUTS.
Keywords: Detection rate, lower urinary tract symptoms, prostate specific antigen, prostate cancer
Prostate cancer is the most commonly diagnosed cancer in men and is one of the leading causes of cancer-related deaths.  Many men with lower urinary tract symptoms (LUTS) are screened for prostate cancer with prostate specific antigen (PSA) testing and a digital rectal examination (DRE) as a part of a routine prostate assessment. There is general agreement among clinicians that the PSA test has the highest predictive value for prostate cancer as compared to DRE or trans-rectal ultrasound sonography (TRUS) alone. , Hence, patients with LUTS who have PSA levels higher than 4 ng/ml are advised to undergo prostate biopsy to rule out cancer. 
The PSA-based prostate cancer detection is fraught with high false-positive rate. Many confounding variables such as benign prostate hyperplasia, inflammation/ infection or traumatic maneuvers of the prostate gland influences serum PSA levels leading to many unnecessary biopsies.
Unlike Western countries, in India, the use of serum PSA in routine screening of men above 50 years age for prostate cancer is rare. However, this strategy is frequently used in the presence of LUTS. As data regarding the utility of serum PSA in detection rate of prostate cancer in asymptomatic or symptomatic men presenting with LUTS in the Indian population is still lacking, we undertook a study to evaluate the case detection rate of prostate cancer using PSA in patients presenting with LUTS and to assess the scope of prostate biopsy in patients presenting with abnormal PSA levels (>4 ng/ml).
In this retrospective study, clinical and histological data of patients presenting with LUTS attending the outpatient clinic at the tertiary teaching hospital over a period of five years was retrieved from the medical record section by one of the contributing authors. The ethical clearance was obtained from the Institutional ethics committee.
Inclusion and exclusion criteria
Men above 40 years of age presenting with LUTS specifically attributed to prostate problems were included in the study. From the registry the data collected showed 922 cases of LUTS over the period of five years. The data obtained included demographic details, clinical findings and assessment of LUTS as per the seven-item American Urological Association Symptoms Index (AUA- SI). , Responses to these seven items were summed up to produce an overall score (0-35). Standardized classification of symptoms severity as given by Barry et al.,  was applied to designate the symptoms as mild (0-7), moderate (8-19) or severe (> 20).
Men with LUTS caused by any urological malignancy other than prostate, those who had previous prostatic surgery or pelvic radiotherapy or complications of urinary obstruction (due to bladder stones, renal failure, recurrent urinary tract infection or residual urinary volumes >200 ml) were excluded from the study.
Serum PSA and DRE
Reports of serum PSA levels by ELISA (Enzyme Linked Immunosorbent Assay kit, Diagnostic System Laboratories)and findings of systematic DRE performed by the attending urologist were noted. The prostate was regarded as abnormal on DRE, if there was evidence of nodularity, indurations, asymmetry or absence of the median sulcus. As a routine practice DRE examination was scheduled after collection of blood sample to avoid an increase in serum PSA that may follow digital manipulation of the gland. PSA levels less than 4ng/ml were considered as normal, those between 4-10 ng/ml as diagnostic gray zone and above 10 ng/ml as indicative of cancer. Other routine investigations like renal profile, blood sugar, complete blood count, erythrocyte sedimentation rate, urine routine and culture were also performed.
The results of prostate biopsies that were performed on patients with abnormal DRE examination and/ or serum PSA levels >4 ng/ml were also noted. As per the then institutional practice, the prostate was mapped by systematic sextant TRUS-guided biopsies using an 18-gauge True-Cut, 20-cm long needle driven by a spring-loaded gun (Magnum Bard Biopsy Instruments) with a cutting length of 22 mm. The biopsies were taken at the apex, middle and base of the right and left lobes in the para-sagittal plane. If a hypoechoic lesion was detected in the peripheral or central zone on TRUS, additional three to six biopsies of that area were also taken in order to avoid the exclusion of carcinoma outside this locus. Since patients with prostatitis presenting with high PSA levels may show a significant decline after six weeks of antibiotic treatment, serum PSA levels were re-ordered in such patients. Subsequently, patients with sustained high PSA levels after the follow-up of six weeks were subjected to a confirmatory biopsy as deemed necessary.
A team of histopathologists including a dedicated genitourinary pathologist of senior faculty level reviewed all biopsies.
Based on clinical and histopathogical diagnoses the patients presenting with LUTS were categorized into two groups: Group I, Prostate cancer and Group II, non-malignant group (benign prostate hyperplasia with or without prostatitis). In specimens diagnosed as prostate cancer additional review was performed by the Gleason Score. 
Data were analyzed using the statistical software packages SPSS 16 and Medcalc for windows.
Detection rate was calculated as percentage of cases with prostate cancer among patients presenting with LUTS. For comparison of serum PSA levels between non-malignant and malignant group unpaired t test was used. Receiver operating characteristic (ROC) curve analysis was performed for assessing the sensitivity, specificity, positive predictive value and negative predictive value at various serum PSA cutoff values for detection of prostate cancer. P <0.05 was considered significant.
In this study, 922 patients (age range of 40-95 years) presenting with LUTS were screened for prostate malignancy. The demographic features are mentioned in [Table 1]. Based on abnormal DRE findings and/ or high PSA levels, 440 patients underwent prostate biopsy. Of these cases, 80 cases were diagnosed to have prostate cancer, the rest showed non-malignant conditions (n=842). Thus, the detection rate of prostate cancer among LUTS cases observed was 8.7% irrespective of their PSA values.
DRE examination revealed abnormalities in 158 (17.1%) subjects. The relationship between the number of subjects with abnormal DRE findings and their PSA levels is shown in [Table 2]. As the PSA levels increased the number of prostate cancer cases detected by DRE also increased.
The mean and standard deviation of PSA levels observed among the malignant and non-malignant group were 267.9± 558.1 and 14.5± 54.4 respectively. On comparison, the PSA levels observed in the prostate cancer group were significantly (t=12.6, df=920 P = <0.001) higher than the non-malignant group. Mean difference observed was -253± 20.08 with 95% confidence interval of -292.6 to -213.9.
[Table 2] shows the detection of prostate cancer at various PSA ranges. It can be seen that the percentage of subjects diagnosed to have prostate cancer increased with rising levels of PSA. However, the detection rate of prostate cancer observed in the PSA range of 0-4 (0.61%), 4-10 (2.34%) and 10-20 (2.54%) ng/ml remains almost similar. The maximum number of prostate cancer cases was detected beyond 20 ng/ml of PSA level.
We performed ROC curve analysis in patients with malignant and non-malignant conditions for the complete serum PSA range (0.01-3849 ng/ml). When the entire PSA range was considered, area under curve observed was 0.938 ( P <0.001) with 95% confidence interval of 0.910- 0.965 [Figure 1]. The diagnostic validity criteria such as the sensitivity, specificity at different decision limits of the ROC curve is shown in [Table 3]. As can be seen from this Table, with the upward trend of PSA cutoff value the specificity showed significant improvement whereas the sensitivity showed marginal decline.
In this study, according to histological findings, Gleason score observed in the prostate cancer patients was in the range of 2-9. No malignant case was observed with Gleason Score 1 and 10. The distribution of patients was 2.5%, 5%, 11.3%, 12.5%, 18.8%, 31.3%, 12.5%, and 6.3% in Gleason Score 2, 3, 4, 5, 6, 7, 8 and 9, respectively.
The low incidence of prostate cancer, lack of awareness about the disease and availability of limited resources preclude institution of a screening program for its detection in a country like India. In addition, recommendations regarding the appropriate level of PSA for undertaking diagnostic biopsy have been changing over time. Our study has established that the cancer detection rates at PSA levels 0-4, 4-10 and 10-20 ng/ml are low and do not vary significantly. The study findings also indicate that a serum PSA cutoff value of over 20 ng/ml provides optimal levels of sensitivity, specificity and positive and negative predictive values for detection of prostatic cancer in patients with LUTS.
It is well established that with the increase in serum PSA levels the detection rate of prostate cancer also increases. The probability of detecting prostate cancer in the range of 4-10 ng/ml is around 25% which increases to more than 30% with serum PSA level above 10 ng/ml. , Our analysis demonstrated detection rate of 0.6%, 2.3%,2.5% and 34.1% in the PSA range of <4,4-10,10-20 and 20-50 ng/ml. According to international literature American men have higher detection rate of prostate cancer than Asian men. ,, In Korean men the prostate cancer detection rate observed with PSA <4 ng/ml or between 4-10ng/ml was 13.8% and 15.95.  This is similar to the 15.8% rate in Japanese men reported by Egawa et al .,  Our study demonstrated a still lower detection rate of prostate cancer as compared to our Asian counterparts i.e. the Koreans or Japanese. Worldwide the incidence of prostate cancer is observed to be varying widely, with Americans showing the highest incidence and Asian men showing the lowest incidence. This variation in the incidence may in part be explained by dietary habits and the racial differences. Also, the high detection rates observed in the Western population may be contributed to an intense screening procedure. In the same vein, in the Indian scenario, lack of awareness, education and facilities for routine health checkups seem to be the main factors contributing to the low detection rates reported hitherto.
A report about Korean men demonstrated that a serum PSA level greater than or equal to 20 ng/ml, independent of DRE findings was 81% accurate in predicting the presence of prostate cancer on initial biopsy.  Similarly, studies by Gerstenbluth et al., demonstrated that a serum PSA level of 20 ng/ml or greater, independent of DRE findings was 87.2% accurate in predicting the cancer and a serum PSA level of 50 ng/ml or greater had a positive predictive value of 98.5%.  Our data showed 80.6% accuracy in detection of prostate cancer at the PSA cutoff of 10 ng/ml whereas at 20 ng/ml 91.3% specificity was observed. The need for detection rate and the accuracy of cutoff value is important in counseling the patients for the chances of positive finding prior to the biopsy.
It appears that the recently advocated PSA cutoff value of 2.5 ng/ml for recommending prostate biopsies would result in many Indian patients unnecessarily undergoing the procedure.
Although the prostate cancer detection rate of 8.7% observed in our study is far higher than those reported in other surveys, , it is lower than the rate of 14.6% reported by Cooner et al .,  in clinical urological practice using PSA, DRE and prostate ultrasonography. The higher rate in our study could be attributed to the fact that ours was a "super-selected" set of symptomatic patients attending a referral urological center, while the surveys were carried out in population-based screening programs or hospital-based screening programs. A multi-centric study in Korean men reported a detection rate of 32.7% of the 4967 men undergoing biopsy for detection of prostate cancer. Further interpretation revealed that 19.6% of subjects with PSA levels between 4 and 9.9 ng/ml were diagnosed with cancer, whereas 53.7% were diagnosed at PSA levels greater than 10 ng/ml. 
Histologically acute and chronic-active prostatitis is considered an important factor for inducing the high increase in serum PSA values via the leak phenomenon,  which causes confusion in the diagnosis of prostate cancer.  However, patients with high PSA levels showing significant decline in their PSA level after six weeks of antibiotic treatment for prostatitis is often considered as confirmation for absence of prostate cancer.
Our study is limited by the following factors. Firstly, being a retrospective study use of more specific forms of PSA such as PSA velocity, PSA density, age-specific PSA reference range, the measurement of free - PSA%, free to total PSA ratio, and complexed PSA which are being used lately could not be utilized. In the event that these were adopted, the positive predictive value of the PSA test would have further increased thus altering the scope of biopsy in case detection. Yet till date basic serum PSA levels determination remains the test of choice widely available and universally accepted not only in many centers in India but all over the world. Despite all these our study becomes relevant specifically as LUTS patients shows higher serum PSA levels in whom the standard normal range of 0-4 ng/ml cannot be applied. This necessitates a higher cutoff level (at 20 ng/ml), which shows more accurate detection rate for prostate cancer.
The study could also be scrutinized critically for using PSA range of 0-4 ng/ml, when the literature is recommending 2.5 ng/ml as the cutoff level. It may be stated that most Indian centers still use 4 ng/ml as the cut-off level and convincing Indian data that would support the cutoff level of 2.5 ng/ml is still lacking. As can be made out from the findings, even the adoption of a cutoff level of 2.5 ng/ml would not have affected the outcome of our study. The third limitation of our study points to the use of sextant biopsies instead of the current norm of 10-12 core biopsies. We were limited in this regard by the retrospective nature of the study design dating back to as early as 2001 wherein the practice of 12 core biopsies was still not in vogue. The execution of a more extensive biopsy regimen would have positively impacted the case detection rate.
Lastly, the retrospective nature of the study once again eliminated the scope for any longitudinal follow-up. The follow-up could have allowed for serial determination of PSA estimations and enhanced case detection. Despite the above mentioned limitations, the detection rate of prostate cancer reported in this study offers a compelling and stimulating insight into the need for more stringent and thorough investigation of all cases of LUTS in order to maximize detection. The cutoff value of PSA recommended i.e. 20 ng/ml in this study is particularly for a select group of patients presenting with LUTS attending the urology out patient department and not for the general population or for screening purpose. In conclusion the applicability of the proposed PSA cutoff for detection of prostate cancer in patients presenting with LUTS needs to be prospectively studied using more sample size of such patients.
[Table 1], [Table 2], [Table 3]