ACCURACY OF PSA TEST IN THE PROSTATE CANCER SCREENING PROGRAM OF LITHUANIA

ACCURACY OF PSA TEST IN THE PROSTATE CANCER SCREENING PROGRAM OF LITHUANIA Romualdas Gurevičius 1, Renata Šturienė 2, Arvydas Šilys 3 1 Institute of Hygiene, 2 Vilnius Šeškinė Outpatient Clinics, 3 Vilnius University Clinics Summary In Lithuania, prostate cancer accounts more than a quarter (29%) of newly diagnosed cases of cancer in males. Since 2006 Prostate cancer screening programme in Lithuania began inviting, asymptomatic men aged 50-70 for a PSA test, which uses a cut-off value of greater than 3ng/ml. Aim of the study. To evaluate PSA test accuracy of a clearly defined screened population with application different statistical indicators. Material and methods. The outpatient clinic in which investigation was carried out serves 97,000 of Vilnius city residents, including 45 thousand men, of which about 11 thousand are 50-75 years of age. During the period of 2008-2010 information on all screened males (N=8810) were collected, covering 80% of the target population. Results and conclusions. In our study, the sensitivity of the PSA test was 74.8% (95% CI: 70-79.1), the specificity of the PSA test was 90.7% (95% C90.1-91.3). False positive rate shows 9,3 percent of the participants in our study did not have prostate cancer but the test was positive. False negative rate indicates, that 25.2 percent of our participants had prostate cancer whiel th test was negative. Consequently, quarter of participants with false-negative results were given unfounded reassurance and 9.3% of particpants with false-positive results might have experienced unnecessary anxiety and inappropriate treatment. Area under ROC curve was 82,8% (95% CI: 80.1-85.5), statisticaly significant, showing that discriminatory ability of the test can be interpretted as relatively good. Unfortunately, prostate cancer incidence growing, and mortality decreasing slowly, especially during recent years. This indirectly shows that national early detection (screening) program is not effective enough. One of possible reasons of this phenomenon is a lack of accessibility to the first-class treatment for the cases, diagnosed durring the screening. The fact, that only 25% prostate biopsies was made for the participants of the programme, with PSA >3ng/ml, indirectly confirming this. Keywords: prostate cancer, screening, PSA, cut-of 3ng/ml, sensitivity, specificity. INTRODUCTION In Lithuania, prostate cancer accounts more than a quarter (29%) of newly diagnosed cases of cancer in males. In 2012, last data, available from the Lithuanian cancer registry, 2606 Lithuanian men (incidence rate of 189,4/100,000) were diagnosed with prostate cancer and 585 of those diagnosed died, making it the second most common cause of cancer death in Lithuanian men after lung cancer. Prostate cancer differs from any other epithelial cancer in many ways. Due to still clearly unknown reasons, sometimes this hormone-dependant cancer remain in a latent state, without any clinical symptoms. In addition, it differs greatly within diagnosed males with regard Correspondence to Romualdas Gurevičius Institute of Hygiene, Center of Health Information Didžioji str. 22, LT-01128 Vilnius, Lithuania E-mail: romualdas.gurevicius@hi.lt to aggressiveness. Currently, the main problem is unknown reasons of this phenomenon, and the lack of sensitive diagnostic tools to assess aggressiveness. All those characteristics of prostate cancer directly influence early detection policy, especially the cancer screening strategy. It must be remembered that the main aim of population-based screening program is not only early detection (increased risk of diagnosis), but to decrease mortality, or in another, words, to extend years of life. Unfortunately, according the latest evidence, prostate cancer screening results in doubling the lifetime risk of prostate cancer diagnosis with little if any decrease in the risk of dying from this disease [1]. The prostate specific antigen (PSA) screening test for early prostate cancer has been surrounded by controversy ever since it was introduced by pathologist Richard Ablin, who discovered the PSA antigen 40 years ago. Scientists repeatedly emphasize 110 2018/4(83)

it should never have been used as a cancer screening tool (for asymptomatic persons) for all men [2]. On the other hand, there is no doubt PSA is enormously important and valuable tool in urological clinical practice, for which the test was primarily invented. Numerous approaches have been proposed to improve diagnostic accuracy of the PSA test, including measuring PSA velocity (change over time), levels of free and protein bound PSA, PSA density (the PSA level divided by prostate volume) and the cut-off values for PSA levels that are specific to the patients age, race or ethnic group [3]. Unfortunately, clinical usefulness of those strategies still remains unproved. Also, more recent studies have shown that some men with PSA levels below 4.0 ng/ml have prostate cancer and that many men with higher levels do not have prostate cancer. In addition, various factors can cause PSA level to fluctuate. For example, PSA level often rises due to prostatitis or a urinary tract infection. Prostate biopsies and prostate surgery also increase PSA level [3]. Several randomized trials of prostate cancer screening have been carried out. One of the largest is the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial 1. In the trial, PSA test and DRE were evaluated for their ability to decrease mortality of diagnosed prostate cancer patients. The PLCO investigators found that men who underwent annual prostate cancer screening had a higher incidence of prostate cancer than men in the control group but the same rate of deaths from the disease [4]. The aim of the study was to evaluate PSA test accuracy of a clearly defined and screened population with application different statistical indicators. MATERIAL AND METHODS The primary Health Care center in which investigation was carried out serves 97,000 residents of Vilnius city population, including 45 thousand men, of which about 11 thousand are 50-75 years of age. During the period of 2008-2010 information on all screened males (N=8810) was collected, covering 80% of the target population. According to Lithuanian law, PSA screening is offered free of charge to anyone eligible, that comes to the outpatient clinic or family doctor due to the health problem. If test shows PSA level of 3 ng/ml, the subject is directed to onco-urologist for a cascade of procedures (biopsy, treatment and follow up). The 1 https://prevention.cancer.gov/major-programs/prostate-lung-colorectal main data file with information on screened males, containing variables age, date of testing, PSA level 3 ng/ml (positive) or < 3 ng/ml (negative) was linked with the information at Lithuanian cancer registry and State cause of death registry. In case of new prostate cancer detection, or death from prostate cancer the main data file was updated. Other cancers or other causes of death were interpreted that participating person does not have prostate cancer. Data processing was done using MS excel and IBM SPSS v. 19. Different statistical methods, showing full scale of screening test parameters (accuracy and reliability) was calculated using statistical analysis package Winpepi, starting from sensitivity and specificity of the test, ending with Youden index, positive/negative likelihood/diagnostic Odds ratios and Receiver Operating Characteristic (ROC) curves [5]. RESULTS Lithuanian national programme for early detection of prostate cancer, covered by National Health insurance fund has started in 2006. General Practioners in ambulatory care have a duty to inform eligible symptom-free male population (50 to 75 years old for general male populations and 45 to 75 years od age if father or brother had this disease) and propose them to take part in by the use of a PSA test. Above mentioned persons are eligible for free PSA test every two ears. After the test result is available, general practitioner informs patient about the level of PSA. If the level is 3 ng/ml patient is then redirected to oncourologist for additional investigations, which includes DRE, trans-rectal prostate biopsy with the ultrasound control (not less than 6 columns), measuring volume of prostate, histological investigation of the biopsy sample. Also, urologist informs the patient about the results of investigation and gives recommendations for the prevention, and, if cancer of prostate confirmed - official committal for the treatment. All patients, to whom PSA test is < 3 ng/ml, are eligible to repeat the test every 2 years. PSA test, and all cascade of events following a positive test (biopsy, treatment and follow up) are covered by insurance. In the period of 2008-2010, 74.8% of males who actually have prostate cancer were correctly identified by the test. This indicates that 25.2% of screening subjects were wrongly identified as being diseasefree. It s important to mention that sensitivity of PSA test in our study gradually increased from 71.9 in 2008 to 73.8 in 2009, and reached 81 in 2010, showing that 81% of males, who have prostate cancer were 2018/4(83) 111

correctly identified by the PSA test. This means, that 19% of subjects have been incorrectly identified as being prostate cancer-free. PSA test specificity parameter shows, that during the three-year period, 90.7% of males who do not have a prostate cancer, were correctly identified by the PSA test. The result indicates, that 9.3% of the subjects without prostate cancer have been wrongly identified as having prostate cancer. Sensitivity parameter showed very small changes in level during each year of the screening (table 1). False positive rate shows, that 9.3 percent of the participants did not have the prostate cancer but the test was positive. False negative rate indicates, that 25.2 percent of participants ad prostate cancer, but the PSA test was negative. Consequently, quarter of participants with false-negative results will be given unfounded reassurance and 9.3% of particpants with falsepositive results might experience unnecessary anxiety and inappropriate treatment. Positive likelihood ratio shoes that during the 2008-2010 we have observed an 8.1 increase of the probability of prostate cancer if PSA test was positive. In other words, positive PSA test is about 8,1 times more likely to be made in presence of prostate cancer, than in the absence of it. There was no statistically significant variation of this parameter in time, but positive likelihood in 2010 increased to 9.7 (table1). At the same time negative likelihood ratio is shoes that in 2008-2010 there was 0.3 of decrease probability of prostate cancer if PSA test was negative. In other words, if prostate cancer was not present, the likelihood for the negative test was 0.3. Again, in 2010, negative likelihood ratio reached 0.2, showing 0.2 decrease in probability of prostate cancer if PSA test is negative. It s a matter of fact, that unlike sensitivity and specificity, the PPV and NPV are dependent on the population being tested and are influenced by the prevalence of the disease. Diagnostic Odds ratio (DOR) shows, that the odds of obtaining PSA test positive result in prostate cancer patient is 29 times higher, rather than non-diseased person. Consequently, our data shows that in our prostate PSA screening program, discriminatory ability of the test is 29. It must be mentioned, that in 2010, DOR almost doubled and reached 46.4, which showing, that the odds of obtaining test positive result in prostate cancer patient is 46 times higher, rather than non-diseased person. Consequently, the discriminatory ability of PSA test in 2010 was 46. DOR is a global measure that is easy to interpret and does not depend on the prevalence of the prostate cancer, although it must be emphasized that it can vary among groups of patients within severity of their disease. Unfortunately, in our study information on prostate cancer staging was not complete and was not used. Another positive feature of DOR is that DOR correlates with the area under the ROC curve of the test. Furthermore, it can be used for comparisons in meta-analysis. The area under this curve (AUC) represents the overall accuracy of a test, with a value approaching 1.0 indicating a high sensitivity and specificity. Diagonal line represents of zero discrimination (AUC=0.5); (the test is no better than tossing a coin). Our PSA screening test results show, that area under ROC curve is 82.8 percent, statisticaly signifficant, and varies from 80.1 to 85.5 with 95 percent confidence. 1 Table. Main PSA (borderline 3ng/ml) prostate cancer screening test parameters Sensitivity Sensitivity Specificity Specificity False False negative Years Corrected corrected Positive P (95% CI) P (95% CI) P (95% CI) P (95% CI) P (95% CI) P (95% CI) 2008 71.9 (63.7-78.7) 69.1 (59.7-76.3) 90.8 (89.9-91.7) 87.3 (85.3-89.0) 9.2 (8.3-10.1) 28.1 (21.3-36.0) 2009 73.8 (65.5-80.7) 70.8 (60.8-78.2) 89.7 (88.4-90.9) 86.0 (83.6-88.0) 10.3 (9.2-11.6) 26.2 (19.3-34.5) 2010 81.0 (71.3-87.9) 79.2 (67.7-86.6) 91.6 (90.4-92.7) 89.6 (87.7-91.3) 8.4 (7.3-9.6) 19.1 (12.1-28.7) 2008-2010 74.8 (70.0-79.1) 72.2 (66.7-76.8) 90.7 (90.1-91.3) 87.6 (86.4-88.7) 9.3 (8.7-9.9) 25.2 (21.0-30.0) 1 table continued +Likelihood -Likelihood Diagnostic odds Youden s index Area under Theil s Years ratio Ratio ratio ROC curve coefficient P (95% CI) P (95% CI) P (95% CI) P (95% CI) P (95% CI) P 2008 7.9 (6.8-9.1) 0.3 (0.2-0.4) 25.4 (17.3-37.4) 0.6 (0.6-0.7) 81.4 (77.6-85.2) 0.250 2009 7.1 (6.1-8.4) 0.3 (0.2-0.4) 24.5 (16.1-37.1) 0.6 (0.6-0.7) 81.7 (77.9-85.6) 0.259 2010 9.7 (8.2-11.4) 0.2 (0.1-0.3) 46.4 (26.4-81.6) 0.7 (0.6-0.8) 86.3 (82.0-90.5) 0.328 2008-2010 8.1 (7.4-8.8) 0.3 (0.2-0.3) 29.0 (22.5-37.3) 0.7 (0.6-0.7) 82.8 (80.5-85.1) 0.272 112 2018/4(83)

Sensitivity 1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 ROC Curve 1 - Specificity 0.6 0.8 1.0 Diagonal segments are produced by ties. Picture 1. Receiver Operating Characteristic curve of PSA prostate cancer screening test (borderline 3ng/ml) DISCUSSION In practice, screening tests are never completely accurate. The specificity of a screening test is the ability of that test to exclude those who do not have disease. There will always be a number of falsepositives results (in which the test indicates that subject has the disease, when in reality they do not). False-negative results can also occur (in which the test indicates that there is no disease present, when in reality the subject does have a disease. A good screening test should keep false-positive and false negative results to an absolute minimum. In the ERSPC 2, the sensitivity or ability of the PSA test to detect prostate cancer among men who agreed to biopsy was 71%. In this study, only about one prostate biopsy in four found prostate cancer, and around 15% of men with a normal PSA level (measured as less than 4 ng/ ml) had prostate cancer [6]. Our study (PSA cut-off value of greater than 3 ng/ml) showing, that during the three-year period, 90.7% of males who do not have a prostate cancer was correctly identified by this test. There is a dearth of evidence into how men view their participation in health screening. Our previous study showed that knowledge of men about prostate cancer 2 European Randomized Study of Screening for Prostate Cancer. 2 Table. Area under ROC curve statistical characteristics of PSA prostate cancer screening test AsymptoticSig. b Lower Bound Upper Bound Asymptotic 95% Confidence Interval Area Std. Error a,828,014,000,801,855 a. Under the nonparametric assumption b. Null hypothesis: true area = 0.5 and its prevention was relatively low. Knowledge of the prostate cancer risk factors, clinical symptoms, treatment outcomes and examination possibilities were not dependent on social and demographic characteristics [7]. The main problem with prostate cancer PSA screening is that the biology of this type of cancer is not clearly known. Many years is known, that sufficient fraction from all prostate cancer, especialy in older ages, remain latent without any clinical symptoms [8]. Prostate cancer can be aggressive and life threatening, but can grow so slowly that it might not produce significant symptoms before a man dies from another cause. The treatments may cause unpleasant side effects, such as erectile dysfunction, urinary incontinence, bowel dysfunction or even death. There is no clear benefit in population-based screening when the cancer does not actually pose a risk of death. In Lithuania, after introducing of this screening program incidence increased dramatically. In the period of 1998-2007 prostate cancer incidence and mortality are growing, especially during recent years. Signifficant mortality decrease has not been observed, which indirectly shows that national early detection (screening) program is not effective enough [9, 10]. One of possible reasons of this phenomenon is a lack of accessibility to the firstclass treatment for the cases, diagnosed durring the screening. The fact, that for only 25% of the participants of the prgogramme, with PSA >3 ng/ml, prostate biopsies was made [11], indirectly confirming this. Some experts suggest that prostate cancer screening test using PSA has minimal value when it comes to finding prostate cancer, especially in older men. It was noticed, that routine PSA blood tests result in overdiagnosis of prostate cancer, resulting in expensive unnecessary invasive diagnostic procedures, unnecessary treatments, psychosocial harm, related to insignificant prostate cancer tumors [12]. According to the American Cancer Society, There can be different reasons for an elevated PSA level, including prostate cancer, BPH, inflammation, infection, age, and race, all factors that make 2018/4(83) 113

PSA test results confusing, leading to potential for unnecessary biopsies, treatment and suffering. Currently, the serum PSA test is the best screening test available for prostate cancer, but unfortunately, the screening per se is increasing the risk of diagnosis of prostate cancer but not sufficiently reducing the risk of dying from this cancer. CONCLUSIONS 1. The typical characteristics of prostate cancer screening PSA test, when cut-of 3ng/ml, was following: The sensitivity of the PSA test is 74.8 (70-79.1), The specificity of the PSA test is 90.7 (90.1-91.3). 2. False positive rate shows, that 9.3 percent of the participants did not have the prostate cancer but the test was positive. False negative rate indicates that 25.2 percent of the patients had prostate cancer but the test was negative. 3. Quarter of participants with false-negative results will be given unfounded reassurance and 9.3 with false-positive results might experience unnecessary anxiety and inappropriate treatment. 4. Area under ROC curve is 82.8 percent, statisticaly signifficant, and varies from 80.1 to 85.5 with 95 percent confidence. 5. Currently, the serum PSA test is the best screening test available for prostate cancer, but, unfortunately, the screening per se mostly increasing the risk of diagnosis of prostate cancer but not sufficiently reducing the risk of dying from this cancer, which is the main target of any cancer screening. Straipsnis gautas 2018-11-14, priimtas 2018-12-15 References 1. Boyle PO. Brawley Prostate Cancer: Current Evidence Weighs Against Population Screening, CA. A Cancer Journal for Clinicians. July/August 2009;59(4):220-224. 2. Tiffany O Callaghan Prostate cancer test has been misused for money. New Scientist. 17 February 2014;2956. 3. Greene KL, Albertsen PC, Babaian RJ, et al. Prostate specific antigen best practice statement: 2009 update. J Urol. 2009;182:2232-41. 4. Thompson IM, Pauler DK, Goodman PJ, et al. Prevalence of prostate cancer among men with a prostate-specific antigen level < or =4.0 ng per milliliter. New England Journal of Medicine. 2004; 350(22):2239-46. 5. Abramson JH. WINPEPI updated: Computer programs for epidemiologists, and their teaching potential. Epidemiologic Perspectives and Innovations. 2011;8:1. DOI: 10.1186/1742-5573-8-1 6. Andriole GL, Crawford ED, Grubb RL, et al. Prostate cancer screening in the randomized Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial: mortality results after 13 years of followup. Journal of the National Cancer Institute. 2012;104(2):125-132. 7. Schröder FH, Hugosson J, Roobol MJ, et al. Screening and prostatecancer mortality in a randomized European study. New England Journal of Medicine. 2009;360(13):1320-8. 8. Šturienė R, Kalibatienė D, Gurevičius R. The relationship between men s knowledge on prostate cancer, it s prevention and the socialdemographical characteristics. Public Health. 2012;1(56):62-72 (in lithuanian). 9. Gurevičius R, Lazutka V. Incidence and future prediction of prostate cancer in Lithuania in 1973-1978 m. Onkologija-5, Mokslas. A. Telyčėnas, Edt. 1983;13-20. ISSN 0208-2535 (in lithuanian). 10. Čepanauskienė R, Gurevičius R. Incidence and mortality from prostate cancer in Lithuania before PSA Era and after. Public Health. 2009;1(44):19-25 (in lithuanian). 11. Source from internet: Valstybinė ligonių kasa prie Sveikatos apsaugos ministerijos. <http://www.vlk.lt/veikla/veiklos-sritys/prevencinesprogramos/priesines-liaukos-vezio-ankstyvosios-diiagnostikosprograma/puslapiai/default.aspx> [accessed 2018-12-1]. 12. Source from internet: <http://www.cancer.gov/cancertopics/ factsheet/detection/psa>. 114 2018/4(83)

PSA testo tikslumas Lietuvos prostatos vėžio organizuotos patikros programos metu Romualdas Gurevičius 1, Renata Šturienė 2, Arvydas Šilys 3 1 Higienos institutas, 2 VšĮ Šeškinės poliklinika, 3 Vilniaus universiteto klinika Santrauka Priešinės liaukos vėžio ankstyvosios diagnostikos programa (toliau Programa) Lietuvoje pradėta vykdyti 2006 m. Šiuo metu Programa skirta vyrams nuo 50 iki 69 metų ir vyrams nuo 45 metų, jeigu jų tėvai ar broliai sirgo prostatos vėžiu. Programos tikslinės grupės vyrams gali būti atliekamas kraujo tyrimas, parodantis prostatos specifinio antigeno (toliau PSA) koncentraciją kraujyje. Jei PSA kiekis yra 3 ng/ml, šeimos gydytojas išduoda siuntimą konsultuotis su urologu, o šis prireikus atlieka tikslesnį tyrimą prostatos biopsiją. Tyrimo tikslas. Nustatyti ir įvertinti PSA testo tikslumą atlikus apibrėžtos populiacijos, dalyvausios organizuotoje patikroje, ištisinį tyrimą. Tyrimo medžiaga ir metodai. Poliklinika, kurioje atliekamas tyrimas, aptarnauja 97 000 Vilniaus miesto gyventojų, tarp jų 45 tūkstančiai vyrų, iš jų apie 11 tūkstančių yra 50 75 metų amžiaus. 2008 2010 m. laikotarpiu buvo surinkta informacija apie visus patikrintus vyrus (N = 8 810), apimanti 80 proc. tikslinės populiacijos. Duomenų statistiniam apdorojimui taikyti įvairiausi, dažnai vienas kitą papildantys metodai, skaičiuota naudojantis Winpepi ver. 11.65. Rezultatai ir išvados. Mūsų tyrime PSA tyrimo jautrumas buvo 74,8 proc. (95 proc. PI 70 79,1), PSA tyrimo specifiškumas 90,7 proc. (95 proc. PI 90,1 91,3). Klaidingai teigiamos diagnozės tikimybė parodė, jog 9,3 proc. mūsų tyrimo dalyvių nesirgo prostatos vėžiu, tačiau testas buvo teigiamas. Klaidingai neigiamos diagnozės tikimybė parodė, jog 25,2 proc. mūsų turimo dalyvių sirgo prostatos vėžiu, nors testas buvo neigiamas. Todėl ketvirtis programos dalyvių, kurių buvo klaidingai neigiami rezultatai, gavo nepagrįstas garantijas, kad jie sveiki, o 9,3 proc. dalyvių, kurių rezultatai buvo klaidingai teigiami, galėjo patirti nereikalingą nerimą ir gauti netinkamą gydymą. Tyrimo rezultatai parodė, jog ROC kreivės plotas buvo 82,8 proc. (95 proc. PI 80,1 85,5), statistiškai reikšmingas. Visa tai rodo problemas, susijusias su pacientų nerimu ir (ar) bereikalingai atliktu radikaliu gydymu, todėl pacientams turėtų būti paaiškintos teigiamos ir neigiamos tokių patikrų pasekmės. Organizuotos prostatos vėžio patikros šalyje rezultatai kol kas rodo tik didėjantį sergamumą ir labai nežymiai mažėjantį mirtingumą. Tai gali priklausyti nuo įvairių priežasčių. Vis dėlto tai, kad tik ketvirtadaliui ištirtų asmenų, kurių PSA testas viršijo ribinį lygį, atliktos biopsijos, rodo, jog 75 proc. patikros dalyvių net nebuvo bandyta patvirtinti / atmesti diagnozę ir, suprantama, kvalifikuotai gydyti. Reikšminiai žodžiai: prostatos vėžys, organizuota patikra, PSA, jautrumas, specifiškumas. Adresas susirašinėti: Romualdas Gurevičius Higienos instituto Sveikatos informacijos centras Didžioji g. 22, 01128 Vilnius El. p. romualdas.gurevicius@hi.lt Received 14 November 2018, accepted 15 December 2018 2018/4(83) 115