- Email: [email protected]
The usefulness of urinary cytology testing in the evaluation of irritative voiding symptoms
American Journal of Obstetrics and Gynecology (2005) 192, 1554–9
www.ajog.org
The usefulness of urinary cytology testing in the evaluation of irritative voiding symptoms Eric R. Sokol, MD,a,* Sutchin R. Patel, BS,a Jeffrey L. Clemons, MD,b Vivian W. Sung, MD,a Charles R. Rardin, MD,a Deborah L. Myers, MDa Department of Obstetrics and Gynecology, Division of Urogynecology and Pelvic Reconstructive Surgery, Women and Infants Hospital, Brown Medical School, Providence, RI,a Madigan Army Medical Center, Tacoma, Washb
KEY WORDS Urinary cytology Urothelial cancer Irritative voiding symptoms
Objective: The purpose of this study was to assess the clinical usefulness of urinary cytology testing for the evaluation of urothelial cancer in women with irritative voiding symptoms who were examined at a urogynecology service. Study design: Urinary cytology studies results that were obtained from January 1, 2000, to December 31, 2002, were cross-matched with the Rhode Island Department of Health Cancer Registry to identify those women who were diagnosed with urinary tract malignancies. The prevalence of urothelial cancer was determined, and the sensitivity, specificity, and positive and negative predictive values of urinary cytologic testing were calculated for 2 common classification strategies: (1) consideration of atypical cytologic test results to be normal and (2) consideration of atypical cytologic test results to be abnormal. Results: Among 1516 cross-matched cytologic test results from 1324 patients, 5 urothelial cancers were identified. Two of the 5 malignancies were associated with positive cytology results. The prevalence of urothelial cancer was 0.38% (95% CI, 0.1%, 0.9%). When atypical cytology studies were classified as normal, the sensitivity of urinary cytology was 40% (95% CI, 7.2%, 83.0%); the specificity was 99.9% (95% CI, 99.5%, 100%); the positive predictive value was 66.7% (95% CI, 12.5%, 98.2%), and negative predictive value was 99.8% (95% CI, 99.2%, 100%). In contrast, when atypical cytology results were classified as abnormal, the sensitivity and negative predictive value remained the same, but the specificity declined to 93.6% (95% CI, 92.1%, 94.8%), and the positive predictive value decreased to 2.3% (95% CI, 0.4%, 8.8%). Conclusion: The low prevalence of urothelial cancers and low sensitivity of urinary cytology studies severely limit the usefulness of this test in the evaluation of women with irritative voiding symptoms. Ó 2005 Elsevier Inc. All rights reserved.
Supported by a grant from The Center for Women’s Surgery Endowment Fund. Presented at the Joint Scientific Meeting of the American Urogynecologic Society and the Society of Gynecologic Surgeons, San Diego, California, July 29-31, 2004. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense. * Reprint requests: Eric R. Sokol, MD, The Center for Women’s Surgery, 695 Eddy St, Suite 12, Providence, RI 02903. E-mail: [email protected] 0002-9378/$ - see front matter Ó 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ajog.2004.10.627
Sokol et al Irritative urinary voiding symptoms that include urinary urgency, frequency, nocturia, and dysuria are the second most common presentation of bladder cancer, with hematuria being the most common complaint.1-3 However, only one third of patients who are diagnosed with bladder cancer have irritative voiding symptoms, which makes treatment decisions that are based on symptoms alone difficult.2 Urinary cytology is often used in the evaluation of patients with irritative voiding symptoms to aid in the detection of urothelial tract malignancies, but its role remains controversial, with some authors arguing for cytology only in high-risk patients.4 The presence of neoplastic urothelial cells in the urine was first reported by Sanders5 in 1864, but the use of urinary cytology for the diagnosis of urothelial malignancy was not described until 1945 by Papanicolaou and Marshall.6 Since then, the subject of urinary cytology as a test for urothelial cancer has been discussed in numerous reports.7,8 Although the sensitivity and specificity of urinary cytology is thought to be higher in women with a history of urinary tract malignancy or hematuria, the clinical usefulness of this test in the evaluation of women with irritative voiding symptoms is uncertain. The objective of this study was to assess the clinical usefulness of urinary cytology for the detection of urothelial cancer in women with irritative voiding symptoms at an academic urogynecology service. Specifically, we sought to determine the prevalence of urothelial cancer in our study population and to determine the sensitivity, specificity, and positive and negative predictive values of urinary cytology for the diagnosis of urinary tract malignancy in these women using the 2 most common classification strategies. Our hypothesis was that urinary cytology has limited usefulness for the evaluation of urinary tract malignancy in women with complaints of irritative voiding symptoms.
Material and methods This study was approved by the Institutional Review Boards at Women and Infants’ Hospital of Rhode Island and by the Rhode Island Department of Health. All urinary cytology studies that were sent from the Division of Urogynecology at Women and Infants’ Hospital of Rhode Island from January 1, 2000, to December 31, 2002, in the course of the evaluation of women with irritative voiding symptoms were included in this study. We defined irritative voiding symptoms as urinary urgency, pain or burning with urination, urinary frequency, and/or nocturia. Definitions conform to the standards recommended by the International Continence Society.9 In accordance with accepted nomenclature, final urinary cytologic testing results were classified
1555 Table I Common classification strategies for grouping atypical urinary cytology studies Strategy 1
Strategy 2
Low risk
High risk
Normal
Abnormal
Normal Atypical
Suspicious Malignant
Normal
Atypical Suspicious Malignant
by 1 of 2 board-certified attending cytopathologists into 1 of 4 categories: normal, atypical/indeterminate, suspicious, or malignant.10 Fresh urine specimens that were analyzed in the cytopathology laboratory were centrifuged, and the precipitate was put into ‘‘thin preparation’’ vials. After being processed, the specimens were read in a standardized fashion in a manner that was similar to the reading of ‘‘thin preparation’’ Papanicolaou smears. It has been our standard practice to obtain urinary cytology studies for women with irritative voiding symptoms and to send repeat cytologic tests for women with atypical or suspicious results. Treatment decisions are then based on the most abnormal cytologic test reading. Urinary cytology studies that were sent from our division during the study period were identified through the Women and Infants’ Hospital cytology laboratory and compiled into a master list. Social security numbers and cytologic testing dates were then extracted from this list and electronically cross-matched with the Cancer Registry database at the Rhode Island Department of Health to identify those women who were diagnosed subsequently with urinary tract malignancies during the study period. We set search parameters for the crossmatch using International Classification of Diseases Oncology incidence codes for bladder (C67.0-C67.9), ureter (C66.9), urothelial (C67), urethra (C68.0, C67.5), and kidney (C64.9, C65.9). Results from the database cross-match were then confirmed with the master cytologic test list and individual medical records to ensure accuracy. Through the database cross-match, types of urothelial cancer and tumor behavior (malignant or in situ) were noted, and the prevalence of urothelial cancer in our study population was determined. We then calculated the sensitivity, specificity, and positive and negative predictive values of urinary cytology (with 95% CIs) using 2 different common classification strategies: (1) grouping atypical/indeterminate urinary cytologic results into a low-risk category with ‘‘normal’’ cytologic testing readings and (2) grouping atypical/indeterminate cytology studies into a high-risk category with all ‘‘abnormal’’ cytology test readings (including atypical/ indeterminate, suspicious, and malignant cytologic results; Table I). To evaluate the impact of varying
1556 Table II
Sokol et al Distribution of urinary cytology studies (n = 1516)
Test result
Patients (n)
Normal Atypical/indeterminate Suspicious Malignant Unsatisfactory
1407 101 3 2 3
Table III Patient cancer status grouped by urinary cytology studies (n = 1324*)
(92.8%) (6.7%) (0.2%) (0.13%) (0.2%)
Test result
Cancer (n)
No cancer (n)
Normal Atypical/indeterminate Suspicious or malignant
3 0 2
1234 84 1
* Patients classified by most abnormal result in cases of repeat urinary cytology studies.
Table IV
Anticipated urinary cytology performance for each classification strategy, stratified by disease prevalence
Variable
Prevalence (%)
Sensitivity (%)
Specificity (%)
Positive predictive value (%)*
Negative predictive value (%)*
Strategy 1 (atypical urinary cytology studies that were considered normal)
0.2
40
99.9
51.4
99.9
0.4 0.6
40 40
99.9 99.9
67.9 76.1
99.8 99.6
0.2
40
93.6
1.2
99.9
0.4 0.6
40 40
93.6 93.6
2.4 3.6
99.7 99.6
Strategy 2 (atypical urinary cytology studies that were considered abnormal)
* Calculated with Bayes’ theorem.
prevalence on estimated test performance, we recalculated the sensitivities and specificities of urinary cytologic testing at different possible disease prevalences and used Bayes’ theorem to calculate the test’s anticipated negative and positive predictive values. In patients who had repeat urinary cytologic results, the most abnormal reading was used, according to our standard clinical practice. Statistical analysis was performed with Stata software (version 8.0; Stata Corporation, College Station, Tex).
Results A total of 1562 urinary cytology studies were sent from the Division of Urogynecology between January 1, 2000, and December 31, 2002. Forty-six cytologic studies lacked social security numbers, which left 1516 cytologic studies from 1324 patients available for crossmatch with the Rhode Island Department of Health Cancer Registry database. The distribution of urinary cytology readings, as coded by the cytopathologist, is listed in Table II. The Rhode Island Department of Health Cancer Registry database contained 98% of the records of all state cancers from the year 2000, 98% of all state cancers from 2001, and 90% of all state cancers from 2002. Four urothelial cancers were identified through this cross-match. One additional case of malignant bladder
cancer, not otherwise specified, was absent from the Cancer Registry database but was identified on the master cytology laboratory list by a malignant urinary cytology studies. Urothelial cancer cases that were identified through the cross-match included 1 case of malignant bladder cancer (not otherwise specified) and 3 cases of papillary transitional cell carcinoma (in situ). Only 2 of the 5 identified urothelial malignancies were associated with positive urinary cytology studies. The numbers of patients with and without cancer are listed by cytologic reading in Table III. The prevalence of urothelial cancer in our study population was 0.38% (95% CI, 0.1%, 0.9%). When urinary cytology studies were grouped into low-risk (normal and atypical/indeterminate) and high-risk (suspicious and malignant) categories, the sensitivity of urinary cytology was 40% (95% CI, 7.2%, 83.0%); the specificity was 99.9% (95% CI, 99.5%, 100%); the positive predictive value was 66.7% (95% CI, 12.5%, 98.2%), and the negative predictive value was 99.8% (95% CI, 99.2%, 100%). Alternatively, if urinary cytology studies were grouped into normal and abnormal categories (including atypical/indeterminate, suspicious, and malignant results), the sensitivity remained unchanged, but the specificity declined to 93.6% (95% CI, 92.1%, 94.8%); the positive predictive value dropped to 2.3% (95% CI, 0.4%, 8.8%), and the negative predictive value remained unchanged. The test properties of urinary cytology for each classification strategy,
Sokol et al at a range of possible disease prevalences, are listed in Table IV and show the effects of changing prevalence on anticipated test performance. None of the 84 patients with only atypical cytology results had or experience the development of an urothelial malignancy during the study period.
Comment Until this study, previous literature regarding the role of urinary cytology in the evaluation of women with irritative lower urinary tract symptoms has been scarce. Although urinary cytology is used commonly in the urology setting, the usefulness of this test in the evaluation of women with these symptoms has been unclear. Numerous reports in the urology literature have been dedicated to the study of urinary cytology as a screening tool for urothelial malignancy,7,8 but only 1 report has specifically addressed the test’s role in the evaluation of women with irritative voiding symptoms. Duldulao et al11 examined 202 urinary cytology specimens from women with irritative voiding symptoms at a urology office and found that 80.2% of the cytology studies were normal, that 14.8% were atypical, that 2.5% were suspicious, and that 2.5% were malignant. A total of 5 patients with lower urinary tract symptoms were diagnosed with transitional cell carcinoma, of which 3 patients had positive urinary cytology studies, 1 patient had atypical urinary cytology studies, and 1 patient had normal urinary cytology studies. The prevalence of urothelial cancer in our study population was 0.38%, one sixth of the 2.5% rate of cancer in the study of Duldulao et al.11 Also, only 0.3% of our cytology studies were read as suspicious or malignant, compared with 5% in the report of Duldulao et al. These discrepancies are possibly due to different referral patterns between urologists and non-urologists, which is reflected by the different cytology positivity rates, as reported by other investigators.12 These discrepancies may also be due to differences in specimen collection, interlaboratory variation in specimen processing, and a lack of a standardized method of reporting.13 Whereas the routine evaluation of urinary cytology may be more clinically useful in those settings with a high prevalence of cancer, our study suggests that the test’s usefulness is limited by the low prevalence of urothelial malignancy in our patient population. Our calculated sensitivity for urinary cytology of 40% is consistent with the low sensitivity reported in previous literature14 and further highlights the limited usefulness of this test in the evaluation for urothelial cancer in an unselected population of women with irritative voiding symptoms. Debate in the urology literature exists regarding the classification and subsequent treatment of patients with atypical or indeterminate cytologic testing results.15,16 A
1557 cost-effectiveness analysis by Novicki et al16 determined that patients with indeterminate urinary cytology studies who were not smokers and who had no history of hematuria or urothelial cancer were at low risk for malignancy and, thus, did not warrant complete urinary tract evaluations. Our report examined the performance of urinary cytology as a test for urothelial cancer with the use of 2 different common classification strategies: considering atypical/indeterminate readings to be ‘‘normal’’ and considering atypical/indeterminate readings to be ‘‘abnormal.’’ Both scenarios revealed the low sensitivity of the test, with the positive predictive value being only 2.3% when atypical readings were grouped with the other high-risk (suspicious and malignant) cytologic test readings. Furthermore, none of the 84 patients with only atypical/indeterminate urinary cytologic test readings had or experienced an urothelial malignancy. These findings support the notion that atypical/indeterminate readings, in an unselected population of women with irritative voiding symptoms, should be considered low risk and may not warrant further evaluation for urothelial malignancy. Our study did not address those risk factors (such as smoking and hematuria), which might identify patients at higher risk for urothelial malignancy and therefore warrant testing with urinary cytology in the urogynecologic setting. It is noteworthy that only 2 of 5 urinary tract malignancies in our study were associated with positive cytologic testing results. Only the 2 patients with bladder cancers with malignant behavior designations (by International Classification of Diseases Oncology coding) had corresponding suspicious or malignant urinary cytology readings. The other 3 patients with transitional cell carcinomas were associated with normal urinary cytology studies. Studies have shown a progression in the sensitivity of urinary cytology with increasing tumor stage, with the strongest association between urinary cytology and histologic tumor grade. Although only 40% of well-differentiated tumors are associated with positive cytology, 82% to 86% of poorly differentiated tumors show a positive association.17 This relationship between tumor grade and cytology findings highlights the inherent difficulties in the use of urinary cytology as a test for urothelial tract malignancy in women with irritative voiding symptoms; women with lower grade tumors often will be missed. Most strikingly, of the 1324 the patients who underwent cytologic evaluation over a 3-year period, only 2 patients had suspicious or malignant results that were associated with an actual cancer. In our population of women with irritative voiding symptoms, O660 patients would have to be screened to obtain 1 true positive cytology result. Because patients with urothelial malignancy often experience hematuria that prompts cystoscopic evaluation, these cancers likely would be diagnosed regardless of cytology findings.1 Therefore,
1558 the role of urinary cytology in the current cost-conscious medical environment is questionable. A recent study by Hofland and Mariani18 determined that urinary cytologic testing was comparable to excretory urography, creatinine, and cystoscopy for the evaluation of gross or microscopic hematuria in terms of cost analysis, although cytology provided unique information in only 0.6% of patients. A cost-effectiveness analysis is underway at our institution to evaluate the cost of urinary cytology in the initial evaluation of irritative voiding symptoms (plus the costs of the ensuing work-up, diagnosis, and treatment of urothelial cancer) compared with the costs and risks of delayed diagnosis. Limitations of this study include the possible lack of reporting of all cancer cases to the Rhode Island Department of Health Cancer Registry and the reliance on recorded social security numbers to perform the database cross-match. Some cases of urothelial cancer may have been missed because of the underreporting of cancer cases to the Department of Health by the hospitals, as evidenced by the 2% rate of missing data for 2000 and the 10% rate of missing data from 2001 and 2002. Indeed, 1 case of urothelial cancer in our study was listed on the master urine cytologic test list but was missing from the Cancer Registry database. A number of the social security numbers were missing, and it is possible that some of the numbers were reported incorrectly by patients or erroneously recorded by the cytology laboratory. In addition, patients who had urothelial cancer could have moved from Rhode Island before diagnosis or could have been out-of-state residents. Also, certain malignancies develop over time before they are detectable. Thus, some patients who had urinary cytology results evaluated within the study period could have cancers in the future. Although our study population consisted primarily of unselected white, postmenopausal women who underwent an initial evaluation for irritative voiding symptoms, we were not able to assess individual patient characteristics that would be helpful in judging the applicability of our findings to other populations. To perform this largescale database cross-match, the Rhode Island Department of Health Cancer Registry mandated that only social security numbers and cytologic testing dates of patients who underwent urinary cytologic testing could be used. Thus, individual demographic characteristics of the 1324 patients who underwent urinary cytology could not be analyzed for this study. Despite these limitations, our prevalence is similar to the 0.5% prevalence of patients with urothelial malignancy in a population-based study of asymptomatic microhematuria and urologic disease among residents of Minnesota.19 We calculated the estimated test properties of urinary cytology for a range of disease prevalences to address these limitations and to increase the generalizability of our findings. Finally, the power of this study
Sokol et al was limited by the rare outcome of urothelial cancer. However, this study is much larger than the only other report that addressed this subject, and our CIs are narrow enough to support our level of inference. In conclusion, the low prevalence of urothelial cancers and low sensitivity of urinary cytology severely limit the usefulness of this test in the evaluation of women who have irritative voiding symptoms. Further research is warranted to identify risk factors that should prompt evaluation with urinary cytology in women with these complaints. Ultimately, practice guidelines that delineate the clinical role for urinary cytology, based on patient risk characteristics, may be helpful in the evaluation and treatment of these patients. Given the large amount of medical resources that are expended currently in the evaluation these patients, a cost-effectiveness analysis, which is underway at our institution, is justified to further elucidate the role of urinary cytology in the evaluation of irritative voiding symptoms.
References 1. Messing EM. Urothelial tumors of the urinary tract. In: Walsh PC, Retik AB, Vaughan ED, Wein AJ, editors. Campbell’s urology. Philadelphia: Saunders; 2002. p. 2732-84. 2. Varkarakis MJ, Gaeta J, Moore RH, Murphy GP. Superficial bladder tumor: aspects of clinical progression. Urology 1974;4:414-20. 3. Messing EM, Vaillancourt A. Hematuria screening for bladder cancer. J Occup Med 1990;32:838-45. 4. Gamarra MC, Zein T. Cytologic spectrum of bladder cancer. Urology 1984;23:23-6. 5. Sanders WR. Cancer of the bladder: fragments forming urethral plugs discharged in the urine: concentric colloid bodies. Edinb J Med 1864;10:273-4. 6. Papanicolaou GN, Marshall VF. Urine sediment smears as a diagnostic procedure in cancers of the urinary tract. Science 1945;101:519-20. 7. Curling M, Broome G, Gendry WF. How accurate is urine cytology? J R Soc Med 1986;79:336-8. 8. Rife CC, Farrow GM, Utz DC. Urine cytology of transitional cell neoplasms. Urol Clin N Am 1979;6:599-612. 9. Abrams P, Cardozo L, Fall M, Griffiths D, Rosier P, Ulmsten U, et al. The standardisation of terminology of lower urinary tract function: report from the standardisation subcommittee of the International Continence Society. Am J Obstet Gynecol 2002; 187:116-26. 10. Murphy WM. Current status of urinary cytology in the evaluation of bladder neoplasms. Hum Pathol 1990;21:886-96. 11. Duldulao KE, Diokno AC, Mitchell B. Value of urinary cytology in women presenting with urge incontinence and/or irritative voiding symptoms. J Urol 1997;157:113-6. 12. Nabi G, Greene DR, O’Donnell M. How important is urinary cytology in the diagnosis of urological malignancies? Eur Urol 2003;43:632-6. 13. Wiener HG, Mian C, Haitel A, Pycha A, Schatzl G, Marberger M. Can urine bound diagnostic tests replace cystoscopy in the management of bladder cancer? J Urol 1998;159:1876-80. 14. Chahal R, Gogoi NK, Sundaram SK. Is it necessary to perform urine cytology in screening patients with haematuria? Eur Urol 2001;39:283-6. 15. Schwalb DM, Herr HW, Fair WR. The management of clinically unconfirmed positive urinary cytology. J Urol 1993;150:1751-6.
Sokol et al 16. Novicki DE, Stern JA, Nemec R, Lidner TK. Cost-effective evaluation of indeterminate urinary cytology. J Urol 1998;160: 734-6. 17. Utz DC, Zincke H. The masquerade of bladder cancer in situ as interstitial cystitis. J Urol 1974;111:160-1.
1559 18. Hofland CA, Mariani AJ. Is cytology required for a hematuria evaluation? J Urol 2004;171:324-6. 19. Mohr DN, Offord KP, Owen RA, Melton LJ. Asymptomatic microhematuria and urologic disease: a population-based study. JAMA 1986;256:224-9.
www.ajog.org
The usefulness of urinary cytology testing in the evaluation of irritative voiding symptoms Eric R. Sokol, MD,a,* Sutchin R. Patel, BS,a Jeffrey L. Clemons, MD,b Vivian W. Sung, MD,a Charles R. Rardin, MD,a Deborah L. Myers, MDa Department of Obstetrics and Gynecology, Division of Urogynecology and Pelvic Reconstructive Surgery, Women and Infants Hospital, Brown Medical School, Providence, RI,a Madigan Army Medical Center, Tacoma, Washb
KEY WORDS Urinary cytology Urothelial cancer Irritative voiding symptoms
Objective: The purpose of this study was to assess the clinical usefulness of urinary cytology testing for the evaluation of urothelial cancer in women with irritative voiding symptoms who were examined at a urogynecology service. Study design: Urinary cytology studies results that were obtained from January 1, 2000, to December 31, 2002, were cross-matched with the Rhode Island Department of Health Cancer Registry to identify those women who were diagnosed with urinary tract malignancies. The prevalence of urothelial cancer was determined, and the sensitivity, specificity, and positive and negative predictive values of urinary cytologic testing were calculated for 2 common classification strategies: (1) consideration of atypical cytologic test results to be normal and (2) consideration of atypical cytologic test results to be abnormal. Results: Among 1516 cross-matched cytologic test results from 1324 patients, 5 urothelial cancers were identified. Two of the 5 malignancies were associated with positive cytology results. The prevalence of urothelial cancer was 0.38% (95% CI, 0.1%, 0.9%). When atypical cytology studies were classified as normal, the sensitivity of urinary cytology was 40% (95% CI, 7.2%, 83.0%); the specificity was 99.9% (95% CI, 99.5%, 100%); the positive predictive value was 66.7% (95% CI, 12.5%, 98.2%), and negative predictive value was 99.8% (95% CI, 99.2%, 100%). In contrast, when atypical cytology results were classified as abnormal, the sensitivity and negative predictive value remained the same, but the specificity declined to 93.6% (95% CI, 92.1%, 94.8%), and the positive predictive value decreased to 2.3% (95% CI, 0.4%, 8.8%). Conclusion: The low prevalence of urothelial cancers and low sensitivity of urinary cytology studies severely limit the usefulness of this test in the evaluation of women with irritative voiding symptoms. Ó 2005 Elsevier Inc. All rights reserved.
Supported by a grant from The Center for Women’s Surgery Endowment Fund. Presented at the Joint Scientific Meeting of the American Urogynecologic Society and the Society of Gynecologic Surgeons, San Diego, California, July 29-31, 2004. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense. * Reprint requests: Eric R. Sokol, MD, The Center for Women’s Surgery, 695 Eddy St, Suite 12, Providence, RI 02903. E-mail: [email protected] 0002-9378/$ - see front matter Ó 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ajog.2004.10.627
Sokol et al Irritative urinary voiding symptoms that include urinary urgency, frequency, nocturia, and dysuria are the second most common presentation of bladder cancer, with hematuria being the most common complaint.1-3 However, only one third of patients who are diagnosed with bladder cancer have irritative voiding symptoms, which makes treatment decisions that are based on symptoms alone difficult.2 Urinary cytology is often used in the evaluation of patients with irritative voiding symptoms to aid in the detection of urothelial tract malignancies, but its role remains controversial, with some authors arguing for cytology only in high-risk patients.4 The presence of neoplastic urothelial cells in the urine was first reported by Sanders5 in 1864, but the use of urinary cytology for the diagnosis of urothelial malignancy was not described until 1945 by Papanicolaou and Marshall.6 Since then, the subject of urinary cytology as a test for urothelial cancer has been discussed in numerous reports.7,8 Although the sensitivity and specificity of urinary cytology is thought to be higher in women with a history of urinary tract malignancy or hematuria, the clinical usefulness of this test in the evaluation of women with irritative voiding symptoms is uncertain. The objective of this study was to assess the clinical usefulness of urinary cytology for the detection of urothelial cancer in women with irritative voiding symptoms at an academic urogynecology service. Specifically, we sought to determine the prevalence of urothelial cancer in our study population and to determine the sensitivity, specificity, and positive and negative predictive values of urinary cytology for the diagnosis of urinary tract malignancy in these women using the 2 most common classification strategies. Our hypothesis was that urinary cytology has limited usefulness for the evaluation of urinary tract malignancy in women with complaints of irritative voiding symptoms.
Material and methods This study was approved by the Institutional Review Boards at Women and Infants’ Hospital of Rhode Island and by the Rhode Island Department of Health. All urinary cytology studies that were sent from the Division of Urogynecology at Women and Infants’ Hospital of Rhode Island from January 1, 2000, to December 31, 2002, in the course of the evaluation of women with irritative voiding symptoms were included in this study. We defined irritative voiding symptoms as urinary urgency, pain or burning with urination, urinary frequency, and/or nocturia. Definitions conform to the standards recommended by the International Continence Society.9 In accordance with accepted nomenclature, final urinary cytologic testing results were classified
1555 Table I Common classification strategies for grouping atypical urinary cytology studies Strategy 1
Strategy 2
Low risk
High risk
Normal
Abnormal
Normal Atypical
Suspicious Malignant
Normal
Atypical Suspicious Malignant
by 1 of 2 board-certified attending cytopathologists into 1 of 4 categories: normal, atypical/indeterminate, suspicious, or malignant.10 Fresh urine specimens that were analyzed in the cytopathology laboratory were centrifuged, and the precipitate was put into ‘‘thin preparation’’ vials. After being processed, the specimens were read in a standardized fashion in a manner that was similar to the reading of ‘‘thin preparation’’ Papanicolaou smears. It has been our standard practice to obtain urinary cytology studies for women with irritative voiding symptoms and to send repeat cytologic tests for women with atypical or suspicious results. Treatment decisions are then based on the most abnormal cytologic test reading. Urinary cytology studies that were sent from our division during the study period were identified through the Women and Infants’ Hospital cytology laboratory and compiled into a master list. Social security numbers and cytologic testing dates were then extracted from this list and electronically cross-matched with the Cancer Registry database at the Rhode Island Department of Health to identify those women who were diagnosed subsequently with urinary tract malignancies during the study period. We set search parameters for the crossmatch using International Classification of Diseases Oncology incidence codes for bladder (C67.0-C67.9), ureter (C66.9), urothelial (C67), urethra (C68.0, C67.5), and kidney (C64.9, C65.9). Results from the database cross-match were then confirmed with the master cytologic test list and individual medical records to ensure accuracy. Through the database cross-match, types of urothelial cancer and tumor behavior (malignant or in situ) were noted, and the prevalence of urothelial cancer in our study population was determined. We then calculated the sensitivity, specificity, and positive and negative predictive values of urinary cytology (with 95% CIs) using 2 different common classification strategies: (1) grouping atypical/indeterminate urinary cytologic results into a low-risk category with ‘‘normal’’ cytologic testing readings and (2) grouping atypical/indeterminate cytology studies into a high-risk category with all ‘‘abnormal’’ cytology test readings (including atypical/ indeterminate, suspicious, and malignant cytologic results; Table I). To evaluate the impact of varying
1556 Table II
Sokol et al Distribution of urinary cytology studies (n = 1516)
Test result
Patients (n)
Normal Atypical/indeterminate Suspicious Malignant Unsatisfactory
1407 101 3 2 3
Table III Patient cancer status grouped by urinary cytology studies (n = 1324*)
(92.8%) (6.7%) (0.2%) (0.13%) (0.2%)
Test result
Cancer (n)
No cancer (n)
Normal Atypical/indeterminate Suspicious or malignant
3 0 2
1234 84 1
* Patients classified by most abnormal result in cases of repeat urinary cytology studies.
Table IV
Anticipated urinary cytology performance for each classification strategy, stratified by disease prevalence
Variable
Prevalence (%)
Sensitivity (%)
Specificity (%)
Positive predictive value (%)*
Negative predictive value (%)*
Strategy 1 (atypical urinary cytology studies that were considered normal)
0.2
40
99.9
51.4
99.9
0.4 0.6
40 40
99.9 99.9
67.9 76.1
99.8 99.6
0.2
40
93.6
1.2
99.9
0.4 0.6
40 40
93.6 93.6
2.4 3.6
99.7 99.6
Strategy 2 (atypical urinary cytology studies that were considered abnormal)
* Calculated with Bayes’ theorem.
prevalence on estimated test performance, we recalculated the sensitivities and specificities of urinary cytologic testing at different possible disease prevalences and used Bayes’ theorem to calculate the test’s anticipated negative and positive predictive values. In patients who had repeat urinary cytologic results, the most abnormal reading was used, according to our standard clinical practice. Statistical analysis was performed with Stata software (version 8.0; Stata Corporation, College Station, Tex).
Results A total of 1562 urinary cytology studies were sent from the Division of Urogynecology between January 1, 2000, and December 31, 2002. Forty-six cytologic studies lacked social security numbers, which left 1516 cytologic studies from 1324 patients available for crossmatch with the Rhode Island Department of Health Cancer Registry database. The distribution of urinary cytology readings, as coded by the cytopathologist, is listed in Table II. The Rhode Island Department of Health Cancer Registry database contained 98% of the records of all state cancers from the year 2000, 98% of all state cancers from 2001, and 90% of all state cancers from 2002. Four urothelial cancers were identified through this cross-match. One additional case of malignant bladder
cancer, not otherwise specified, was absent from the Cancer Registry database but was identified on the master cytology laboratory list by a malignant urinary cytology studies. Urothelial cancer cases that were identified through the cross-match included 1 case of malignant bladder cancer (not otherwise specified) and 3 cases of papillary transitional cell carcinoma (in situ). Only 2 of the 5 identified urothelial malignancies were associated with positive urinary cytology studies. The numbers of patients with and without cancer are listed by cytologic reading in Table III. The prevalence of urothelial cancer in our study population was 0.38% (95% CI, 0.1%, 0.9%). When urinary cytology studies were grouped into low-risk (normal and atypical/indeterminate) and high-risk (suspicious and malignant) categories, the sensitivity of urinary cytology was 40% (95% CI, 7.2%, 83.0%); the specificity was 99.9% (95% CI, 99.5%, 100%); the positive predictive value was 66.7% (95% CI, 12.5%, 98.2%), and the negative predictive value was 99.8% (95% CI, 99.2%, 100%). Alternatively, if urinary cytology studies were grouped into normal and abnormal categories (including atypical/indeterminate, suspicious, and malignant results), the sensitivity remained unchanged, but the specificity declined to 93.6% (95% CI, 92.1%, 94.8%); the positive predictive value dropped to 2.3% (95% CI, 0.4%, 8.8%), and the negative predictive value remained unchanged. The test properties of urinary cytology for each classification strategy,
Sokol et al at a range of possible disease prevalences, are listed in Table IV and show the effects of changing prevalence on anticipated test performance. None of the 84 patients with only atypical cytology results had or experience the development of an urothelial malignancy during the study period.
Comment Until this study, previous literature regarding the role of urinary cytology in the evaluation of women with irritative lower urinary tract symptoms has been scarce. Although urinary cytology is used commonly in the urology setting, the usefulness of this test in the evaluation of women with these symptoms has been unclear. Numerous reports in the urology literature have been dedicated to the study of urinary cytology as a screening tool for urothelial malignancy,7,8 but only 1 report has specifically addressed the test’s role in the evaluation of women with irritative voiding symptoms. Duldulao et al11 examined 202 urinary cytology specimens from women with irritative voiding symptoms at a urology office and found that 80.2% of the cytology studies were normal, that 14.8% were atypical, that 2.5% were suspicious, and that 2.5% were malignant. A total of 5 patients with lower urinary tract symptoms were diagnosed with transitional cell carcinoma, of which 3 patients had positive urinary cytology studies, 1 patient had atypical urinary cytology studies, and 1 patient had normal urinary cytology studies. The prevalence of urothelial cancer in our study population was 0.38%, one sixth of the 2.5% rate of cancer in the study of Duldulao et al.11 Also, only 0.3% of our cytology studies were read as suspicious or malignant, compared with 5% in the report of Duldulao et al. These discrepancies are possibly due to different referral patterns between urologists and non-urologists, which is reflected by the different cytology positivity rates, as reported by other investigators.12 These discrepancies may also be due to differences in specimen collection, interlaboratory variation in specimen processing, and a lack of a standardized method of reporting.13 Whereas the routine evaluation of urinary cytology may be more clinically useful in those settings with a high prevalence of cancer, our study suggests that the test’s usefulness is limited by the low prevalence of urothelial malignancy in our patient population. Our calculated sensitivity for urinary cytology of 40% is consistent with the low sensitivity reported in previous literature14 and further highlights the limited usefulness of this test in the evaluation for urothelial cancer in an unselected population of women with irritative voiding symptoms. Debate in the urology literature exists regarding the classification and subsequent treatment of patients with atypical or indeterminate cytologic testing results.15,16 A
1557 cost-effectiveness analysis by Novicki et al16 determined that patients with indeterminate urinary cytology studies who were not smokers and who had no history of hematuria or urothelial cancer were at low risk for malignancy and, thus, did not warrant complete urinary tract evaluations. Our report examined the performance of urinary cytology as a test for urothelial cancer with the use of 2 different common classification strategies: considering atypical/indeterminate readings to be ‘‘normal’’ and considering atypical/indeterminate readings to be ‘‘abnormal.’’ Both scenarios revealed the low sensitivity of the test, with the positive predictive value being only 2.3% when atypical readings were grouped with the other high-risk (suspicious and malignant) cytologic test readings. Furthermore, none of the 84 patients with only atypical/indeterminate urinary cytologic test readings had or experienced an urothelial malignancy. These findings support the notion that atypical/indeterminate readings, in an unselected population of women with irritative voiding symptoms, should be considered low risk and may not warrant further evaluation for urothelial malignancy. Our study did not address those risk factors (such as smoking and hematuria), which might identify patients at higher risk for urothelial malignancy and therefore warrant testing with urinary cytology in the urogynecologic setting. It is noteworthy that only 2 of 5 urinary tract malignancies in our study were associated with positive cytologic testing results. Only the 2 patients with bladder cancers with malignant behavior designations (by International Classification of Diseases Oncology coding) had corresponding suspicious or malignant urinary cytology readings. The other 3 patients with transitional cell carcinomas were associated with normal urinary cytology studies. Studies have shown a progression in the sensitivity of urinary cytology with increasing tumor stage, with the strongest association between urinary cytology and histologic tumor grade. Although only 40% of well-differentiated tumors are associated with positive cytology, 82% to 86% of poorly differentiated tumors show a positive association.17 This relationship between tumor grade and cytology findings highlights the inherent difficulties in the use of urinary cytology as a test for urothelial tract malignancy in women with irritative voiding symptoms; women with lower grade tumors often will be missed. Most strikingly, of the 1324 the patients who underwent cytologic evaluation over a 3-year period, only 2 patients had suspicious or malignant results that were associated with an actual cancer. In our population of women with irritative voiding symptoms, O660 patients would have to be screened to obtain 1 true positive cytology result. Because patients with urothelial malignancy often experience hematuria that prompts cystoscopic evaluation, these cancers likely would be diagnosed regardless of cytology findings.1 Therefore,
1558 the role of urinary cytology in the current cost-conscious medical environment is questionable. A recent study by Hofland and Mariani18 determined that urinary cytologic testing was comparable to excretory urography, creatinine, and cystoscopy for the evaluation of gross or microscopic hematuria in terms of cost analysis, although cytology provided unique information in only 0.6% of patients. A cost-effectiveness analysis is underway at our institution to evaluate the cost of urinary cytology in the initial evaluation of irritative voiding symptoms (plus the costs of the ensuing work-up, diagnosis, and treatment of urothelial cancer) compared with the costs and risks of delayed diagnosis. Limitations of this study include the possible lack of reporting of all cancer cases to the Rhode Island Department of Health Cancer Registry and the reliance on recorded social security numbers to perform the database cross-match. Some cases of urothelial cancer may have been missed because of the underreporting of cancer cases to the Department of Health by the hospitals, as evidenced by the 2% rate of missing data for 2000 and the 10% rate of missing data from 2001 and 2002. Indeed, 1 case of urothelial cancer in our study was listed on the master urine cytologic test list but was missing from the Cancer Registry database. A number of the social security numbers were missing, and it is possible that some of the numbers were reported incorrectly by patients or erroneously recorded by the cytology laboratory. In addition, patients who had urothelial cancer could have moved from Rhode Island before diagnosis or could have been out-of-state residents. Also, certain malignancies develop over time before they are detectable. Thus, some patients who had urinary cytology results evaluated within the study period could have cancers in the future. Although our study population consisted primarily of unselected white, postmenopausal women who underwent an initial evaluation for irritative voiding symptoms, we were not able to assess individual patient characteristics that would be helpful in judging the applicability of our findings to other populations. To perform this largescale database cross-match, the Rhode Island Department of Health Cancer Registry mandated that only social security numbers and cytologic testing dates of patients who underwent urinary cytologic testing could be used. Thus, individual demographic characteristics of the 1324 patients who underwent urinary cytology could not be analyzed for this study. Despite these limitations, our prevalence is similar to the 0.5% prevalence of patients with urothelial malignancy in a population-based study of asymptomatic microhematuria and urologic disease among residents of Minnesota.19 We calculated the estimated test properties of urinary cytology for a range of disease prevalences to address these limitations and to increase the generalizability of our findings. Finally, the power of this study
Sokol et al was limited by the rare outcome of urothelial cancer. However, this study is much larger than the only other report that addressed this subject, and our CIs are narrow enough to support our level of inference. In conclusion, the low prevalence of urothelial cancers and low sensitivity of urinary cytology severely limit the usefulness of this test in the evaluation of women who have irritative voiding symptoms. Further research is warranted to identify risk factors that should prompt evaluation with urinary cytology in women with these complaints. Ultimately, practice guidelines that delineate the clinical role for urinary cytology, based on patient risk characteristics, may be helpful in the evaluation and treatment of these patients. Given the large amount of medical resources that are expended currently in the evaluation these patients, a cost-effectiveness analysis, which is underway at our institution, is justified to further elucidate the role of urinary cytology in the evaluation of irritative voiding symptoms.
References 1. Messing EM. Urothelial tumors of the urinary tract. In: Walsh PC, Retik AB, Vaughan ED, Wein AJ, editors. Campbell’s urology. Philadelphia: Saunders; 2002. p. 2732-84. 2. Varkarakis MJ, Gaeta J, Moore RH, Murphy GP. Superficial bladder tumor: aspects of clinical progression. Urology 1974;4:414-20. 3. Messing EM, Vaillancourt A. Hematuria screening for bladder cancer. J Occup Med 1990;32:838-45. 4. Gamarra MC, Zein T. Cytologic spectrum of bladder cancer. Urology 1984;23:23-6. 5. Sanders WR. Cancer of the bladder: fragments forming urethral plugs discharged in the urine: concentric colloid bodies. Edinb J Med 1864;10:273-4. 6. Papanicolaou GN, Marshall VF. Urine sediment smears as a diagnostic procedure in cancers of the urinary tract. Science 1945;101:519-20. 7. Curling M, Broome G, Gendry WF. How accurate is urine cytology? J R Soc Med 1986;79:336-8. 8. Rife CC, Farrow GM, Utz DC. Urine cytology of transitional cell neoplasms. Urol Clin N Am 1979;6:599-612. 9. Abrams P, Cardozo L, Fall M, Griffiths D, Rosier P, Ulmsten U, et al. The standardisation of terminology of lower urinary tract function: report from the standardisation subcommittee of the International Continence Society. Am J Obstet Gynecol 2002; 187:116-26. 10. Murphy WM. Current status of urinary cytology in the evaluation of bladder neoplasms. Hum Pathol 1990;21:886-96. 11. Duldulao KE, Diokno AC, Mitchell B. Value of urinary cytology in women presenting with urge incontinence and/or irritative voiding symptoms. J Urol 1997;157:113-6. 12. Nabi G, Greene DR, O’Donnell M. How important is urinary cytology in the diagnosis of urological malignancies? Eur Urol 2003;43:632-6. 13. Wiener HG, Mian C, Haitel A, Pycha A, Schatzl G, Marberger M. Can urine bound diagnostic tests replace cystoscopy in the management of bladder cancer? J Urol 1998;159:1876-80. 14. Chahal R, Gogoi NK, Sundaram SK. Is it necessary to perform urine cytology in screening patients with haematuria? Eur Urol 2001;39:283-6. 15. Schwalb DM, Herr HW, Fair WR. The management of clinically unconfirmed positive urinary cytology. J Urol 1993;150:1751-6.
Sokol et al 16. Novicki DE, Stern JA, Nemec R, Lidner TK. Cost-effective evaluation of indeterminate urinary cytology. J Urol 1998;160: 734-6. 17. Utz DC, Zincke H. The masquerade of bladder cancer in situ as interstitial cystitis. J Urol 1974;111:160-1.
1559 18. Hofland CA, Mariani AJ. Is cytology required for a hematuria evaluation? J Urol 2004;171:324-6. 19. Mohr DN, Offord KP, Owen RA, Melton LJ. Asymptomatic microhematuria and urologic disease: a population-based study. JAMA 1986;256:224-9.