Section 1 -

Urinary Cytopathology: How We Got Where We Are William M Murphy University Florida

Urinary cytopathology (UC) hasbeen used to detect bladder cancers for more than 150 years. The story is told that, after examining his own urine, the great pathologist James Ewing handed the slide to an assistant saying, 'take this to Freddie (Stewart), he will find cancer cells in it'. Dr. Ewing died of bladder cancer 4 years later. The tale emphasizes two important features of urinary cytopathology: the method is good at detecting true bladder cancers; and true bladder cancers are killers, even today being associated with a 40-65% rate of disease-specific death. George Papanicolaou was an early advocate of the method, publishing several papers on its usefulness for neoplasms of the bladder and kidney. In fact, urinary cytopathology has been used to evaluate patients for conditions as varied as papillomavirus infection and renal transplant rejection. In the 1960's, when screening the otherwise "normal" population for clinically inapparent diseases became popular, UC was applied to industrial workers at high risk for bladder neoplasms. In the 1970's and early 1980's, after urothelial carcinoma in situ (CIS) was established as an important entity, UC was "high-tech" and the literature on the subject blossomed. And yet, the application of cytopathology to urinary samples has never been completely satisfactory. Practitioners have been searching for better methods as long as I have been in medicine. We have looked at red cell adherence, carcinoembryonic antigens, glycoproteins, and most recently chromosomes and genes. Why do we continue to perceive UC as a flawed methodology necessitating an urgent search for a replacement? Of course, UC has drawbacks. It applies primarily to epithelial neoplasms and has trouble differentiating urothelial from squamous from glandular types. The method cannot localize the tumors. Oftentimes, immediate correlation with tissue samples is not achieved, creating the impression of falsely positive interpretations. Neither the minimum criteria for adequacy nor the minimum number of cells required for a positive interpretation has been generally accepted. All of these factors are valid but in my opinion, the decades long dissatisfaction with UC as a method for detecting urothelial neoplasms can be condensed into three somewhat related themes: 1) inaccurate histological classification and terminology; 2) inappropriate clinical approach to patient care; 3) lack of confidence among pathologists in general and cytopathologists in particular. Inaccurate histological classification and terminology. Quite naturally, the value of UC has been inextricably connected to the debate over the nature of carcinoma. There are two major schools of thought. One group holds (even today) that carcinomas must be invasive by definition; they have little concern with cellular anaplasia. To these pathologists, non-invasive lesions are not carcinoma regardless of how ugly their cells look. Applied to urothelial tumors, these pathologists use terms like papilloma, atypical papilloma, and (invasive) carcinoma. They consider CIS an oxymoron. When pressed to describe flat lesions, they use terms such as "denuding cystitis" and "atypia", sometimes including the modifier "severe". A competing group contends that the potential for recurrence defines carcinoma. Pathologists holding to this view tend to consider all papillary and invasive urothelial lesions carcinoma, since even the lowest grade papillary tumors can recur. Recognizing that urothelial lesions vary in their degree of cellular anaplasia, these pathologists grade them on a scale of 1 to 3, 1 to 4, and even 1 to 5. They often do not recognize the term papilloma. When pressed to describe flat lesions, they tend to fall back on cellular anaplasia and have tried to grade using a 3 step scale. Failing this, this group uses terms such as CIS for the high end of the morphologic spectrum; they have problems agreeing on terms for the rest of the spectrum. This is how things stood for the first 100 years and many elements continue today. Enter the world of consensus pathology, where science is important but the facts are in dispute so those with the most agreeable personalities have the most influence. What if we could get the world to agree on a single classification of urothelial neoplasms? Wouldn't that be a positive influence on clinical research and patient care? In 1973, the World Health Organization (WHO) classification of urothelial neoplasms was born. Not surprisingly, being a committee decision heavily influenced by the most authoritatively suave individual present, this classification represented a compromise of the two age-old approaches. It recognized a papilloma, but defined it so narrowly that the lesion became rare. Thus, nearly all urothelial tumors were classed as carcinomas, grades 1, 2, or 3. No particularly specific definition of the grades was forthcoming. Flat lesions were not included, even though they had been described and illustrated for the previous 20 years. The good that might have been achieved by the 1973 WHO scheme is debatable. The effect on urinary cytopathology was devastating. The problem? The concept of recurrence as the basis for malignancy was wrong, at least for the lowest grade tumors. Patients presenting with grade 1 "carcinomas" tended to develop new tumors all right, but the new tumors were usually not in the exact same place as the primaries (actually new occurrences) and they were almost always of the same grade. They lacked the capacity to invade or metastasize and killed no one. The consensus panel had taught us to diagnose the stalk, not the cells! When only the cells were available for interpretation, as in a urinary sample, low-grade "carcinomas" could not be detected with any degree of reliability among cytopathologists. Similarly, grade 2 tumors were hard to detect when only the cells were available. When flow cytometry became available in the 1970's and 1980's, very few G1 and only about 50% of G2 urothelial tumors in WHO were aneuploid, tending to confirm the non-malignant nature of these lesions. In fact, the cells of low-grade, papillary urothelial tumors usually lack features of malignancy, regardless of the technology used. And yet, the tumors continued to be termed "carcinoma" in 1973 WHO and cytopathology was denigrated because the method could not detect this type of "cancer". To be sure, a certain number (<10%) of patients presenting with G1 tumors developed progressive disease and about 60% of these (4-6% of total) died of bladder cancer. Believers in 1973 WHO continue to use these figures to support the classification, seemingly unable to understand that all those patients with G1 who progressed had developed a new, high-grade lesion and none had died of metastatic low-grade disease. The disaggregated cells in the urinary specimens had been telling us something about the nature of urothelial neoplasms for 30 years but we were not prepared to accept the knowledge. Like the well digger convinced that there was water in his dry hole, we did not abandon the effort but instead asked for a larger shovel. Eventually, the facts became overwhelming and the classification as well as the terminology had to be altered. The 1998 WHO/ISUP scheme is now widely accepted, despite the cumbersome term "papillary urothelial neoplasm of low malignant potential" (PUNLMP). This approach has the following virtues: 1) the term "carcinoma", with all of its connotations, has been removed from the low-grade papillary tumors; 2) it is increasingly recognized that most true carcinomas are high-grade; 3) urinary cytology is recognized, albeit in a footnote. When UC is applied to the detection of true urothelial carcinomas, most authors have achieved sensitivities and specificities in the range of 90%. Clinicians and patients receiving an interpretation of "high-grade neoplasm" from the assessment of a urinary sample can rely on the result (PPV >90%). And yet we clearly have work to do. The current enthusiasm for ancillary testing emphasizes the increased sensitivity of these techniques primarily for low-grade neoplasms, as if the early detection of lesions that are not particularly harmful to patients and will eventually be detected cystoscopically is important. After all, they are bladder cancer, aren't they. Inappropriate clinical approach to patient care. Clinicians should know better and many of them do. The great urologist Willet Whitmore once jokingly told me, "Kash Mostofi cures more bladder cancer than I do". Clinicians know that patients presenting with low-grade urothelial neoplasms are at low risk for progressive disease. Most do not treat other than to excise the lesion and many will simply fulgurate any "recurrences". Yet almost to a man, urologists call all urothelial neoplasms "bladder cancer". Some use the modifier "superficial" but it is unlikely that patients' fears are assuaged by this nuance. Having put the fear of the big C in the patient's mind, one can only imagine how the clinician can tell them that this is the good kind of cancer and he does not need aggressive follow-up. It must be nearly impossible to defend UC for follow-up when the literature concludes that UC is poor at detecting this particular kind of "cancer". Urologists were mightily confused but ultimately grateful when UC could detect flat lesions that they could not see cystoscopically. How could UC detect lesions that we cannot see when the method can't find the papillary lesions that we so readily visualize? In fact, it was CIS that put UC on the urologists' radar. When I first asked a group of urologists and pathologists in 1979 how many used UC, only about 20% raised their hands. When I followed with how many believed the results, only a few hands remained up. When I ask the same questions today, nearly all hands go aloft and most stay there. We have made some progress. Even when clinicians are believers, however, the clinical facts seem to undermine their faith. It turns out that cytopathologists do more poorly when in possession of relevant clinical information than when in a research setting. When asked to evaluate urinary specimens outside the patient care setting, the literature confirms a high diagnostic yield for high-grade neoplasms and an acceptable yield for low-grade lesions. The corresponding figures in actual practice are not as good. How could this be when the cytopathologist in a practice setting can know important details, even the results of biopsies, before rendering a diagnosis? One can only speculate but my hunch is that we fret too much about the implications of our interpretations. Many pathologists have been taught to subtly manipulate patient care by adjusting their terminology to the perceived behavior of their clinical colleagues. How many of us have been taught that 'you have to know your clinicians'? Another factor is the perceived adversity associated with an incorrect interpretation. In the community setting, the adversity is perceived as missing a lesion and, perhaps because of this, it is better to overinterpret so long as all cases are detected. A high sensitivity is the goal even though the positive predictive value may be low. In the referral and cancer center settings, where the diagnosis of "bladder cancer" has already been made in a large percentage of the patients, the opposite tends to prevail. The perceived adversity associated with missing a lesion is not as great as that occurring if no tumor is present and the patient receives unnecessary treatment or increased testing. A low specificity is acceptable so long as the positive predictive value is high. Paradoxically, we would do better diagnostically if we were to trust ourselves and let the cells speak to us. One further problem surfaces from an analysis of UC in the clinical setting. It is the rare urologist who performs a good bladder washing. When the urologist asks his resident or junior colleague to review the charts for the diagnostic yield of UC, it probably never occurs to him that a negative interpretation in a case of histologically documented "bladder cancer" could mean that the tumor cells were not present in the specimen. In a study of our experience over a single year's time, slightly more than 20% of bladder washings from patients with histologically documented bladder tumors lacked neoplastic cells in the specimen. Urologists have their own problems. Too many apply large amounts of anesthetic gel, thus obscuring the cellular elements in the resulting sample. Most allow the residual urine found at the introduction of the cystoscope to escape in the mistaken belief that it contains uninterpretable degenerated cells. Most like to orient themselves in the bladder before washing, letting all the fluid used for this purpose flow down the drain. Modern flexible cystoscopes issue a jet of fluid that often injures the urothelium, creating erythematous areas that can be confused with CIS. Knowing this, urologists are reluctant to wash first and examine afterwards. By the time they get to the bladder washing, a good many tumor cells are gone, leaving the least abnormal elements to be pried loose with the lavage. The yield is especially affected when few cells were present to begin with, e.g. after topical therapy. A good bladder washing consists of the residual urine plus an immediate lavage, before any other manipulation, in a patient who has been given as little anesthetic gel as possible. Voided urines suffer from less manipulation but their hypocellularity is well known. All of these mitigating factors could be known to practitioners if they had a more consultative relationship with their cytopathologists. Unfortunately, pathologists are increasingly viewed as laboratorians whose function is to record a series of observations in the form of a list rather than to render consultations based on tissues and cells. Lack of confidence among cytopathologists. I have often said that the major drawback to the widespread application of UC to patient care is the pathologists themselves. Many urologic pathologists do not claim to have proficiency in cytology and many cytopathologists have not been specifically trained in urinary cytopathology. When assessing a urinary specimen, it is not uncommon for cytopathologists to attempt an extrapolation of criteria learned for uterine cervical lesions. Nuclear anaplasia, for example, must be assessed in light of the amount of cytoplasm in urinary specimens, lest a reactive superficial cell be misinterpreted as cancer. This misunderstanding is the primary cause of the oft-heard complaint that topical therapy confounds the subsequent interpretation of urinary specimens for recurrent cancer. So-called papillary aggregates in urinary samples are not reliable indicators of the presence of a low-grade urothelial neoplasm, a factor that has confounded many an expert, as judged by the literature. The cells of urothelial CIS, unlike those of cervical CIS, cannot be distinguished from cells of invasive carcinoma in urinary specimens. Many urinary samples contain cells infected by the BK papillomavirus, a potential pitfall that rarely if ever occurs in cervical samples. Most cytopathologists are reluctant to make an unequivocal diagnosis on the basis of 4 or less abnormal cells, a problem that apparently does not affect the analysis of Vysis FISH specimens. The historic problem of inaccurate histological classification must be factored into all of this. Imagine the frustration of a cytopathologist who is constantly criticized for being unable to recognize G1 "bladder cancer" cells in a urinary specimen when the tumor is readily apparent to both the cystoscopist and the histopathologist? Constantly being reminded of the deficiencies of your craft while being subtly threatened by the latest new ancillary methodology? It's enough to make most perfectly good cytopathologists cower in mea culpas. Many of us have attempted to assuage our insecurities by using temporizing diagnostic terminology. Cytoatypia, what is the urologist supposed to do with that? Does it help if we say moderate or severe cytoatypia? If the meaning is crystal clear in any particular practice, good. But many if not most urologists interpret the word atypia in a diagnostic report as unhelpful. They came to us for help and aren't getting any. No wonder they agitate for some other way to follow their patients. These observations on the background and current application of urinary cytopathology to patient care are intended to offer perspective. They should not be construed as a defense of the cytologic approach in the face of the current onslaught of ancillary tests. Urinary cytopathology doesn't need an apologist. Urinary cytopathology has deficiencies but it is very good at its primary mission, i.e. to detect high-grade urothelial neoplasms. Urinary cytopathology is the only currently available method that can distinguish these high-grade, aggressive neoplasms from low-grade, non-aggressive neoplasms in patient follow-up. Even histopathology can be less informative, depending as it does on the cystoscopist's ability to biopsy the proper site. In the patient care setting, the vagaries of cytopathologic detection of low-grade lesions is actually a plus, the insistence of urologists on calling these lesions "cancer" notwithstanding. In the patient care situation, it is the reliability more than the sensitivity of the result that matters most. The PPV of UC for high-grade lesions is at least 90%, compared to about 50% for Vysis FISH. And the "test" is 7-10 times cheaper. Of course, it would help if the medical community would more accurately label urothelial neoplasms but this is unlikely to occur. In the meantime, cytopathologists can understand the problems, take comfort in their discipline, and develop the confidence that the cytopathologic approach deserves. References Long S, Cohen MB . Classics in cytology. V: William Sanders and early urinary tract cytology. Diagn Cytopathol 1991;8:135-6. Harrison JH, Botsford TW, Tucker MR. The use of the smear of the urinary sediment in the diagnosis and management of neoplasms of the kidney and bladder. J Urol 2002;167:864-71. Papanicoloau GN. Cytology of the urine sediment in neoplasms of the urinary tract. J Urol 1947;57:375-9. Dean PJ, Murphy WM: Importance of urinary cytology and future role of flow cytometry. Urology (Suppl) 1985;26:11-15. Lokeshwar VB, Habachi T, Grossman HB, et al. Bladder tumor markers beyond cytology: International Consensus Panel on Bladder Tumor Markers. Urology 2005;66(suppl):35-63. Dukes CE, Masina F. Classification of epithelial tumours of the bladder. Br J Urol 1949;21:273-95. Broders AC. Epithelium of the genito-urinary organs. Ann Surg 1922;75:574-604. Mostofi FK, Sobin LH, Torloni H. International histological classification of tumours, No 10. Geneva: World Health Organization, 1973. Bastacky S, Ibrahim S, Wilczynski SP, Murphy WM. The accuracy of urinary cytology in daily practice. Cancer(Cancer Cytopathol)1999;87:118-28. Malik S, Murphy WM. Monitoring patients for bladder neoplasms: what can be expected of urinary cytology consultations in clinical practice? Urology 1999;54:62-6. Melicow MM. Histological study of vesical urothelium intervening between gross neoplasms in total cystectomy. J Urol 1952;68:261-79. Tribukait B. Flow cytometry in surgical pathology and cytology of tumors of the genitourinary tract. In Koss LG, Coleman DV, eds. Advances in clinical cytology. Vol II. New York: Masson; 1984:163-89. Epstein JI, Amin MB , Reuter VR, Mostofi FK. The World Health Organization/International Society of Urological Pathology consensus classification of urothelial (transitional cell) neoplasms of the urinary bladder. Bladder Consensus Conference Committee. Am J Surg Pathol 1998;22:1435-48. Murphy WM. Urinary Cytopathology. ASCP Press, Chicago, 2000.