Fax: (410) 614-9556
Part I—Creating the template
Article first published online: 28 NOV 2012
Copyright © 2012 American Cancer Society
Volume 121, Issue 1, pages 15–20, January 2013
How to Cite
Rosenthal, D. L., VandenBussche, C. J., Burroughs, F. H., Sathiyamoorthy, S., Guan, H. and Owens, C. (2013), The Johns Hopkins Hospital template for urologic cytology samples . Cancer Cytopathology, 121: 15–20. doi: 10.1002/cncy.21255
Presented in part as a poster at the American Society of Cytopathology 58th Annual Scientific Meeting; November 12-16, 2010; Boston, MA.
- Issue published online: 14 JAN 2013
- Article first published online: 28 NOV 2012
- Manuscript Accepted: 2 OCT 2012
- Manuscript Revised: 1 OCT 2012
- Manuscript Received: 1 AUG 2012
- bladder cancer;
- urologic cytopathology;
- urine cytology;
The most important indicator for urologic surgeons at The Johns Hopkins Hospital to have a patient undergo cystoscopy is a cytologic diagnosis of high-grade urothelial carcinoma. The template was designed to standardize diagnostic categories so clinicians can manage their patients uniformly. The template was based in part on the Bethesda System for cervical cytology.
According to the template, reactive/inflammatory changes were included in the negative group (no urothelial atypia or malignancy identified). The category atypical urothelial cells of undetermined significance (AUC-US) was akin to atypical squamous cells of undetermined significance (ASC-US), as was the category of atypical urothelial cells, favor high-grade carcinoma (AUC-H). The categories high-grade urothelial carcinoma (HGUC) and low-grade urothelial carcinoma also were added.
The Pathology Data System at the Johns Hopkins Hospital was searched for cases that met the following criteria over a period from July 1, 2007 to June 30, 2009: all cytologic specimens from the urinary tract and all surgical specimens with a diagnosis of HGUC, regardless of invasion status. All cytologic specimens were then matched with biopsies during the same period, and all surgical specimens from patients who had a cytologic diagnosis of AUC-US or AUC-H were retrieved for 18 months after the end of the 2-year study period. Greater than 50% of patients who had biopsy-confirmed HGUC had a preceding cytologic diagnosis of AUC-H or HGUC. When patients with AUC-US were added to the analysis, 80% of patients with HGUC had at least 1 abnormal urinary cytology result. Of those patients who had a diagnosis of AUC-H, 38% had urothelial cancer discovered at biopsy compared with only 10% of those with an AUC-US diagnosis.
The authors concluded that their template is effective in targeting those patients who need to undergo cystoscopy. Cancer (Cancer Cytopathol) 2013;121:15–20 © 2012 American Cancer Society.
In cytopathology laboratories that process a variety of samples, those from the urinary tract are among the most frequent. With such familiarity, it would seem reasonable to expect that performance parameters of urine cytology would be exceptionally high. Quite the contrary!1 Consequently, faith in the results from urinary samples is low. Much effort has been spent developing alternative tests to screen patients with hematuria and those who have a high risk because of industrial exposure and to follow patients who already are diagnosed with urothelial malignancies for new or recurrent lesions. Unfortunately, the quest for the near-perfect screening test for bladder cancer is not over, and urologic surgeons continue to rely on cytology as the first line of diagnosis and as an indicator for cystoscopy. The reader is referred to the review article by Owens et al in this issue of Cancer (Cancer Cytopathology) for an expanded discussion of these issues and an extensive literature review.2 The Johns Hopkins Hospital template for urologic cytology samples was designed to standardize diagnostic categories, because 7 cytopathologists rotate to sign out these samples. The purpose of the template is to enable Johns Hopkins Hospital clinicians to uniformly manage their patients based on the reports generated by samples that are submitted for interpretation.
MATERIALS AND METHODS
Construction of the Template
Discussions with Johns Hopkins Hospital urologists revealed that the most important indicator for cystoscopy was a cytologic diagnosis of high-grade urothelial carcinoma (HGUC). In fact, most surgeons said that they would be happy with a binary scheme: high grade versus no high grade. We rejected that suggestion; however, we recognized the importance of emphasizing those samples with undisputed evidence of or suspicion of HGUC. The diagnostic template (Table 1) was based in part on the Bethesda System for gynecologic cytology.3 Reactive/inflammatory changes are included in the negative group (no urothelial atypia or malignancy identified [NUAM]); atypical urothelial cells of undetermined significance (AUC-US) is considered akin to atypical squamous cells of undetermined significance (ASC-US), as is the category atypical urothelial cells, favor high-grade lesion (AUC-H), which is akin to atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion (ASC-H); the category of urothelial carcinoma requires the choice of low grade or high grade; and inadequate samples are rare and usually are washings with low cellularity or obscuring noncellular material, such as lubricant.
|No urothelial atypia or malignancy identified (NUAM)|
|Urothelial carcinoma, specify|
|High grade (HGUC)|
|Low grade (LGUC)|
|Atypical urothelial cells of uncertain significance (AUC-US)|
|Atypical urothelial cells, cannot exclude HGUC (AUC-H)|
|Other (squamous carcinoma, adenocarcinoma, etc)|
The Pathology Data System at the Johns Hopkins Hospital was searched for all urinary cytologic specimens between July 2007 and July 2009. Each year was tallied separately to make sure that the data from the 2 years could be pooled; the data indeed were comparable and, thus, were combined (Table 2). The database was searched for those samples that met the following criteria over the period from July 1, 2007 to June 30, 2009: all cytologic specimens with a diagnosis of AUC-US and AUC-H and all surgical pathology specimens with a diagnosis of HGUC, regardless of invasion status. Cytologic specimens were matched with biopsies by a medical record number during the same period (Table 3). All surgical pathology specimens that were collected after June 30, 2009 provided follow-up to a cytologic diagnosis of AUC-US or AUC-H.
|Surgical Pathology Diagnosis: No. of Specimens|
|Cytopathology Diagnosis||Unsatisfactory||No Lesion||Low-Grade Atypia||High-Grade Atypia||Cancer||No Tissue||Total|
|No. of specimens (%)|
|Diagnosis||Cases||Biopsied||Cancer on Biopsy|
|NUAM||2894 (60)||180 (6)||58 (32)|
|AUC-US||1246 (26)||183 (14)||82 (45)|
|AUC-H||237 (5)||82 (34)||58 (71)|
|LGUC/HGUC||283 (6)||143 (50)||110 (77)|
|Inadequate||193 (4)||13 (6)||2 (15)|
|Total||4853 (100)||601 (12)||310 (52)|
Once it was determined that there were sufficient cases of AUC-US (n = 607) and AUC-H (n = 118) to yield convincing results (Table 4), permission from the Johns Hopkins Hospital Institutional Review Board was obtained to examine electronic patient records for demographic data, such as age and sex, and to link cytopathology results with surgical pathology diagnoses and long-term clinical follow-up. Rather than tally only the number of cytology cases (Tables 2 and 3), the final analysis was based on the number of patients who had HGUC discovered (Table 4). In patients who had multiple cytology studies, the most severe cytologic diagnosis was recorded as the referring cytology study.
|No. of patients (%)|
|No. of patients||607||118||725|
|No. biopsied||114 (18)||62 (53)||176 (24)|
|Benign||52 (8)||17 (14)||69 (10)|
|Low-grade UC||14 (2)||6 (5)||20 (3)|
|High-grade UC||49 (8)||39 (33)||88 (12)|
The project was immediately recognized as highly work-intensive. To facilitate data capture and review of slides, the project was divided as described below.
The objectives of part I (the current study) were to describe the template and the results from 2 years of experience with it and to produce a descriptive article. Attention went immediately to the AUC-US category, which comprised 26% of all samples (Table 3, Fig. 1). This is too large for an indeterminate category, is probably not clinically useful, and may inflate the cost of patient management. Instead of addressing this large category immediately, work proceeded to Part II.
The objective of part II was to review all AUC-H slides and establish which morphologic features best predict the patients who will have HGUC identified on biopsy. Once the most predictive cytomorphologic features were identified, a portion of the AUC-US samples was used in part III.
The objective of part III (the training set) was to split the AUC-US slides into a training set and a test set. For this part, all training set slides were reviewed. The features identified as the best predictors of HGUC on biopsy after referral cytology for AUC-H (part 2) were to be applied to determine whether these features are useful in recategorizing cases originally reported as AUC-US. The goal was to identify cytomorphologic criteria in AUC-US cases that predicted HGUC on biopsy (eg, clearly identify the risky cytologic features if present that were not originally recognized in these under called cases).
Both parts II and III were conducted with the reviewers masked to clinical background and outcomes. A report on these 2 parts is published elsewhere in this issue of Cancer (Cancer Cytopathology).4
The objective of part IV (the test set) will be to apply the experience from part III, the training set, to the remaining AUC-US cases, the test set, and observe how the criteria that were selected as best predictors of HGUC actually perform. Reviewers will be Johns Hopkins Hospital cytopathologists who were not involved in the training set study as well as Johns Hopkins Hospital cytotechnologists. The results will be published in a subsequent article.
To avoid confusion that may arise from reporting variations produced by multiple pathologists at the microscope, the cytopathologists at the John K. Frost Laboratory of Johns Hopkins Hospital devised the template described above (Table 1). After 2 years, an evaluation of the success or failure of the standardized terminology was considered prudent, because we realized that the criteria were not as standardized as they might be and, thus, could account for variability in performance of the terminology. The time had come to define the performance and morphologic characteristics of each category in the template.
Table 2 describes the entire case load of urinary cytologic samples during the initial 2 years of the study at Johns Hopkins Hospital. The surgical pathology categories are simplified on the table and can be explained as follows: No lesion includes benign tissue or denuded surface; low-grade atypia includes papilloma, papillary hyperplasia, and papillary urothelial neoplasm of low malignant potential; high-grade atypia includes atypia that falls short of carcinoma in situ but with marked focal cytologic atypia and disorganized architecture. Biopsy results were in the database from the surgical pathology reports in the Pathology Data System, and slides were not reviewed.
Over half of specimens with biopsy-confirmed HGUC had a preceding cytologic diagnosis of AUC-H or HGUC (Table 3); when patients with AUC-US were added to the analysis, 80% of patients with HGUC had at least 1 abnormal urinary cytology result. Another way of looking at the data in Table 3 is to compare those specimens with biopsy-proven HGUC after an AUC-US diagnosis (82 of 183 patients; 45%) with those after an AUC-H diagnosis (58 of 82 patients; 71%). The difference in the 2 cytology categories was significant (P < 0.0001; Fisher exact t test).
Very few patients (n = 20) (Table 4) were detected in the low-grade cancer category, as expected, for the well known reason that cytologic changes in low-grade urothelial lesions are very subtle and are essentially normal in many instances. Of those patients who had biopsies with a diagnosis of AUC-H, 72% had urothelial cancer, either high grade or low grade, discovered at biopsy, compared with only 55% of those who had an AUC-US diagnosis (Table 4). Those samples that were diagnosed as low-grade urothelial carcinoma on cytology almost invariably were washings in which tissue fragments with central fibrovascular cores were present.
A subset of patients includes those whose cytology interpretations were inadequate or negative but who had clinical indications that resulted in cystoscopy and biopsy. An impressive percentage of these patients had HGUC on biopsy (Table 3, Fig. 2). Their clinical profiles will be included in another article.
Although the annual incidence of new cases of bladder cancer and death rates in the United States are relatively low compared with the incidence of cancers at the top of the list (lung, breast, colon, and prostate cancers), some 500,000 bladder cancer survivors in the United States require regular surveillance to monitor for new or recurrent lesions. In the absence of a better test, urologic surgeons still rely on morphologic changes in exfoliated cells recovered from voided urines as a screening test.5-7 Once an abnormal cytology sample is identified, the patient undergoes cystoscopy, during which another cytology sample and/or biopsy is obtained to identify and confirm the source of the first abnormal cell population. From that point, the patient is managed by a surgical procedure and/or intravesical chemotherapy or immunotherapy to eradicate the tumor cells. Surveillance after treatment begins with urinary cytologic examination every 3 months for the first year or 2, depending on the patient's risk factors and staging. Cytologic screening intervals are extended, depending on previous cellular and cystoscopic findings. Cytology continues to be used for the remainder of the patient's life.8, 9 Because of the successful management of these patients, bladder cancer is the most expensive cancer per patient from diagnosis to death, usually from other causes.10
Thus, urinary cytology remains important in the diagnosis of bladder cancer and follow-up despite the well known low sensitivity for low-grade lesions.5, 6, 11, 12 We believe that the perceived poor performance of urinary cytology derives from the way in which it is used. Instead of attempting to identify all lesions originating in the bladder and upper collecting system, we propose using the test to its best advantage: identify the high-grade lesions, which the test does very well! Such an approach will reap the greatest benefit from the test, because HGUC is life threatening if allowed to invade. Low-grade lesions increase a patient's risk of developing a high-grade lesion but, in and of themselves, pose no danger to the patient's life.
The review article written by Owens and colleagues2 and published in this issue describes the history of terminology for urinary tract samples. To our knowledge, there has never been a National Institutes of Health consensus conference for urine cytology terminology, like that for gynecologic and thyroid cytologic specimens. However, in 2003, the Papanicolaou Society for Cytopathology convened a panel to reach a consensus for processing and reporting cytologic specimens from the genitourinary tract.13 Most schemes have tried to be over-reaching, ie, attempting to achieve exact concordance with biopsy results. The senior author of the current article did so for most of her career, including the publication of a volume with Raab,14 carefully defining and illustrating the criteria applicable to each diagnostic category. Another more recent atlas by Ali et al15 continued in the same way, attempting to correlate cytology and histology for all grades of lesions. However, when all samples are tallied over time, the result is a poor performance of cytology in predicting the presence of low-grade lesions but a superior and reliable performance in predicting high-grade lesions, either flat or papillary.
The Hopkins template for urologic cytology is novel in approaching cytologic diagnosis from a vantage point of strength, ie, emphasizing the critical target: HGUC. Although there will be times when a specimen contains a minibiopsy, and unequivocal evidence of a low-grade urothelial lesion is present, those events will be rare.
In conclusion, the Johns Hopkins Hospital cytopathologists' enthusiasm for the Hopkins template for urologic cytology samples is shared by the urologic surgeons, effectively targeting those patients who need to undergo cystoscopy. Because of the unacceptably large numbers in the AUC-US category, the major objective in subsequent studies will be to define that category to include only those cases that have clinical relevance. The AUC-H category performed remarkably well and was small enough to be meaningful (Fig. 1). In fact, it performed so well that it could be folded into the HGUC cytologic category. We hope to meaningfully define AUC-US, reducing it to 10% to 15% of diagnoses, so that the remaining AUC-US samples can be included in the negative (NUAM) category. Those results will be reported separately in subsequent articles.
Once the morphologic criteria are defined for each category based on the performance of the criteria, other institutions in various clinical settings will be recruited to establish how the template performs in the “real world.” Until the molecular and genetic profiles of bladder cancer are defined, our best defense against HGUC continues to be urinary cytology.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 2A review of reporting systems and terminology for urine cytology. Cancer (Cancer Cytopathol). 2013; 000: 000-000., , , .
- 4The Johns Hopkins Hospital Template for Urologic Cytology Samples—parts II and III: improving the predictability of indeterminate results in urinary cytologic samples: an outcomes and cytomorphologic study. Cancer (Cancer Cytopathol). 2013; 000: 000-000., , , , , .
- 9Cytology: Diagnostic Principles and Clinical Correlates. Philadelphia, PA: Saunders; 2008., .
- 11Is urinary tract cytology still useful for diagnosis of bladder carcinomas? A large series of 592 bladder washings using a 5-category classification of different cytological diagnoses. Cytopathology. 2007; 18: 79-83., , .
- 13Review of the state of the art and recommendations of the Papanicolaou Society of Cytopathology for urinary cytology procedures and reporting: the Papanicolaou Society of Cytopathology Practice Guidelines Task Force. Diagn Cytopathol. 2004; 30: 24-30., , , , .
- 14Cytologic Detection of Urothelial Lesions. New York: Springer; 2006., .
- 15Atlas of Urinary Cytology with Histopathologic Correlations. New York: DemosMedical; 2010., , , .