Evaluation of atypical urine cytology progression to malignancy
In urine cytology, the diagnosis of atypia is subjective and clinical management based on these results can be difficult to determine. In this study, the authors determined the percentage of atypical urine diagnoses that progressed to positive cytology or surgical pathology results over an 11-year period.
In a retrospective review of the authors’ institution, 1320 atypical urine cytology diagnoses were identified in specimens from 851 patients obtained from January 2000 through December 2010. All subsequent pathology reports were reviewed to determine which patients developed positive cytology/surgical pathology diagnoses. In total, 4106 cytology and surgical pathology specimen reports were reviewed.
At the authors’ institution, 1320 of 16,299 of urine cytology specimens (8.1%) were diagnosed as atypical during the 11-year period. Overall, 271 of 1320 initial atypical urine specimens (21%) progressed to positive cytology or surgical pathology results with a mean time to progression of 155 days. Of the cases that progressed to malignancy, 118 were high-grade urothelial carcinoma and 92 were low-grade urothelial carcinoma.
The rate of atypia in urine specimens at this institution was 8.1%. Of the specimen types, atypia was the most common in urinary diversion specimens (16%) and the least common in upper tract cytology (3.8%). When diagnosed as atypical, upper tract specimens had the highest percentage of progression to high-grade carcinoma. Therefore, the authors concluded that the diagnosis of atypia in this specimen group has higher clinical significance and should be managed more aggressively. Cancer (Cancer Cytopathol) 2013;121:387–391. © 2013 American Cancer Society.
Urine cytology is an important modality both in screening for new urothelial carcinomas and for surveillance of cancer recurrence, yet there is no strong consensus regarding a comprehensive classification system for urine cytology like that for cervical cancer screening. The most commonly used method for urine cytology is the 5-category system, which includes unsatisfactory, negative, atypical, suspicious, and positive for high-grade urothelial carcinoma and this is the system used at our institution. Yet, this system often has challenges when trying to manage patients with inconclusive cytology, particularly those with atypical diagnoses.
One of the most difficult challenges faced in urine cytology is the wide interobserver variability of urine cytology diagnoses. Glatz et al tested 246 pathologists and reported that high-grade urothelial carcinoma was diagnosed correctly on cytology 76% of the time, whereas pathologists rendered a low-grade carcinoma diagnosis only 33.9% of the time. Furthermore, 54.5% of pathologists misdiagnosed decoy cells as malignant. This demonstrates the extensive variability in interpretation and misinterpretation of urine cytology specimens on a routine basis and the wide interobserver variability that exists.
Another challenge to urine cytology is the wide cytologic variability observed in atypical specimens and many modifications to the 5-category system have been suggested to improve diagnosis and interpretation of atypical results. Renshaw has suggested an extensive subclassification of atypical urine diagnoses into 7 pattern subtypes: cell clusters, reactive/repair, well preserved with diffuse mild atypia, well preserved with diffuse significant atypia, well preserved with focal mild atypia, poorly preserved with focal degenerated atypia, and poorly preserved with focal pseudodegenerated atypia. Brimo et al suggested a simpler approach by dividing the atypical category into 2 groups: “atypical reactive” and “atypical unknown.” Piaton et al suggested applying the Bethesda terminology of “atypical cells of undetermined significance” and “atypical cells cannot exclude high grade” to better explain the risk of progression in urothelial atypia. To date, none of these approaches have proven to give clinicians better insight into malignant potential of a specimen and it remains difficult to determine clinical management. Therefore, clinicians at our institution most often choose to do routine follow-up, treating an atypical urine diagnosis as negative or inconclusive. Yet previous smaller studies have indicated that a significant portion of atypical results do signify an underlying malignancy that needs to be addressed.[5-8]
We studied patients who were diagnosed with atypical urinary cytology at our academic center over an 11-year period then reviewed their follow-up pathology results to determine the significance of an atypical urine cytology specimen. We analyzed the malignant potential of atypical urine cytology results to provide better guidance to manage these patients more effectively.
MATERIALS AND METHODS
In total, 16,299 urine specimens were collected from January 2000 to December 2010 in a retrospective review of our institution. Of these specimens, we identified 1320 atypical urine cytology diagnoses. We recorded the initial atypical urine cytology diagnosis then documented each patient's follow-up cytology and histology results up to June 2011. The length of patient follow-up ranged from 6 months to 11.5 years. The demographic data for each patient were recorded, including age, sex, clinical history, and presenting symptoms (when available). In total, 4106 cytology and surgical pathology specimen diagnoses were retrieved from 851 patients. All initial cytology data were gathered from the CoPath anatomic pathology database (1994-2003; Cerner DHT, Inc., Waltham, Mass) and all follow-up data were gathered from our institution's electronic medical records (1979-2011; Epic Systems Corporation, Verona, Wis). All cytology and histology diagnoses recorded were the originals and none of the diagnoses were changed.
In this study, an atypical diagnosis that progressed to positive urine cytology or malignancy on surgical pathology was considered a single event. Then, if the same patient had a new atypical result after a positive diagnosis, that new atypical diagnosis was treated as a separate new event. Thus, there were 1320 initial atypical urine specimen events from a total of 851 patients.
Previous studies have included only 1 year of follow-up histology to assess cytologic/histologic correlation and to prevent the inclusion of new tumors. However, in the current study, we wanted to assess the presence of both current malignancies and precursor dysplastic lesions, because both have the potential to be identified on cytology. Furthermore, we chose to include all 1320 atypical results in our study, including those with no follow-up urology results. Because the follow-up for an atypical urine diagnosis is largely based on the clinical findings and clinician's suspicion, we did not want to discount any atypical results by excluding them from this study. Also, because we had no accurate way of determining which patients did not require follow-up given their clinical history versus those patients who chose not to seek further treatment at our institution, we chose to include all of these patients in our results.
For statistical purposes, the various categories for atypical results were compared using a 2 × 2 contingency table. A 2-tailed P value was computed using the Fisher exact test and the chi-square test on GraphPad software (San Diego, Calif; available at: http://www.graphpad.com; [Accessed February 17, 2013]).
When analyzing the demographic information, we observed that our cohort of patients with atypical urine diagnoses was composed of 604 men and 247 women; just slightly less than the 3:1 ratio expected for patients with urothelial carcinoma. The mean age at first atypical urine diagnosis was 66 years (range 15-99 years).
At our institution, the total urine atypia rate was 8.1% (1320 of 16,299 specimens) among urine cytology specimens from January 2000 to December 2010. When divided according to specimen type, instrumented bladder washings had an atypia rate of 8.9% and voided urines had a similar rate at 8.7%. Catheterized urine had a slightly lower rate of atypia at 5.2%. Most notably, the lowest rate of atypia was observed in upper tract urine specimens (3.3%), whereas the highest rate of atypia was observed in urinary diversion specimens (16%), as indicated in Table 1.
Table 1. Atypical Cytology by Specimen Type
|Upper tract urine||2482||82||3.3|
Overall, 271 of the 1320 initial atypical urine cytology specimens (21%) progressed to a malignant diagnosis, with either positive cytology results or positive surgical pathology results. The mean time to progression was 155 days. This included 57 cases (21%) in which a positive specimen was gathered and diagnosed within 2 days of the atypical cytology diagnosis. With these cases, we presumed that the atypical cytology specimen was from an already present malignancy, not from a precursor dysplastic lesion, and that the atypical diagnosis was due to either poor sampling of cells in the bladder or pathology under calling the specimen.
When we analyzed all specimens by subtype, although upper tract urine specimens were diagnosed least often as atypical, they had the highest rate of progression to a positive diagnosis (38%). In contrast, urinary diversion specimens were diagnosed most commonly as atypical on cytology, but had the lowest rate of progression to a positive diagnosis (12%). Voided urine specimens also had a low rate of progression to malignancy (14%). Instrumented bladder washings had a 22% malignancy rate and catheterized urine had a 17% malignancy rate, as indicated in Table 2.
Table 2. Specimens Low-Grade and High-Grade Malignancy by Subtype After Atypical Cytology
|Total: No. (%)||1320||271||(21)||92 (34)||118 (44)|
Of the 271 cases that progressed to malignancy, 118 were high-grade urothelial carcinoma (which included high-grade urothelial carcinoma, carcinoma in situ, invasive carcinoma, and urothelial metastasis) and 92 were low-grade urothelial carcinoma. Furthermore, 44 cases had positive cytology results with no surgical pathology follow-up and 17 were identified as other primary cancers or metastases (9 prostate, 7 renal clear cell, and 1 lung). Again, upper tract urine specimens had the highest rate of progression to a positive diagnosis. Upper tract specimens also had the highest percentage of high-grade malignancy, which was identified in 55% of positive upper tract specimens. Data on progression to high-grade and low-grade malignancy are provided in Table 2.
Two-tailed P values were statistically significant when upper tract cytology progression was compared to bladder washings (P = .0023), voided urine (P = .0001), and urinary diversion specimens (P = .0005). This comparison was not statistically significant only for catheterized urine cytology (P = .1048), likely due to the low number of catheterized urine specimens. Similar results were obtained with the chi-square test.
The focus of this study was to determine the progression to a malignant diagnosis after an atypical diagnosis was made on urine cytology. Previous studies have examined similar facets of this topic, however, to our knowledge, none have included such an extensive retrospective review. We determined that there is an 8.1% atypia rate at our institution with a 21% rate of progression to a positive diagnosis. Of the positive diagnoses, 118 were high grade, 92 were low grade, 44 were positive cytology only, and 17 were other types of malignancy.
Our atypia rate is comparable with the national averages, although it has been observed that atypia rates vary widely among institutions, ranging from 1.9% to 23.2%.[3, 6] Furthermore, our rate of progression to malignancy at 21% (271 of 1320 specimens) is comparable to the rates reported in previous studies. Deshpande and McKee observed that 47 of 201 atypical samples in their study progressed to a positive diagnosis at a rate of 23.4% and Bhatia et al reported that 18 of 58 specimens (31%) progressed to cancer on follow-up. The difference between our results and these previous studies is that we did not exclude patients with atypical urine specimens who did not have histologic correlation on follow-up. The previous studies only included patients who had surgical pathology results to correlate with cytology. Yet, given the important role that clinical suspicion plays on the part of the urologist, we believed that all patients with atypical diagnoses should be included in our study. Although there may be some patients who chose not to continue treatment at our institution, we believe that the majority of patients did not have histologic follow-up because the clinician did not believe it was warranted. We believe this is an important subset within the cohort of patients studied, thus we chose to include all atypical specimens in our data. That is likely why our rate of progression is slightly lower than those reported in the other studies, because we did not select only the patients with histologic follow-up results.
When comparing the diagnoses by subtype, we observed that upper tract specimens had the lowest rate of atypia (3.8%). However, once diagnosed as atypical, they were most likely to progress to a positive cancer diagnosis (38%). This is not surprising, because upper tract specimens are notoriously more cellular and may appear to have greater cytologic atypia compared with their lower tract counterparts, therefore the threshold for diagnosing atypia is much higher. Furthermore, upper tract urine specimens had the greatest percentage of high-grade lesions, with over half of the positive specimens diagnosed as some type of high-grade carcinoma, including high-grade urothelial carcinoma, carcinoma in situ, invasive carcinoma, and urothelial metastases. These results indicate that the clinician and cytopathologist should have a much higher clinical suspicion for malignancy when presented with an atypical diagnosis on upper tract specimens and they should consider managing these patients more aggressively.
Conversely, urinary diversion specimens had the highest rate of atypia (16%). This is not surprising given that the patient already has a history of cancer, so the threshold was much lower for diagnosing a specimen as atypical. Yet, although urinary diversion specimens had the highest rate of atypia, they also progressed to a positive diagnosis the least often (12%). These results indicate that clinicians may lower their clinical suspicion for malignancy in this subset and perhaps choose to manage with routine follow-up.
Our current results appear to differ compared with previous studies on the prevalence of malignancy. A study by Raab et al indicated that voided urines had the highest rate of progression to a positive diagnosis (68%), whereas upper tract urines had the lowest rate of progression (50%). Another study by Kapur et al compared bladder barbotage to voided urine cytology and indicated that voided urine cytology had a higher rate of progression to a positive cancer diagnosis (46.6%), but that study did not include either upper tract or ileoconduit cases. With both of those studies, a significant point is that they only chose to analyze atypical cytology with histologic correlation. Their results differ compared to our data where voided urine had one of the lowest rates of progression to malignancy (14%) and upper tract urines had the highest rate of progression (38%). Furthermore, when comparing high-grade and low-grade lesions detected on surgical pathology, Raab et al observed more low-grade malignancies on histologic follow-up than high-grade lesions whereas our study showed the opposite. More of our malignancies were high-grade carcinomas. Given the morphology and cytology of high-grade lesions, it would be reasonable to expect to observe more high-grade lesions on follow-up, as was observed in this study.
Although our current study was extensive, one of the newer aspects of urine cytology is the correlation with fluorescent in situ hybridization (FISH) results. At our institution, FISH data were obtained for all atypical urine samples from 2008 to 2010. We hope to further review FISH results from the 2008 to 2010 subset of patients to determine the significance of FISH data in predicting which atypical diagnoses actually are positive for malignancy.
Many previous studies have investigated similar data regarding atypical urine cytology, but we believe this is the largest retrospective review on the subject. However, we did identify a few challenges when interpreting our results. First, although these data encompass a large review, they reflect only the experiences at 1 urban academic institution and considerations must be made when extrapolating these data to different types of institutional settings like community-based practices and rural areas. Second, the atypical diagnoses were made by trained cytopathologists at an academic center which has higher percentages of cancer patients and where the pathologists subspecialize almost exclusively in cytology. Yet, overall, we believe the results of this study can have an impact across institutions in guiding the clinical follow-up required for patients with atypical urine cytology.
The focus of this study was a large retrospective review of 1320 atypical urine cytology cases over an 11-year period. Our results indicate that higher clinical suspicion should be noted on the part of the cytopathologist and the urologist after an atypical diagnosis on upper tract specimens is made and these patients should be managed more aggressively. Not only do upper tract specimens appear to have the highest rates of progression to malignancy, they also have particularly high percentages of high-grade lesions discovered on follow-up.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.