We observed the prevalence of LOH to be approximately 40% that is lower that rates published in most series. This difference can be explained by the variation of microsatellite markers used (in type and in number) and by the ratio values retained to be indicative of LOH, essentially. The cut-off of the allelic-imbalance ratio varied from 20% to 50% [6
] in the series or was not indicated. Interestingly, Schneider et al have calculated the cut-off value for each microsatellite at 2 SD to obtain a better sensitivity [16
]. The prevalence of LOH seemed to be correlated with this cut-off: 94-97%, 87% or 75% for a retained signal decrease superior to 30%, 40% or 50%, respectively [3
]. The prevalence of AL (minimal reduction of 30% in the allelic-imbalance ratio) was in line with published values. It has become important to fix a single cut-off to be able to relate one series to another.
In addition, we solely included patients with superficial bladder carcinoma which may participate with the somewhat lower sensitivity in our study. Globally, comparison of a breakdown of the cohort in groups by stage of disease and by differentiation resulted in detection increasing with more advanced and aggressive disease [20
The present study confirms the highest rate of LOH affecting the chromosome 9 [2
]. Interestingly, allelic losses on chromosome 9p were more frequently observed than on chromosome 9q that was not consistent with recent findings [22
]. Difference of prevalence of 9p LOH between low-stage/grade and high-stage/grade tumors was not significant (except at 9p22.1 concerning the stage). Even tumors of early stage and low grade displayed microsatellite alterations at chromosome 9 and tumors with intermediate grade exhibited the higher prevalence of 9p LOH, essentially at 9p31.3. Comparatively, LOH at 17p13.1 is strongly associated with high-grade, high-stage tumors and was not found TaG1/G2. The results confirm that chromosome 9 LOH and 17p13 LOH represent two distinct pathways of tumor progression to high stage TCC, one being from Ta tumors and exhibiting 9p LOH [24
]. Chromosome 9 losses occur early during urothelial tumorigenesis [9
], whereas mutations in the p53 gene, located at 17p13, were commonly seen in high-grade tumors [27
]. The invasive tumor population would be a combination of tumors with different backgrounds. Interestingly, LOH at 8p21.1 and 11p15.5 were also identified in this study as frequent progression-associated alterations [20
No locus was correlated with tobacco carcinogens, LOH at markers studied may reflect a different way of carcinogenesis. Prevalence of LOH in the smoker subgroup was lower (30% versus 50% in the non-smoker subgroup, NS). These findings can limit the interest of microsatellite analysis as non-invasive method for screening smoker high-risk population.
In the majority of cases, results of tumor tissue were identical or more sensitive than urine sediment analyses. The lack of detection in urine sediment could be attributable to a limited amount of tumor cells in the collected urines and to the presence of a mixture of cancer, normal and inflammatory cells [8
]. For about 20% of cases, allelic losses were detected in urine sediment but not in tumor tissue [3
]. One possibility is that the urine sediment contained a more advanced tumor cells clones, which were not sampled during transurethral resection. Close surveillance is required for these patients because LOH may bear witness of flat lesions, such as dysplasia or carcinoma in situ that exfoliate easily. In this case the risk of recurrence within short delay seems to be probably high [7
]. Furthermore, normal looking mucosa adjacent to or distant from urothelial tumor displays frequently genetic alterations [28
]. In addition, tumors originating from the upper urothelial tract also may cause microsatellite alterations. The other hypothesis is that LOH in tumor specimen have been masked by cells of the normal mucosa.
Results showed the high specificity of LOH achieved by microsatellite analysis for diagnosis of superficial tumors in a population with symptoms suggestive for bladder cancer, in line with published studies [16
]. Comparatively, application of urinary cytology is largely limited by low sensitivity and operator dependency, especially for low-grade tumors. Reducing the cut-off of the allelic-imbalance ratio provided better sensitivity, but a strong loss of specificity [30
]. These results indicate that microsatellite analysis may serve as an important and complementary tool to cystoscopy for non-invasive diagnosis of bladder cancer. For the molecular detection of tumor cells, the interest is to provide the urologist with the highest sensitivity. Moreover the low sensitivity and some discrepancies between alterations observed in tumor tissue and urine don't suggest that a cystoscopic examination may be omitted if microsatellite analysis is negative. Nevertheless, the very high specificity makes microsatellite of urine sediment a good adjunct to cystoscopy and/or cytology. Presence of LOH as defined previously in our series must force the physician to look for malignant urinary tract disease. If investigations are negative, it is reasonable to propose that these patients suspicious for cancer lesions would benefit from careful follow-up. One of the interest of our series was that sensitivity and specificity calculations were performed among a population consulting for symptoms suggestive for bladder cancer, and not among healthy controls, as described previously [20
Interestingly, LOH at chromosome 9 seemed to be specific of urothelial neoplastic origin whereas losses at chromosomes 5, 17 and 18 were detected in non-urothelial urinary tract tumors. However, no yet-defined set of modified markers can specifically identify cellular types or tumor localization.
Recently, van der Aa et al. have studied the utility of urine microsatellite analysis during the follow-up of bladder cancer patients [31
]. They found that consecutive positive microsatellite analysis results were a strong predictor for future recurrences. In a 228 patient cohort, the risk to develop a recurrence at 24 months reached 83% if microsatellite analysis was persistently positive compared with only 22% when microsatellite analysis was persistently negative. However, in line with our results, they emphasized that sensitivity needed to be improved, for example, by patient selection and testing of additional genetic markers in urine samples.
Further study is warranted to evaluate the prognostic value of LOH on recurrence, progression and muscle invasion. Our group has recently showed that LOH at 9p22 was a significant and independent predicting factor of global progression and progression to muscle-invasive bladder cancer [32
]. Prognostic evaluation of a combination of different molecular alterations, such as LOH, FGFR3 and p53 mutations, seems to be interesting to better identify patients at high risk of progression and muscle invasion [33