This study was population based within a large, defined geographic region. The case collection methodology, utilizing a state cancer registry to which cancer reporting is mandated by state law, ensured that the biopsies reviewed were from the time of the initial diagnostic work-up. Cooperation from pathology laboratories was excellent, with a greater than 90% slide retrieval rate. Virtually all cases were reviewed by a single urological pathologist to maximize standardization and uniformity of diagnostic criteria.
All these parameters suggest that the distribution of tumors in the present population-based study more accurately reflects prevalence for bladder cancer grade and stage than do studies based on clinic- or hospital referral-based patient series. Referral populations are potentially skewed by patients with higher stage, difficult-to-treat or otherwise problematic tumors that may constitute a significant proportion of the patient population in academic or specialized medical centers. The present findings suggest that the distribution of tumor types may vary by gender and age, and thus the demographics of the selected referral population could affect the distribution of tumor types.
compares pathological stage distribution between the present population-based study and representative previously reported studies [8
], all but one of which [14
] were based on selected, hospital- or clinic-based patient populations. The hospital- or clinic-based series show a range of non-invasive (stage pTa) tumors between 30 and 49%, compared with 66% in this population-based sample. Conversely, muscle invasive tumors (stage T2 or greater) in hospital- or clinic-based series range from 12–54%, compared with only 8% in the present study. One of these studies [21
], although stated as unselected, comprised a consecutive group of “randomly” referred patients to a university hospital, and the data from that study are entirely in keeping with all of the other hospital- or clinic-based series. By contrast, the one previous study with a population-based design comparable to the current study shows a very similar stage distribution. These data highlight the higher proportion of non-invasive (stage Ta) tumors and lower proportion of high-stage (≥T2) tumors in general population series than depicted in selected series.
Comparison of pathological stage at presentation with eight selected, hospital- or clinic-based studies.
In theory, the effect of patient selection could be diminished by limiting comparisons to low-stage tumors. presents the WHO/ISUP classification of non-invasive papillary tumors across several studies [23
]. The data on tumor distribution of this study’s patients is close only to the one study [23
] that was limited to patients with primary non-invasive (stage Ta) carcinomas that was derived from the general population. In the other two studies, patients of all stages were examined, and the distributions differed. Thus, this comparison would tend to support the premise that patient selection affects the grade as well as the stage distribution.
WHO/ISUP tumor distribution of stage pTA (non-invasive) papillary tumors: comparison between population-based and hospital- and clinic-based studies.
presents a comparison of the WHO (1973) grading system between hospital- and clinic-based [21
] and population-based [28
] studies. The two studies with population-based design (the current study and that of Holmäng et al. [28
]) show very similar grade distribution results. The two hospital- or clinic-based studies [21
], in contrast, show considerable variation. Such differences could potentially be explained by a non-standard application of grading criteria. For instance, the tendency to lump tumors into the middle grade of a three-tiered grading scheme is one of the flaws of the WHO (1973) system that was cited to help bolster the adoption of the newer WHO/ISUP system [18
]. The grade distributions, however, are probably at least as likely to reflect differences in referral patterns and patient case-mix in the patient populations at tertiary care institutions.
Comparison of population-based vs hospital- or clinic-based studies by WHO (1973) grade (all stages).
The data presented in this report are derived from a geographic location in the continental USA with high incidence and mortality rates from bladder cancer [29
], and this could potentially limit the generalizability of the findings. New Hampshire is less industrialized and more uniformly Caucasian in population than many other states. Further, this study was undertaken in the context of a larger study exploring possible environmental influences, including arsenic, on the high incidence of bladder cancer. This could possibly explain the high percentage of cases of urothelial carcinoma in situ in this study in comparison to other studies ( and ); this effect could be related to an excessive environmental exposure, e.g. to arsenic, but this requires further investigation. It is conceivable, although unlikely, that the similarity of the results to those of Holmäng et al. [14
] from Scandinavia could in part be related to similar ethnic mix, levels of industrialization, lifestyle or other demographic characteristics, rather than to the population-based character of the study population. Future analyses to evaluate potential histological patterns associated with exposure history may provide helpful etiological clues.
The exclusion from the present study of patients older than 74 years means that the population cannot be described as completely non-selected. The age range was determined by the criteria used for an epidemiological case-control study [16
]. The rationale for exclusion was based on the facts that not only do response rates tend to be lower among older patients, but also the diagnoses of cancer may be obscured and hence underreported among older patients in whom multiple diagnoses coexist. In the present data, no marked differences in histological characteristics by age were detected. In another study, the mean age at diagnosis of a stage T2 cancer was 75 years [14
], while the mean ages for Ta and T1 tumors were 70 and 73 years, respectively. Thus, the possibility cannot be excluded that omission of cases in patients older than 74 years in the current study could have led to an underestimation of advanced (stage 2 or greater) disease.
Another limitation is that in this study about 10% of bladder cases diagnosed during the study period could not be retrieved. Although there is no reason to suspect a hidden selection bias, this possibility cannot be completely excluded. In addition, the pathological review of cases by a single pathologist maximizes uniformity of diagnostic criteria and standardization, but may result in subjectivity and an idiosyncratic detection bias. The published diagnostic criteria of the WHO/ISUP classification system, however, have been profusely illustrated and described in unusual detail; this was done expressly to reduce diagnostic variability and subjectivity [18
]. Also, the similarity of the data to those of other studies with a population-based design [14
] suggests similar application of diagnostic criteria. In addition, internal checks for intraobserver variability supported the intrastudy stability of diagnoses.
In conclusion, this study provides grading and staging data from the time of initial bladder cancer diagnosis that cannot be derived from selected, hospital-based population studies. Thus, a considerably higher proportion of non-invasive and low-grade urothelial tumors, and a much lower percentage of muscle-invasive tumors likely exist in the general population than what is inferred from clinical or hospital referral-based patient series.