This study examines a population-based cancer registry, comparing survival among bladder cancer patients with single and multiple malignancies. Using SEER data collected from 1973–2004, we determined that AB and ABS patients have a distinct survival disadvantage when compared to BO patients. Several factors may account for this finding. First, patients with antecedent cancers who develop bladder cancer may have inherent deficiencies in immune surveillance, DNA repair, or epigenetic mechanisms leading to the development of multiple malignancies [7
]. Similar acquired defects may also play a role in tumorgenesis in patients with previous cancer diagnoses [8
]. Moreover, the development of these deficiencies may be accelerated by smoking, obesity, and/or treatment burden (chemotherapy or radiation) associated with antecedent malignancies. Several well-known examples illustrate this point.
Cyclophosphamide has a clear association with bladder cancer—it is for this reason that urologists perform bladder cancer screening in this population [10
]. In addition, patients who receive radiation for the treatment of prostate cancer are more likely to be diagnosed with bladder cancer, and more likely to develop high-grade disease and have diminished survival [11
]. The development of subsequent cancer of the bladder and other sites is seen predominantly among patients who have undergone external beam radiation, not brachytherapy [14
]. These observations may partially account for the negatively affected survival among the AB and ABS cohorts in this study. Of note, the most common nonbladder malignancy in all multiple malignancy groups was prostate cancer. Many patients in the BS group likely had their prostate removed at the time of cystectomy. We suspect that many of the prostate cancers diagnosed in this group may have been found incidentally upon review of surgical specimens. This may explain the high incidence of subsequently diagnosed prostate cancers among a group where many have had their prostate removed. Future work might employ SEER-Medicare data to examine the correlation between prostate radiation and bladder carcinoma and differential survival among patients with both these malignancies.
Aside from acquired genetic or immune surveillance deficiencies, treatment burden from previous cancers may also predispose patients to noncancer related comorbidities such as chronic renal insufficiency and cardiovascular disease, leading to decreased survival [15
]. These comorbidities may also preclude the use of optimal cancer therapies in patients with antecedent cancers, as they may be unable or unwilling to undergo recommended treatments. This factor almost certainly contributes to our observation of diminished survival in the AB and ABS patient cohorts. Although seemingly intuitive, these findings are important to document because they can and should play a role in clinical practice. Treatment decisions may be significantly influenced by such information, and counseling a patient with only bladder cancer is quite different than a patient with bladder cancer and an antecedent malignancy.
Contrary to our observations in the AB and ABS cohorts, we observed that patients who develop a malignancy subsequent to high-stage bladder cancer had improved overall, cancer-specific and bladder cancer-specific survival when compared to patients with only high-stage bladder cancer. Although this specific finding supports our initial hypothesis, we believe it is more likely the result of statistical and not biological phenomena. The BS group has the longest mean survival from the time of diagnosis of bladder cancer of any group analyzed. This finding raises the possibility that the reason for the BS group's apparent survival advantage is that the only patients included in the BS cohort are those who live long enough to develop another cancer following their bladder cancer diagnosis. This effect is specifically seen amongst high-stage cancer patients since they are most likely to die of their disease; those receiving a second cancer diagnosis may become part of the BS group since they have outlived other high-stage bladder cancer patients, not because their second malignancy confers a protective effect. While it is conceivable that a true biological phenomenon is responsible for improved survival in some patient cohorts with an increased cancer burden, this appears unlikely and is not possible to fully determine given our data.
However, our study is not the first to speculate on improved outcomes among patients multiple malignancies patients. Duchateau and Stokkel examined the prevalence of multiple malignancies among patients with nonsmall cell lung cancer (NSCLC) and compared survival between patients with only NSCLC and various cohorts with NSCLC and additional cancers [17
]. They found that patients with only NSCLC had the least favorable outcome, whereas those with at least two cancers in addition to a primary diagnosis of NSCLC had the greatest survival, and patients with one cancer in addition to a primary NSCLC diagnosis had an intermediate outcome. One weakness of this study may be that approximately 25% of patients had multiple NSCLC tumors that were included in the analysis as separately diagnosed malignancies. The authors note that TNM stage and treatment were comparable between all groups, and they speculate that tumors in patients with multiple malignancies may actually have distinct growth habits that are responsible for the finding in this study.
Our analysis sought to control for a number of important confounding variables, including: age, sex, race, grade, stage, and year of diagnosis. It should be noted that while we used the reported cause of death in SEER for our analyses, it may be especially difficult to assign a single cause of death to a patient with multiple cancers. Also, since SEER records cancer diagnoses just as they are reported, it is possible that some reported second malignancies were not diagnosed histologically and were actually mestastatic lesions. Finally, we considered examining whether or not differences in treatment existed between cohorts; however data was inconsistent and not comprehensive within this SEER dataset (SEER-Medicare allows more complete access to treatment data), and so we chose not to include it as a variable. It would be useful to investigate possible treatment and practice pattern differences between these cohorts in a future study, as determining if such differences do exist may help elucidate reasons for variable outcome between cohorts.
The SEER Summary Staging classification, which differentiates low stage (local) and high-stage (regional or distant) disease, is an important limitation of this study. The American Joint Committee on Cancer (AJCC) method is more commonly used in clinical settings, while SEER has standardized and simplified staging to ensure consistent definitions over time. AJCC staging is available for more recent years through SEER, however, we chose to use the SEER Summary Staging classification because it was the most consistent over the entire time period examined (1973–2004). The limitation exists due to the variability of patients in the high- and low-stage groups. For example, “low-stage” encompasses noninvasive and localized invasive disease (Ta, Tis, T1, and localized T2) in the same cohort, even though the patient population within this cohort is heterogeneous, as are treatment strategies; this makes comparison between BO, AB, ABS, and BS cohorts by stage more difficult to interpret. Part of the explanation for the survival advantage among high-stage BS patients relative to high-stage BO patients may stem from this limitation. Comparisons between overall cohorts may also be affected by these groupings. Unfortunately, the extent of such an effect is hard to measure. Nevertheless, we believe that the sample size of each cohort is large enough that the observations of diminished survival among multiple malignancy patients are valid.
Finally, we included all bladder cancer reported in SEER as part of this investigation and did not differentiate urothelial carcinoma from other bladder malignancies. Given that urothelial carcinoma comprises a large proportion of bladder cancer (>90%), the effect on this study is likely minimal, but worthy of mention. Because cancers such as small cell carcinoma or squamous cell carcinoma of the bladder typically present at higher stages and carry worse prognoses, it is possible that these are disproportionately represented in the “high-stage” groupings, thereby negatively skewing survival data for these patients. Again, however, the proportion of bladder cancers represented by these rare diagnoses is small and unlikely to significantly affect the study outcome.
It should also be noted that epidemiological information regarding the differential survival of cancer cohorts may not be appropriate to directly inform individual patient care. However, these findings are important to shape future research, better understand multiple malignancy patients, and ultimately improve oncologic practice.