The Impact of Primary Versus Secondary Muscle-invasive Bladder Cancer at Diagnosis on the Response to Neoadjuvant Chemotherapy

Take Home Message Our study generates the hypothesis that patients with a primary diagnosis of muscle-invasive bladder cancer (MIBC) might have a better pathologic response to cisplatin-based neoadjuvant chemotherapy in comparison to patients with a history of non–muscle-invasive bladder cancer that progressed to MIBC.

Outcome measurements and statistical analysis: The outcome measures were pathologic objective response (pOR; ypT1 N0), pathologic complete response (pCR; ypT0 N0), overall mortality, and cancer-specific mortality. Results and limitations: The primary MIBC group had higher pOR (51% vs 34%; p = 0.02) and pCR (33% vs 17%; p = 0.01) rates in comparison to the secondary MIBC group. On multivariable logistic regression analysis, primary MIBC was independently associated with both pOR (odds ratio [OR] 0.49, 95% confidence interval [CI] 0.26-0.87; p = 0.02) and pCR (OR 0.41, 95% CI 0.19-0.82; p = 0.02). However, on multivariable Cox regression analysis, primary MIBC was not associated with overall mortality (hazard ratio 1.70, 95% CI 0.84-3.44; p = 0.14) or cancer-specific mortality (hazard ratio 1.50, 95% CI 0.66-3.40; p = 0.3). Genomic analyses revealed a significantly higher ERCC2 mutation rate in primary MIBC than in secondary MIBC (12.4% vs 1.3%; p < 0.001). Conclusions: Patients with primary MIBC have better pathologic response rates to NAC in comparison to patients with secondary MIBC. Chemoresistance might be related to the different genomic profile of primary versus secondary MIBC. Patient summary: We investigated the treatment response to neoadjuvant chemotherapy (NAC; chemotherapy received before the primary course of treatment) and survival for patients with a primary diagnosis of muscle-invasive bladder cancer (MIBC) in comparison to patients with a history of non-muscle-invasive bladder cancer that progressed to MIBC. Patients with primary MIBC had a better response to NAC but this did not translate to better survival after accounting for other tumor characteristics.

Introduction
Neoadjuvant chemotherapy (NAC) with a cisplatin-based combination followed by radical cystectomy (RC) is the standard of care for patients with cT2-4N0M0 muscleinvasive bladder cancer (MIBC) who are cisplatin-eligible [1]. Although current evidence supports an overall improvement in survival, not every patient will benefit from NAC [2,3]. Patient selection is of paramount importance in order to identify those who will not respond to NAC to avoid NAC toxicity and a delay to definitive therapy. To address this issue, several working groups have investigated the predictive role of clinical, pathologic, and molecular characteristics [3][4][5][6][7]. In this context, retrospective series have shown that non-muscle-invasive bladder cancer (NMIBC) progressing to MIBC (secondary MIBC) might have differential oncologic outcomes in comparison to primary MIBC [8]. However, the role of tumor status, especially in patients treated with NAC, has not been fully elucidated [8][9][10]. To fill this gap in knowledge, we investigated the association of primary versus secondary MIBC with pathologic response to NAC and survival using data from an international collaborative group.

Patient characteristics
We reviewed our multi-institutional database to identify 1002 patients None of the patients was treated with split-dose NAC.

Endpoints
The primary endpoint of the study was the association of tumor status (primary vs secondary) with pathologic objective response (pOR), defined as stage ypT1N0 at RC after NAC [11]. Secondary endpoints were the association of tumor status with pathologic complete response (pCR), overall mortality (OM), and cancer-specific mortality (CSM).

Statistical analyses
Point estimates and 95% confidence intervals (CIs) for pathologic response were generated using exact binomial distributions. The association of tumor status with pOR and pCR was investigated using univariable and multivariable logistic regression analyses. The association of tumor status with OM and CSM was investigated using univariable and multivariable Cox regression analyses. The multivariable models were adjusted for clinicopathologic characteristics known to be associated with the outcomes investigated. Survival was plotted using the Kaplan-Meier method and compared using the log-rank test. The discrimination of the multivariable models was assessed using the Harrel c index. The frequency of somatic genomic mutations was compared between groups using Fisher's exact test. All tests were two-sided and statistical significance was set at p < 0.05. We used R (R Foundation for Statistical Computing, Vienna, Austria) for statistical analyses.
In the cohort used for genomic analysis, 78 patients had secondary MIBC and 398 had primary MIBC. Overall, the number of gene mutations was comparable between the two cohorts (Fig. 2B, C). We found a significantly higher ERCC2 mutation rate in primary MIBC than in secondary MIBC (12.4% vs 1.3%; p < 0.001; Fig. 2D). There was no significant difference in overall survival between the primary and secondary MIBC groups (p = 0.4; Fig. 2A).

Discussion
We investigated differential oncologic outcomes for patients with primary and secondary MIBC. We found a significant association between tumor status and pathologic outcomes. Our findings have several significant implications for clinical decision-making and patient counseling. While NAC is recommended for all patients with cT2-4N0 MIBC, not every patient is likely to respond to this therapy [15]. Therefore, tools for patient selection are needed in daily    practice. Molecular subtyping and biomarkers might help clinicians in the near future [6,16]. However, until their clinical impact and cost effectiveness are evaluated in prospective trials, physicians must rely on clinical and pathologic characteristics and patient comorbidities in the decision-making process [3,5,17,18]. Our study adds to the relevant information that might help during this process. We showed that tumor status might be associated with NAC response and this could help in patient selection for upfront RC if confirmed in prospective trials. Several studies have investigated the association of tumor status with oncologic outcomes, with conflicting results [10,[19][20][21][22]. However, these studies did not assess the effect of tumor status in patients treated with NAC. A recent systematic review and meta-analysis found similar 5-yr and 10-yr overall survival and cancer-specific survival (CSS) rates between primary MIBC and secondary MIBC [8]. Interestingly, in subgroup analyses of patients treated with NAC, those with secondary MIBC had worse 5-yr CSS (HR 1.5; p = 0.04) but not 10-year CSS. These findings generate the hypothesis that there might be a differential response to NAC according to tumor status and that worse survival might be attributable to a delay in RC. However, these data are from a small number of retrospective single-center studies with a limited number of patients. Moreover, these series did not report on pathologic response rates to NAC [17,23,24].
We found that patients with secondary MIBC had higher rates of non-organ-confined disease (stage ypT3/4 and/or N +) at RC, were less likely to respond to NAC, and had shorter survival in comparison to patients with primary MIBC. Our findings are in accordance with results reported for a retrospective series of 288 patients [9]. The authors hypothesized that one of the reasons for this detrimental effect could be clonal cell selection resulting from previous intravesical chemotherapies and bacillus Calmette-Guérin immunotherapy.
Alterations in DNA damage repair genes have been associated with chemosensitivity to cisplatin-based NAC [12][13][14]. Therefore, we investigated genomic differences between primary and secondary MIBC as a possible rationale to explain the differential response. Our genomic analysis of the MSK/TCGA cohort showed that the ERCC2 mutation rate was higher in primary MIBC than in secondary MIBC. Somatic missense mutations in ERCC2 have emerged as clinically significant biomarkers for chemotherapy response in bladder cancer in several trials [13,14,25]. Our study confirms the findings of a previous report showing a significantly higher rate of ERCC2 mutation in primary MIBC [9] and reinforces the theory that this might correlate with better chemosensitivity and therefore better pathologic response and survival.
Currently, the majority of the evidence shows that for patients treated with RC alone, tumor status is not associated with oncologic outcomes [8]. However, our analysis generates the hypothesis that this might not to be true for patients treated with NAC and RC. Chemoresistance related to disparate genomic characteristics between primary and secondary MIBC might lead to a delay in RC, disease progression, and therefore worse oncologic outcomes. This could partly explain our findings of worse pathologic responses and survival for patients treated with NAC but comparable survival in the MSK/TCGA cohort of patients treated with RC only.
There are several limitations to our study that should be considered. We acknowledge the selection bias inherent to the retrospective design and the lack of a control cohort of patients treated with RC only. We could not account for patient performance status, comorbidities, surgical quality, and other nonmeasurable confounders. We had no granular information on previous intravesical therapies in the secondary MIBC cohort. There was no central pathology review of the specimens. The preoperative staging and follow-up were not standardized but were based on guidelines [1] and institutional protocols. The median follow-up in our cohort was relatively short and this might have limited the ability to detect a significant difference in multivariable survival analyses. Finally, we acknowledge the long study period resulting in cohort heterogeneity and different treatment protocols between centers over the years.
Despite all the limitations, our findings could serve as rationale for patient counseling in clinical practice and a basis for planning prospective trials with the aim of improving patient selection for NAC.

Conclusions
Our study generates the hypothesis that patients with secondary MIBC are less likely to respond to NAC in comparison to patients with primary MIBC. This might be related to genomically driven chemoresistance mechanisms that have emerged over the tumor natural history from NMIBC to MIBC, partly as a result of therapy-related clonal selection. These findings suggest that tumor status could be applied to prospectively guide therapy decisions regarding NAC versus immediate RC. However, further evaluation in prospective trials is warranted.
Financial disclosures: David D'Andrea certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.
Funding/Support and role of the sponsor: None.