Intravesical mitomycin-C administered immediately before transurethral resection of bladder tumor in non–muscle-invasive bladder cancer: Clinical outcomes and molecular predictors from over 3 years of extended follow-up in a phase II trial Authors: Hye Won Lee, Geehyun Song, Hyung Ho Lee, Eui Hyun Jung, Seo-Young Kim, Hanbyeol Kim, Weon Seo Park, Joonha Kwon, Ho Kyung Seo Published: 2026 ABSTRACT PurposeTo evaluate the long-term outcomes and molecular correlates of response after immediate preoperative intravesical chemotherapy (IPeIC) with mitomycin-C (MMC) in patients with non–muscle-invasive bladder cancer (NMIBC).Materials and MethodsIn this single-center, open-label, randomized phase II trial, 33 patients received two split doses of IPeIC/MMC (40 mg/20 mL), whereas 38 patients underwent transurethral resection of bladder tumor (TURBT) alone. The primary endpoint was 3-year recurrence-free survival (RFS), and secondary endpoints included progression-free survival (PFS). Exploratory RNA sequencing was performed on IPeIC-treated patients (three with recurrence, 25 without) using a Monte Carlo-based resampling strategy.ResultsThe median follow-up durations were comparable between the intervention (60.0 months) and control arms (60.4 months). IPeIC/MMC reduced recurrence risk by 76.8% versus TURBT alone (p=0.024), yielding a 3-year RFS rate of 90.7% versus 78.6%. On multivariable analysis, IPeIC/MMC independently improved RFS (hazard ratio [HR] 0.266, p=0.044). IPeIC was associated with superior PFS, with 3-year and 5-year rates of 100% versus 92.1% and 85.8%, respectively, in the controls (HR 0.078, p=0.014). Exploratory transcriptomics identified low Glutathione S-transferase Mu 1 (GSTM1) expression as the factor most strongly associated with recurrence.ConclusionsIPeIC/MMC is associated with improved long-term oncological outcomes compared with TURBT alone and represents a safe prophylactic option for patients with NMIBC who are unable to receive standard immediate postoperative intravesical chemotherapy because of safety concerns or practical constraints. The GSTM1 findings are hypothesis-generating and support future biomarker-driven validation studies. INTRODUCTION Transurethral resection of bladder tumor (TURBT) remains foundational to the diagnosis and initial management of non–muscle-invasive bladder cancer (NMIBC) [12]. However, NMIBC poses a considerable clinical challenge, characterized by high recurrence rates of 15%–61% and 31%–78% at 1 and 5 years, respectively, and progression rates of up to 45% at 5 years post-TURBT [1234]. TURBT alone has inherent limitations in controlling recurrence, including the release of free-floating tumor cells during piecemeal resection and the potential for residual tumor [4]. Early recurrence of NMIBC is primarily driven by intravesical tumor seeding via freefloating tumor cell reimplantation and residual tumor regrowth [1245]. A single immediate postoperative intravesical chemotherapy (IPOIC), typically incorporating mitomycin-C (MMC), significantly reduces NMIBC recurrence by approximately 15% by preventing reimplantation of free-floating tumor cells and destroying residual tumors [1246789]. Despite strong guideline recommendation, the real-world utilization of IPOIC remains remarkably low. MMC’s prototypical vesicant nature and deep tissue penetration raise concerns about potentially severe toxicities, including extravesical extravasation and transmural necrosis [101112]. Consequently, IPOIC is strictly contraindicated in approximately 15% of TURBT cases involving extensive resection, suspected perforations, or significant bleeding [12913]. As a result, many patients receive TURBT alone, thereby maintaining their exposure to high recurrence risk. Immediate preoperative intravesical chemotherapy (IPeIC) with MMC (IPeIC/MMC), administered immediately before TURBT, represents a logical and safer alternative. IPeIC/MMC allows for cytotoxic exposure prior to resection, maximizing the anti-seeding effect without the safety concerns of postoperative extravasation in patients with deep or suspicious resection beds. Early-phase studies and meta-analyses have suggested that IPeIC/MMC may effectively reduce early recurrence while maintaining an excellent safety profile [1415161718]. However, the optimal regimen and, critically, the long-term efficacy and molecular mechanisms underlying IPeIC treatment response have not been sufficiently validated. The favorable short-term clinical outcomes and excellent tolerability of a two-dose IPeIC/MMC regimen in a preliminary analysis of a single-center, randomized phase II trial has been reported, showing a trend toward reduced 1-year recurrence risk compared to TURBT alone (63% reduction, p=0.11) [14], though statistical significance was not reached due to limited follow-up. Building upon these encouraging initial data, this study presents the definitive 5-year oncological outcomes for the entire patient cohort, establishing IPe-IC/MMC as a safe and highly effective prophylactic strategy superior to TURBT alone. Furthermore, to transition from empirical observation to mechanistic understanding, we employ tissue-based transcriptomic profiling to identify novel molecular biomarkers, such as Glutathione S-transferase Mu 1 (GSTM1) loss, that govern IPeIC/MMC efficacy and enable the development of truly personalized intravesical chemotherapy. MATERIALS AND METHODS 1. Study design and participants This updated analysis builds upon a previous singlecenter, open-label, randomized phase II trial [14] by including the same patient cohort enrolled between August 2016 and December 2020 at the National Cancer Center, Korea. The original study received approval from the Institutional Review Board of the Research Institute and Hospital of the National Cancer Center in Korea (approval number: NCC2016-0168; ClinicalTrials.gov: NCT01149174) and was conducted in strict accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment. The inclusion and exclusion criteria are comprehensively described in the prior publication and in Supplementary Table 1 [14]. 2. Randomization and interventions Randomization and intervention procedures were conducted according to those detailed in the preceding study [14]. Patients in the intervention arm received two doses of 40 mg MMC (Kyowa Hakko Kogyo Co.) reconstituted in 20 mL sterile water intravesically via an 8-Fr urethral catheter into an empty bladder 1 day and 4 hours before TURBT and retained for 1–2 hours. In the previous study, MMC was detectable in tumors and matched normal tissues of nearly all patients in the intervention arm, confirming effective MMC delivery into bladder tissue before TURBT [14]. This protocol improved patient convenience by delivering intensive, short-term neoadjuvant therapy without postponing the planned TURBT procedure. Participants in the control arm underwent standard TURBT only; no patients in either arm received IPOIC after TURBT. All endoscopically visible tumors, including bases and surrounding mucosa, were resected during the index TURBT using both white- and blue-light cystoscopy, ensuring complete tumor removal and confirming detrusor muscle presence. In this study, complete TURBT was defined as the total excision of all visible tumor tissue, indicated by the absence of any macroscopic residual tumor at the base or margins of the resection site, as determined by the surgeon’s intraoperative endoscopic evaluation following TURBT. Pathological evaluation was performed as per the World Health Organization (WHO) 1973 and WHO/ISUP (World Health Organization/International Society of Urological Pathology) 2004 guidelines, and TNM staging system. Within 4–6 weeks of index TURBT, patients with carcinoma in situ (CIS), T1, or high-grade tumors without detrusor muscle underwent repeat TURBT [121920]. 3. Follow-up and outcome measures After index TURBT, follow-up and adjuvant intravesical instillations with Bacillus Calmette–Guérin or MMC, including induction and maintenance, were administered based on risk stratification according to the current guidelines for NMIBC, as previously described [14]. The risk-adjusted follow-up strategy included urine cytology and cystoscopy until the initial bladder cancer recurrence was identified [14]. Intermediate- or high-risk patients underwent computed tomography or magnetic resonance imaging annually for 5 years post-initial treatment to monitor distant metastases and upper urinary tract tumor recurrence. All recurrent lesions were histopathologically confirmed during followup. Event definitions and censoring rules followed standard criteria and are detailed in the prior report [14]. The primary endpoint was 3-year recurrence-free survival (RFS); secondary endpoints included progression-free survival (PFS) and identifying molecular predictive biomarkers for IPeIC with MMC through transcriptomic analysis of TURBT tissue post-intervention. 4. Sample collection for transcriptomic profiling and bioinformatic analysis Fresh-frozen tumor samples were collected intraoperatively from 28 patients in the intervention arm (three with recurrence, 25 without). Total RNA was extracted and sequenced to generate approximately 50 million pairedend reads per sample. Raw reads were quality-checked with FastQC, trimmed with Trimmomatic, aligned to the reference genome using HISAT, and quantified with feature-Counts. To address the sample imbalance, we employed a Monte Carlo resampling approach for multiple-testing. In each iteration, three tumor samples were compared with three randomly selected, non-overlapping non-recurrent samples. This procedure was repeated 1,000 times to calculate gene-level statistical probabilities. Significant genes in ≥25% of iterations were identified as candidate predictive biomarkers associated after two doses of IPeIC/MMC before TURBT; subsequently, they were selected for further analysis. An empirical p-value for each gene was defined as (1,000-k)/1,000, where k represents the number of iterations with a false discovery rate ≤0.05. Genes were subsequently ranked based on empirical p-values, and following rank-based adjusted p-values were calculated. 5. Statistical analyses Our prior pilot study aimed to assess whether administering two doses of IPeIC/MMC before TURBT would enhance the 1-year RFS rate from 50% to 70%. A sample size of 70 patients, corresponding to 28 recurrence events, was determined to achieve over 80% statistical power to detect a 20% difference in RFS using a log-rank test with a one-sided alpha level of 0.2 [14]. Statistical analyses were performed on the per-protocol (PP) population. Secondary outcomes, including PFS and molecular predictors, were evaluated as part of a post hoc analysis, as these endpoints were not pre-specified in the original study protocol but were defined prior to the present analysis. Categorical variables were analyzed using chi-square or Fisher’s exact tests, and continuous variables with Student’s t-test. RFS and PFS were estimated via Kaplan–Meier methods, with survival curves compared using log-rank tests. Hazard ratios (HRs) and 95% confidence intervals (CIs) for recurrence were estimated using Cox proportional hazards models with the Wald test. As no progression events occurred in the intervention arm, Firth’s penalized likelihood method was used to analyze progression outcomes. Univariable and multivariable Cox regression assessed the effect of two IPeIC doses with MMC before TURBT on RFS. Given the limited number of outcome events (n=15), the number of variables included in the multivariable model was intentionally limited to reduce the risk of overfitting. Specifically, in accordance with the events-per-variable (EPV) guideline—where an EPV of 10 is generally considered the minimum acceptable threshold—a stringent significance level (p<0.05) was applied, and only clinically relevant variables identified in the univariate analysis were included in the final multivariable model. Statistical significance was set at two-sided p<0.05. All analyses were conducted using SAS 9.4 (SAS Institute) and R version 3.6.2 (R Software for Statistical Computing). RESULTS 1. Patient characteristics Patient enrollment, randomization, and allocation procedures have been detailed in our previous report [14]. Of the 103 patients screened, 99 were randomized (49 to the IPeIC/MMC group and 50 to the control group). After postrandomization exclusions and withdrawals specified for the PP population, 71 patients (33 in the IPeIC/MMC group and 38 in the control group) were included in the PP analysis (Supplementary Fig. 1). Baseline characteristics were well-balanced between arms, with no significant differences in demographics, tumor features, risk classifications, or adjuvant intravesical therapies (Table 1; all p>0.05). Most patients presented with primary tumors (90.9% vs. 92.1%), and high-risk disease comprised 57.6% and 52.6% of cases in the intervention and control arms, respectively. 2. Updated clinical outcomes Median follow-up durations were comparable between the intervention (60.0 months) and control arms (60.4 months) (p=0.649). Recurrence occurred in 9.1% (3/33) of the intervention arm compared with 31.6% (12/38) in the control arm. Median RFS was not reached in either arm (Table 2, Fig. 1A). Two doses of IPeIC/MMC significantly reduced the risk of recurrence by 76.8% compared with TURBT alone (HR 0.23, 95% CI 0.06–0.83, p=0.024). This translated to superior 3-year and 5-year RFS rates of 90.7% in the intervention, versus 78.6% and 75.3% in the control arm, respectively (Table 2). In the univariable Cox regression analysis, IPeIC/MMC was significantly associated with a reduced risk of recurrence (HR 0.232, p=0.024), whereas concomitant CIS was associated with a markedly increased risk (HR 8.510, p=0.001) (Table 3). Multivariable analysis adjusting for concomitant CIS confirmed that IPeIC/MMC was the independent predictor of lower recurrence risk (HR 0.266, 95% CI 0.073–0.967, p=0.044) (Table 3). Other factors, including tumor stage, grade, size, multifocality, prior recurrence rate, and risk stratification, were not significant predictors of RFS. Notably, no disease progression occurred in the intervention arm (0/33), whereas six of 38 patients (15.8%) in the control arm experienced progression. Consequently, the intervention arm demonstrated significantly superior PFS, with rates of 100% at both 3 and 5 years, compared to 92.1% and 85.8%, respectively, in the control arm (HR 0.078, 95% CI 0.001–0.662, p=0.014) (Table 2, Fig. 1B). 3. Molecular correlates of recurrence in the IPeIC/MMC cohort Exploratory transcriptomic analysis was performed on the available tumor tissues from the IPeIC cohort to identify molecular features associated with recurrence after treatment. Using the Monte Carlo-based resampling strategy described above, we identified candidate genes robustly associated with recurrence (Fig. 2A). Recurrent tumors showed significant upregulation of MUC2 and PCDHA4. Notably, low GSTM1 expression showed the strongest association signal with recurrence (Fig. 2B-D). As reduced GSTM1 expression may reflect genomic deletion and/or transcriptional downregulation (potentially consistent with a null-genotype status), patients with low GSTM1 expression had a higher likelihood of recurrence than those with preserved GSTM1 expression. These findings should be regarded as exploratory and hypothesis-generating, as the transcriptomic analysis was confined to tumor tissues from the intervention arm. Therefore, it remains uncertain whether low GSTM1 expression is specifically linked to diminished IPeIC/MMC efficacy or functions as a more general prognostic marker in NMIBC. Validation in larger, independent cohorts encompassing both intervention and control arms is required to clarify this distinction. DISCUSSION This study provides definitive long-term validation of two doses of IPeIC/MMC before TURBT, demonstrating a sustained reduction in NMIBC recurrence and progression over a 5-year follow-up. Unlike single-dose regimens that may be insufficient to suppress free-floating tumor cells, our two-dose protocol reduced recurrence risk by 76.8% and achieved a 90.7% 3-year RFS, with results remaining stable beyond 5 years. Notably, the IPeIC arm showed zero progression events (0% vs. 15.8%), highlighting its potential to alter the natural history of the disease without increasing complications. The rationale for IPeIC stems from the critical limitations of standard IPOIC. While IPOIC reduces recurrence by preventing tumor cell reimplantation, its realworld utilization is strikingly suboptimal (<10%) owing to logistical barriers and safety concerns, such as bladder perforation or deep resection [9212223]. The 4-hour preoperative instillation may present logistical challenges in certain clinical settings, particularly for early morning operating room schedules. Nonetheless, this protocol was feasible within our institutional workflow without causing delays to the surgical timetable, as TURBTs in the intervention arm were generally scheduled after 10:00 a.m. The two-dose regimen was designed to maintain a high local drug concentration at two preoperative time points to maximize the prophylactic anti-seeding effect while minimizing patient discomfort and avoiding disruption to the planned TURBT schedule. However, a simplified single-dose neoadjuvant regimen may represent a pragmatic approach for future research. Furthermore, this phase II trial lacked sufficient statistical power to conduct subgroup analyses across low-, intermediate-, and high-risk categories as defined by the American Urological Association (AUA) risk classification; consequently, formal comparisons based on these categories were not performed. The limited sample size and the small number of outcome events precluded meaningful statistical analyses within these subgroups. To avoid potentially misleading interpretations, underpowered subgroup comparisons have been deliberately omitted. Importantly, the distribution of AUA risk groups was well balanced between the two study arms at baseline, indicating that the observed treatment effect is unlikely to be confounded by significant imbalances in risk composition. Additionally, since IPeIC/MMC is administered prior to TURBT, precise postoperative risk stratification is not always available at the time of treatment, thereby justifying the initial evaluation of this approach in a broad NMIBC population rather than in narrowly defined postoperative risk subsets. Larger-scale studies are warranted to determine whether the efficacy of IPeIC/MMC differs among low-, intermediate-, and high-risk NMIBC populations. This stark underutilization is primarily ascribed to (1) contraindications for safe administration (e.g., extensive resection, suspected bladder perforation, severe bleeding), which affect approximately 15% of patients, and (2) logistical and workflow barriers within clinical practice, coupled with clinician concerns over potential severe complications. Consequently, the majority of patients receive TURBT alone, a suboptimal standard. IPeIC addresses this unmet need by delivering cytotoxic agents prior to resection, thereby bypassing the contraindications of IPOIC while effectively targeting tumor seeding. The fact that all patients in this trial received standardized continuous bladder irrigation further confirms that the superior outcomes in the intervention arm are directly attributable to the cytotoxic efficacy of IPeIC. Our findings align with and expand upon previous studies. Carrión et al. [16] demonstrated reduced early recurrence with preoperative MMC, although their long-term benefit was limited by a short follow-up. In contrast to a recent negative phase III trial employing neoadjuvant MMC weeks prior to surgery [18], our study highlights that the timing of administration—immediately before TURBT—is crucial for targeting perioperative reimplantation. Furthermore, while Di Stasi et al. [15] showed efficacy with electromotive drug administration, our passive diffusion IPeIC approach offers a comparable oncological benefit but with significantly greater accessibility and lower cost, avoiding the need for specialized equipment. Furthermore, this study provides preliminary transcriptomic insight into the molecular feature associated with recurrence after IPeIC/MMC. Notably, low GSTM1 expression emerged as a prominent candidate associated with recurrence, which is biologically plausible given the known role of GSTM1 as a phase II detoxification enzyme for electrophilic compounds, including carcinogens derived from tobacco exposure, that may be present in urine [242526]. We interpret this finding through the concept of “field cancerization”. Patients exhibiting a GSTM1-deficient transcriptional state may lack sufficient enzymatic capacity to detoxify urinary carcinogens effectively, leading to prolonged exposure and accumulation of DNA damage throughout the entire urothelium. This may contribute to the formation of a biologically unstable mucosal field that is prone to multifocal recurrence. Therefore, in these patients, the background carcinogenic pressure may partially limit the prophylactic effect of intravesical MMC. This suggests that patients with functional deficiency of GSTM1 may represent a biologically distinct subgroup that could benefit from closer surveillance or consideration of additional therapeutic strategies beyond local chemotherapy. Although these findings are exploratory, the consistent biological relevance of GSTM1 documented in prior studies supports its consideration as a candidate biomarker warranting further investigation. However, since the transcriptomic data were derived exclusively from the intervention group, it remains possible that low GSTM1 expression serves as a general prognostic indicator rather than a treatment-specific predictive marker at this stage. This study has some limitations: it has a single-center design, and a modest sample size. Furthermore, the omission of a comparator group receiving standard IPOIC with MMC is acknowledged as a limitation of this study. However, this design directly reflects the core objective of our research to demonstrate the clear clinical benefit of IPeIC for a patient population with limited treatment options, such as those in whom IPOIC is contraindicated. The utilization of a PP population may have disrupted the balance established by randomization, potentially introducing selection bias and leading to an overestimation of the treatment effect. Additionally, transcriptomic profiling was limited by the small number of analyzed samples, lack of pre-IPeIC biopsies, and no validation cohort. Furthermore, the lack of matched transcriptomic data from the control group constitutes a major limitation of the exploratory molecular analysis. In the absence of such a comparator, it remains challenging to ascertain whether the observed associations—especially the decreased GSTM1 expression—are attributable to treatmentspecific effects or represent inherent prognostic characteristics. To overcome this limitation, future multicenter studies should incorporate parallel transcriptomic profiling of both treatment and control cohorts. CONCLUSIONS In conclusion, this 5-year analysis demonstrates that IPeIC/MMC is a safe and durable prophylactic strategy that significantly reduces recurrence and is associated with excellent long-term disease control compared with TURBT alone. IPeIC/MMC may be particularly useful for patients with NMIBC who are unable to receive standard IPOIC because of safety concerns or practical constraints. In addition, our exploratory transcriptomic analysis suggests that low GSTM1 expression is associated with an increased risk of recurrence after IPeIC/MMC, warranting cautious interpretation and independent validation. These findings provide a rationale for biomarker-informed follow-up strategies and future studies. A multicenter phase III trial comparing IPe- IC to standard IPOIC is warranted to validate these findings and further define the role of IPeIC in the contemporary management of NMIBC.