Alignment of molecular subtypes across multiple bladder cancer subtyping classifiers

BACKGROUND

Treatment of patients with muscle-invasive bladder cancer (MIBC) includes cisplatin-based neoadjuvant chemotherapy (NAC) followed by radical cystectomy (RC). Molecular subtypes have been associated with patient outcomes after NAC and RC, but the reported results have been highly inconsistent.

OBJECTIVE

To evaluate the association of molecular subtypes from different classifiers with overall survival (OS) among patients with MIBC who underwent RC.

MATERIALS AND METHODS

We analyzed gene expression data generated from transurethral resection of MIBC from a previously assembled and published meta-cohort, NACmeta (N = 601, 247 treated with NAC+RC and 354 RC without NAC), where extended follow-up was available. Molecular subtypes were assigned using the Genomic Subtyping Classifier (GSC), the Consensus Classifier, The Cancer Genome Atlas (TCGA) Classifier, and the Lund Classifier. For survival analysis, inverse probability weighting was used to balance the clinical NAC and non-NAC patient groups.

RESULTS

A high consistency in gene expression patterns and nomenclature was observed between luminal-like subtypes, defined as GSC-Luminal, Consensus-Luminal Papillary (LumP), TCGA Luminal-Papillary (LumP) and Lund-UroA, but not for basal-like subtypes such GSC-Basal, Consensus Basal/Squamous, TCGA-Basal/Squamous and Lund-Basal/Squamous. Patients with luminal-like subtypes demonstrated no difference in 3-year OS when treated with or without NAC (P = 0.7 for GSC, P = 0.94 for Consensus, P = 0.87 for TCGA and P = 0.66 for Lund-UroA, respectively).

CONCLUSION

Luminal-like molecular subtypes identify a subgroup of MIBC patients who do not appear to benefit from current NAC regimens, even for locally advanced disease. In addition, we were able to illustrate differences in subtyping nomenclature that are not reflected in the underlying biological definition of the subtypes.

PATIENT SUMMARY

Muscle-invasive bladder cancer exhibits molecular diversity, and various classifications identify different groups who do not benefit from chemotherapy. On the other hand, there is a high inconsistency in the way cancer groupings are named.

A Genomic Urine Assay for Surveillance of Patients with Bladder Cancer Treated with Radiotherapy

Background

Patients with muscle-invasive bladder cancer (MIBC) who receive radiotherapy with curative intent are followed by imaging, cystoscopy, and urine cytology. However, interpretation of cytology and cystoscopy is hampered by the impact of ionizing radiation on cells.

Objective

To assess the diagnostic performance of a genomic urine assay to detect urinary tract recurrences in patients with MIBC treated by (chemo)radiation.

Design setting and participants

Patients with nonmetastatic MIBC who underwent (chemo)radiation with curative intent from 2016 to 2020 were prospectively included. Follow-up consisted of cystoscopy and upper tract imaging. Prior to cystoscopy, a urine sample was analyzed to assess mutations in the genes FGFR3, HRAS, and TERT and methylation of OTX1, TWIST1, and ONECUT2. The treating physician was blinded for the assay result.

Outcome measurements and statistical analysis

The primary endpoint was a urinary tract recurrence. Cross-sectional sensitivity, specificity, and negative predictive value (NPV) were analyzed using a previously developed logistic regression model for the detection of bladder cancer with this assay. The secondary endpoint was the risk of a future urinary tract recurrence following a positive test and negative cystoscopy/imaging, using a time-dependent Cox proportional hazard analysis.

Results and limitations

A total of 143 patients were included, and 503 urine samples were analyzed. The median study duration was 20 mo (interquartile range [IQR] 10-33), and the median time to a recurrence was 16 mo (IQR 12-26). In 27 patients, 32 urinary tract recurrences were diagnosed, including three upper tract tumors. Of 32 recurrences, 18 (56%) had a concomitant urine test available. The diagnostic model had an area under the curve of 0.80 (95% confidence interval [CI] 0.69-0.90) with corresponding sensitivity, specificity, and NPV of 78 (95% CI 52-94), 77% (95% CI 73-81), and 99% (95% CI 97-100). When taking into account the anticipatory effect of the test, 28/32 (88%) recurrences were detected. A Cox regression analysis showed a hazard ratio of 14.8 for the development of a future recurrence (p < 0.001). A major limitation was the lack of a concomitant urine test result in 14/32 (44%) recurrences.

Conclusions

A genomic urine assay detected urinary tract recurrences after (chemo)radiation in patients with MIBC, and a positive test was strongly associated with future recurrences. Although validation in a large cohort is warranted, the test has the potential to limit frequent cystoscopies.

Patient summary

Radiotherapy is a bladder-sparing treatment in patients with bladder cancer. After treatment, these patients undergo visual inspection of the bladder by cystoscopy to detect possible recurrences. However, interpretation of cystoscopy is difficult due to the effects of radiation on the bladder lining. Hence, we analyzed the diagnostic value of a molecular urine test to detect recurrent disease in bladder cancer patients treated by radiotherapy, and we showed that the urine test has the potential to limit the number of cystoscopies.

The Diagnostic Accuracy of Cystoscopy for Detecting Bladder Cancer in Adults Presenting with Haematuria: A Systematic Review from the European Association of Urology Guidelines Office

CONTEXT

Haematuria can be macroscopic (visible haematuria [VH]) or microscopic (nonvisible haematuria [NVH]), and may be caused by a number of underlying aetiologies. Currently, in case of haematuria, cystoscopy is the standard diagnostic tool to screen the entire bladder for malignancy.

OBJECTIVE

The objective of this systematic review is to determine the diagnostic test accuracy of cystoscopy (compared with other tests, eg, computed tomography, urine biomarkers, and urine cytology) for detecting bladder cancer in adults.

EVIDENCE ACQUISITION

A systematic review of the literature was performed according to the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy and Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) extension for diagnostic test accuracy studies' checklist. The MEDLINE, Embase, Cochrane CENTRAL, and Cochrane CDSR databases (via Ovid) were searched up to July 13, 2022. The population comprises patients presenting with either VH or NVH, without previous urological cancers. Two reviewers independently screened all articles, searched reference lists of retrieved articles, and performed data extraction. The risk of bias was assessed using Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2).

EVIDENCE SYNTHESIS

Overall, nine studies were included in the qualitative analysis. Seven out of nine included trials covered the use of cystoscopy in comparison with radiological imaging. Overall, sensitivity of cystoscopy ranged from 87% to 100%, specificity from 64% to 100%, positive predictive value from 79% to 98%, and negative predictive values between 98% and 100%. Two trials compared enhanced or air cystoscopy versus conventional cystoscopy. Overall sensitivity of conventional white light cystoscopy ranged from 47% to 100% and specificity from 93.4% to 100%.

CONCLUSIONS

The true accuracy of cystoscopy for the detection of bladder cancer within the context of haematuria has not been studied extensively, resulting in inconsistent data regarding its performance for patients with haematuria. In comparison with imaging modalities, a few trials have prospectively assessed the diagnostic performance of cystoscopy, confirming very high accuracy for cystoscopy, exceeding the diagnostic value of any other imaging test.

PATIENT SUMMARY

Evidence of tests for detecting bladder cancer in adults presenting with haematuria (blood in urine) was reviewed. The most common test used was cystoscopy, which remains the current standard for diagnosing bladder cancer.

A Urine-based Genomic Assay Improves Risk Stratification for Patients with High-risk Hematuria Stratified According to the American Urological Association Guidelines

BACKGROUND

According to the recent American Urological Association (AUA) guideline on hematuria, patients are stratified into groups with low, intermediate, and high risk of urothelial carcinoma (UC). These risk groups are based on clinical factors and do not incorporate urine-based tumor markers.

OBJECTIVE

To evaluate whether a urine-based genomic assay improves the redefined AUA risk stratification for hematuria.

DESIGN, SETTING, AND PARTICIPANTS

We selected patients with complete biomarker status, as assessed on urinary DNA, from a previously collected prospective Dutch hematuria cohort (n = 838). Patients were stratified into the AUA risk categories on the basis of sex, age, and type of hematuria. Biomarker status included mutation status for the FGFR3, TERT, and HRAS genes, and methylation status for the OTX1, ONECUT2, and TWIST1 genes.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

The primary endpoint was the diagnostic model performance for different hematuria risk groups. Further analyses assessed the pretest and post-test UC probability in the hematuria subgroups using a Fagan nomogram.

RESULTS AND LIMITATIONS

Overall, 65 patients (7.8%) were classified as low risk, 106 (12.6%) as intermediate risk, and 667 (79.6%) as high risk. The UC incidence differed significantly between the gross hematuria (21%, 98/457) and microscopic hematuria (4%, 14/381) groups (p < 0.001). All cancer cases were in the high-risk group, which had UC incidence of 16.8% (112/667). Application of the diagnostic model revealed robust performance among all risk groups (area under the receiver operating characteristic curve 0.929-0.971). Depending on the risk group evaluated, a negative urine assay was associated with post-test UC probability of 0.3-2%, whereas a positive urine assay was associated with post-test UC probability of 31-42%.

CONCLUSIONS

This study shows the value that a urine-based genomic assay adds to the AUA guideline stratification for patients with hematuria. It seems justified to safely withhold cystoscopy for patients with AUA low risk who have a negative urine assay. In addition, evaluation should be expedited for patients with AUA intermediate or high risk and a positive urine assay.

PATIENT SUMMARY

Patients who have blood in their urine (hematuria) can be classified as having low, intermediate, or high risk of having cancer in their urinary tract. We found that use of a urine-based genetic test improves the accuracy of predicting which patients are most likely to have cancer. Patients with a negative test may be able to avoid invasive tests, while further tests could be prioritized for patients with a positive test.

Optimization of Preoperative Lymph Node Staging in Patients with Muscle-Invasive Bladder Cancer Using Radiomics on Computed Tomography

Approximately 25% of the patients with muscle-invasive bladder cancer (MIBC) who are clinically node negative have occult lymph node metastases at radical cystectomy (RC) and pelvic lymph node dissection. The aim of this study was to evaluate preoperative CT-based radiomics to differentiate between pN+ and pN0 disease in patients with clinical stage cT2-T4aN0-N1M0 MIBC. Patients with cT2-T4aN0-N1M0 MIBC, of whom preoperative CT scans and pathology reports were available, were included from the prospective, multicenter CirGuidance trial. After manual segmentation of the lymph nodes, 564 radiomics features were extracted. A combination of different machine-learning methods was used to develop various decision models to differentiate between patients with pN+ and pN0 disease. A total of 209 patients (159 pN0; 50 pN+) were included, with a total of 3153 segmented lymph nodes. None of the individual radiomics features showed significant differences between pN+ and pN0 disease, and none of the radiomics models performed substantially better than random guessing. Hence, CT-based radiomics does not contribute to differentiation between pN+ and pN0 disease in patients with cT2-T4aN0-N1M0 MIBC.

Gene Expression Profiling of Muscle-Invasive Bladder Cancer With Secondary Variant Histology

OBJECTIVES

To determine the potential impact of the presence of secondary variant histology on the gene expression profiles of muscle-invasive bladder cancer (MIBC) tumors.

METHODS

For six tumors, revised samples were collected from urothelial and secondary variant components (cohort A). The commercial cohort (cohort B) consisted of the anonymized gene expression profiles of 173 patients with MIBC. Samples were obtained from the clinical use of the Decipher Bladder test that were available as part of the Decipher GRID prospective registry (NCT02609269). Secondary variant presence in cohort B was abstracted from institutional pathology reports. For the commercial cohort, only the urothelial carcinoma component was profiled.

RESULTS

Molecular subtyping of both urothelial and variant components found micropapillary and nested cases were classified as a luminal subtype. Conversely, the sarcomatoid and small cell cases were classified as basal/squamous or neuroendocrine-like, respectively. For cohort B, 50 (29%) of 173 cases had reported secondary variant histology. Cases with squamous variant had basal profiles, small cell cases expressed neuronal markers, and micropapillary cases were classified as luminal. Sarcomatoid tumors had robust epithelial-mesenchymal transition marker expression.

CONCLUSIONS

Our data suggest that in MIBC with secondary variant, the urothelial component can demonstrate an expression profile that closely resembles the variant component.

Long-term efficacy of hyperthermic intravesical chemotherapy for BCG-unresponsive non-muscle invasive bladder cancer

BACKGROUND

The recommended treatment for patients with Bacillus Calmette-Guérin (BCG) unresponsive non-muscle invasive bladder cancer (NMIBC) is radical cystectomy (RC). However, many patients refuse, or are unfit for RC. Therefore, alternative bladder-sparing treatment modalities are needed for BCG-unresponsive NMIBC. In this study we sought to assess the long-term efficacy of hyperthermic intravesical chemotherapy (HIVEC) as alternative to radical cystectomy in BCG-unresponsive non-muscle invasive bladder cancer patients.

METHODS AND MATERIALS

Retrospectively collected data from 56 patients with BCG-unresponsive NMIBC who received ≥5 HIVEC instillations between October 2014 and March 2020 was analyzed. All patients met the BCG-unresponsive criteria according to the current EAU guideline on NMIBC 2020. Patients were followed-up with cystoscopy and/or bladder biopsies, urine cytology and annually CT-urography. The Primary outcome was the high grade (HG) recurrence-free survival (RFS), defined as the time from the first HIVEC instillation until histologically confirmed intravesical recurrence or last follow-up. The Kaplan Meier method was used to estimate survival outcomes. Secondary outcomes were: complete response rate (CR), adverse events (AE), assessed by the Common Terminology Criteria for Adverse Events v5.0 (CTCAE) and tumor progression to muscle invasive disease or distant metastases.

RESULTS

The median follow-up was 32.2 months (IQR 13.7-44.8). The 1- and 2-year HG-RFS was 53% (SE:6.8) and 35% (SE:6.9), respectively. The CR for patients with CIS was 70% (21/30) at 6 months. Overall, 80% of the population developed an AE, only 1 was classified as CTCAE ≥3. Limitation of this study was the small sample size.

CONCLUSION

HIVEC resulted in a 2-year HG-RFS of 35% for BCG-unresponsive NMIBC patients without severe side-effects and therefore HIVEC seems to be an alternative treatment option for patients who refuse or are unfit for RC.

Molecular subtyping and immune-gene signatures identify a subset of early bladder tumors as candidates for single-agent immune-checkpoint inhibition

PURPOSE

Clinical high-grade (HG) T1 non-muscle invasive bladder cancer (NMIBC) represents a significant risk to patients, but these patients are not typically offered neoadjuvant therapies, including immune therapy. In this study, we determine whether patients with HG clinical T1 or T2 bladder urothelial carcinoma (UC) have profiles that predict the potential effectiveness of immune-checkpoint inhibitors (ICI).

MATERIALS AND METHODS

Data from transurethral resection of bladder tumor (TURBT) specimens from 2 studies was evaluated. The molecular upstaging (MOL) cohort included HG cT1N0M0 (n = 87) and cT2N0M0 (n = 119) bladder UC who underwent radical cystectomy (RC) without any neoadjuvant therapy. The PURE-01 cohort (n = 102) was used as ICI-treated reference. Specimen collection and sample processing were conducted using a clinical-grade whole-transcriptome assay (Decipher). Immune-signatures scores and molecular subtyping were evaluated. Kaplan-Meier curves and log-rank tests were used for exploratory analyses of recurrence-free survival (RFS) and overall survival (OS).

RESULTS

In both the PURE-01 and MOL cohorts, the Immune190 signature, stratified by subtype, showed the highest scores in basal-type, but also in luminal-infiltrated tumors, but the lowest scores in the luminal tumors. However, in HG cT1 tumors the Immune190 scores were the lowest for luminal papillary tumors (Consensus, TCGA) and luminal tumors (GSC), with less distinct differences between other subtypes. RFS was significantly longer for luminal vs non-luminal tumors in MOL (P = 0.04) but not in PURE-01 (P = 0.8). In the MOL cohort, OS was inferior in HG cT1 tumors for Immune190-high vs low tumors (median split, P = 0.042).

CONCLUSION

We identified a population of cT1-T2N0M0 tumors in the MOL cohort that shared molecular features with tumors included in PURE-01. These profiles suggest that treatment with ICI could be proposed to more selected HG cT1N0M0 tumors, identified with a gene expression assay.

PD-L1 expression in urothelial bladder cancer varies more among specimen types than between companion assays

Urothelial bladder cancer (UBC) patients ineligible to platinum-based chemotherapy can be treated with immune-checkpoint inhibitors (ICI) in Programmed Death Ligand 1 (PD-L1) positive cases. Although concordance exists between different PD-L1 assays, little is known on PD-L1 expression variability in matched UBC samples. We compared PD-L1 expression in whole slides of matched transurethral resections (TURBT), radical cystectomies (RC), and lymph node metastasis (LN). Immunohistochemistry using the VENTANA PD-L1 (SP263) assay was performed on 115 patients and scored positive if expression occurred in ≥25% immune cells (IC), ≥25% tumour cells (TC), or both. PD-L1 was positive in 42.7% TURBT, 39.8% RC, and 27.3% LN specimens. Concordance was moderate (κ=0.52; P<0.001) between TURBT and RC, and fair between LN and TURBT (κ=0.31; P=0.048) or RC (κ=0.25; P=0.075). Comparison with the VENTANA PD-L1 (SP142) assay which had been performed previously on the same cohort showed moderate to substantial inter-assay agreement (κ=0.42-0.66). Although TC staining is not part of the SP142 scoring algorithm, discordant PD-L1 assay outcome could be attributed to SP263 TC≥25% staining in only 41% of cases. These results show that PD-L1 expression variability between matched specimens is higher than that between individual assays. Optimal specimen determination for PD-L1 testing needs to be addressed in future studies.

Molecular Characterization of Residual Bladder Cancer after Neoadjuvant Pembrolizumab

BACKGROUND

In patients with muscle-invasive urothelial bladder cancer (MIBC), molecular alterations in immunotherapy-resistant tumors found at radical cystectomy (RC) remain largely unstudied.

OBJECTIVE

To investigate the biology of pembrolizumab-resistant tumors in comparison to an RC cohort treated without any systemic therapy and a cohort of neoadjuvant chemotherapy (NAC)-treated tumors.

DESIGN, SETTING, AND PARTICIPANTS

Transcriptome-wide expression profiling was performed on 26 RC samples from patients with ypT2-4 disease after pembrolizumab treatment, of which 22 had matched pretherapy samples. Unsupervised consensus clustering (CC) was performed to compare 26 post-pembrolizumab samples with 94 RC samples without neoadjuvant treatment and 21 samples collected from the former tumor bed of NAC-treated patients (scar tissue). Clusters were investigated for their biological and clinical characteristics and were compared to a cohort of post-NAC tumors (n = 133).

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

Patient and tumor characteristics were compared between subgroups using χ² tests and two-sided Wilcoxon rank-sum tests. The primary endpoint was recurrence-free survival.

RESULTS AND LIMITATIONS

Molecular subtyping of pre- and post-pembrolizumab samples revealed significant differences: only 36% of samples had a concordant subtype according to the consensus classifier. Unsupervised CC revealed three distinct post-pembrolizumab clusters (basal, luminal, and scar-like). A scar-like subtype was present in 50% of the post-pembrolizumab cases (n = 13) and expressed genes associated with wound healing/scarring. This subtype had higher luminal marker expression in the post-pembrolizumab setting compared to CC scar-like tumors from the other cohorts. Patients with the scar-like subtype showed favorable prognosis after systemic therapy, but not in the RC-only setting. The small numbers in each subgroup represents the major study limitation.

CONCLUSIONS

This study expands our understanding of the biology of pembrolizumab-resistant MIBC and provides a framework for defining molecular subtypes after treatment. The results further support the hypothesis that luminal-type tumors may be resistant to immunotherapy or that this treatment may select for, or induce, a luminal phenotype.

PATIENT SUMMARY

We carried out genetic analysis for bladder cancer tumors from patients who had received an immunotherapy agent called pembrolizumab and compared them to tumors treated with standard chemotherapy or just bladder removal. We found differences in gene expression between the treatment types and between tumor tissue from the same patient before and after treatment. These results may be helpful in personalizing therapy strategies for patients with bladder cancer.

Abstract IA06: A urine assay to triage patients with hematuria for cystoscopy

Objectives: To assess the accuracy of a previously developed molecular assay to detect bladder cancer in a large prospective cohort of patients referred for hematuria.

Design, Setting and Participants: A urine sample was collected prior to cystoscopy and mutation/methylation status of 6 genes was determined in DNA from urinary cells and combined with age. This existing diagnostic model was validated on this cohort and optimized using logistic regression. Clinical usefulness was determined by the net benefit approach. Main outcome measures: Primary endpoint was the model performance to detect the presence of urothelial cancer, as indicated by the area under the receiver operating characteristic (ROC) curve (AUC). Secondary endpoints were the sensitivity, specificity, and negative predictive value (NPV) of the optimal diagnostic model.

Results: In 838 patients, the mutation/methylation status could be determined for all genes. Urothelial cancer was observed in 112 patients (98/457 in the gross and 14/381 in the microscopic hematuria group). Application of the existing model resulted in an AUC of 0.93. Combining the assay with the type of hematuria resulted in the final optimal model with an AUC of 0.96 (96% sensitivity, 73% specificity, 99% negative predictive value). The assay also detected all six upper tract urothelial tumors that are not visible by cystoscopy. Net benefit analysis showed that the urine test should be preferred over “cystoscopy for all.” Application of the optimal model on patients whose biomarker status was incomplete resulted in the identification of 5 additional tumors. Implementing the urine test as a triage tool could lead to a 53% reduction in cystoscopies.

Conclusions: The urine test detects urothelial cancer with a very high accuracy and is ready for clinical implementation.

Citation Format: Kim E.M. van Kessel, Joep J. de Jong, Angelique C.J. Ziel-van der Made, Hossain Roshani, Stefan M. Haensel, Josien H. Wolterbeek, Egbert R. Boevé, Eric H.G.M. Oomens, Niels J. van Casteren, Manuel Krispin, Joost L. Boormans, Ewout W. Steyerberg, Wim van Criekinge, Ellen C. Zwarthoff. A urine assay to triage patients with hematuria for cystoscopy [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2019 May 18-21; Denver, CO. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(15_Suppl):Abstract nr IA06.

Abstract B07: A long noncoding RNA-based genomic classifier identifies a subset of luminal muscle-invasive bladder cancer patients with favorable prognosis

Background: Muscle-invasive bladder cancer (MIBC) is a heterogeneous disease and gene expression profiling has identified different molecular subtypes, each having distinct biologic and clinicopathologic characteristics. Subtyping MIBC has primarily been messenger RNA (mRNA)-based, although noncoding RNAs, including long noncoding RNAs (lncRNAs), have potential utility in providing additional resolution to current molecular subtyping models.

Materials and Methods: The expression profiles of over 6,000 lncRNAs were quantified from whole-transcriptome microarray data of a MIBC patient cohort treated by neoadjuvant chemotherapy (NAC) and radical cystectomy (N=223). Unsupervised consensus clustering of the most highly variant lncRNAs identified a four-cluster solution, which was further characterized using a panel of MIBC biomarkers, gene signatures, and survival analysis. The four-cluster consensus was validated using the publicly available The Cancer Genome Atlas (TCGA) radical cystectomy cohort (N=405). A single-sample genomic classifier (GC) was trained using ridge-penalized logistic regression and then validated in two independent patient cohorts (N=255 and N=94).

Results: In the NAC and TCGA cohorts, survival analysis of a lncRNA-based consensus cluster solution revealed a lncRNA-cluster (LC3) with strikingly good prognosis that was enriched for tumors of the luminal-papillary mRNA subtype. In both cohorts, the luminal-papillary tumors from this cluster (LPL-C3) were clinically less aggressive than other luminal-papillary tumors. Patients having LPL-C3 tumors were younger and had more frequent organ-confined, node-negative disease than other luminal-papillary tumors. LPL-C3 tumors were characterized by enhanced FGFR3 pathway activity, wild-type P53 expression, and robust SHH signaling. In the TCGA cohort, LPL-C3 tumors were also enriched for FGFR3 mutations and depleted for TP53 and RB1 mutations. A GC trained to identify these LPL-C3 patients showed robust performance in two validation cohorts.

Conclusions: Using lncRNA expression profiling, we identified a biologically distinct subgroup of luminal-papillary MIBC with less-aggressive molecular characteristics and favorable prognosis. These data suggest that lncRNAs can provide additional information in resolving higher-resolution subtypes for more precise patient management strategies.

Citation Format: Joep J. de Jong, Yang Liu, Roland Seiler, A. Gordon Robertson, Michiel S. van der Heijden, Jonathan L. Wright, James Douglas, Marc Dall'Era, Simon J. Crabb, Bas W.G. van Rhijn, Kim E.M. van Kessel, Elai Davicioni, Yair Lotan, Ellen C. Zwarthoff, Peter C. Black, Joost L. Boormans, Ewan A. Gibb. A long noncoding RNA-based genomic classifier identifies a subset of luminal muscle-invasive bladder cancer patients with favorable prognosis [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2019 May 18-21; Denver, CO. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(15_Suppl):Abstract nr B07.

Distribution of Molecular Subtypes in Muscle-invasive Bladder Cancer Is Driven by Sex-specific Differences

Muscle-invasive bladder cancer (MIBC) is a sex-biased cancer with a higher incidence in men but worse outcomes in women. The root cause behind these observations remains unclear. To investigate whether sex-specific tumor biology could explain the differences in clinical behavior of MIBC, we analyzed the transcriptome profiles from transurethral resected bladder tumors of 1000 patients. Female tumors expressed higher levels of basal- and immune-associated genes, while male tumors expressed higher levels of luminal markers. Using molecular subtyping, we found that the rates of the basal/squamous subtype were higher in females than in males. Males were enriched with tumors of the luminal papillary (LumP) and neuroendocrine-like subtypes. Male MIBC tumors had higher androgen response activity across all luminal subtypes and male patients with LumP tumors were younger. Taken together, these data confirm differences in molecular subtypes based on sex. The role of the androgen response pathway in explaining subtype differences between men and women should be studied further. PATIENT SUMMARY: We explored the sex-specific biology of bladder cancer in 1000 patients and found that women had more aggressive cancer with higher immune activity. Men tended toward less aggressive tumors that showed male hormone signaling, suggesting that male hormones may influence the type of bladder cancer that a patient develops.

Genomic Subtyping in Bladder Cancer

PURPOSE OF REVIEW

Molecular characterization of cancer allows us to understand oncogenesis and clinical prognosis as well as facilitates development of biomarkers and treatment. Our aim was to review the current literature on genomic characterization of bladder cancer, and how far we are in implementing genomics into clinical practice.

RECENT FINDINGS

Bladder cancers are molecularly diverse tumors with a high mutational rate. On molecular level, bladder cancer can be categorized into at least six subtypes called luminal-papillary, luminal-unstable, luminal non-specified, basal-squamous, neuroendocrine-like, and stroma-rich. These subtypes have characteristic genomic and transcriptomic profiles and appear to have different prognoses. Several molecular subtypes have been identified in bladder cancer. Prospective trials are underway to validate the applicability of genomic subtypes for clinical decision making. Further integrative analyses of genomic alterations, gene expression, epigenetics, and proteomics need to be performed before genomic subtyping can be attained in clinical practice.

Impact of Molecular Subtyping and Immune Infiltration on Pathological Response and Outcome Following Neoadjuvant Pembrolizumab in Muscle-invasive Bladder Cancer

BACKGROUND

The PURE-01 study (NCT02736266) evaluated the use of pembrolizumab before radical cystectomy (RC) in muscle-invasive bladder cancer (MIBC).

OBJECTIVE

To evaluate the ability of molecular signatures to predict the pathological complete response (CR: ypT0N0) and progression-free survival (PFS) after pembrolizumab and RC.

DESIGN, SETTING, AND PARTICIPANTS

We analyzed the expression data from patients with T2-4aN0M0 MIBC enrolled in the PURE-01 study (N=84) and from patients of a retrospective multicenter cohort treated with cisplatin-based neoadjuvant chemotherapy (NAC; N=140).

INTERVENTION

Neoadjuvant pembrolizumab or NAC and RC.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

Immune signatures and molecular subtyping (The Cancer Genome Atlas, consensus model, and genomic subtyping classifier [GSC]) were evaluated in relation to CR and PFS. Multivariable logistic regression analyses for CR were used, adjusting for gender and clinical T stage.

RESULTS AND LIMITATIONS

The Immune190 signature was significant for CR on multivariable logistic regression analyses (p= 0.02) in PURE-01, but not in the NAC cohort (p= 0.7). Hallmark signatures for interferon gamma (IFNγ; p= 0.004) and IFNα response (p= 0.006) were also associated with CR for PURE-01, but not for NAC (IFNγ: p= 0.9 and IFNα: p= 0.8). In PURE-01, 93% of patients with the highest Immune190 scores (>1st quartile) had 2-yr PFS versus 79% of those with lower scores; no difference was observed in NAC patients, as well as for the other hallmarks in both groups. The neuroendocrine-like subtype had the worst 2-yr PFS in all three subtyping models (33%) and the GSC claudin-low subtype had the best, with no recurrences in 2 yr. Basal subtypes (across classifications) with higher Immune190 scores showed 100% 2-yr PFS after pembrolizumab therapy (p = 0.04, compared with basal-Immune190 low). Statistical analyses are limited by the small number of events and short follow-up.

CONCLUSIONS

Higher RNA-based immune signature scores were significantly associated with CR and numerically improved PFS outcomes after pembrolizumab, but not after NAC. These data emphasize that RNA profiling is a potential tool for personalizing neoadjuvant therapy selection.

PATIENT SUMMARY

We used gene expression profiling to evaluate the association between immune gene expression and response to neoadjuvant immunotherapy, compared with standard chemotherapy, in patients with muscle-invasive bladder cancer (MIBC). We found a significant association between immune gene expression and response to pembrolizumab, but not chemotherapy. We conclude that gene expression profiling has the potential to guide personalized neoadjuvant therapy in MIBC.

Long non-coding RNAs identify a subset of luminal muscle-invasive bladder cancer patients with favorable prognosis

BACKGROUND

Muscle-invasive bladder cancer (MIBC) is a heterogeneous disease, and gene expression profiling has identified several molecular subtypes with distinct biological and clinicopathological characteristics. While MIBC subtyping has primarily been based on messenger RNA (mRNA), long non-coding RNAs (lncRNAs) may provide additional resolution.

METHODS

LncRNA expression was quantified from microarray data of a MIBC cohort treated with neoadjuvant chemotherapy (NAC) and radical cystectomy (RC) (n = 223). Unsupervised consensus clustering of highly variant lncRNAs identified a four-cluster solution, which was characterized using a panel of MIBC biomarkers, regulon activity profiles, gene signatures, and survival analysis. The four-cluster solution was confirmed in The Cancer Genome Atlas (TCGA) cohort (n = 405). A single-sample genomic classifier (GC) was trained using ridge-penalized logistic regression and validated in two independent cohorts (n = 255 and n = 94).

RESULTS

NAC and TCGA cohorts both contained an lncRNA cluster (LC3) with favorable prognosis that was enriched with tumors of the luminal-papillary (LP) subtype. In both cohorts, patients with LP tumors in LC3 (LPL-C3) were younger and had organ-confined, node-negative disease. The LPL-C3 tumors had enhanced FGFR3, SHH, and wild-type p53 pathway activity. In the TCGA cohort, LPL-C3 tumors were enriched for FGFR3 mutations and depleted for TP53 and RB1 mutations. A GC trained to identify these LPL-C3 patients showed robust performance in two validation cohorts.

CONCLUSIONS

Using lncRNA expression profiles, we identified a biologically distinct subgroup of luminal-papillary MIBC with a favorable prognosis. These data suggest that lncRNAs provide additional information for higher-resolution subtyping, potentially improving precision patient management.

Morphologic and genomic characterization of urothelial to sarcomatoid transition in muscle-invasive bladder cancer

INTRODUCTION

The sarcomatoid morphology of muscle-invasive bladder cancer (MIBC) is associated with unfavorable prognosis. However, the genomic, transcriptomic, and proteomic relationship between conventional urothelial and synchronous sarcomatoid morphology is poorly defined.

METHODS

We compiled a cohort of 21 MIBC patients with components of conventional urothelial and adjacent sarcomatoid morphology within the same tumor focus. We performed comprehensive pathologic and immunohistochemical characterization and in 4 selected cases, subjected both morphologic components to targeted DNA sequencing and whole transcriptome analysis.

RESULTS

Synchronous sarcomatoid and urothelial morphology from the same MIBC foci shared truncal somatic mutations, indicating a common ancestral clone. However, additional mutations or copy number alterations restricted to the either component suggested divergent evolution at the genomic level. This was confirmed at the transcriptome level since while the urothelial component exhibited a basal-like subtype (TCGA2014: cluster III, LundTax: basal/squamous-like), the sarcomatoid morphology was predominantly cluster IV (claudin-low). Protein expression was consistent with a basal-like phenotype in both morphologies in 18/21 of cases. However, most cases had evidence of active epithelial-to-mesenchymal transition (E-Cad ↓ and Zeb1 or TWIST1 ↑) from urothelial toward the sarcomatoid morphology. Drug response signatures nominated different targets for each morphology and proposed agents under clinical investigation in liposarcoma or other sarcoma. PD-L1 expression was higher in the sarcomatoid than the urothelial component.

CONCLUSIONS

Conventional urothelial and adjacent sarcomatoid morphologies of MIBC arise from the same common ancestor and share a basal-like phenotype. However, divergence between the morphologies at the genome, transcriptome, and proteome level suggests differential sensitivity to therapy.

Morphologic and genomic characterization of urothelial to sarcomatoid transition in muscle-invasive bladder cancer

INTRODUCTION

The sarcomatoid morphology of muscle-invasive bladder cancer (MIBC) is associated with unfavorable prognosis. However, the genomic, transcriptomic, and proteomic relationship between conventional urothelial and synchronous sarcomatoid morphology is poorly defined.

METHODS

We compiled a cohort of 21 MIBC patients with components of conventional urothelial and adjacent sarcomatoid morphology within the same tumor focus. We performed comprehensive pathologic and immunohistochemical characterization and in 4 selected cases, subjected both morphologic components to targeted DNA sequencing and whole transcriptome analysis.

RESULTS

Synchronous sarcomatoid and urothelial morphology from the same MIBC foci shared truncal somatic mutations, indicating a common ancestral clone. However, additional mutations or copy number alterations restricted to the either component suggested divergent evolution at the genomic level. This was confirmed at the transcriptome level since while the urothelial component exhibited a basal-like subtype (TCGA2014: cluster III, LundTax: basal/squamous-like), the sarcomatoid morphology was predominantly cluster IV (claudin-low). Protein expression was consistent with a basal-like phenotype in both morphologies in 18/21 of cases. However, most cases had evidence of active epithelial-to-mesenchymal transition (E-Cad ↓ and Zeb1 or TWIST1 ↑) from urothelial toward the sarcomatoid morphology. Drug response signatures nominated different targets for each morphology and proposed agents under clinical investigation in liposarcoma or other sarcoma. PD-L1 expression was higher in the sarcomatoid than the urothelial component.

CONCLUSIONS

Conventional urothelial and adjacent sarcomatoid morphologies of MIBC arise from the same common ancestor and share a basal-like phenotype. However, divergence between the morphologies at the genome, transcriptome, and proteome level suggests differential sensitivity to therapy.

Hyperthermic Intravesical Chemotherapy for BCG Unresponsive Non-Muscle Invasive Bladder Cancer Patients

Background:

Adjuvant intravesical instillations with bacillus Calmette-Guérin (BCG) is the recommended treatment option for patients with intermediate-and high-risk non-muscle invasive bladder cancer (NMIBC). Despite adequate BCG treatment, a large proportion of patients experience a recurrence. Although radical cystectomy is the gold standard for BCG unresponsive NMIBC, some patients are unfit or unwilling to consider this option.

Objective:

To assess the effectiveness of Hyperthermic IntraVEsical Chemotherapy (HIVEC^®) in BCG unresponsive NMIBC patients.

Methods:

A post-hoc analysis was conducted of prospectively included intermediate-and high-risk NMIBC patients who were planned to receive HIVEC^® treatment between October 2014 and November 2017. For the present analysis, only patients who met the BCG unresponsive definition were included. Patients were followed by cystoscopy and cytology every 3 months and a CT-urography scan yearly. The primary outcome was the disease-free survival (DFS). The Common Terminology Criteria for Adverse Events (CTCAE) was used to assess side-effects.

Results:

The study population consisted of 55 BCG unresponsive NMIBC patients of whom 52 underwent≥5 HIVEC^® treatments. The median age and follow-up were 73 years and 14.0 months (IQR 7.6 – 24.6). The median DFS was 17.7 months (SE 6.72) and progression occurred in four patients. The 1-year cumulative incidence rate of disease recurrence/progression was 53%. Two patients experienced severe side-effects (CTCAE≥3).

Conclusions:

HIVEC^® seems a valid treatment option for BCG unresponsive NMIBC patients. We report a median DFS of 17.7 months (SE 6.72), potentially avoiding or postponing the need for radical surgery in a proportion of these patients.

Concordance of PD-L1 expression in matched urothelial bladder cancer specimens

AIMS

Programmed death ligand 1 (PD-L1) expression has predictive value for response to immune-checkpoint inhibitor treatment in urothelial cancer patients. The consistency of PD-L1 expression among different specimen types, however, is unknown. The aim of this study is to compare PD-L1 expression in matched transurethral resections of the bladder (TURB), cystectomy specimens and lymph node metastases of urothelial cancer patients.

METHODS AND RESULTS

We performed PD-L1 (SP142) immunohistochemistry on whole tissue slides of 115 urothelial carcinoma patients who had undergone TURB, followed by radical cystectomy and/or pelvic lymph node dissection. The PD-L1 assay was positive if PD-L1 expression in immune cells occupied ≥5% of the tumour area. PD-L1 was positive in 15 of 97 (15.5%) TURB, 17 of 98 (17.3%) cystectomies and nine of 49 (18.4%) lymph node metastases. Agreement of PD-L1 assay outcome between cystectomy and TURB (kappa = 0.34; P = 0.002) and cystectomy and lymph node metastasis (kappa = 0.35; P = 0.034) was fair; there was no agreement between TURB and lymph node metastasis (kappa = 0.045; P = 0.82). Discordance of PD-L1 outcome in matched TURB and cystectomy specimens occurred more frequently after neoadjuvant therapy (53.3% versus 25.4%; P = 0.03), and was not associated with other clinicopathological parameters.

CONCLUSIONS

Urothelial bladder cancer patients showed fair agreement of PD-L1 assay outcome in cystectomies and matched TURB or lymph node specimens. PD-L1 expression was discordant more often after neoadjuvant therapy. Therefore, immune-checkpoint inhibitor studies should take into account specimen type and neoadjuvant therapy in assessing the predictive value of PD-L1 expression.