BIOMARKERS OF THERAPEUTIC RESPONSE IN BLADDER CANCER by
ANDREW EDWARD GOODSPEED
B.S. State University of New York at Brockport, 2012 M.S. State University of New York at Brockport, 2013
A thesis submitted to the
Faculty of the Graduate School of the University of Colorado in partial fulfillment of
the requirements for the degree of Doctor of Philosophy
Pharmacology Program 2018
ii This thesis for the Doctor of Philosophy degree by
Andrew Edward Goodspeed has been approved for the
Pharmacology Program by
Scott Cramer, Chair Heide Ford Dan Theodorescu
Aik-Choon Tan Joshua Black James Costello, Advisor
iii Goodspeed, Andrew Edward (PhD, Pharmacology Program)
Biomarkers of therapeutic response in bladder cancer Thesis directed by Professor James C Costello
ABSTRACT
Treatment of muscle-invasive bladder cancer (MIBC) typically includes systemic cisplatin-based chemotherapy. Despite being a mainstay first-line treatment for 3 decades, the response rate to this chemotherapy is only 50%. Genomic biomarkers of response are opportunities to improve cancer treatments, where specific therapies are selected because of patient’s individual traits. In this study, a combination of in vitro and clinical data are used to expand pharmacogenomics knowledge in MIBC in an attempt to lay the groundwork of biomarker identification with the goal to eventually influence patient care.
We performed an unbiased, whole-genome CRISPR screen to identify mediators of cisplatin resistance in a bladder cancer cell line. The mismatch repair pathway and the mismatch repair gene, MSH2, were strongly enriched in the plasmids increasing cisplatin resistance. We hypothesized that bladder cancer cell lines and tumors with low levels of MSH2 protein would be more resistant to cisplatin-based chemotherapy
regimens. shRNA-mediated reduction in MSH2 reduced cisplatin-mediated apoptosis in two bladder cancer cell lines. Publicly available data from 340 patients with MIBC was used to determine the clinical impact of low MSH2 on the response to platinum-based chemotherapy. As hypothesized, MIBC patients with tumors expressing low levels of MSH2 had poorer survival when treated with platinum-based chemotherapy compared to patients expressing higher levels of MSH2. There was no association between MSH2
iv and survival in patients without pharmacologic or radiation treatment, supporting MSH2 as a predictive biomarker of response to therapy and not prognostic of patient survival. Another portion of this work was to use pharmacogenomics from another cancer type to rationally select MIBCs for alternative therapies. We hypothesized that a gene expression signature derived from colorectal tumors would predict the response to EGFR inhibitors. A gene expression signature, the EGFRi67, was derived from training data of colorectal tumor response to the EGFR antibody cetuximab. This EGFRi67 was applied to bladder cancer cell lines to predict their response to several EGFR inhibitors. Using newly generated and publically available pharmacologic data, we found that EGFR inhibitors were more effective in the predicted sensitive cell lines, consistent with our hypothesis.
Future validation studies in bladder cancer patients are needed to fully evaluate MSH2 and the EGFRi67 as effective biomarkers of therapeutic response in bladder cancer. Both of these biomarkers have the potential to improve clinical decision making by predicting response to treatment. MSH2 has the potential to identify patients unlikely to benefit from first-line therapy and may pursue other, perhaps more effective,
alternative therapies. The EGFRi67 could assist in this decision-making process by identifying patients likely to respond to a specific targeted therapy, informing an alternative therapy for patients where platinum-therapy is ineffective. Collectively, this work expands our knowledge of pharmacogenomics in MIBC and may be used to improve the use of precision medicine.
The form and content of this abstract are approved. I recommend its publication. Approved: James C Costello
v
ACKNOWLEDGEMENTS
This work was funded by the University of Colorado Cancer Center Front Range Cancer Challenge Fellowship awarded to Andrew Edward Goodspeed, the Denver Chapter of the Golfers Against Cancer and a Boettcher Foundation Award to James C. Costello, and the T32GM007635.
This thesis could not be possible without the support and insight of so many people. I thank Laurie Cook for seeing something special and challenging me to attend Graduate School. I would like to thank the entire Costello laboratory for their work and suggestions over the years. I owe deep gratitude to Rani Schwindt and Nicolle Witte for their patient assistance during my early work in coding and bioinformatics. I cannot speak enough to how my conversations and brainstorming with Bob Jones and Rani have kept me sane and helped my science. The Pharmacology training program and department are an amazing group of smart, fun, and supportive students, faculty, and staff who I will always remember for their immediate acceptance and continuous support. I would also like to thank the laboratories of Dan Theodorescu, Joaquin
Espinosa, and Joshua Black for their expertise, willingness to help, and kind sharing of resources.
Graduate school requires a great collection of great friends and I am so thankful for all of the hiking, camping, rock climbing, and pub trips that have made Colorado feel like home. My thesis work and so much more could not have been conceivable without my partner Pilar, whose never ending support and belief in me is a surprise every day. Finally, none of my accomplishments would be possible without the long-term support of
vi my family and particularly my parents, Ed and Cheryl. I am thankful for everyone who contributed to this work and who I am today.
vii
TABLE OF CONTENTS CHAPTER
I: INTRODUCTION ... 1
Overview of muscle-invasive bladder cancer ... 1
Muscle-invasive bladder cancer in the United States ... 1
Progression of muscle-invasive bladder cancer ... 2
Genomic landscape of muscle-invasive bladder cancer ... 3
Somatic DNA mutations ... 5
Somatic alterations in the TP53/Cell cycle pathway ... 6
Somatic alterations in the RTK/Ras/PI3K pathway ... 6
Somatic alterations in DNA repair pathways ... 6
Somatic alterations in chromatin epigenetics pathways ... 7
MIBC transcriptional subtypes ... 7
Treatment of muscle-invasive bladder cancer ... 11
Diagnosis of bladder cancer ... 11
Standard-of-care in bladder cancer ... 12
Emerging therapies for muscle-invasive bladder cancer ... 16
Biomarkers of chemotherapy response in muscle-invasive bladder cancer ... 23
Potential of biomarkers to enhance treatment of MIBC ... 23
Biomarkers involved in DNA-repair ... 24
DNA-damage response mutational panels ... 30
APOBEC3 expression and mutational signature ... 32
Molecular subtypes as biomarkers ... 32
Biomarkers associated with apoptosis ... 34
viii
Gene expression signatures ... 36
RTK biomarkers ... 37
Precision medicine ... 38
Promise of precision medicine ... 38
Benefit of precision medicine ... 39
Dissertation overview ... 42
II: MISMATCH REPAIR PATHWAY AND CISPLATIN RESISTANCE IN MUSCLE- INVASIVE BLADDER CANCER ... 45
Introduction ... 45
Motivation ... 45
Mismatch repair and cisplatin resistance ... 46
Mismatch repair in bladder cancer ... 49
Project outline ... 50
Materials and methods ... 51
Cell culture, shRNA knockdown, and drug treatments ... 51
Performing the CRISPR resistance screen ... 51
Sequencing of sgRNA ... 52
Resistance screen enrichment analysis ... 53
Western blots ... 53
Caspase activation ... 53
qRT-PCR ... 54
Analysis of bladder tumors ... 54
Results ... 56
Whole-genome CRISPR screen identifies mediators of cisplatin resistance in a bladder cancer cell line ... 56
ix
The mismatch repair pathway is a mediator of cisplatin sensitivity ... 59
Bladder cancer cell lines with knockdown of MSH2 are resistant to cisplatin ... 59
MSH2 knockdown impairs cisplatin-induced DNA-damage response ... 61
Cells with low levels of MSH2 are equally sensitive to oxaliplatin and other chemotherapies ... 65
Bladder tumors with low MSH2 protein have reduced expression of PI3K/mTOR/PKC- ζ pathway components ... 65
MSH2 protein levels do not correlate with clinicopathologic features in bladder cancer .... 68
MSH2 levels correlate with survival in platinum-treated bladder cancer patients ... 70
Discussion ... 72
III: A GENE EXPRESSION SIGNATURE TO PREDICT EGFR INHIBITOR SENSITIVITY IN BLADDER CANCER CELL LINES ... 80
Introduction ... 80
Motivation ... 80
Project outline ... 82
Materials and methods ... 82
Data preparation ... 82
Generation of the EGFRi67 gene expression signature ... 83
Prediction of cell line sensitivity by the EGFRi67 ... 84
Evaluation of the EGFRi67 in the Genomics of Drug Sensitivity (GDSC) ... 85
Drug enrichment analysis in the GDSC ... 85
Evaluation of the EGFRi67 in publically available bladder cancer cell line response to erlotinib and lapatinib ... 86
Cell culture and gefitinib dose response ... 86
x
Code and data availability ... 87
Results ... 87
Defining the EGFRi67 gene expression signature ... 87
Evaluation of the EGFRi67 in bladder cancer cell lines from the GDSC ... 91
Cancer cell lines classified as resistant by the EGFRi67 are more sensitive to PI3K and mTOR pathway inhibition ... 94
EGFRi67 predicts bladder cancer cell lines sensitive to EGFR TKI inhibition ... 94
Basal-like bladder cancer cell lines are not more sensitive to EGFR inhibition ... 97
Discussion ... 99
IV: DISCUSSION ... 107
Shortcomings of precision medicine ... 107
Role of resistance biomarkers ... 107
Lack of patients with precision medicine opportunity ... 109
More than genomics ... 111
Requirement of multiple arms in biomarker testing and discovery ... 113
Predictive or prognostic? ... 113
Evaluation of MIBC biomarkers for prognosis ... 114
Gaps in knowledge ... 117
NAC biomarkers and adjuvant treatment ... 117
Mechanism of MSH2 downregulation ... 118
Separation of functions in DNA repair and cisplatin response ... 120
EGFR signaling and mismatch repair ... 121
Overall summary ... 121
xi
LIST OF TABLES
Table 1.1. Proposed biomarkers of platinum-based response in MIBC with clinical
evidence. ... 25
Table 2.1. Sequences of shRNA sequences used in this study. ... 55
Table 2.2. Specific antibodies used in this study. ... 55
Table 2.3. Sequences of qRTPCR primers used in this study. ... 55
Table 2.4. Patient characteristics of platinum-treated patients by MSH2 group. ... 75
Table 2.5. Patient characteristics patients with an unrecorded treatment by MSH2 group…… ... 75
Table 2.6. Patient characteristics of platinum-treated and lymph+ patients by MSH2 group ... 75
Table 3.1. The nearest shrunken centroids derived from the PAM algorithm and the genes that define the EGFRi67 signature. ... 89
xii
LIST OF FIGURES
Figure 1.1. Bladder cancer progression. ... 4
Figure 1.2. Overlap of bladder cancer transcriptional subtypes across studies. ... 9
Figure 1.3. Distribution of recent pharmacologic-based clinical trials in MIBC. ... 17
Figure 1.4. The promise of pharmacogenomics and thesis project outline. ... 44
Figure 2.1. A whole-genome CRISPR screen to identify mediators of cisplatin resistance in a bladder cancer cell line. ... 58
Figure 2.2. sgRNA constructs targeting genes of the mismatch repair pathway are enriched in the cisplatin-treated MGHU4 cells. ... 60
Figure 2.3. Knockdown of MSH2 increases cisplatin resistance in bladder cancer cell lines ... 62
Figure 2.4. The DNA-damage response is slightly reduced in bladder cancer cell lines with knockdown of MSH2 compared to non-targeting controls treated with cisplatin. ... 64
Figure 2.5. Cell viability of MGHU4 cells when treated with several chemotherapies is unaffected by MSH2 knockdown. ... 66
Figure 2.6. Cell viability of 253J cells when treated with several chemotherapies is unaffected by MSH2 knockdown. ... 67
Figure 2.7. Expression of PI3K/mTOR/PKC-zeta pathway components in low MSH2 protein bladder cancer. ... 69
xiii Figure 2.9. Bladder cancer patients with low levels of MSH2 protein correlate with poorer
survival in platinum-treated patients. ... 73
Figure 2.10. Low MSH2 protein levels correlate with poorer survival in patients with aggressive disease treated with platinum-based chemotherapy. ... 74
Figure 3.1. Stratification of colorectal cancer using the EGFRi67 gene signature. ... 90
Figure 3.2. Experimental outline of this project. ... 92
Figure 3.3. Data normalization. ... 93
Figure 3.4. Evaluation of the EGFRi67 gene signature on EGFR inhibitor sensitivity in bladder cancer cell lines from the GDSC. ... 95
Figure 3.5. Evaluation of the EGFRi67 gene signature for all tested drugs in GDSC bladder cancer cell lines. ... 96
Figure 3.6. Prediction of EGFR inhibition sensitivity by the EGFRi67 gene signature in bladder cancer cell lines from the BLA40. ... 98
Figure 3.7. EGFR inhibition sensitivity according to transcriptional subtype in bladder cancer cell lines. ... 100
Figure 3.8. Evaluation of transcriptional subtype in the GDSC. ... 102
Figure 4.1. Precision medicine after the adoption of resistance biomarkers. ... 108
Figure 4.2. Evaluation of the ERCC2 and DDR mutation biomarkers in the bladder cancer TCGA cohort. ... 116
xiv
LIST OF ABBREVIATIONS
ABC, ATP binding cassette
ABCC3, ATP Binding Cassette Subfamily C Member 3 AKT, AKT Serine/Threonine Kinase
ALL, Acute lymphoblastic leukemia AMVAC, Accelerated MVAC
APOBEC3A, Apolipoprotein B MRNA Editing Enzyme Catalytic Subunit 3A APOBEC3B, Apolipoprotein B MRNA Editing Enzyme Catalytic Subunit 3B APOBEC3D, Apolipoprotein B MRNA Editing Enzyme Catalytic Subunit 3D APOBEC3H, Apolipoprotein B MRNA Editing Enzyme Catalytic Subunit 3H AREG, Amphiregulin
ARID1A, AT-Rich Interaction Domain 1A ATM, Ataxia Telangiectasia Mutated
ATR, Ataxia Telangiectasia And Rad3-Related Protein AURKA, Aurora kinase A
BAX, BCL2 Associated X, Apoptosis Regulator BBC3, BCL2 Binding Component 3, encodes PUMA BCG, Bacillus Calmette-Guerin
BLA40, Collection of 40 bladder cancer cell lines
BRAF, B-Raf Proto-Oncogene, Serine/Threonine Kinase BRCA1, BRCA1 DNA Repair Associated
BRCA2, BRCA2 DNA Repair Associated Cas9, CRISPR associated protein 9
CDKN1A, Cyclin Dependent Kinase Inhibitor 1A CDKN2A, Cyclin Dependent Kinase Inhibitor 2A cDNA, Complementary DNA
CHK2, Checkpoint Kinase 2
CMV, Cisplatin, methotrexate, and vinblastine COXEN, Coexpression extrapolation
CREBBP, CREB binding protein
CRISPR, Clustered Regularly Interspaced Short Palindromic Repeats CTLA-4, Cytotoxic T-Lymphocyte Associated Protein 4
CTR1, Copper Transporter 1, encoded by SLC31A1
DAVID, The Database for Annotation, Visualization and Integrated Discovery dd, Dose-dense
DDR, DNA-damage response E2F3, E2F Transcription Factor 3
EGFR, Epidermal Growth Factor Receptor EMT, Epithelial mesenchymal transition EP300, E1A Binding Protein P300
ERBB2, Erb-B2 Receptor Tyrosine Kinase 2 ERBB3, Erb-B2 Receptor Tyrosine Kinase 3
xv ERCC2, ERCC Excision Repair 2, TFIIH Core Complex Helicase Subunit
ERCC5, ERCC Excision Repair 5, Endonuclease EREG, Epiregulin
EXO1, Exonuclease I
FANCC, Fanconi Anemia Complementation Group C FDA, Food and Drug Administration
FGFR1, Fibroblast Growth Factor Receptor 1 FGFR2, Fibroblast Growth Factor Receptor 2 FGFR3, Fibroblast Growth Factor Receptor 3 FGFR4, Fibroblast Growth Factor Receptor 3 FOXA1, Forkhead Box A1
GATA3, GATA Binding Protein 3
GC, Gemcitabine and cisplatin regimen
GDSC, Genomics of Drug Sensitivity in Cancer GeCKO, Genome-Scale CRISPR Knock-Out GEO, Gene Expression Omnibus
GO, Gene ontology
GSEA,Gene set enrichment analysis HDAC, Histone deacetylase
HER2, Erb-B2 Receptor Tyrosine Kinase 2 HER3, Erb-B2 Receptor Tyrosine Kinase 3 HER4, Erb-B2 Receptor Tyrosine Kinase 4 HRAS, HRas Proto-Oncogene, GTPase IDL, Insertion deletion loop
IHC, Immunohistochemistry
KANSL1, KAT8 Regulatory NSL Complex Subunit 1 KDM6A, Lysine Demethylase 6A
KEGG, Kyoto Encyclopedia of Genes and Genomes KMT2C, Lysine Methyltransferase 2C
KMT2D, Lysine Methyltransferase 2D KRAS, KRAS Proto-Oncogene, GTPase
LCK, Lymphocyte Cell-Specific Protein-Tyrosine Kinase MAPK, Mitogen-Activated Protein Kinase
MCT1, Monocarboxylic Acid Transporter 1, encoded by SLC16A1 MDM2, MDM2 Proto-Oncogene
MDR1, Multidrug Resistance Protein 1, encoded by ABCB1 MEK, Mitogen-Activated Protein Kinase Kinase 1
MIBC, Muscle-invasive bladder cancer MLH1, MutS Homolog 2
MSH2, MutS Homolog 2 MSI, Microsatellite instability
mTOR, Mechanistic Target Of Rapamycin Kinase
xvi NAC, Neoadjuvant chemotherapy
NCI-60, National Cancer Institute 60 Human Tumor Cell Lines Screen
NCI-MATCH, National Cancer Institute-Molecular Analysis for Therapy Choice NGS, Next-generation sequencing
NMIBC, Non-muscle-invasive bladder cancer NRAS, NRAS Proto-Oncogene, GTPase ORR, Overall response rate
OS, Overall survival
PALB2, Partner And Localizer Of BRCA2 PAM, Prediction Analysis for Microarrays PD-1, Programmed Death Protein 1 PD-L1, Programmed Death Ligand 1 PI3K, Phosphoinositide 3-kinase
PIK3C2B, Phosphoinositide 3-Kinase-C2-Beta
PIK3CA, the Phosphatidylinositol 3-Kinase, Catalytic, Alpha Polypeptide PKC-ζ, PRKCZ, Protein Kinase C Zeta
PMI, Precision Medicine Initiative
PMS2, PMS1 Homolog 2, Mismatch Repair System Component POLE, DNA Polymerase Epsilon, Catalytic Subunit
PPARy, Peroxisome Proliferator Activated Receptor Gamma
qRT-PCR, Quantitative reverse transcription polymerase chain reaction RAD51B, RAD51 Paralog B
RB1, RB Transcriptional Corepressor 1 RC, Radical Cystectomy
RPPA, Reverse phase protein array
RSEM, RNA-Seq by Expectation Maximization RTK, Receptor tyrosine kinase
SDS-PAGE, Sodium dodecyl sulfate polyacrylamide gel electrophoresis sgRNA, Short guide RNA
shRNA, Short hairpin RNA SLC, Solute carrier
STAG2, Stromal Antigen 2
SWI/SNF, SWItch/Sucrose Non-Fermentable TBS-T, Tris-buffered saline-tween
TCGA, The Cancer Genome Atlas TKI, Tyrosine kinase inhibitor TP53, Tumor Protein P53 TSC1, TSC Complex Subunit 1
TURBT, Transurethral Resection of Bladder Tumor VEGF, Vascular Endothelial Growth Factor
VEGFR2, Vascular Endothelial Growth Factor Receptor2 WES, Whole-exome sequencing
1
CHAPTER I INTRODUCTION Overview of muscle-invasive bladder cancer
Bladder cancer is the ninth most common cancer type with an estimated 430,000 new cases and 165,000 deaths worldwide in 2012 (1). This cancer type represents all cancers that arise within the bladder with 80-90% of cases being urothelial carcinomas (2). The depth of tissue invasion defines the two main types of bladder cancer: non-muscle-invasive (NMIBC) and non-muscle-invasive bladder cancer (MIBC). While NMIBC is typically not immediately life threatening it does require long-term surveillance, has a high recurrence rate, and can progress into MIBC (3). In contrast, MIBC can rapidly metastasize and accounts for the majority of deaths caused by bladder cancer (4). The response rates to first-line treatment of MIBC has remained stagnant at 50% for three decades (5–9). However, several biomarkers of response and new therapeutic targets have recently been identified that may finally advance the treatment options and outcome of this disease (10–12).
Muscle-invasive bladder cancer in the United States
When all cancer types are considered, there are projected to be 1.7 million new cancer cases and 600,000 cancer-related deaths in the United States in 2018 (13). This places cancer as the second leading cause of death behind only heart diseases. While its continuous challenge to healthcare, society, and research cannot be understated, overall death rates have declined since 1990. It is estimated that the reduction in death rate has saved nearly 2.4 million lives over this time span. Since 1975, cancer incidence has also declined for many cancer types, but bladder cancer incidence has remained stagnant over this period (13).
2 Projections for 2018 suggest 81,000 new bladder cancer cases and 17,000
deaths in the United States, making bladder cancer the sixth most common cancer type in the country (13). In the last 5 years, total new cases of bladder cancer have
increased by 11.9% and total deaths have increased by 13.3%. In contrast, all cancer cases have increased by only 4.5% and deaths by 5% over the same time span. Bladder cancer is failing to see the reduction in incidence and death rates that many cancer types are now experiencing.
Bladder cancer is 3-4 times more likely to occur in men than women (14,15), and while definitive causes of this relationship are unclear, various environmental, hormonal, and sexual differences have been implicated (16). This cancer type is primarily a
disease of the elderly, with over 72% of cases diagnosed over the age of 65 (17). Additionally, the likelihood of developing bladder cancer is nearly three times greater in developed compared to less-developed countries (15). However, this trend may differ in the future as tobacco usage has significantly declined in developed countries faster than in less-developed countries. Tobacco usage is the main risk factor for bladder cancer (18) and the decline in bladder cancer incidence rate reflects the tobacco usage decline observed 20-30 years prior (19). Therefore, the incidence, mortality, and health
challenges of bladder cancer in less-developed countries is expected to increase (15).
Progression of muscle-invasive bladder cancer
Nearly 80% of patients with bladder cancer will be diagnosed with NMIBC (Figure 1.1 (20)) (21). NMIBC is a heterogeneous disease with one type being papillary tumors (Ta), which are low grade tumors limited to the epithelial tissue of the bladder (22–24). Higher grade NMIBC includes carcinoma in situ (CIS or Tis) and tumors that invade the
3 basement membrane into the subepithelial tissue (T1). While Ta tumors typically do not progress into the advanced stages of bladder cancer (25), CIS tumors have a 31% chance of becoming MIBC (26) and T1 tumors have around a 50% chance of progressing (23,27,28). Currently, no molecular biomarkers accurately predict the progression of NMIBC to MIBC, although risk tables have shown some promise (29).
MIBC is comprised of tumors that are either confined to the bladder muscle (T2), have progressed into the fat tissue (T3), or beyond (T4) (Figure 1.1 (20)). At diagnosis, between 20-30% of patients present with MIBC (30), which is more prone to metastasis and death compared to NMIBC (4).
Genomic landscape of muscle-invasive bladder cancer
Exciting additions to the field of cancer research are the massive molecular characterization studies that have been made possible by technological advancements and the sheer willpower of scale. The Cancer Genome Atlas (TCGA) has characterized 33 cancer types consisting of over 11,000 cancer patients in the form of DNA
sequencing, RNA-sequencing, DNA methylation, Reverse phase protein array (RPPA), miRNA, and clinical data. The TCGA has molecularly analyzed 412 patients with MIBC while others have characterized NMIBC (31–33). Unbiased genomic characterization has the advantage of identifying relatively low penetrance but potentially important alterations in addition to the frequent alterations that may have already been identified. This type of characterization also allows for the discovery of large subtype classes based on molecular measurements, most often by expression level of many genes (31,32).
4
Figure 1.1. Bladder cancer progression.
This figure is adopted from Sanli et al. 2017 (20). It demonstrates the tissue invasion as bladder cancer progresses as well as diagnosis rates and treatment by bladder cancer type. At diagnosis: ~80% ~20% Treatment: TURBT Intravesical therapy Radical cystectomy Systemic chemotherapy NMIBC MIBC
5 NMIBC is generally has a stable genome aside from frequent deletions on
chromosome 9 (~50%) (34). Several genes of interest have been located in this region including CDKN2A and TSC1. Another common alteration in NMIBC is activating point mutations in FGFR3 (53%) (33). FGFR3 mutations have been proposed to be mutually exclusive to other alterations in the same pathway such as the Ras genes HRAS, KRAS, and NRAS (35). Molecular characterization has revealed NMIBC have frequent mutations in several other cancer-related genes including, PIK3CA (23%), KDM6A (22%), KMT2C (19%), STAG2 (18%), ATM (15%), ARID1A (13%), CREBBP (12%), KMT2D (12%), ERBB3 (10%), ERBB2 (8%), TP53 (8%), and RB1 (7%), all of which have also been identified in MIBC (31–33).
To date, the TCGA has performed the most complete molecular characterization of MIBC by performing analyses on 412 cases (31,32). Most patients were treatment naïve but 35 were previously treated with the tuberculosis vaccine Bacillus Calmette-Guerin (BCG). All but 13% have urothelial carcinoma. Over 67% of the tumors were T3 or higher and nearly all are high grade. This represents a later stage and highly
aggressive cohort with the 5-year survival for the entire cohort being just 42%. Somatic DNA mutations
The TCGA identified 4 common causes of the mutational signatures observed in MIBC: APOBECA, APOBECB, 5-methylcytosine deamination, and ERCC2. The
APOBEC family encodes cytidine deaminases, introducing mutations that correlate with gene expression (36,37). APOBEC alone was found to contribute 67% of all SNVs in the TCGA cohort. Patients whose tumors exhibited high APOBEC-related mutational load strongly correlated with better overall survival (31). ERCC2 is a component of the
6 nuclear excision repair pathway. This mutational signature correlated with mutations in the ERCC2 gene (31). Mutations are also caused by 5-methylcytosine deamination, which is a spontaneous process (38,39).
Somatic alterations in the TP53/Cell cycle pathway
The TP53/Cell cycle pathway is altered in 89% of MIBC cases (31). This is largely driven by mutations in TP53 (48%) and deletions of CDK2NA (22%), which are mutually exclusive. Also contributing to alterations in this pathway are mutations and deletions in RB1 (10 and 4%, respectively), E2F3 amplifications (12%), ATM mutations (14%) and MDM2 amplifications (6%). TP53 mutations are also mutually exclusive from MDM2 amplification and CDKN2A deletions while they co-occur with RB1 mutations. Somatic alterations in the RTK/Ras/PI3K pathway
The RTK/Ras/PI3K pathway is altered in 71% of MIBC cases (31). Genes encoding the tyrosine-kinase receptors FGFR3 (14%), ERBB2 (12%), and ERBB3 (10%) are commonly mutated while EGFR (4%) and ERBB2 (4%) are amplified. Additionally, several mediators of downstream signaling are altered including PIK3CA (22%) and Ras (HRAS, NRAS, and KRAS) (11%) mutations. FGFR3 mutations were more frequently identified in lower-stage tumors and associate with better survival. Additionally, a low number of tumors (4%) had a gene fusion involving FGFR3. Somatic alterations in DNA repair pathways
Several genes involved in DNA repair are altered including mutations in ATM (14%) and ERCC2 (9%) and deletions in RAD51B (2%) (31). As mentioned above, ERCC2 mutations contribute to one of the mutational signatures identified in MIBC.
7 Somatic alterations in chromatin epigenetics pathways
Several genes involved in chromatin structure and epigenetics are also altered in MIBC (31). These include mutations in several histone methyltransferases (KMT2D (28%), KMT2C (18%), and KMT2A (11%)), the histone demethylase, KDM6A (26%), and several histone acetylases (EP300 (15%), CEBBP (12%), and KANSL1 (6%)). ARID1A (25%), a member of the SWI/SNF remodeling complex is also frequently mutated.
MIBC transcriptional subtypes
Traditionally, cancer subtypes have been classified on histological markers. Perou et al. (40), followed shortly by Golub et al. (41), presented a new way to classify molecular subtypes by gene expression patterns measured using microarrays. This was an advancement to the field because molecular subtyping could be done in an unbiased manner based on gene expression patterns. Therapeutic response to some therapeutic agents can be predicted based on molecular subtype both in vitro (42,43) and clinically (10,44–47). Modeling and analyzing subtypes has been the subject of many studies because of its clinical impact (48).
The first transcriptional subtyping study performed in bladder cancer was
published in 2012 (49). It became known as the LUND study after the university where the work took place. This study clustered 308 bladder cancer samples that ranged from Ta to T2 tumors. To date, this is still the only bladder cancer study to include both NMIBC and MIBC in its cluster generation. Many of the primary findings of this study were clearly fundamental in the field as the results are consistent with several other more recent studies.
8 The LUND study identified 5 subtypes in bladder cancer. Excitingly, some of these subtypes correlated with prognosis as the urobasal A subtype had good survival while the urobasal B and SCC-like subtypes had poorer survival. The LUND study also found that the subtypes differed in their expression of genes relating to immune
infiltration, and the expression of keratins and RTKs.
Several groups have subsequently identified 2-5 transcriptional subtypes in MIBC. Further computational analyses have indicated strong overlap between all of the subtypes that have been identified and that for many inter-study subtypes, the only difference is nomenclature (Figure 1.2) (50). Accordingly, the two major recognized subtypes are known as ‘basal’ and ‘luminal’.
The basal subtype is often defined by its expression of high molecular weight keratins, particularly keratins 5, 6B, and 14 as well as CD44 (10,49,51). All of these markers are expressed at high level in urothelial basal cells. In contrast, the luminal subtype is defined by high expression of uroplakins, including uroplakin 1B, 2, and 3A in addition to keratin 20. These are markers of urothelial luminal differentiation cells. The basal subtype is driven by the transcriptional factor p63 while the luminal subtype is driven by PPARy, GATA3, and FOXA1 (10,52). Histologically, luminal tumors often have papillary-like morphology while basal tumors have squamous-like features (10,53). These subtypes are also linked to common mutations in bladder cancer as the luminal subtype is enriched in FGFR3 mutations while the basal subtype is enriched in TP53 mutations (10,31,49,51,54). The latest TCGA analysis has subdivided the luminal subtype by differences in wild-type p53, epithelial-mesenchymal transition (EMT), and immune cell gene signatures (31).
9
Figure 1.2. Overlap of bladder cancer transcriptional subtypes across studies.
This figure is adopted from Aine et al. 2015 (50). It demonstrates the similarities in the identified subtypes from the UNC (51), MDA (10), TCGA (54), and LUND (49) studies.
10 In addition to molecular characterization, bladder cancer subtypes have been linked to prognosis and response to therapy. Beginning with the LUND study, the basal subtype was found to correlate with poorer survival (49), a finding that was confirmed in subsequent studies (10,31,32). Despite its poor prognosis, there is mounting evidence that the basal subtype responds the best to the addition of chemotherapy to the
treatment regimen. The first evidence came in 2014, when Choi et al. showed that a novel subtype, p53-like, was resistant to chemotherapy (10). Bioinformatics analysis suggests that the majority of these tumors are luminal, likely reducing the benefit of chemotherapy in the luminal subtype (20,55). Since this work, several studies have confirmed that basal tumors are aggressive if untreated or only treated with radical cystectomy. However, when chemotherapy is added to the regimen (either neoadjuvant or adjuvant), these are the tumors that benefit the most in survival (56–58). This is consistent with breast cancer, where basal tumors are also sensitive to chemotherapy (59). Despite evidence that basal tumors benefit from chemotherapy, one recent meeting abstract suggests they are resistant to treatment compared to luminal tumors that show a strong response (60). Therefore, further research is necessary to cement transcription subtypes as a means to predict patient response to chemotherapy.
Molecular subtyping has also lead to the identification of several targetable alterations that are enriched in various subtypes. These alterations represent additional therapeutic targets that may be differentially beneficial across transcriptional subtypes. EGFR was found to be expressed more highly and to be more active in basal bladder cancers (31,32). Accordingly, basal bladder cancer cell lines have more active EGFR signaling and are more sensitive to EGFR inhibition (61). The serine/threonine kinase,
11 Aurora kinase A (AURKA) is also more highly expressed and active in basal bladder cancer representing an additional kinase target in this subtype (62). Meanwhile, the enrichment of FGFR3 mutations in the luminal subtype is targetable and inhibitors of FGFR3 are being studied in bladder cancer (NCT03473743 and NCT03390504) (31,54,63,64). As discussed above, an important feature of luminal bladder cancer is the activation of PPARγ. Preclinical studies have suggested that PPARγ agonists can inhibit growth of some bladder cancer cell lines (65). HER2 (ERBB2 gene product) is more highly expressed in luminal bladder cancer (31,32). Targeting of HER2 has clinical benefit in breast cancer and its benefit in bladder cancer should be further explored.
Treatment of muscle-invasive bladder cancer
Diagnosis of bladder cancer
The most common symptom of bladder cancer is the presence of visible blood in urine, which occurs for 78% of patients (66). However, only 2-5% of patients with this symptom actually have bladder cancer (67). Initial diagnosis requires an endoscopic resection by a procedure called transurethral resection of bladder tumor, or TURBT. Diagnosis includes a cystoscopy to image the inner layer of the bladder (68), which may be combined with histological analysis and cytology of cells in the urine to identify any abnormal cells that may have been missed during the cystoscopy (69). In addition to being an important diagnostic tool, TURBT also plays a role in the treatment of the bladder tumor. This technique is used to physically remove both small and large tumors in single or repeated fractions.
Following diagnosis, it is important to classify the cancer under the Tumor, Node, Metastasis (TNM) classification system because bladder cancer prognosis and
12 management depends on the histology of the cancer (70). Under staging the bladder tumor is an important risk because it can lead to mismanaged treatment. A complete TURBT that contains the muscular layer provides the best opportunity to properly diagnose the pathology of a tumor. However, even when the muscular layer is present, there is up to a 20% chance that patient initially diagnosed with T1 will be upstaged to T2. This increases to up to a 40% chance if the muscular layer is not included (71). Because accurate staging is so important to proper treatment, it is recommended to repeat TURBT 2-6 weeks after the initial procedure. Repeating TURBT increases pathologic accuracy and recurrence-free survival by 25% in patients initially diagnosed with T1 disease (72–74). Molecular characterization of bladder cancer may indicate improperly staged tumors as the frequency of some alterations differ from NMIBC to MIBC (31,33,54).
Standard-of-care in bladder cancer
NMIBC is typically treated with TURBT alone or coupled with intravesical therapy if there is high risk of recurrence or progression. Risk groups are based on the number of tumors, recurrence rate, presence of CIS, tumor size, grade, and stage (75).
Intravesical therapy typically includes chemotherapy or BCG. A randomized controlled clinical trial and meta-analysis suggests BCG is superior to chemotherapy (76,77). If relapse or no benefit from BCG is observed, the best option for NMIBC is radical cystectomy, although other considerations can lead to the use of intravesical chemotherapy, typically mitomycin or doxorubicin (78).
Meanwhile, MIBC is typically treated more aggressively with a cystectomy and potentially systemic chemotherapy. Radical cystectomy involves the surgical removal of
13 the bladder, a lymphadenectomy, and typically a prostatectomy in men and a
hysterectomy in women. Radical cystectomy reduces the quality of life for patients in part because of the requirement of urinary diversion (79). In a large bladder cancer cohort treated with radical cystectomy, 5-year survival of NMIBC was 79.1% compared to just 56.2% in patients with MIBC. In the lymph node positive MIBC patients, 5-year survival fell to 29.1% (80).
In addition to radical cystectomy, systemic chemotherapy is commonly given to patients with MIBC. This typically includes either the MVAC (Methotrexate, vinblastine, doxorubicin, cisplatin) or GC (Gemcitabine, cisplatin) chemotherapy regimen (9). Clinical trials indicate there is no survival difference between MVAC or GC but GC provides a better safety and tolerability profile (9,81). The addition of neoadjuvant chemotherapy (NAC) has been shown to increase 5-year survival of MIBC patients by 5% (82). This benefit is even greater for patients who experience complete down staging following NAC. The 5-year survival rate for these MIBC patients is 88% compared to ~50% in patients without complete down staging (83). This also
demonstrates the validity of response rate as a surrogate for survival. The benefit of adjuvant chemotherapy is less clear than NAC, however, it is suggested to
independently increase overall and disease-specific survival (84).
Despite the improvement in survival when NAC is administered, adoption in the clinic has been poor due to the limited improvement in survival and the toxicity
associated with chemotherapy. A retrospective study of MIBC from 2003-2008 suggests that as few as 17% of patients that receive radical cystectomy are also treated with NAC
14 but the rate of NAC usage has been increasing (84,85). Biomarkers that are predictive of the response to chemotherapy may improve clinical adoption of NAC (86).
Before the recent approval of immunotherapy in bladder cancer, patients unresponsive or ineligible for cisplatin-based chemotherapy were treated with non-cisplatin chemotherapy regimens. Typical chemotherapy regimens after progression include taxanes or vinfluine while cisplatin ineligible patients may also be treated with a combination including the less toxic cisplatin analog carboplatin. Unfortunately, the response rate to secondary chemotherapy is typically only 5-20% (87–94).
In 2016, secondary treatment options were expanded with FDA-approval of the PD-L1 inhibitor atezolizumab for patients with locally advanced or metastatic bladder cancer that has progressed during or after platinum-based chemotherapy1. Since this announcement, the FDA has approved 4 other PD-1 and PD-L1 (pembrolizumab, nivolumab, durvalumab, and avelumab) for use in patients who have worsened
following platinum-based chemotherapy. Additionally, atezolizumab and pembrolizumab are approved as first-line therapy for use in bladder cancer patients ineligible to receive cisplatin-based chemotherapy. One mechanism of immune evasion by cancer is
increased expression of PD-L1, which is the ligand of PD-1 expressed on cytotoxic T cells. Binding of PD-L1 by PD-1 reduces the immune response and T cell antigen receptor-mediated lysis (95). PD-1 and PD-L1 inhibitors enhance the immune response by blocking this suppressive interaction.
The development of effective immunotherapy for treatment of bladder cancer is exciting because it is one of the few therapeutic changes in bladder cancer treatment in
1
15 nearly three decades. However, the benefit of PD-1 and PD-L1 inhibitors in bladder cancer is controversial. While the overall response rate (ORR) of these inhibitors in previously treated patients is fairly strong (15-24.4%) (96–101), these results have not always translated to improved overall survival compared to the previous standard of secondary chemotherapy. Atezolizumab versus chemotherapy (physician's choice of vinflunine, paclitaxel, or docetaxel) was evaluated in bladder cancer patients who progressed past platinum-based chemotherapy. There were no statistical differences in the atezolizumab and chemotherapy groups in ORR (23% vs. 22%) or overall survival (median 11.1 months vs 10.6 months), although the atezolizumab group did have fewer adverse effects (102). Pembrolizumab was also directly compared to the same
chemotherapy arm in a similar patient population. In this study, the immunotherapy arm outperformed chemotherapy in ORR (21.1% v 11.0%) and overall survival (median 10.3 v 7.3 months) (103). While this study of direct comparison suggests pembrolizumab is more effective than secondary chemotherapy, several discussion points are raised. First, not all studies show immunotherapy is more effective than secondary
chemotherapy in platinum-progressed patients. Second, the benefit of pembrolizumab is a minor improvement compared to secondary chemotherapy. Additionally, the ORR to pembrolizumab and atezolizumab for patients ineligible for cisplatin chemotherapy is only 23-24% (104,105), which is considerably lower than less toxic cisplatin analogs such as oxaliplatin (ORR 48%) in a similar patient population (106). Comparative studies testing immunotherapy versus cisplatin analogs are underway but results are currently unavailable. Therefore, while the initial approval of immunotherapy in bladder
16 cancer is exciting, additional advancements and improvements to overall survival are still necessary for this patient population.
Emerging therapies for muscle-invasive bladder cancer
The treatment of MIBC would benefit from more effective and less toxic therapies compared to standard platinum-based chemotherapy. The investigation of targeted agents and immunotherapies in bladder represents hope that new therapeutic options will become available that will be effective for at least a subset of patients.
Since 2015, 139 therapeutic intervention clinical trials containing drugs and/or biologicals have begun in advanced (muscle-invasive or metastatic) bladder cancer (Figure 1.3, data downloaded from clinicaltrials.gov on June 30, 2018). Remarkably, 81 (58%) of these clinical trials include the use of PD-1 or PD-L1 inhibitors. Roughly 37% of all trials primarily include chemotherapy, radiation, surgery, and/or imaging (denoted as ‘Standard’). Other popular treatment regimens currently explored include therapies that target non-PD-1/PD-L1-immunology (17%), growth signaling (6%), angiogenesis (6%), and DNA-damage repair (5%) (Figure 1.3).
As discussed above, PD-1 and PD-L1 inhibitors have been approved for use in advanced bladder cancer ineligible for or that have progressed past platinum-based chemotherapy. Several current clinical trials are exploring these agents for use in other indications but none have published results (Figure 1.3). Pembrolizumab
(NCT02736266) and Atezolizumab (NCT02662309) are being investigated as
preoperative single agents prior to planned cystectomy in MIBC. Meanwhile, Avelumab (NCT02603432) and Pembrolizumab (NCT02500121) are undergoing testing as
17
Figure 1.3. Distribution of recent pharmacologic-based clinical trials in MIBC.
All MIBC clinical trial data with a start data after January 1, 2015 were downloaded from clinicaltrials.gov on June 30, 2018. The treatments for these clinical trials were manually curated and categorized. The majority (58%) of these recent MIBC trials contained PD-1 or PD-L1 inhibitors. Standard 31 38.3% PD-1/PD-L1 inhibitor alone 18 22.2% Immunotherapy (Non-PD-1/PD-L1) 16 19.8% Angiogenesis 6 7.4% DNA-damage repair 3 3.7% Growth signaling 3 3.7% Cell cycle 2 2.5% Conjugate Ab 1 1.2% Other 1 1.2% Standard 21 30.0% Immunotherapy (Non-PD-1/PD-L1) 8 11.4% Growth signaling 5 7.1% Metabolism 6 8.6% Other 4 5.7% DNA-damage repair 4 5.7% Sex hormone 4 5.7% Angiogenesis 3 4.3% Conjugate Ab 3 4.3% Standard therapies indicate that only chemotherapy, radiation,
surgery (including heat and laser), and imaging is used
Trials with PD-1/PD-L1 inhibitors (n = 81)
Trials without PD-1/PD-L1 inhibitors (n =58) Conjugate Ab Other Cell cycle DNA−damage repair Growth signaling Angiogenesis Immunotherapy (Non−PD−1/PD−L1) PD−1/PD−L1 inhibitor alone Standard Standard Growth signaling Metabolism Other DNA−damage repair Sex hormone Angiogenesis Conjugate Ab Immunotherapy (Non−PD−1/PD−L1)
Containing PD-1 or PD-L1 (n = 81, 58%) Not containing PD-1 or PD-L1 (n = 58, 42%)
18 (NCT02632409) and Atezolizumab (NCT02450331) are under investigation as adjuvant therapy following surgery. There is mounting evidence that combination therapies may be the key to increasing the number of patients that benefit of PD-1 and PD-L1
inhibitors (107). As shown in Figure 1.3, over three quarters of MIBC clinical trials involving PD-1 or PD-L1 inhibitors are in combination with a targeted agent or a standard bladder cancer treatment. Similarly with these inhibitors in other settings of bladder cancer, results of these combinations are not yet available.
Another immunotherapy investigated in MIBC are antibodies against CTLA-4. CTL-4 is a surface receptor found on regulatory T cells. This receptor competes with other T cells for binding to CD80 and CD86 on antigen-presenting cells. CTLA-4 inhibits T cell responses by sequestering CD80 and CD86 (108,109). Antibodies to CTLA-4 inhibit this inhibition in a similar manner as PD-1 and PD-L1 inhibitors. Unfortunately, the combination of the CTLA-4 antibody, ipilimumab, with GC failed to see any added benefit compared to GC alone in metastatic MIBC (110). However, there are 9 studies currently investigating the combination of CTLA-4 antibodies and PD-1 or PD-L1 inhibitors (Figure 1.3). This combination has shown an improvement to long-term response in melanoma compared to single agents (111,112).
Therapeutic targeting of growth signaling is another emerging treatment in advance bladder cancer. This typically consists of targeting fibroblasts growth factor receptors (FGFR1-4) the Erbb receptor family (EGFR and HER2-4) or the PI3K/mTOR pathway.
FGFR3 mutations and gene fusions are found in 14% and 3% of MIBC (31) and up to 70% of superficial bladder cancer harbor FGFR3 mutations (113). Frequent
19 FGFR3 alterations have lead to several clinical trials evaluating agents that target FGF receptors. The first FGFR3 inhibitor evaluated in bladder cancer was dovitinib. This study was terminated because only 1 out of 44 patients obtained a partial response. Since this failure, more potent inhibitors have performed better in clinical trials (114). The pan-FGFR inhibitor, JNJ-42756493, was tested in 65 patients with advanced bladder cancer, 23 of which had alterations in FGFR1-4 genes (amplifications,
mutations, translocations). While none of the FGFR wild-type patients responded to this therapy, 3 of the patients with FGFR fusions had a partial response, suggesting that response may be improved in patients with FGFR alterations. Preliminary results from another pan-FGFR inhibitor, BGF398, also provide evidence that FGFR inhibition is more effective in tumors with FGFR alterations. Four out of the five bladder cancer patients with FGFR3 mutations in one trial cohort had tumor regression (63). Another study of 25 advanced bladder cancers with FGFR3 mutations or fusions found that 8 patients demonstrated partial response while another had a complete response to BGF398 (64). These trials suggests that FGFR inhibitors have clinical activity in some bladder cancer patients and that FGFR3 alterations may serve as an effective
biomarker to identify those most likely to benefit. Two current clinical trials are
evaluating the combination of an FGFR inhibitor with a PD-1 inhibitor in FGFR-altered bladder cancer (NCT03473743 and NCT03390504), while others are evaluating FGFR inhibitors as single agents (Figure 1.3).
The Erbb family of receptors contains 4 members that form heterodimers. The targeting of EGFR and HER2 is successful in several cancer types. In MIBC, EGFR is amplified in 4% of tumors and is also a dominant feature of bladder cancers of the basal
20 transcriptional subtype (31). This has lead to the observation that basal bladder cancer cell lines are more sensitive to EGFR inhibition (61). Meanwhile, ERBB2, the gene that encodes HER2, is mutated (14%) and amplified (4%) in MIBC (31). Because of the similarities between EGFR and HER2, some dual-tyrosine kinase inhibitors (TKIs) have been developed. Targeting of EGFR and HER2 is also accomplished using monoclonal antibodies targeting the extracellular region of each receptor. Several clinical trials targeting EGFR and HER2 have also been performed in bladder cancer with mixed results.
When added to GC, the EGFR antibody Cetuximab performed similarly to chemotherapy alone (115). Gefitinib (EGFR TKI) was administered in combination with GC in advanced bladder cancer. While this additional treatment was well-tolerated, the response rate and overall survival was consistent with historical averages of GC alone (116). A study of lapatinib (dual HER2/EGFR irreversible TKI) found no benefit as maintenance therapy after first-line chemotherapy in EGFR1/HER2 positive metastatic bladder cancer (117). In advanced bladder cancer patients, the HER2 antibody
Trastuzumab was added to standard chemotherapy. While the ORR and survival was relatively high, it was consistent with historical averages of chemotherapy as a single treatment (118).
While these studies show that EGFR inhibition may have limited efficacy in bladder cancer, several trials have found at least some patients in fact benefit from EGFR and HER2 inhibition (119–122). The EGFR TKI erlotinib was administered as neoadjuvant therapy to 20 patients with stage T2 muscle-invasive bladder cancer. Prior to radical cystectomy, 5 (25%) and 7 (35%) patient tumors were down staged to pT0 or
21 pT1, respectively (121). This study demonstrates the potential of erlotinib as a single agent to provide at least short-term benefit in the neoadjuvant setting for select patients with bladder cancer. Another promising trial evaluated gefitinib as a single agent for patients with metastatic bladder cancer patients who have progressed past first-line chemotherapy. In this trial, 3 of the 31 patients (9.7%) demonstrated stable disease or partial response (122). In a study of 49 advanced bladder cancer patients who have failed platinum-based therapy, lapatinib did not reach its primary endpoint of ORR > 10%. However, 17/19 patients with stable disease overexpressed EGFR or HER2 by IHC. This population also experienced significantly increased median survival (30.3 weeks vs. 10.6 weeks) (119). The pan-Erbb family TKI, Afatinib, was administered to 23 platinum-refractory patients. Five out of the six patients with HER2 amplification or ERBB3 mutation reached the primary endpoint of PFS of 3 months and they were the only patients to do so. The median PFS for altered patients was 6.6 months compared to 1.4 months in wild-type patients (120). While the response rates for these trials are relatively low, there are some patients that respond to EGFR inhibition and there is evidence that some of these patients can be identified by alterations in the Erbb receptor family.
Potentially due to the mixed results of past trials, only 1 study since 2015 is investigating the benefit of EGFR inhibition in advanced bladder cancer
(NCT02780687). Similarly, just 1 study (NCT03507166) is investigated the use of an antibody-drug conjugate that links the toxin Monomethyl auristatin E to an antibody targeting HER2. The background of this study is HER2+ advanced bladder cancer and perhaps testing this agent in an altered background will increase the likelihood of benefit
22 as there is strong evidence from other cancer types that targeting HER2 is beneficial in patients with ERBB2 amplifications (123,124).
Due to growth and hypoxic conditions, tumors often rely on the generation of new blood vessels, or angiogenesis, to delivery nutrients and oxygen (125). Preclinical
studies have implicated the role of angiogenesis in bladder cancer growth (126). A standard way to target angiogenesis is to inhibit VEGF and VEGF receptors. Single arm studies of VEGFR2 TKIs in both platinum-treated and metastatic bladder cancer have shown relatively high response rates (11 and 72%, respectively) but the benefit of VEGFR2 inhibition cannot be determined without a comparative study arm (127,128). Another angiogenesis agent tried in bladder cancer is the VEGFR2 monoclonal antibody, ramucirumab. This antibody was added in combination with docetaxel and compared to docetaxel alone in 90 patients previously treated with platinum-based therapy. The combination impressively resulted in a higher ORR (19.6% vs. 4.5%) and improved overall survival (5.1 vs. 2.4 months) (129). These studies demonstrate the need to further investigate the use of angiogenesis inhibitors in advanced bladder cancer. Enthusiasm continues to be high for angiogenesis agents, as 6 studies since 2015 are investigating various angiogenesis inhibitors in combination with PD-1 and PD-L1 inhibitors and another 3 studies are studying these as single agents or in combination with chemotherapy (Figure 1.3).
The poly (adenosine-diphosphate ribose) polymerase (PARP) family is a
prevalent target in cancer treatment (130). The most dominant isoform is PARP1 and it mediates DNA repair of both single- (131) and double-stranded DNA breaks (132). PARP inhibitors have been shown to be particularly beneficial in cancers with
23 alterations in DNA repair pathways (133,134). Accordingly, the PARP inhibitor,
Olaparib, is being investigated as a single agent in advanced bladder cancer with mutations in DNA-repair genes. PARP inhibitors have also been shown to increase PD-L1 expression in breast cancer cell lines (135), and 3 current studies are investigating the combination of Olaparib with the PD-L1 antibody Durvalumab in bladder cancer. There have been few results of PARP inhibitors in bladder cancer but a recent pan-cancer study found that one bladder pan-cancer patient with a mutation in the BRCA-interacting protein, PALB2, had a partial response to the PARP inhibitor, talazoparib (136).
Biomarkers of chemotherapy response in muscle-invasive bladder cancer
Potential of biomarkers to enhance treatment of MIBC
There are several pieces of evidence that strongly support NAC as a component of first-line treatment in MIBC. Most importantly, the addition of NAC increases cancer-specific survival by 5% compared to radical cystectomy alone (82). Additionally,
treatment with radical cystectomy alone is inadequate for many patients with MIBC with organ-confined disease. Five-year survival rates are only 72% in this population which demonstrates even in patients where the only detectable disease is in the bladder, removing the organ is inadequate for many of these patients (80). Systemic therapy has the potential to target micrometastases that play a major role in the progression of this disease. Finally, NAC is favorable to adjuvant chemotherapy because it is better tolerated and there is less data surrounding adjuvant therapy in MIBC.
Despite the clear advantages of NAC, there is not wide acceptance in the
24 treatment. A second concern is of the high level of toxicity associated with
chemotherapy, which may affect surgical morbidity and mortality. These valid concerns may be partially alleviated with accelerated chemotherapy regimens and the
advancement of predictive biomarkers. Modifications to MVAC maintain historical response to chemotherapy but shorten length of treatment and the time to surgery compared to standard chemotherapy regimens (137,138). Predictive biomarkers in cancer treatment include any measurement that can identify which patients will respond to a given treatment. In regards to NAC, biomarkers could inform which patients are most likely to benefit from this treatment and provide confidence to the clinician that delaying surgery may be beneficial. Conversely if a biomarker suggests a patient is unlikely to benefit from NAC, the clinician would have that information to decide on immediate surgery or an alternative treatment without going through the toxicity and time associated with standard chemotherapy. The adoption of effective biomarkers is expected to increase survival and reduce the overall cost to treat MIBC (139). Several genomic and protein biomarkers have been suggested to predict the response and benefit of platinum-based chemotherapy in the neoadjuvant and adjuvant setting during the treatment of advanced bladder cancer (Table 1.1).
Biomarkers involved in DNA-repair
Several components of DNA-repair have been implicated in the response to NAC. The nuclear excision repair pathway (NER) typically repairs thymine dimers caused by ultraviolet light but it is also capable of repairing DNA adducts, such as the intrastrand crosslinks formed by cisplatin, a major component of chemotherapy
regimens in MIBC. Several components of the NER pathway are altered in cancer,
h
25 Ta b le 1.1. P roposed bioma rke rs of p latinum -based respons e in M IBC w ith c linica l ev idenc e .
Table 1.1. Proposed biomarkers of platinum-based response in MIBC with clinical evidence.
Pathway Biomarker Broad suggestion Author Evaluated patients (n) Patient
population Chemotherapy Technology Main findings DNA-repair
ERCC2 ERCC2 mutant tumors are more responsive Van Allen et al. 2014 50 MIBC GC (62%), ddMVAC (32%), GC+Sunitinib (4%), ddGC (2%) WES
9/25 responders (40%) had ERCC2 mutations compared to 0/25 (0%) non-responders ERCC2 mutant tumors are
more responsive
Liu et al.
2016 48 MIBC
ddGC + pegfilgrastim and AMVAC
WES (only ERCC2 visible)
8 of 20 responders (40%) and 2 of 28 nonresponders (7%) had ERCC2 mutations. ERCC2 mutants had better OS.
ERCC2 mutant tumors are more responsive
Groenendijk
et al. 2016 71 MIBC Platinum-containing NGS of 178 genes
ERCC2 mutations trended to be higher in complete responders 6/38 (16%) responders compared to 2/33 (6%) nonrespnders (p = 0.27) in platinum-treated patients. When carboplatin-treated patients are removed, ERCC2 mutations are significant (p = 0.03). ERCC1 Low ERCC1 expression
results in a better response
Bellmunt et
al. 2007 57 Metastatic MIBC GC or GC+paclitaxel qRTPCR
Patients with low ERCC2 (37/57 or 65%) had better overall survival (25.4 vs. 15.4 months, p = 0.03) P = 0.03
High ERCC2 has poorer response
Hoffman et
al. 2010 108 MIBC
Adjuvant
platinum-treated qRTPCR
High ERCC2 expression (top 25%) resulted in poorer PFS (p = 0.03)
ERCC1 expression was not associated with response
Choueri et al. 2015 31 Non-metastatic MIBC ddMVAC + pegfilgrastim IHC
With few patients, a significant association could not be reached but the response rate was higher in ERCC1 negative patients (60% vs. 43%). BRCA1
High BRCA1 expression is associated with a poorer response
Font et al.
2011 57 MIBC CMV or GC qRTPCR
The top third of patients with high BRCA1 mRNA levels had reduced response rate (22% vs. 66%) and poorer median survival (34 vs. 168 months) BRCA1 mRNA expression
did not associate with survival
Bellmunt et
al. 2007 56 Metastatic MIBC GC or GC+paclitaxel qRTPCR
BRCA1 mRNA expression did not correlate with survival
DNA-damage response mutational panels
ATM/FANCC/RB1 Patients with at least one mutation in ATM, FANCC, or RB1 have improved response Plimack et al. 2015 Discovery: 34 Validation: 24 Non-metastatic MIBC Discovery: AMVAC Validation: ddGC NGS of 287 genes
Discovery: 13/15 responsive patients had an ATM/FANCC/RB1 mutation compared to 0 nonresponders. Validation: 7/11 responders had an ATM/FANCC/RB1 mutation compard to 2/13 nonresponders (p = 0.03). Improved PFS (p = 0.0085) and OS (p = 0.007) in discovery cohort.
34 DDR panel
Patients with at least one mutation in a DDR gene have improved survival
Teo et al.
2017 100
Locally advanced or metastatic
bladder cancer Platinum-containing NGS of 341 genes
47/100 (47%) of patients had at least one DDR gene mutation correlating with improved PFS (9.3 vs. 6.0 months, p = 0.007) and overall survival (23.7 vs. 13.0 months, p = 0.006).
55 DDR panel
Patients with at least one mutation in a DDR gene have improved survival
Galsky et al.
2018 28 Metastatic MIBC GC + ipilimumab WES
All 10 patients with predicted deleterious mutations in any of the 54 DDR genes had partial or complete response (p = 0.03) and trended to have better PFS and OS but did not reach statistical significance. Apoptosis
26
Table 1.1 continued.
Pathway Biomarker Broad suggestion Author Evaluated patients (n) Patient
population Chemotherapy Technology Main findings
DNA-repair55 DDR panel have improved survival 2018 28 Metastatic MIBC GC + ipilimumab WES and OS but did not reach statistical significance.
Apoptosis
p53 p53 mutant tumors are more responsive Cote et al. 1997 56 Non-metastatic MIBC Adjuvant cisplatin-based chemotherapy IHC
Adjuvant chemotherapy improved survival patients with p53 mutations (p = 0.005) but not WT-p53 (p =0.94 )
p53 WT tumors are more responsive to NAC Kakehi et al. 1998 Group 1: 32 Group 2: 28 Group 1: Non-metastatic MIBC Group 2: Unresectable MIBC Cisplatin-based (Group 2 - No RC) IHC
In Group 1, p53-negtive (WT) tumors had improved DFS (P=0.009) but there was no association in Group 2
No association between p53 status and response
Qureshi et
al. 1999 83
Non-metastatic MIBC
Cisplatin-based
chemotherapy IHC No association between p53-staining and response p53 mutant tumors are
more responsive
Sarkis et al.
1995 90
Non-metastatic
MIBC MVAC IHC
p53 overexpression (mutant) increased overall survival (p = 0.001)
p53 mutant tumors are more responsive
Watanabe
et al. 2004 13
Non-metastatic
MIBC Cisplatin-based p53 sequencing
6/7 responders were p53-mutant compared to 2/6 nonresponders (p = 0.05)
No association between p53 status and response
Plimack et
al. 2014 39 MIBC AMVAC NGS of 287 genes No association between p53 status and response No association between
p53 status and response
Stadler et al.
2011 58 (37 MIBC)
21 NMIBC, 37
MIBC Adjuvant MVAC IHC No association between p53 expression and survival No association between
p53 status and response
Van Allen et
al. 2014 50 MIBC
GC (62%), ddMVAC (32%), GC+Sunitinib (4%), ddGC (2%) WES
P53 mutations were found in 56% of tumors but no association with response
No association between p53 status and response
Groenendijk
et al. 2016 32 MIBC Platinum-containing NGS of 178 genes
72% of patients had a p53 mutation but no association with response
No association between p53 status and response
Plimack et al. 2015 Discovery: 34 Validation: 24 Non-metastatic MIBC Discovery: AMVAC Validation: ddGC NGS of 287 genes
P53 mutations were found in 20% of tumors but no association with response
Bcl-2 Bcl-2 negative patients have improved survival
Cooke et al.
2000 25 MIBC Cisplatin IHC
BCL-2 negative tumors results in better median survival (72 months compared to 17 months, p = 0.03)
Drug transporters MCT1 (SLC16A3)
+ CD147 (BSG) MCT1 and CD147 positive tumors are resistant
Afonso et al. 2015
114 (31
platinum-treated) MIBC Platinum-treated IHC
Survival was worse in tumors positive for MCT1 and CD147 (p = 0.03) but this was also true for non-platinum-treated patients
CTR1 (SLC31A1) High CTR1 expression improves response
Kilari et al.
2016 44 MIBC Platinum-treated IHC
High CTR1 expression had improved response (p = 0.008)
MDR1 (ABCB1) High MDR1 has poorer response
Hoffman et
al. 2010 108 MIBC
Adjuvant
platinum-treated qRTPCR
High MDR1 expression (top 25%) resulted in poorer PFS (p = 0.0006)
Molecular subtypes
Basal subtype Basal tumors have better response*
Seiler et al.
2017 343 MIBC Cisplatin-based
Whole-genome mRNA transcription
Survival of patients with basal tumors improves the most with treatment
p53-like subtype p53-like tumors rarely benefit from NAC**
Choi et al.
2014 57 MIBC Platinum-containing
Whole-genome mRNA transcription
2/21 (10%) of the p53-like tumors responded to chemotherapy compared to 17/36 (47%, Fisher’s exact test, p = 0.003) of the basal or luminal tumors Gene expression signatures
27
Table 1.1 continued.
Pathway Biomarker Broad suggestion Author Evaluated patients (n) Patient
population Chemotherapy Technology Main findings
DNA-repairGene expression signatures
Takata 12-gene signature
A 14-gene signature predicts response
Takata et al.
2005 27 (18 in training set) MIBC MVAC
Whole-genome mRNA transcription
In the test set, the 14-gene signature correctly predicted 8/9 cases by actual response Takata et al.
2007 22 MIBC MVAC
Whole-genome mRNA transcription
19/22 samples were correctly predicted by actual response
Kato A 12-gene signature predicts response Kato et al. 2011 37 (18 in training set) Non-metastatic MIBC GC Whole-genome mRNA transcription
In the test set, the 12-gene signature correctly predicted 18/19 cases by actual response
COXEN
The COXEN algorithm predicts response
Williams et al. 2009
Group 1: 45
Group 2: 14 MIBC MVAC
Whole-genome mRNA transcription
In cohort 1, 3-year survival for predicted responders was 88% vs 33% in nonresponders (p = 0.002). In cohort 2, 3-year survival was 61% in predicted responders compared to 16% in nonresponders (p = 0.02)
APOBEC3 expression and mutational signature APOBEC
expression
Tumors with higher expression of APOBEC3A, 3D, and 3H are more responsive
Mullane et
al. 2016 73 Metastatic MIBC
Platinum-containing (some adjuvant)
Targeted expression by Nanostring technology
Of the APOBEC3 genes tested, higher expression of APOBEC3A, APOBEC3D, and APOBEC3H correlated with improved survival (p = 0.01, p = 0.02, and p = 0.004, respectively)
RTK biomarkers
ERBB2 ERBB2 mutant tumors are more responsive
Groenendijk
et al. 2016 71 MIBC Platinum-containing NGS of 178 genes
Complete responders (9/38) had a higher frequency of ERBB2 missense mutations compared to nonresponders (0/33) (p = 0.003) *Also supported by meeting abstracts (Sundi 2017, Metcalfe 2017, Choi 2017)
28 including ERCC1, ERCC2, and ERCC5. Loss of heterozygosity of ERCC5 improves the response to chemotherapy in ovarian cancer while osteosarcoma tumors with some polymorphisms in ERCC2, are more sensitive to cisplatin-based chemotherapy
(140,141). Without fully functioning NER, a tumor is less likely to repair cisplatin adducts and will be more susceptible to treatment. ERCC1 and ERCC2 have been associated with chemotherapy response in MIBC.
The most well established biomarker of response to cisplatin-based
chemotherapy in MIBC is ERCC2 mutations. In the original discovery study, Van Allen et al. performed whole-exome sequencing (WES) on 25 MIBC patients who responded to cisplatin-based NAC and 25 who did not (12). Response was defined as pathologic down staging at the point of cystectomy. This study aimed to find alterations that were more common in the responders or the nonresponders. A mutation in ERCC2 is the only gene to reach statistical significance (9 mutants in responders compared to 0 in nonresponders). The authors validated their finding by depleting and replacing bladder cancer cell lines with mutated ERRC2 and found them to be more sensitive to cisplatin. This finding was retrospectively validated in an independent cohort of 48 patients with MIBC treated with platinum-based neoadjuvant chemotherapy (11). As hypothesized, ERCC2 mutations were more commonly found in responders and than nonresponders (40% vs. 7%), and also associated with a statistical increase in overall survival. Finally, ERCC2 mutations trended to be higher in complete responders 6/38 (16%) responders compared to 2/33 (6%) nonresponders although this did not reach statistical
significance in this discovery dataset (p = 0.27) (142). However, the 2 nonresponders that had ERCC2 mutations were treated with carboplatin- rather than cisplatin-based