THE SEARCH FOR BIOMARKERS IN BLADDER CANCER

CDDP and IO WORLD

ALEJO RODRÍGUEZ-VIDA MD PhDConsultant Medical Oncologist

Associate ProfessorHospital del Mar, Barcelona

November 23rd 2018

DISCLOSURE OF INTEREST

� Advisory role:

◦ MSD, Pfizer, BMS, Astellas, Janssen, Clovis, Bayer, Roche

� Speaker role:

◦ Pfizer, MSD, Astellas, Sanofi Aventis, Janssen, Bayer, BMS,

Roche, AstraZeneca

� Research funding:

◦ Takeda, Pfizer, MSD

CHEMOTHERAPY (CDDP)Biomarkers

Molecular determinants of response to cisplatin-based neoadjuvant chemotherapy

Bellmunt J, Curr Opin Urol 2013, 23:466–471

• N=50

• DNA extracted from pre-NAC specimens.

• WES + correlation with pathologic response to NAC.

• Identification of genes selectively mutated in

responders vs non-responders

• Nucleotide excision repair genes among others

Mutated ERCC2 only gene significantly

enriched in the responder cohort

Van Allen EM, et al. Somatic ERCC2 mutations correlate with cisplatin sensitivity in muscle-invasive urothelial carcinoma. Cancer Discov. 2014 Oct;4(10):1140-53.

Van Allen EM, et al. Somatic ERCC2 mutations correlate with cisplatin sensitivity in muscle-invasive urothelial carcinoma. Cancer Discov. 2014 Oct;4(10):1140-53.

* p<0.05

Markers of pathologic response to neoadjuvantcisplatin-based chemotherapy

Liu D, JAMA Oncol 2016

ERCC2 mutations associated with pCR �

Bladder sparing approach for those with

somatic ERCC2 mutations planned to beprospectively investigated

Data from 2 phase II trials NACN=48, WES

pCR: 80% in ERCC2mut31% in ERCC2wt(p=0.01)

DNA repair gene variants associated with pCR

Plimack. GU-ASCO 2015. Eur Urol. 2015

Dec;68(6):959-67.

Patients with pCR had more

genomic alterations than those

with RD (p=0.024)ATM/RB1/FANCC alterations also

predictive of better OS

N=34 (NAC)

Markers of pathologic response to neoadjuvant cisplatin-based chemotherapy

Is the ERCC2 genomic test ready for prime time?

� ERCC2 is linked to pCR in only around 40% of patients across studies

� No “hot spot” mutation. Not all the mut are drivers

� pCR is seen in ERCC2 negative pts

� Tumor heterogeneity is an issue in all tumors

� We can only enrich the positive prediction

Takata R, et al. Predicting Response to Methotrexate,Vinblastine, Doxorubicin, and Cisplatin Neoadjuvant Chemotherapy for Bladder Cancers through Genome-Wide Gene Expression Profiling. Clin Cancer Res 2625 2005;11(7), 2005

RNA extracted from pre-NAC specimens. N=27.

cDNA microarray obtaining via RT-PCR consisting of 27,648 genes.

Identification of genes expressed differently between responders andnon-responder tumors.

We developed a geneexpression model topredict the pathologicalnode status from primarytumor tissue in 3independent cohorts ofpatients who wereclinically node negative

• Cutoff system identified patients with highRR (1.74) and low RR (0.70) of N+disease.

• Multivariate analysis: GEM predictorindependent of age, sex, pathological Tstage, and lymphovascular invasion

• Potential to select high-risk patients forNAC while sparing the rest from patientstoxic effects and delay to cystectomy

GEM-CIS

DD- MVAC

Muscle-Invasive Bladder

Cancer SWOG 8710 criteria -T2-T4a N0M0,

cisplatin eligible

Cystectomy

to assess

pT0 or pT1

pathology

Correlate with

COXEN prediction

pT0 /pT1

SWOG TRIAL: COXEN-directed neoadjuvant chemotherapyProspective validation of the COXEN biomarker to predict pT0 /pT1

COXEN

NCI-60 Cell

Line Panel (IC50)

575ASensitive

1 1

1

1 1

2 2

2

2 2

3 3

3

3 3

Bladder Cancer

patient samples

Human Bladder

Cancer Cell Lines

0.0

0.2

0.4

0.6

0.8

1.0

Kaplan-Meier Analysis

Survival Time (Months)

| | |

0 12 24 36 48 60 72 84 96

Predicted Responders (5)Predicted Nonresponders (9)

P = 0.0469

0.0

0.2

0.4

0.6

0.8

1.0

Kaplan-Meier Analysis

Survival Time (Months)

| | |

0 12 24 36 48 60 72 84 96

Predicted Responders (4)Predicted Nonresponders (10)

P = 0.0303

MVAC (N=16)

GC (N=14)

Pro

po

tio

n S

urv

ivin

gP

rop

oti

on

Su

rviv

ing

Gene Expression Model

COXEN Model Predicting response to chemotherapyTheodorescu et al. Clin Can Res 2007;4407 13(15):4407

Cisplatin, Gemcitabine, Methotrexate, Doxorubicin, Vinblastine

R

Eur Urol. 2017 Oct;72(4):544-554.

• Whole transcriptome profiling

• Pre-NAC TURb samples

• N= 343 MIBC

• Classified according to four published molecular subtyping methods.

•

•

•

Outcome after NAC varies by molecular subtype

Eur Urol. 2017 Oct;72(4):544--554.

Patients with claudin-low

tumors had the worst

prognosis irrespective of

treatment strategy,

suggesting also that

these patients derived

little or no benefit from

NAC

Luminal immune-infiltrated

tumors did significantly worse

than those with luminal non-

infiltrated tumors. Patients with

luminal-infiltrated tumors

appear to have poor prognosis

with and without NAC

Luminal tumors had the best

prognosis, irrespective of the

treatment strategy, implying

that these patients may not

need to receive NAC.

Patients with basal tumorsappear to derive the mostbenefit from NAC

IMMUNOTHERAPYBiomarkers

Summary of FDA-Approved and Investigational PD-L1 Assays in Urothelial Carcinoma*

Ab clone/epitope

Cell type scored

Scoring method

FDA status for urothelial carcinoma

PD-L1 thresholds under evaluation

1Bellmunt, J et al. N Engl J Med. 2017; 2Loriot Y, et al. Poster presentation at ESMO 2016. Abstract 83P; 3Ventana. Roche receives FDA Approval for novel PD-L1 biomarker assay [press release]. May 18, 2016; 4Sharma P, et al. Lancet Oncol. 2017; 5Powles T, et al. Poster presentation at ASCO GU 2017. Abstract 286; 6Patel M, et al. Poster presentation at ESMO 2017. Abstract 777PD.

Nivolumab4

28-8

TCs

% of PD-L1 expressing TCs

NA

≥1% ≥5%

Durvalumab5

SP263

TCs or ICs

% of PD-L1 expressing TCs or ICs

NA

≥25%

Atezolizumab2,3

SP142

ICs

% of PD-L1 expressing ICs

Complementary

IC2/3 (≥5%), IC1 (≥1% but <5%),

IC0 (<1%)

Pembrolizumab1

22C3

TCs and ICs

CPS: % of PD-L1 positive TCs and ICs relative to

the total number of tumor cells

NA

≥1%≥10%

Avelumab6

73-10

TCs

% of PD-L1expressing TCs

NA

≥5%

* No head-to-head studies have been conducted and direct comparisons cannot be made between these studies.

� Powles, et al. Nature. 2014. Phase I Atezolizumab

� Rosenberg, et al. Lancet. 2016. Phase II Atezolizumab

� Powles, et al. Lancet 2018. Phase III Atezolizumab

� Balar, et al. Lancet. 2017. Phase II Atezolizumab

� Massard, et al. J Clin Oncol. 2016. Phase I Durvalumab

� Sharma, et al. Lancet Oncol. 2016. Phase I/II Nivolumab

� Sharma, et al. Lancet Oncol. 2017. Phase I/II Nivolumab

� Balar, et al. Lancet Oncol. 2017. Phase II Pembrolizumab

� Bellmunt, et al. N Engl J Med. 2017. Phase III Pembrolizumab

PD-L1 Expression as a Predictor of Checkpoint Blockade Sensitivity in UC

6/9 studies reported positive association with PD-L1 staining

Opposite results in the cis-ineligible 1st line single arm trials

Vuky J, et al. J Clin Oncol 36, 2018 (suppl; abstr 4524)

Balar AV, et al. J Clin Oncol 36, 2018 (suppl; abstr 4523)

https://www.fda.gov/Drugs/InformationOnDrugs/ApprovedDrugs/ucm612484.htm

Novel Biomarkers: Beyond PD1

Early data suggests the following may enrich for response to PD1 pathway

inhibition:

� Higher mutational load

� TCGA Molecular Subtypes (Luminal II vs basal III)

� CD8 infiltration

� Immune related gene expression signatures (Nanostring)

� Peripheral expansion of certain TCR clones

16.625.4

21.7

15.1

22.7

30.939.1

24.2

59.1

41.830.4

60.6

0%

25%

50%

75%

100%

Cluster 1 (Luminal1) n=66

Cluster 2 (Luminal2) n=55

Cluster 3 (Basal1) n=23

Cluster 4 (Basal2) n=33

Complete Responsea

PartialResponse

StableDisease

ProgressiveDisease

50

25

0

75

100

Pe

rce

nta

ge

Luminal 2(Cluster 2)

n=55

Basal 1(Cluster 3)

n=23

Basal 2(Cluster 4)

n=33

Luminal 1(Cluster 1)

n=66

8.7

Galsky et al. LBA 31. ESMO 2016

But in TCGA:

- Luminal II have high immune markers expression

- Basal subtypes have the strongest immune expression

phenotype (T eff cell markers)

TCGA Subgroups (mRNA)

Nivolumab phase II CheckMate 275 study

Rosenberg et al. Lancet 2016; 387: 1909–20

Atezolizumab 2nd line

Phase II Trial

Biomarkers beyond PD-L1: Mutation load is associated with OS and RR with ICI

Atezolizumab 1st line Unfit Phase II Trial

Balar AV, et al. Lancet. 2016 Dec 7.

Nivolumab 2nd line Phase II Trial

Galsky et al. LBA 31. ESMO 2016

Atezolizumab Phase III Trial

Powles, et al. GU ASCO 2018

Teo, et al. J Clin Oncol. 2018 Jun 10;36(17):1685-1694.

Biomarkers beyond PD-L1: Alterations in DDR genes as marker of benefit with ICI

Atezolizumab Phase III Trial

Powles, et al. GU ASCO 2018

ORR: DDR+ vs wt: 67.9% vs 18.8% (p <0.001)

Nivolumab or atezolizumab

retrospective analysis

N=60, WES

Biomarkers beyond PD-L1: Alterations in DDR genes as marker of benefit with ICI

PURE-01 Pembrolizumab

Neoadjuvant Phase II Trial

Necchi et al. ASCO 2018

TGF-β: A potential new resistance mechanismand therapeutic target?

The impact of checkpoint inhibition on patient outcome in mUC isdictated by three core biological pathways:

(i)Pre-existing T-cell immunity

(ii)TMB, which is positively associated with outcome,(iii) Absence of TGFβ expression, which is associated with lack

of response and reduced survival

The enrichment of the fibroblast TGFβ -responsesignature in non- responding immune-excluded tumours,combined with preclinical models showing that co-inhibition of TGFβ and PD-L1 converted tumours from anexcluded to an inflamed phenotype, support a model inwhich TGFβ signalling may counteract anti-tumourimmunity by restricting the movement of T-cells in theTME.

Future Directions: Microbiota as

Biomarkers

MICROBIOME CULTUROMICS

16S RNA SEQUENCING

TRANSCRIPTOMICS

SHOTGUN METAGENOMICS

Ig-SEQ

Cell. 2017 Feb 9;168(4):707-723.

Analysis of 113 fecal samples of patients withMM treated with anti PD-1:

• The gut microbiota of responders had a greater

diversity

• Responders had increased abundance of fecal

Clostridiales (specifically Ruminoccocaceae

family)

• No association between oral microbiome and

response to therapy

Take Home Messages� Several predictive biomarkers to chemotherapy mainly in the neoadjuvant setting to predict

response rate

� Most validated: ERCC2 mutations, DDR alterations, gene expression profiles and molecularsubtypes

� Even a more preliminary situation regarding immunotherapy biomarkers

� Controversial role of PD-L1 expression

� Promising: TMB, DDR alterations, molecular subtypes, TGF-β, microbiota,…

Currently, no biomarkers have been translated into daily clinical

practice in bladder cancer

arodriguezvida@hospitaldelmar.catThank You

Acknowledgement: Joaquim Bellmunt