Feasibility Study of Genomic Biomarker Profiling for Patients

with Metastatic Colorectal Cancer

Bradley L. Smith1, Philip Breitfeld1, Jennifer Cubino1, Victor Weigman1, Donald P. Richards2, Ki Y. Chung3.1

1 Quintiles, Durham, NC;

2 Texas Oncology - Tyler, TX;
3 Cancer Center of the Carolinas, Spartanburg, SC

Abstract
Background: The adoption of Next-Generation Sequencing (NGS) platforms

AmpliSeq Cancer Hotspot Panel v2 assay, enriching for hotspots within 50 cancer-

therapies in other indications. 84.3% of patients had variant associated with open

and development of targeted oncology drugs have enabled matching of patients

related genes. Clinical annotation and reporting to the doctors was provided by

clinical trials. Of these 43 patients, 32 had multiple biomarkers with associated

and drugs. The authors undertook an observational, clinical study to explore the

N-of-One. Basic demographic and clinical information was collected but formal

trials. Overall, more than 100 mutations were identified including alterations in

feasibility and potential clinical benefits of an upfront approach to the genomic

disease monitoring and follow-up was not performed. Clinicians were asked to

KRAS, BRAF, EGFR, PIK3CA, GNAS, TP53, APC and other genes. The number

profiling of tumors from metastatic colorectal cancer (mCRC) patients. The study

report the impact of the genomic test report on patient recommendations.

of actionable mutations was not associated with progressor status. Doctors

sought to determine the number of drug targetable genomic changes, which

Results: The study enrolled and profiled 51 stage IV mCRC patients from July

recommended clinical trials following profiling and reporting of genomic alterations

occur within mCRC patients including a comparison of patients who progress early

2013 to October 2013 from 14 sites in the U.S.; one additional patient was enrolled

in 15 out of the 43 patients (35%) that had actionable mutations.

over the targeted number. Subjects were stratified by time to progression prior to

Conclusions: The outcome of this observational study demonstrates the

Methods: The study targeted enrollment of 50 mCRC patients within the U.S.

entering the study. The study population was evenly distributed across early

feasibility of rapid screening and reporting of NGS genomic results targeting

Oncology Network followed by collection of archival formalin-fixed paraffin

(<1 yr) and late progressors (>1 yr) with a median age of 62. Test turn-around time

actionable mutations in mCRC. The lack of an association between early and late

embedded (FFPE) samples and genomic testing. Sample collection and

averaged 15 days. 98% of the bases sequenced in the genomic analysis reached

progressors, suggests that a greater sample size will be required for future studies.

processing was performed at Quintiles Central Laboratories followed by testing and

the target coverage necessary to identify 5% variant frequency in the sample.

The reported impact on clinician recommendations indicates the value of the

bioinformatic analysis at the Quintiles EA Genomic Laboratory. Genomic profiling

Genomic variants associated with approved therapies in mCRC were observed

results to inform treatment and clinical trial decisions.

was performed on the Ion Torrent PGM following enrichment of tumor DNA via the

in 7.8% of patients while 64.7% of patients had variants associated with approved

versus late.

Background
Introduction
Cancer genomics is moving into practice driven by the increased molecular
complexity of cancer and drugs that target those genomic alterations.
Explosion of targeted agents in cancer; 22% of the pipeline agents currently in
pivotal trials are being developed in a biomarker-defined patient population.

Potential solution:
Remove barriers to patient participation in clinical trials: multiplex testing allows
for efficient use of scarce tumor samples and rapid testing of the sample ensures
patients are not delayed in receiving treatment.
Genomic profile patients to match study criteria prior to site startup.

Recent technical development of genomic platforms enables rapid and broad

genomic profiling.

Clinically annotate and report results to clinician.

Patient pre-profiling platform

Summary:
mCRC 50 patient trial with 50-gene NGS profiling and reporting to clinicians
performed in collaboration with U.S. Oncology Research (USOR) Network

The Challenge: efficient execution of programs targeting niche oncology

populations with specific genomic alterations.

Feasibility study to demonstrate pre-profiling operational platform

How do I cost-effectively develop a drug with an anticipated high screen failure

rate in a timely fashion?

Primary objective:
To determine the number of drug targetable genetic changes, which occur within
patients with mCRC. The number of clinically actionable mutations with FDAapproved drugs will be compared to the number without FDA- approved drugs
and to actionable mutations with associated clinical trials.
Secondary objectives:
To determine the feasibility of collecting tumor samples and having those patients
undergo tumor genomic analysis.
To investigate the number and cause of failed analyses after registration and
tissue submission.
To determine if physicians took into consideration regimens that were suggested
by the results of the sequencing analysis when deciding their patients next line
of therapies.

Operational results
Pilot study description

Ion Torrent PGM NGS Platform

Introduction
50 genes; Hot Spot coverage (not total open reading frames)
2800 COSMIC mutations
~26kbp of captured sequence

Study Characteristics
Total months of enrollment: Approx. 3 months (planned for 6 months)
Number of Sites: 14 (all within U.S. Oncology Research Network and the U.S.)
Sample type: FFPE archival slides or blocks
Total patients in Early (recurrence <1 year) Occurrence arm: 25 patients
Total patients in Late (recurrence after 1 year) Occurrence arm: 26 patients
Total patients diagnosis: 100% Stage IV Colorectal Cancer
Both recurrence and gender were distributed evenly across patient age groups
80% of enrolled patients were Caucasian

Performance
> 400X depth of coverage for 95% of the targeted regions
98.4% of samples in study were powered to identify alleles
at 5% frequency (i.e. callable bases)
10% of patients in this study had callable bases with
between 90-95% coverage, potentially due to biological
factors

Genomic test turn-around-time (TAT) summary

PGM - GENE LIST
ABL1

EZH2

JAK3

PTEN

AKT1

FBXW7

IDH2

PTPN11

ALK

FGFR1

KDR

RB1

APC

FGFR2

KIT

RET

ATM

FGFR3

KRAS

SMAD4

BRAF

FLT3

MET

SMARCB1

CDH1

GNA11

MLH1

SMO

CDKN2A

GNAS

MPL

SRC

CSF1R

GNAQ

NOTCH1

STK11

CTNNB1

HNF1A

NPM1

TP53

EGFR

HRAS

NRAS

VHL

ERBB2

IDH1

PDGFRA

ERBB4

JAK2

PIK3CA

TAT summary:
Min: 9 Days
Max: 21 Days
Avg: 15 Days

Five samples were excluded due to prospective decision to re-sequence samples

Genomic & clinical results

Results of Genomic Analysis & Clinical Annotation

Distribution of mutations in patient population

Influence of genomic data on clinician decision making

(3 categories of actionable genomic alterations)

Approximately 3% of oncology patients participate in clinical trials in the U.S. The much higher percentage of patients informed
of relevant trials in the preprofiling study suggests preprofiling may improve clinical trial participation. This may be due to:
1. Providing genomic data improves awareness and interest in trial options
2. Patients enrolled in pre-profiling were more likely to be eligible or interested in clinical trials than patients not enrolled in pre profiling

Overall 100 mutations were observed, encompassing all 3 categories of action-ability including mutations in
PIK3CA corresponding to open clinical studies.
Distribution of clinical recommendations and associated therapies or trials was similar across gender and age
Data provided by N-of-One

Number and distribution of mutations was similar between early and late progressors.
However, a few genes/mutations were primarily found in one group and not the other
(BRAF, CDKN2).

Summary
Test / profiling feasibility and performance
High quality genomic data and a clinical report was delivered
to doctors in a reasonable time (average 15 days from sample
submission to report) utilizing archival FFPE samples
Rapid patient enrollment indicates clinician and patient excitement
for this type of information
Profiling results
Overall 100 mutations were observed, encompassing all 3 categories
of action-ability
60 distinctly different actionable alterations were observed in 43 out of
the 51 total patients (84%)

*Physicians response was limited to a single category

**8/51 patient samples had no actionable mutations or clinical recommendations
***Physicians may have reported not influenced due to existing use of standard of care FDA approved drug consistent with
preprofiling report

Discussion
The most frequent actionable mutations had associated clinical
trials, followed by therapeutics approved in other indications and
therapeutics approved in CRC
The number of actionable mutations did not statistically correlate with
patient demographics or progression status
Progression status did not statistically correlate with specific
alterations although further investigation may be warranted
Study impact
Genomic pre-profiling and the genomic report impacted clinician
recommendations for available clinical trials in a significant percentage
of cases that reported actionable mutations (15/43; 35%)

Patient pre-profiling may rapidly identify qualified patients for

biomarker-driven oncology drug development
Pre-profiling may improve trial timelines by increasing the pool of
patients participating and screened for clinical trials
Implementation of pre-profiling will require collaboration across key
stakeholders including sponsors, CROs, clinicians and patients