Transcript SMP Poster for WIN conference

DELIVERY OF HIGH QUALITY MOLECULAR TESTS FOR 9,000 CANCER PATIENTS
ROUTINELY CONSENTED FOR USE OF MOLECULAR AND CLINICAL DATA
Turtiainen
1
T,
Hair
2
J,
3
D,
1
S,
1
LK ,
1
M,
2
K,
1
J,
3
F,
1
R,
Harrison
Johnson
Jones
Jones
Oien
Peach
Rae
Sharpe
1
1
1
1
Shaw E , Telaprolu S , Tuff A , Johnson P.W.M.
1 Cancer Research UK, London, 2 Southern General Hospital, Glasgow, 3 Western General Hospital, Edinburgh
Cancer Research UK is conducting a two year pilot
study demonstrating routine molecular
characterisation of tumours for 9,000 cancer
patients and collection and storage of their
molecular and clinical data for research
Background
Molecular analysis of tumours can be used to identify patients
predicted to benefit from targeted therapies. The Cancer Research
UK Stratified Medicine Programme is piloting plans to apply testing
broadly across the UK healthcare system, linking molecular
phenotype to clinical outcomes enabling further research and patient
cohort identification.
The CR-UK Stratified Medicine Programme has been developed in
partnership with AstraZeneca, Pfizer, the UK Department of Health
and academic researchers.
Year One has demonstrated the acquisition of
tumour tissue for molecular phenotyping and
extraction of clinical data from 24 hospitals and 3
laboratories
Recruitment is on target. In year two we will focus
on improving turnaround times.
6000
5000
4945
4000
4000
Target for samples sent
The English NHS is revising the commissioning of molecular
pathology and we are feeding into the Scottish and Welsh system
based on our findings.
3000
2896
2694
Patients consented
2000
Results returned
Datasets sent to ECRIC
1000
0
Patient accrual has exceeded expectations, with 4945 patients
agreeing to participate by end of August 2012 (2808 from England,
1540 from Scotland and 597 from Wales). The laboratories are
streamlining their processes in order to achieve consistent
turnaround times within clinically relevant timescales from sample
collection to return of results (25 working days).
Phase One is demonstrating a model of routine
genetic testing while collecting and storing clinical
data and genetic data for research use
Researchers
Partners
NHS
Research
Research Database
(ECRIC)
The Programme is creating a database of the
genetic abnormalities detected and this information
will be publically accessible in January 2013
Anonymised
Data
Compiled
Data
Service Delivery
9,000
Cancer Patients
Leeds
Manch
Edinb
Glasgow
Camb
Cardiff
RMH
Samples
Genetic
Results
Cardiff
Birming
ham
Patients are consented for use of surplus diagnostic formalin-fixed
paraffin-embedded tumour tissue and blood samples for research.
Samples are tested for mutations from a panel of genes of potential
clinical interest. Results of molecular analysis are sent back to the
hospital (clinical hubs) and made available to clinical teams involved
in patient care.
Different testing technologies are used across the three laboratories
but all tests are technically validated and are subjected to external
quality assurance monitoring and targets for completion within
clinically relevant timescales.
A preliminary analysis of mutation rates is presented for 787 patients.
The percentage of abnormalities is displayed for the panel tests for
lung (302 patients), colorectal (126 patients), melanoma (56 patients),
breast (183 patients; 140 only for BRAF), ovarian (43 patients; 34 only
rates
higher
for for
theBRAF).
larger
for BRAF) Failure
and prostate
(77are
patients;
47 only
Prostate
carcinoma
Lung carcinoma
Ovarian
carcinoma
Malignant
melanoma
The Programme has high consent rates (average 96%) suggesting
that generic consent for research is acceptable to patients.
It is important to have ‘champions’ at each site who can drive and/or
organise local processes. In some cases an operational lead with
the ability to influence others is key. Support from high-level
surgical/oncology is also crucial.
Building on local systems can increase local buy in
Full and partial failure rates are higher for multiple
exon gene screens such PTEN and TP53
Full and partial failure rates by tumour type and gene test
The Programme model allowed sites to use local consent forms and
clinical data capture systems. This has enabled hospitals and labs
to adopt the model quickly and has demonstrated how it could work
based on different models of use (tertiary referral and academic
centres, network of district general hospitals).
The infrastructure can support recruitment for
clinical trials
% Ovary
15.0%
% Melanoma
% Lung
Breast
carcinoma
Identifying potentially eligible patients early in their clinical pathway
has increased recruitment rates in some sites.
* For sequencing based tests, abnormality includes accepted pathogenic mutations as well as
functionally uncharacterised sequence variants, but excludes known polymorphisms in the gene
of interest.
20.0%
10.0%
Scope of test
Codons 12, 13, 61 and 146
Codons 599-601
Codons 12, 13, 61
Exons 9 and 20
Exons 4-9
Exons 9 and 20
Exons 4-9
Exons 1-9
Microsatellite analysis
Codons 599-601
Exons 1-9
FISH ** for detection of gene rearrangement
Microsatellite analysis
Codons 599-601
Exons 18-21
Codons 12, 13, 61 and 146
FISH** for detection of gene rearrangement
Codons 599-601
Exons 4-9
Exons 1-9
Microsatellite analysis
Exons 9 and 20
Codons 599-601
Codons 599-601
Exons 11, 13 and 17
Codons 12, 13 and 61
Exons 9 and 20
The hospitals have demonstrated the potential to
collect routine clinical data for large-scale research
Accrual to sample acquisition programmes relies
on buy in from local research nurses, oncologists
and pathologists
% Prostate
Colorectal
carcinoma
Data can be recorded and flow for UK wide studies against
published standards and the use of these standards significantly
improves data quality. The data will be made available to UK based
researchers in academia and industry in an anonymised format.
The centrally developed research database will be used to share
findings via a secure online portal in real time. It is delivered in
partnership with Oracle and the University of Oxford.
25.0%
Gene of
interest
KRAS
BRAF
NRAS
PIK3CA
TP53
PIK3CA
TP53
PTEN mutation
PTEN LOH*
BRAF
PTEN mutation
TMPRSS2-ERG
PTEN LOH*
BRAF
EGFR
KRAS
EML4-ALK
BRAF
TP53
PTEN mutation
PTEN LOH*
PIK3CA
BRAF
BRAF
KIT
NRAS
PIK3CA
Lessons learned?
genes such as KIT, PTEN and P53
30.0%
Tumour
A pragmatic approach has been taken to resolve differences in
routine clinical data collection in England, Wales and Scotland. The
largest population is in England so we have used the Cancer
Outcomes and Services Dataset (COSD) as our standard. This
means that for Wales and Scotland we have where possible
mapped at data items or provided proxy values. There are some
issues around releasing demographic data to the central repository,
particularly from Scotland but we have discussed this with the
Caldicott Guardians and applied the appropriate Information
Governance (IG) rules. This will limit some linkage opportunities so
if necessary that will happen before data transfer.
The Programme consent model builds on local systems.
Patient clinical data are recorded in the existing hospital IT systems,
then standardised extracts (diagnostic including molecular
information, treatment and clinical outcome) are transferred via a
secure data transfer mechanism to a central data repository. This is
hosted within the Eastern Cancer Registration and Information
Centre (ECRIC), in Cambridge.
A panel of clinically actionable and research
genes is being tested in each tumour type
How we have tackled these and resolved issues?
Generic consent is feasible and acceptable to
patients
RMH
Genetic Technology Hubs
The devolved administrations of the United Kingdom of Great
Britain and Northern Ireland have responsibility for managing their
own healthcare systems. This means that there are different
information standards applied in England, Wales and Scotland. As
Cancer Research UK is a UK wide organisation we have had to
take this into consideration when designing our dataset and flows.
All hospitals have provided diagnostic clinical data to the
Programme. The main difficulties have been in obtaining pathology
data in the required format. All hubs have migrated to electronic
transmission of request and report of molecular tests using XML
messaging via a secure FTP server.
Clinical Hubs
Birm
UK wide data challenges and cross border issues
3429
Samples sent for testing
Phase One is a two year feasibility study to demonstrate the
collection and molecular analysis of consented sections of surplus
diagnostic tumour tissue and blood samples from 9,000 patients
across 24 hospitals within the Experimental Cancer Medicine Centre
network and three NHS genetic laboratories.
The practical implementation of the model has
identified potential barriers and solutions needed to
embed this model nationally
% Colorectal
Roche and Bristol-Myers Squibb are working with the Programme to
run trials in the UK that rely on the identification of cohorts of
eligible patients whose tumours contain certain molecular
abnormalities.
5.0%
% Breast
0.0%
The combined full and partial failure rates for BRAF, EML4-ALK,
KRAS and PIKC3A are under 5%. Higher rates are seen for KIT,
PTEN mutation and TP53 and the causes of the high failure rates
for KIT are being investigated. More than half (54%) of all
failures are for part of the gene test and therefore a clinically
meaningful result may still be obtained. Of the 2827 samples
tested,1.2% have failed all gene tests.
Funding is acknowledged from Cancer Research UK, AstraZeneca and Pfizer and recent contributions from Roche, Bristol Myers Squibb and Oracle.
Oncologists receive the molecular data and can discuss potential
trials with their patients if clinically appropriate.
There is a need to agree reporting terminology for
molecular pathology
Consistency is required in the nomenclature and reporting of
molecular results. We are establishing agreement across the
Programme for reporting previously described polymorphisms and
mutations in the genes.
We also intend to use neutral language for reporting sequence
changes of uncertain clinical or functional significance.