Precision medicine in oncology aims to provide targeted cancer treatments based on a patient's individual tumor characteristics. The presentation discusses precision oncology initiatives including NCI-MATCH clinical trials which assign cancer therapies based on a tumor's molecular abnormalities rather than location. It outlines plans to expand genomically-based cancer trials, understand and overcome treatment resistance through molecular analysis, and establish a national cancer database integrating genomic and clinical data to accelerate cancer research. Cloud computing platforms are being developed to provide researchers access to large cancer genomic and clinical datasets. The goal is to advance precision cancer treatment by incorporating individual patient genetics and biomarkers into therapeutic decision making.

1. Precision Medicine in
Oncology – Cancer Informatics
October 20th, 2015
Warren Kibbe, PhD
NCI Center for Biomedical Informatics

2. 2
1. Precision Medicine
2. Pre-clinical models
3. TCGA / TCIA
4. Genomic Data Commons
5. Cloud Pilots
Slides are from many sources, but special thanks to
Drs. Harold Varmus, Doug Lowy, Jim Doroshow, Lou Staudt

3. PhotobyF.Collins
President Obama Announces the Precision Medicine Initiative
The East Room, January 30, 2015

4. 4
TOWARDS PRECISION MEDICINE
(IoM REPORT, NOVEMBER 2011)

5. 5
Definition of Precision Oncology
 Interventions to prevent, diagnose, or treat cancer, based
on a molecular and/or mechanistic understanding of the
causes, pathogenesis, and/or pathology of the disease.
Where the individual characteristics of the patient are
sufficiently distinct, interventions can be concentrated on
those who will benefit, sparing expense and side effects
for those who will not.
Modified by D. Lowy, M.D., from IoM’s Toward Precision Medicine report, 2011

6. 6
Understanding Cancer
 Precision medicine will lead to fundamental
understanding of the complex interplay between
genetics, epigenetics, nutrition, environment and clinical
presentation and direct effective, evidence-based
prevention and treatment.

7. 7
What is Cancer?
 Cancer is a disease where cells ‘lose’ the normal controls that enable
them to participate as productive, stable members of tissues, organs
and organ systems. The process of developing cancer is usually
viewed as having three phases
 Initiation, where some of the normal control processes define the cell
type, inter-cellular communication with other cells, and normal
responses to cell-cell signaling are altered
 Proliferation, where cancerous cells ‘take over’ normal processes and
begin to proliferate
 Metastasis, where the cancerous cell leave their normal location and
invade other tissues

8. 8
Cancer Statistics
In 2015 there will be an estimated
1,700,000 new cancer cases
and
600,000 cancer deaths
- American Cancer Society 2015
Cancer remains the second most common cause of death in the U.S.
- Centers for Disease Control and Prevention 2015

9. 9
Survivorship is on the rise
Today there are 14 million Americans alive with a history of cancer.
It is estimated that by 2024, the population of cancer survivors will
increase to almost 19 million: 9.3 million males and 9.6 million females
- American Cancer Society 2015

10. 10
The hope
 We need to do what the HIV/AIDS research community did in the last
30 years, turn a certain death sentence into a managed, chronic
disease.
 Like the HIV/AIDS story for the course of the HIV infection, we believe
that cancer has multiple pathways that are altered during the course of
the disease, and that targeted therapies hitting multiple involved
signaling pathways, metabolic pathways, receptors can block and
prevent disease progression
 Unlike HIV, the course of cancer is more individualized and involves
different molecular actors depending on the cancer and the individual

11. 11
The frustration
We already know how to remove >50% of the cancer burden in the U.S.
Consistent nutrition, exercise & sleep – estimate(10-30%)
Tobacco control (smoking cessation) – ~25%
HPV and HCV immunization – ~20%
Behavior change is hard. Impacting the behavior of a broad population is
harder.

12. 12
Precision Medicine is not new: An early example
First disease for which molecular defect identified
Single substitution at position 6 of ß-globin chain
Abnormal Hb polymerization upon deoxygenation
“I believe medicine is just now entering into a new era when progress will be much more rapid than
before, when scientists will have discovered the molecular basis of diseases, and will have discovered
why molecules of certain drugs are effective in treatment, and others are not effective.”
Linus Pauling 1952 2015: still no good treatment

13. 13
The NCI Mission
& the PMI for
Oncology
• NCI Mission: To help people live
longer, healthier lives by supporting
research to reduce the incidence of
cancer and to improve the outlook for
patients who develop cancer
• Precision Medicine Initiative for
Oncology: Significantly expand our
efforts to improve cancer treatment
through genomics

14. 14
What Problems are We Trying to Solve?
Precision Medicine
Initiative in Oncology
• For most of its 70-year history,
systemic cancer treatment has relied
on therapies that are marginally more
toxic to malignant cells than to normal
tissues
• Molecular molecules that predict
benefit, response, or resistance in the
clinic have been lacking for many
cancers

15. 15
Proposed Solution to these Problems
Use genomics and other high
throughput technologies including
imaging to identify, create
predictive signatures, and
target molecular vulnerabilities
of individual cancers
Precision Medicine
Initiative in Oncology

16. 16
Precision Oncology in Practice
Nature Rev. Clin. Oncol. 11:649-662 (2014)

17. 17
Precision Oncology
Trials Launched
2014:
MPACT
Lung MAP
ALCHEMIST
Exceptional Responders
2015:
NCI-MATCH
ALK Inhibitor
MET Inhibitor
NCI-MATCH: Features
[Molecular Analysis for Therapy Choice]
•Foundational treatment/discovery
trial; assigns therapy based on
molecular abnormalities, not site of
tumor origin for patients without
available standard therapy
• Regulatory umbrella for phase II
drugs/studies from > 20 companies;
single agents or combinations
•Available nationwide (2400 sites)

18. NCI MATCH
•Conduct across 2400 NCI-supported sites
•Pay for on-study and at progression biopsies
•Initial estimate: screen 3000 patients to complete
20 phase II trials
1CR, PR, SD, and PD as defined by RECIST
2Stable disease is assessed relative to tumor status at re-initiation of study agent
3Rebiopsy; if additional mutations, offer new targeted therapy
,2

19. 19
NCI-MATCH: Initial Ten Studies
Agents and targets below grey line are pending final regulatory review; economies of
scale—larger number of agents/genes, fewer overall patients to screen
Agent(s) Molecular Target(s)
Estimated
Prevalence
Crizotinib ALK Rearrangement (non-lung adenocarcinoma) 4%
Crizotinib ROS1 Translocations (non-lung adenocarcinoma) 5%
Dabrafenib and Trametinib BRAF V600E or V600K Mutations (non-melanoma) 7%
Trametinib BRAF Fusions, or Non-V600E, Non-V600K BRAF Mutations (non-
melanoma)
2.8%
Afatinib EGFR Activating Mutations (non-lung adenoca) 1 – 4%
Afatinib HER2 Activating Mutations (non-lung adenoca) 2 – 5%
AZD9291 EGFR T790M Mutations and Rare EGFR Activating Mutations (non-
lung adenocarcinoma)
1 – 2%
TDM1 HER2 Amplification (non breast cancer) 5%
VS6063 NF2 Loss 2%
Sunitnib cKIT Mutations (non GIST) 4%
≈ 35%

20. 20
MATCH Assay: Workflow for 10-12 Day Turnaround
Tissue Fixation
Path Review
Nucleic Acid
Extraction
Library/Template Prep
Sequencing , QC Checks
Clinical
Laboratory aMOI
Verification
Biopsy Received at Quality Control Center
1 DAY
1 DAY
1 DAY
1 DAY
3 DAYS
10-12 days
Tumor content >70%
Centralized Data Analysis
DNA/RNA yields >20 ng
Library yield >20 pM
Test fragments
Total read
Reads per BC
Coverage
NTC, Positive, Negative
Controls
aMOIs Identified
Rules Engine
Treatment Selection
3-5 DAYS

21. 21
The Components of the
Precision Medicine
Initiative for Oncology
• Dramatically expand the NCI-MATCH umbrella
to include new trials, new agents, new genes,
and new drug combinations
• Understand and overcome resistance to therapy
through molecular analysis and development of
new cancer models
• Increase genomics-based preclinical studies,
especially in the area of immunotherapy,
through the creation of patient-derived pre-
clinical models and non-invasive tumor profiling
• Establish the first national cancer database
integrating genomic information with clinical
response and outcome: to accelerate
understanding of cancer and improve its
treatment

22. 22
PMI-O: Expanding
Genomically-Based Cancer
Trials (FY16-FY20)
• Accelerate Launch of NCI-Pediatric MATCH
• Broaden the NCI-MATCH Umbrella:
 Expand/add new Phase II trials to explore novel
clinical signals—mutation/disease context
 Add new agents for new trials, and add new
genes to panel based on evolving evidence
 Add combination targeted agent studies
 Perform Whole Exome Sequencing, RNAseq,
and proteomic studies on quality-controlled
biopsy specimens—extent of research based on
resource availability
 Add broader range of hematologic malignancies
• Perform randomized Phase II studies or hand-off to
NCTN where appropriate signals observed
• Apply genomics resources to define new predictive
markers in novel immunotherapy trials
• Expand approach to ‘exceptional responders’: focus
on mechanisms of response/resistance in pilot

23. 23
PMI-O: Understanding and
overcoming resistance to
therapy
(FY16-FY20)
• Create a repository of molecularly
analyzed samples of resistant disease
• Expand the use of tumor profiling
methods such as circulating tumor cells
(CTCs) and fragments of tumor DNA in
blood to understand and monitor disease
progression
• Develop new cancer models to identify
the heterogeneity of resistance
mechanisms
• Use preclinical modeling to determine the
effectiveness of new combinations of
novel molecularly targeted investigational
agents

24. 24
We need sophisticated
computational models to
understand patient
response, methods of
resistance, and to integrate
pre-clinical model data
Drivers for High Performance
Computing and
Computational Modeling

25. 25
Develop a Cancer
Knowledge System.
Establish a national
database that integrates
genomic information with
clinical response and
outcomes as a resource.
PMI-O: Informatics Goal

26. 26
Develop molecular,
imaging, pathology, and
clinical signatures that
predict therapeutic
response, outcomes, and
tumor resistance
PMI-O: Informatics Goal

27. 27
Build multi-scale,
predictive computational
biology models for
understanding cancer
biology and informing
therapy. Develop detailed
cancer pathway models to
create targeted
combination therapies in
cancer. This approach has
transformed HIV therapy
and has the potential to do
the same in cancer
PMI-O: Informatics Goal

28. 28
The Cancer Genomic Data
Commons (GDC) is an existing
effort to standardize and
simplify submission of genomic
data to NCI and follow the
principles of FAIR -Findable,
Accessible, Interoperable,
Reusable. The GDC is part of
the NIH Big Data to Knowledge
(BD2K) initiative and an
example of the NIH Data
Commons
Genomic Data Commons

29. NCI Cancer Genomic Data Commons (GDC)

30. 30
Genomic Data Commons (GDC) – Rationale
 TCGA and many other NCI funded cancer genomics projects each
currently have their own Data Coordinating Centers (DCCs)
 BAM data and results stored in many different repositories; confusing
to users, inefficient, barrier to research
 GDC will be a single repository for all NCI cancer genomics data
 Will include new, upcoming NCI cancer genomics efforts
 Store all data including BAMs
 Harmonize the data as appropriate
 Realignment to newest human genome standard
 Recall all variants using a standard calling method
 Define data sharing standards and common data elements
 Will be the authoritative reference data set
 Will need to scale to 200+ petabytes

31. 31
Genomic Data Commons (GDC)
 First step towards development of a knowledge system for cancer
 Foundation for a genomic precision medicine platform
 Consolidate all genomic and clinical data from:
 TCGA, TARGET, CGCI, Genomic NCTN trials, future projects
 Project initiated Spring of 2014
 Contract awarded to University of Chicago
 PI: Dr. Robert Grossman
 Go live date: Mid 2016
 Not a commercial cloud
 Data will be freely available for download subject to data access
requirements

32. 32
Integration with Imaging - TCIA
 The Cancer Imaging Archive has imaging data and some imaging
series and cover a number of TCGA disease types
 Breast invasive carcinoma (BRCA)
 Glioblastoma (GBM), lower grade glioma (LGG)
 Head&Neck squamous cell carcinoma (HNSC)
 Kidney renal clear cell carcinoma (KIRC)
 Ovarian serous cystadenocarcinoma (OV)
 Integration through standardized RESTful APIs
 Driven by the need to have better computational models

33. The NCI Cancer Genomics
Cloud Pilots
Understanding how to meet the research
community’s need to analyze large-scale cancer
genomic and clinical data

34. Slide courtesy of Deniz Kural, Seven Bridges Geno

35. 35
NCI GDC and the Cloud Pilots
 Working together to build common APIs
 Working with the Global Alliance for Genomics and Health (GA4GH) to
define the next generation of secure, flexible, meaningful,
interoperable, lightweight interfaces
 Competing on the implementation, collaborating on the interface
 Aligned with BD2K and serving as a part of the NIH Commons and
working toward shared goals of FAIR (Findable, Accessible,
Interoperable, Reusable)
 Exploring and defining sustainable precision medicine information
infrastructure

36. 36
Information problem(s) we intend to solve with the Precision
Medicine Initiative for Oncology
 Establish a sustainable infrastructure for cancer genomic
data – through the GDC
 Provide a data integration platform to allow multiple data
types, multi-scalar data, temporal data from cancer models
and patients
 Under evaluation, but it is likely to include the GDC, TCIA,
Cloud Pilots, tools from the ITCR program, and activities
underway at the Global Alliance for Genomics and Health
 Support precision medicine-focused clinical research

37. 37
NCI Precision Medicine Informatics Activities
 As we receive additional funding for Precision Medicine,
we plan to:
 Expand the GDC to handle additional data types
 Include the learning from the Cloud Pilots into the GDC
 Scale the GDC from 10PB to hundreds of petabytes
 Include imaging by interoperating between the GDC and the
Quantitative Imaging Network TCIA repository
 Expand clinical trials tooling from NCI-MATCH to NCI-MATCH Plus
 Strengthen the ITCR grant program to explicitly include precision
medicine-relevant proposals

38. 38
Project Schedule and Deliverables
Selection
Design/
Build I
Design/Build II Evaluation
6 Months
Initial Design and
Development
9 Months
Completion of Design,
Development and
Implementation
9 Months
Provide cloud to
researchers
NCI evaluations
Community evaluations
Jan
14
Sep
14
Apr
15
Jan
16
Sep
16

39. 39
Three Cancer Genomics Cloud Pilots
• PI: Gad Getz
• Google Cloud
• Firehose in the cloud
• http://firecloud.org
Broad Institute
• PI: Ilya Shmulevich
• Google Cloud
• Interactive visualization and analysis
• http://cgc.systemsbiology.net/
Institute for
Systems
Biology
• PI: Deniz Kural
• Amazon Web Services
• > 30 public pipelines
• http://www.cancergenomicscloud.org
Seven Bridges
Genomics

40. FireCloud is modeled on Firehose,
the cancer genome analysis
platform built by the Getz lab at the
Broad Institute, which supports
both small groups and major
projects (e.g. TCGA, GTEx).
FireCloud significantly expands on
Firehose’s capabilities. Firehose is
used by both production managers
for large-scale analysis and analysts
for interactive analysis, curation
and manual review of data for
publication.
Free trial workspaces
are available
for all new users!
Firehose re-born in the cloud
FireCloud is a collaboration of the Broad
Institute, University of California at
Santa Cruz and University of California
at Berkeley.
This project has been funded by the
National Cancer Institute and
National Institutes of Health.

41. 41
Pre-loaded FireCloud workspaces
FireCloud will be populated with pre-loaded workspaces, which, when cloned,
will allow users to replicate analyses from curated and published works
 A sampling of pre-loaded workspaces:
 Curated TCGA Tumor Type Analysis Working Group workspaces
 TCGA GDAC Analysis Working Group workspaces
 TCGA PanCanAtlas analysis workspace
 Tutorial workspaces containing paired tumor and normal cell lines
 Benchmarking data to enable users to test developing tools and methods
 Synthetic BAMs for testing contamination
 Synthetic BAMs for testing mutation calling
A sampling of pipelines:
TCGA Production Analysis
PCAWG Pipeline
TCGA GDAC Pipeline

42. ISB Cancer Genomics Cloud @ isb-cgc.org
TCGA Data in the Cloud
 Available Now§: over 400 GB of curated data
in open-access BigQuery tables
 Before the end of 2015: over 1 PB of
DNAseq, RNAseq, and SNP6 controlled-
access* data in Google Cloud Storage
Interactive Exploration
 Interactively explore all metadata and open-
access data
 Customize, save, and share data visualizations
with colleagues
 Define, export, and share custom cohorts – based
on:
 clinical information,
 molecular characteristics, and/or
 data type availability
 3rd party tools: IGV and NG-CHM
§ email info@isb-cgc.org for details
*requires dbGaP authorization,
with authentication via NIH Login

43. Programmatic Access backed by the Power of Google
Easy Access to the Data
 ISB-CGC Cloud Endpoint APIs expose a REST interface for search
and retrieval, with responses returned as JSON objects
 Google APIs provide direct access to BigQuery & Cloud Storage
 RStudio and IPython tutorial examples avaiable on github
 Use RStudio or IPython Notebooks to script your own analyses
Your own Google Cloud Platform Project
 Start with $300 to spend on compute & storage – request additional
credits as needed
 Invite your students, post-docs & collaborators to join your project
 Upload your own data into Cloud Storage and BigQuery
 Run your own algorithms and pipelines on Google VMs
 Use and share methods via
 Docker containers
 Galaxy Workflows
 Common Workflow Language

44. 44

45. Finding and exploring TCGA data
Data browser enables complex
queries of 100+ metadata
properties (can also be queried
programmatically)
“Find data from cases with
(RNAseq data from BOTH
Primary tumor and Adjacent
Normal) AND (WGS on Blood
Derived Normal samples).”

46. Release and
Evaluation

47. 47
Cloud Pilots: Coming your way!
Broad
• Version 1.0 – 1/20/2016
• Version 1.1 – 4/2/2016
ISB
• Pre-release – 11/15/2015 (open-access data)
• Version 1.0 – 12/20/2016
• Version 2.0 – 3/20/2016
SBG
• Early access – 11/15/2015
• Version 1 – 12/28/2015
• Version 2 – 3/28/2016

48. 48
NCI-Sponsored Evaluation Activities
Independent Testing
& Evaluation
• Highly structured test of functionality, security, load capability
• Assessment of the key strengths, weaknesses, issues, and
risks
Administrative Supplements
• Active NCI grants: R01, R21, U01, U24, P30
• Proposals due October 18
DREAM Challenge
• Launching early 2016
• Use TCGA RNA-Seq data to identify somatic mutations
Hands-On Workshop
• May 24, 2016 @ NCI Shady Grove
• Half day for each Cloud Pilot
NCI Intramural Evaluation
• Dedicated resources to support intramural investigators’
research

49. CGC Pilot Team Principal Investigators
• Gad Getz, Ph.D - Broad Institute - http://firecloud.org
• Ilya Shmulevich, Ph.D - ISB - http://cgc.systemsbiology.net/
• Deniz Kural, Ph.D - Seven Bridges – http://www.cancergenomicscloud.org
NCI Project Officer & CORs
• Anthony Kerlavage, Ph.D –Project Officer
• Juli Klemm, Ph.D – COR, Broad Institute
• Tanja Davidsen, Ph.D – COR, Institute for Systems Biology
• Ishwar Chandramouliswaran, MS, MBA – COR, Seven Bridges Genomics
GDC Principal Investigator
• Robert Grossman, Ph.D - University of Chicago
Cancer Genomics Project Teams
NCI Leadership Team
• Doug Lowy, M.D.
• Lou Staudt, M.D., Ph.D.
• Stephen Chanock, M.D.
• George Komatsoulis, Ph.D.
• Warren Kibbe, Ph.D.
Center for Cancer Genomics Partners
• JC Zenklusen, Ph.D.
• Daniela Gerhard, Ph.D.
• Zhining Wang, Ph.D.
• Liming Yang, Ph.D.
• Martin Ferguson, Ph.D.

50. 50
National Strategic Computing Initiative
 Executive Order announced July 29, 2015
 Create a cohesive, multi-agency strategic vision and Federal investment strategy
in high-performance computing (HPC)
 Lead agencies
 DOE, DoD and NSF
 Deployment agencies
 NASA, NIH, DHS, and NOAA
 Participate in shaping future HPC systems to meet aims of respective missions and
support workforce development needs
 Implications for NCI
 Work cross agency with DOE and others to expand use of HPC to advance research
and clinical applications impacting cancer
50

51. 51
Possible NCI-DOE initiative aligned with NSCI
Three candidate pilot projects identified:
 Pre-clinical Model Development and Therapeutic Evaluation
(Doroshow)
 Improving Outcomes for RAS Related Cancers (McCormick)
 Information Integration for Evidence-based Cancer Precision Medicine
(Penberthy)
Collaboratively developing project plans with DOE computational
scientists
51

52. 52
Pilot Project 1: Pre-clinical Models
 Pre-clinical Model Development and Therapeutic
Evaluation
 Scientific lead: Dr. James Doroshow
 Key points:
 Rapid evaluation of large arrays of small compounds for
impact on cancer
 Deep understanding of cancer biology
 Development of in silico models of biology and predictive
models capable of evaluating therapeutic potential of
billions of compounds
52

53. 53
Pilot Project 2: RAS Related Cancers
 Improving Outcomes for RAS Related Cancers
 Scientific lead: Dr. Frank McCormick
 Key points:
 Mutated RAS is found in nearly one-third of
cancers, yet remains untargeted with known drugs
 Advanced multi-modality data integration is required for model
development
 Simulation and predictive models for RAS related molecular
species and key interactions
 Provide insight into potential drugs and assays
53

54. 54
Pilot Project 3: Evidence-based Precision Medicine
 Information Integration for Evidence-based Cancer
Precision Medicine
 Scientific lead: Dr. Lynne Penberthy
 Key points:
 Integrates population and citizen science into improving
understanding of cancer and patient response
 Gather key population-wide data on treatment,
response and outcomes
 Leverages existing SEER and tumor registry resources
 Novel avenues for patient consent, data sharing and
participation
54

55. 55
Take homes
 Genomic Data Sharing policy is an opportunity for standardized,
meaningful data exchange
 Open APIs and RESTful interfaces enable ‘purpose focused’ data
sharing
 Genomic Data Commons will be the repository for NCI-funded
genomic data
 New architectures including discoverable identifiers, baked-in
provenance, ad hoc levels of versioning and micro-publication (e.g.
DOIs) enable very different incentives for data sharing
 Computational models for prediction are critical to the future

56. 56
Questions?
Warren A. Kibbe
warren.kibbe@nih.gov
Thank you

57. www.cancer.gov www.cancer.gov/espanol

58. 58
Questions?
Warren A. Kibbe
warren.kibbe@nih.gov
Thank you