Advancing Evidence - Based Indication Prioritization to

Navigate the Segmentation of the Disease Landscape

By Chris McKenna, General Manager, Discovery Science and Richard Harrison, Chief
Scientific Officer
Clarivate Analytics | Advancing Evidence 2

Authors
Chris McKenna
General Manager, Discovery Science

Chris started his career in 1989 in the field of

cheminformatics at Molecular Design Ltd and has
spent the last 25 years working in Pharma R&D
informatics, disease modeling, and multi-disciplinary
data science, analytics, and advisory services for
research intensive industries. At Clarivate Analytics he
manages the Life Science practice areas of Discovery &
Translational Science, Clinical & Regulatory, and
Portfolio & Licensing for Pharmaceutical and Biotech
clients and the Scientific & Academic Research
services which performs research program assessment
for government and academic research funders. Chris
actively participates in non-profit, pre-competitive
alliances, including Pistoia Alliance, and is a
Participating Member on the Pistoia Alliance Board, as
well as pharmaceutical consortia run by Clarivate
Analytics (CBDD, Metaminer, and CMR International).

Dr. Richard Harrison

Chief Scientific Officer

He has over 30 years of experience in the life sciences

industry. His career has focused on all aspects of pre-
clinical drug discovery. During this time he has held
positions of increasing responsibility at Aventis,
Merck, DuPont and Wyeth Pharmaceuticals where he
delivered numerous clinical candidates and worked on
several marketed drugs. In addition, he served as a
founding scientist in two successful venture funded
computational and structure based drug design
companies. Dr. Harrison has authored over 40 peer
reviewed publications and book chapters in all areas of
drug discovery and has presented over 50 invited
lectures on these subjects. He earned his Ph.D. in
Biophysical Chemistry and Enzymology and currently
serves as an adjunct professor at several universities.
 3

The complexity in both the target (molecular)

and patient (phenotypic) dimensions of drug
discovery and development has grown quickly
in the last two decades, presenting enormous
opportunity for pharmaceutical and biotech
companies.
Clarivate Analytics | Advancing Evidence
Landscape is made up of increasingly
segmented diseases
Innovation in genomic technologies over the past two
decades has ushered in an explosion of new drug targets for
drug discovery. And while new opportunities abound in this
larger target landscape, another important trend is
unfolding. The human phenotype dimension is rapidly
expanding to reveal a constantly evolving view of a more
diverse phenotypic landscape. This landscape is made up of
increasingly segmented diseases, with greater molecular
association to clinical symptoms. This diverse landscape is
more challenging to quantify due to lower incidence and
prevalence rates and smaller sample sizes compared with the
major disease classes of the past. A good example is blood
cancers. According to the NCI’s Surveillance, Epidemiology
and End Results Program, what was classified as a “disease of
the blood” 60 years ago, is today 90x more specific, classified
as over 40 different leukemia types and more than 50 unique
lymphoma types.
ICD Diagnosis Code Set meSH Disease Terms (clinical study type)
Figure 1
What was classified as a
“disease of the blood” 60 years
ago, is today 90x more specific,
classified as over 40 different leukemia
types and more than 50 unique
lymphoma types.
4
Advances made in clinical diagnostics
Contributing to this growing complexity in the disease
landscape are advances made in clinical diagnostics. Clinical
science has advanced to exploit this greater genomic and
molecular diagnostic capability, discovering and using
biomarkers to stratify patients in clinical trials more
effectively. In addition, as artificial intelligence and machine
learning techniques continue to advance, patient
stratification is being applied more often to better
understand the molecular basis of the increasingly narrower
patient segments.
Coding has expanded dramatically
As evidence, disease coding has expanded, while scientists
and clinicians have harnessed the molecular basis of that
clinical know-how in several ways:
 The disease terms within meSH, used to index clinical
study types from the biomedical literature, have
expanded to capture more specific disease classification
over the last three decades (Figure 1).
 In the last revision of The International Statistical
Classification of Diseases and Related Health Problems
(10th revision, 2nd ed.), or ICD-10, published by the
World Health Organization (WHO), codes for diagnosis of
disease increased nearly five times from those published
in ICD-9, implemented in 1979 (Figure 1).
 Investigational drugs entering Discovery Phase in Cortellis
Competitive Intelligence with a new indication not
previously recorded in Cortellis have increased recently
(Figure 2).
 Biomarkers are increasingly used in clinical trials to select
patients into studies from among those who are more
likely to respond to the drug under investigation.
Biomarkers are also used as surrogates for clinical
endpoints or toxicity effects (and patient stratification).
Use of biomarkers in clinical studies rose 15% between
2010 and 2015 alone.
Figure 2
Clarivate Analytics | Advancing Evidence
Risk of repeating “shots on goal”
research strategy
The take-home message is clear: The complexity in both the
target (molecular) and patient (phenotypic) dimensions of
drug discovery and development has grown quickly in the
last two decades, presenting enormous opportunity for
pharmaceutical and biotech companies. And while new
indications are making their way at an accelerated pace into
discovery programs, still only 7% of investigational drugs
tested in first human dose studies survive to market launch
(according to the 2015 CMR Pharmaceutical R&D Factbook).
While it is appealing to enter into this new expanding space
of target and indication segmentation, it is important not to
repeat the narrow view of increasing the “shots on goal”
research strategy of the 1990s, which drove R&D spending
higher and fed a declining return on drug R&D investment,
down from 10.1% in 2010 to 4.8% in 2013, with no end to
the decline in 2016 (Figure 3).
Figure 3 | Deloitte 2016
More effective and efficient methods are required to
evaluate this expanding indication space so more meaningful
indications can be selected, resulting in more candidates
likely to survive clinical development and generate
meaningful financial returns.
2.
1. Experimental models
Accepted standard for Fast, un-biased, large disease
Strength
evidence coverage
Subject to biological
hypothesis bias, not reliably Connection to human outcomes
Weakness
predictive of human disease, requires validation
small disease coverage
Applied more as validation, Systematic implementation for
Trending
not solely explanation exploration
5
Methods for indication prioritization
Through an analysis of the literature and qualitative
assessment of the techniques being used by pharmaceutical
and biotech R&D for indication prioritization, those decisions
appear to be relying on these three major methods:
1. The classical, experimental model whereby a compound
is tested and evidence is confirmed in increasingly
relevant models of human disease (differentially
expressed genes -> cellular pharmacologic -> animal
pharmacologic -> human outcomes). While this is still
the gold standard model from an evidence perspective,
the process is costly, slow, and the number of indications
that can be studied is small. Equally problematic, animal
models have not proven to be reliable models of human
disease, the fundamental biological hypotheses are
subject to bias, and the use of a variety of experimental
methods across the drug discovery process can yield
confounding outcomes for decision-making.
2. The computational science approach, whereby systems
biology methods used by bioinformatics analysts
evaluate both the association and mechanistic
relationship between biological pathways and disease.
While this approach is fast and can cover large biological
pathway space, the findings are reliant on our existing
knowledgebase and may be dependent on experimental
validation in either animal models or human trials.
3. The standard management consulting approach,
whereby a select few indications are profiled based on
the assumption that a drug has a role in a small number
of chosen indications, and those chosen indications are
then analyzed against certain development and
commercial criteria, evaluating the market sizing, the
clinical and/or regulatory risks. This approach is useful
for narrowing options and risk, typically applied later in
the R&D cycle; however, its small scale and lack of
molecular mechanistic knowledge limits its utility for
exploration tasks.
Systems Biology 3. R&D Feasibility and
(Computational) Commercial
Narrow options based on known
R&D and market constraints and/or
risks
No mechanistic or biological basis of
disease; relies on mechanism of
action assumptions
Combined with computational
approaches earlier in R&D cycle
Clarivate Analytics | Advancing Evidence
Multidisciplinary and pan-R&D process
for indication prioritization
At Clarivate Analytics, our staff of industry veterans and data
science experts, including the authors of this article, have
worked extensively in all phases of drug development, from
preclinical to commercial forecasting and market sizing
domains with both large pharma and smaller biotech.
Together we have developed an indication prioritization
methodology that brings together the computational and
evidence-based approaches of methods 2 and 3, employing
efficient, multi-disciplinary metrics that are critical to making
good indication prioritization decisions, and coupling them
more cost-effectively with the experimental models of
method 1.
This approach benefits from operating at a large scale, using
evidence-based, manually curated data across many
segments of life science, together with mathematical
algorithms to inform indication prioritization.
Previous attempts at conducting indication prioritization
have concentrated on using only a few methods and
disciplines. However, this approach integrates chemical,
biological, clinical, regulatory, and commercial measures into
a multidisciplinary and pan-R&D process that can operate at
scale, capable of screening compounds against thousands of
diseases simultaneously.
This methodology can operate both in exploratory mode
(using drug-to-target combinations together with biological
networks and their mechanistic relationship with disease)
and in a filtering mode (narrowing indications based on
clinical success rates, R&D predicted timelines, regulatory
pathways and approvals, incidence and prevalence of patient
segments, and predicted commercial outcomes based on
risk-adjusted, financial return metrics).
This approach integrates
chemical, biological, clinical,
regulatory, and commercial measures
into a multidisciplinary and pan-R&D
process that can operate at scale,
capable of screening compounds
against thousands of diseases
simultaneously.
6
Evidence-based approaches in target
and indication prioritization
For large pharma:
 Profiling compounds under investigation as well as re-
examining historical compound collections, quickly, at
scale. By using a consistent method to identify candidates
for new indications from large compound libraries,
scientists and portfolio teams quickly identify candidates
for further investigation.
For biotech companies:
 Starting with a single compound, the extensive biological
pathways and large collection of investigational drugs
with their biological properties can be used to explore
new indications faster and cheaper than experimental
models, and can then be quickly filtered based on
development feasibility and commercial metrics.
Both approaches produce prioritized indications across
thousands of diseases with detailed development,
partnering, and/or commercialization strategies.
Both approaches produce
prioritized indications across
thousands of diseases with detailed
development, partnering, and/or
commercialization strategies.
Clarivate Analytics is excited about the increasing
opportunities to exploit evidence-based approaches in target
and indication prioritization. To learn more about how our
products support target discovery, see our newly developed
target druggability tool.
Clarivate Analytics | Advancing Evidence 7

Who we are
Clarivate Analytics accelerates the pace of innovation by providing trusted insights and analytics
to customers around the world, enabling them to discover, protect and commercialize new ideas,
faster. Formerly the Intellectual Property and Science business of Thomson Reuters, we own and
operate a collection of leading subscription-based businesses focused on scientific and academic
research, patent analytics and regulatory standards, pharmaceutical and biotech intelligence,
trademark protection, domain brand protection and intellectual property management. Clarivate
Analytics is now an independent company with over 4,000 employees, operating in more than
100 countries and owns well‐known brands that include Web of Science, Cortellis, Derwent
Innovation, Derwent World Patents Index, CompuMark, MarkMonitor and Techstreet, among
others.

To learn more, visit:

clarivate.com

Contact us

North America
Philadelphia +1 888 537 5862
+1 215 278 8424
Latin America
Brazil +55 11 8370 9845
+1 215 823 5674

Europe, Middle East and Africa

+44 203 811 4048

Asia Pacific
Japan +81 3 4589 3100
+08 0 08888855

2017
© 2017 Clarivate Analytics clarivate.com