Joseph A. DiMasi, Ph.D.Director of Economic Analysis,
Tufts Center for the Study of Drug DevelopmentTufts University
Partnerships in Clinical Trials USInstitute for International Research
Boston, MA, October 7, 2016
The Changing Economics of New Drug Development: M&A and Other Risk-Sharing Trends
Agenda• New drug development economic factors that can impact M&A activity and
collaborative development
Clinical development and regulatory approval times
Trends in drug development activity
Clinical approval rates and phase transition rates
Total out-of-pocket and capitalized costs per approved compound
Post-approval cost estimates
Cost drivers
Declining rates of return
Analysis of competitive development in pharmacologic classes
• Empirical results on performance
Clinical Development and
Approval Phase Times
0
2
4
6
8
10
Ye
ars
Year of NDA Approval
Total Phase
Clinical Phase
Approval Phase
Points are 3-year moving averages
Mean U.S. Approval and Clinical Phases for U.S. New Drug Approvals, 1963-2015
Source: Tufts CSDD
Trends in Drug Development by
Therapeutic Class
Trends in New Drug and Biologics Approvals Within Four Largest Therapeutic Class
Source: Tufts CSDD
0% 10% 20% 30% 40% 50% 60% 70%
2010-15
2000-09
1990-99
1980-89
Share of US New Drug Approvals
Anti-infective* Antineoplastic Cardiovascular CNS
* Anti-infective excludes AIDS antivirals
Compounds in Development by Therapeutic Area: Oncology Leads
Source: EvaluatePharma (3 May 2012)
Oncology 31%
Systemic Intiinfectives 15%
Central Nervous System 14%
Cardiovascular 7%
Musculoskeletal 5%
Endocrine 5%
Gastrointestinal 4%
Respiratory 4%
Dermatology 3%
Sensory Organs 3%
Genito-Urinary 3%
Blood 3%
Other 4%
R&D Expenditure Trends and
Costs per Approval
New Drug and Biologics Approvals and R&D Spending
0
15
30
45
60
0
15
30
45
60
R&
D E
xp
en
ditu
res
(Billio
ns
of 2
014$)
NM
E/N
BE
Ap
pro
va
ls
Sources: Tufts CSDD, PhRMA, 2015
R&D expenditures are adjusted for inflation; curve is 3-year moving average for NMEs
R&D Expenditures
New Compound Approvals
Drug Development Technical
Risks
Clinical Phase Transition Probabilities and Overall
Clinical Approval Success Rate*
*Therapeutic new molecular entities and new therapeutically significant biologic
entities first tested in humans, 1995-2007
59.5%
35.5%
62.0%
90.4%
11.8%
Phase I-II Phase II-III Phase III-NDA/BLA Sub NDA/BLA Sub-NDA/BLA App
Phase I - NDA/BLAApp
Tran
siti
on
Pro
bab
ility
Source: DiMasi et al., Journal of Health Economics 2016;47:20-33.
12.0%
23.0%21.5%
11.8%
1970s-early 1980sapprovals
1980s-early 1990sapprovals
1990s- early 2000sapprovals
2000s- early 2010sapprovals
Clin
ical
Ap
pro
val S
ucc
ess
Rat
e
Sources: 1970s-early 1980s, Hansen, 1979; 1980s-early 1990s, DiMasi et al., J Health Econ 1991;
1990s-mid 2000s, DiMasi et al., J Health Econ 2003; DiMasi et al., J Health Econ 2016
New Drug Development Risks Have Increased Markedly
R&D Cost per Approved Drug
Out-of-Pocket and Capitalized Cost per Approved New
Compound
430
965
1,395
1,098
1,460
2,558
Pre-human Clinical Total
Mill
ion
s o
f 2
01
3 $
Out-of-Pocket CapitalizedSource: DiMasi et al., Journal of Health Economics 2016;47:20-33.
Pre-approval, Post-approval and Total Lifecycle Cost
per Approved New Compound
1,861
2,870
1,395
2,558
466312
Out-of-Pocket Capitalized
Mill
ion
s o
f 2
01
3 $
Total Pre-approval Post-approval
Source: DiMasi et al., Journal of Health Economics 2016;47:20-33.
Growth in Capitalized R&D Costs
per Approved New Compound
Sources: 1970s, Hansen (1979); 1980s, DiMasi et al. (1991); 1990s-early 2000s, DiMasi et
al. (2003); 2000s-early 2010s, DiMasi et al. (2016)
109 70179278
135
413436608
1,0441,098
1,460
2,558
Pre-human Clinical Total
Mill
ion
s o
f 2
01
3 $
1970s 1980s 1990s-early 2000s 2000s-early 2010s
Cost Drivers: Change in Capitalized Cost per
Approved Compound by Factor (direct cash outlays)*
* Factor impact on current study cost relative to prior study cost ($1,044 million in 2013 dollars)
Factor Category FactorPercentage Change
in Cost
Cash Outlays Out-of-Pocket Clinical Phase Costs 82.5%
Pre-human/Clinical Cost Ratio 1.6%
Overall Out-of-Pocket Costs 85.5%
Source: DiMasi et al., Journal of Health Economics 2016;47:20-33.
Cost Drivers: Change in Capitalized Cost per
Approved Compound by Factor (development risk)*
* Factor impact on current study cost relative to prior study cost ($1,044 million in 2013 dollars)
Factor Category
FactorPercentage
Change in Cost
RiskClinical Approval Success Rate with Prior Study Distribution of Failures
57.3%
Distribution of Failures with Prior Study Clinical Approval Success Rate
-6.0%
Overall Risk Profile: Clinical Approval Success Rate plus Distribution of Failures
47.3%
Source: DiMasi et al., Journal of Health Economics 2016;47:20-33.
Cost Drivers: Change in Capitalized Cost per
Approved Compound by Factor (time and cost of capital)*
* Factor impact on current study cost relative to prior study cost ($1,044 million in 2013 dollars)
Factor Category FactorPercentage Change in
Cost
Time Pre-human Phase -4.9%
Clinical Phase 0.2%
Regulatory Review -3.0%
Overall Development Timeline -5.6%
Cost of Capital Discount Rate -3.1%
Source: DiMasi et al., Journal of Health Economics 2016;47:20-33.
• Increased clinical trial complexity: more procedures per patient
(additional data gathered)
• Patient recruitment and retention
• Life sciences sector inflation (cost of inputs used in development)
• Testing against comparator drugs to meet market (payer)
demands for comparative effectiveness
• Higher failure rates and more indications pursued
• Increased regulatory burden for some classes of compounds
Some Conjectures and Evidence Underlying
Growth in Clinical Costs
Procedures per Protocol
Median Number (2005)
1999-2005 Annual Growth Rate
Phase I Unique Procedures 40 6.1%
Total Procedures* 217 9.5%
Phase II Unique Procedures 35 5.8%
Total Procedures 195 12.1%
Phase III Unique Procedures 33 5.5%
Total Procedures 132 6.1%
Phase IV Unique Procedures 32 9.1%
Total Procedures 99 11.0%
* Defined as the number of unique procedures multiplied by their frequency during the duration of the study
Source: Getz et al., Amer J Ther 2008;15:450-457
Data Points Collected per Patient for a Typical Phase III Protocol
492,000
929,000
2002 2012
Nu
mb
er
of
Dat
a P
oin
ts
Source: Getz and Kaitin, Re-Engineering Clinical Trials 2015:ch 1; Medidata Solutions
• Unexpected cardiovascular risks found for diabetes drug rosiglitazone
(Avandia® )
• FDA issued guidance in Dec 2008 (Guidance for Industry: Diabetes
Mellitus – Evaluating Cardiovascular Risk in New Antidiabetic Therapies
to Treat Type 2 Diabetes)
• Number of randomized patients and patient-years increased more than
2.5 and 4.0 fold before and after guidance, respectively, for diabetes
drugs approved 2005-2010 (Viereck and Boudes, Contemporary
Clinical Trials, 2011;32(3):324-332)
• Clinical costs (particularly for phase III) higher for diabetes drugs in cost
sample
Regulatory Change (Diabetes Drugs) and
Impact on Drug Development Costs
Clinical and Clinical plus Approval Phase Times for
Diabetes Drugs by Period of Approval
(pre- and post-FDA guidance)
4.7
6.36.7
7.9
Clinical Clinical plus Approval
Year
s
2000-2008 (n=8) 2009-2014 (n=9)
Source: DiMasi et al., Journal of Health Economics 2016;47:20-33.
Rates of Return to New Drug
Development
Mean Present Discounted Value of Lifetime After-Tax Net Returns
for New Drugs and Biologics (millions of 2005 dollars)
Launch Period All Drugs and Biologics Small Molecule Drugs Biologics
1995-1999 $725
2005-2009 -$111 -$186 $93
Source: Berndt et al., Health Affairs 2015;34:245-252
Development Within Pharmacologic Classes: Imitation or Racing?
Share of Later-in-Class Drugs with Patent Filed or
Development Phase Initiated Prior to First-in-Class Approval
100%96%
92%
78%
50%
22%
WW Patent US Patent Phase I Phase II Phase III NDA/BLA Filing
Pe
rce
nt
of
Late
r-in
-Cla
ss D
rugs
Source: DiMasi and Chakravarthy, in press, Clin Pharmacol Ther 2016, doi: 10.1002/cpt.502
First-in-class drugs approved from 2005 to 2011; later-in-class drugs approved from 2005 to 2015
Time from Patent Filed or Development Phase Initiated for
Later-in-Class Drugs to First-in-Class Approval
7.36.8
4.74.1
3.1
5.8 5.9
4.23.5
2.5
WW patent US patent Phase I Phase II Phase III
Year
s b
efo
re 1
st-i
n-C
lass
Ap
pro
val
Mean Median
Source: DiMasi and Chakravarthy, in press, Clin Pharmacol Ther 2016, doi: 10.1002/cpt.502
First-in-class drugs approved from 2005 to 2011; later-in-class drugs approved from 2005 to 2015
Priority (n=41)
Standard (n=38)
51.9%
48.1%
First-in-class drugs approved from 1998 to 2011; 43 pharmacologic classes; later-in-class drugs approved from 1998 to 2015
Is First-in-Class the Best-in-Class?:
FDA Therapeutic Significance Ratings for Later-in-Class Drugs
Source: DiMasi and Chakravarthy, in press, Clin Pharmacol Ther 2016 , doi: 10.1002/cpt.502
New Business Models and Emerging R&D Strategies to Deal with the Growing
Challenges of Drug Innovation
• Industry-wide or company-specific shocks
• Economies of scale and scope
• Access to new technologies
• Expansion to foreign markets and other stages of the drug
distribution chain
• Increased market power and size
Economic Theories Advanced to Explain M&A
and Alliance Activity
Source: Grabowski and Kyle, The Economics of the Biopharmaceutical Industry, ch 18, 2012
Alliance (voluntary, enduring, remain separate businesses)
• Most prevalent in Oncology and CNS, next was M&E and CV, then OTC and Generics and GI in 2007-2010
• Best when in-house resources/expertise are limited, speed of entry is important
Buy
• M&A deal values increased by 30%
• Best when there is lack of in-house expertise, but need to acquire share of market lead, and speed of entry important
Build
• Partnerships with academia gaining in popularity
• Best when need to expand or re-establish in-house expertise/resource in core business
Build, Buy or Ally!
Source: SCRIP 100, Elsevier Business Insight & IntelligenceTM
Strategic Alliances…Therapeutic Area (TA)
is Key Factor!
Source: Windover, Burrill and Company (Nat. Biotech., June 2008, 26(6): 602)
• By definition, it is association with others in activity of common interest
• Reasons make common sense and commercial cents
• Maintains flexible business model
• Provides intellectual fertilization
• Responds to changing resource and expertise needs
• Shifts fixed costs to variable costs
• Allows for sharper focus on internal expertise
Partnering for Long-Term Success
• Relationship between academia and industry is naturally complementary, historically lending itself to formation of joint research enterprises
• There are 10m researchers worldwide, 15,000 scientific articles published everyday, 7.8m active patents (Innovation Days 2012)
• In 2010, 25 top R&D universities in US earned $1.5bn in licensing income, and launched 260 start-ups (Burrill Report June 2012)
• Half of all biotechnology firms founded by university scientists, many maintain academic affiliations (Stuart et al 2007)
• Survey of academic medical centers shows over half of researchers already conduct drug and device trials (Zinner & Campbell 2009)
Allure of Academic Alliances
Trends in Industry-Academic
Partnership Models
Commonly used
Increasingly popular in the present
Emerging
Unrestricted grants Fee-for-service
Risk-sharing Competitive grants
Corporate venture capital funds Academic drug discovery centers
Transforming R&D Strategy: Innovation Partnerships
Academic-Industry: (‘upstream/TR focus’) e.g., PFEs-CTIs; J&J Innovation Centers, LLY-Lilly Innovation Fellowship Awards; ADDCs
Multi-company Consortia and PPPs: e.g., Enlight Biosciences (6 cos); ADNI (22 cos); Sage BioNetwork; Mass. Neuroscience Consortium (7 cos); IMI; TransCelerate (17 cos)
Outsourcing providers: (virtual) LLY-Chorus; (functional) LLY-Covance/Advion; BMS-Accenture; AZN-(API)
Patient groups: e.g., VRTX-CFF; Breast Cancer Alliance; Lupus Foundation; Michael J. Fox Foundation
Global partners: e.g., BMS -”String of Pearls”
VC partnerships
Open innovation and Crowdsourcing: e.g., LLY-Open Innovation Drug Discovery; Transparency Life Sciences
Performance Analyses for M&As and
other Risk-Sharing Arrangements
• Under-researched area
• Evidence mixed, but a number of studies suggest lower post-merger
R&D spending, number of projects or patents, and productivity
• The studies, however, examine short-term impacts (two or three years
post-merger) and outcomes are heterogeneous
• Some evidence that alliances and mergers can be complementary
(i.e., alliances pre-merger can help predict which potential mergers
will be successful)
• Some evidence that at an industry level, too much M&A activity can
reduce industry innovation levels (fewer independent sources of
innovation)
Empirical Evidence on the Effects of M&A and
Alliance Activity on Pharmaceutical Innovation
0%
10%
20%
30%
40%
50%
Single Firm Licenced Co-developed M&A
Sh
are
of A
pp
rova
ls
Trends in Collaborative and Risk-
Sharing Arrangements
2000-2003 2004-2007 2008-2011
Source: DiMasi et al., Ther Innov Reg Sci, 2014;48(3):482-487
74.9
17.0
91.5
66.0
15.8
82.0
70.2
16.4
86.5
0
20
40
60
80
100
Clinical Phase* Approval Phase** Total Phase***
Mo
nth
sAverage Clinical, Approval, and Total
Phase Times (2000-2011) and Shared Risk
Multi-firm clin dev Single firm clin dev All
* p=0.0131; ** p=0.4147; ***p=0.0116
Source: DiMasi et al., Ther Innov Reg Sci, 2014;48(3):482-487
Multi-firm=licensed, co-developed, M&A
• New drug development is lengthy, risky, and costly
• Regulatory approval success rates for new drugs have declined
significantly
• R&D costs have continued to increase in an increasingly cost-
conscious market; rates of return have declined
• Biopharmaceutical innovation is highly competitive, with
development within pharmacologic classes occurring largely
contemporaneously
• Biopharmaceutical firms have increasingly engaged in
collaborative discovery and development to share risks and increase innovation
Summary
Websitehttp://csdd.tufts.edu
Tufts Center for the Study
of Drug DevelopmentTufts University, Boston, Massachusetts, USA
Joseph A. DiMasi, Ph.D.
Director of Economic Analysis