某医药公司产品组合优化内部资料

June 4-8, 2004
10
From Pipelines to Portfolios
Research Objectives
Understand how risk and expected reward change as the nature of a pipeline changes.
Today: How Phase III variance and expected reward change as the degree of correlation changes
• e.g., AIDS
We use the terms “pipeline or project” to represent the NPD structure for a given indication We use “approach” to represent a specific way (e.g., a compound, a biological molecule) of drug development tailored for a given indication.
Some theoretical insights An empirical demonstration -- GlaxoSmithKline
June 4-8, 2004
2
Terms Used in Presentation
Indications
Also known as disease Each indication is considered as one business opportunity
– Indication level – Area level
Develop optimal portfolio, given a certain managerial objective
Today: An application for Glaxo-Smith-Kline (GKS)’s Phase III subportfolio
Drug Discovery (CEDDs)
Cardiovascul ar and
Urogenital
Metabolic and Viral Diseases
Microbial, Musculoskelet
al and Proliferative
Diseases
Neurology and
Gastrointesti nal
June 4-8, 2004
3
Optimizing New Drug Pipeline
An Example
Bristol-Myers Squibb wants to develop a preventive HIV vaccine ...
Pre-clinical trial
NPD in Pharmaceutical Industry
Psychiatry
Respirator y and
Inflammati on
Indications Supported
In Each CEDD
9 (Stroke) and 3 (ED)
5 (Obesity) and 7 (HIV)
7 (Sepsis), 2 (osteoporosis) and 9 (cancers)
From Pipelines To Portfolios
Min Ding Jehoshua Eliashberg
Pennsylvania State University University of Pennsylvania
Choice Symposium June 4-8, 2004
Outline
Succeeded:
Correctly recognized the importance of crossproject effects in structuring portfolio Developed the necessary organizational structure (e.g., 3 tier structure at GSK)
– a survey of new drug portfolios in 1450 firms on 12/31/2002
Number of Indications Targeted by a Firm December 31, 2002ຫໍສະໝຸດ Number of Firms
500 450 400 350 300 250 200 150 100
Clinical Phase I
Clinical Phase II
Clinical
1
launch
Phase III
Alternative approaches: subunit vaccine, recombinant vector vaccine,
peptide vaccine, virus-like particle vaccine, plasmid DNA vaccine, “naked DNA”vaccine, wholeinactivated virus vaccine, live-attenuated virus, etc.
Failed:
Risk has not received sufficient attention, ENPV is the most commonly used metric. No rigorous portfolio evaluation process and optimization model exist.
11 (Alzheimer's) and 5 (Irritable bowel synd.)
9 (Depression)
4 (Asthma) and 3
(Rheumatoid arthritis)
June 4-8, 2004
9
From Pipelines to Portfolios
Where does the state of the art practice succeed and fail?
June 4-8, 2004
5
Optimizing New Drug Pipeline
One Key Insight - A Pattern of Underspending? (Ding and Eliashberg, Management Science, 2002)
Therapeutical Class
E[1(n1)] [1 (1 p1)n1 ]E[0 (s1 0)] n1c1
V[1 (n1 )] (1 p1 ) n1 [1 (1 p1 ) n1 ]E[0 (s1 0)]2
L1 (n1 ) 1 (1 p1 ) n1
We solve for the optimal number of approaches to be supported at each phase in development
Question: How many alternative approaches should the firm pursue at each NPD stage?
June 4-8, 2004
4
Optimizing New Drug Pipeline
Model Output For each phase in development, we derived the expected reward, variance, and likelihood of having a successful drug candidate. For example, for Phase III, they are, respectively,
June 4-8, 2004
11
From Pipelines to Portfolios
Indication Level -- Scenario
552
12 10 3
18 17 7
11 11 4
Novartis 1 1 ?
20 19 7
Pfizer
WarnerLambert
BMS
111 11? 111
2 22 16 10 4 17 11 5
June 4-8, 2004
6
Firms Target Multiple Business Opportunities
Hypertension
130 3078 2036 655 6773
94 1484 5174
Lead firm
BMS Monsanto
Merck Eli Lilly
actual pipeline
21? 222 2 2 1 111
pipeline structure recommended by
our model
Therapeutical Classes
Also known as therapeutical areas, groups, or categories
• e.g., Anti-infective products
Each class has many different indications We use the term “portfolio” to represent the NPD structure for many different indications, within or across different therapeutical classes. A portfolio has many different pipelines (or projects).
50 0 0
20
40
60
80