`
`O P I N I O N
`
`Can the pharmaceutical industry
`reduce attrition rates?
`
`Ismail Kola and John Landis
`
`The pharmaceutical industry faces
`considerable challenges, both politically
`and fiscally. Politically, governments around
`the world are trying to contain costs and,
`as health care budgets constitute a very
`significant part of governmental spending,
`these costs are the subject of intense
`scrutiny. In the United States, drug costs
`are also the subject of intense political
`discourse. This article deals with the fiscal
`pressures that face the industry from the
`perspective of R&D. What impinges on
`productivity? How can we improve current
`reduced R&D productivity?
`
`The average life expectancy of humans has
`gone up from about 45 years of age at the start
`of the twentieth century to about 77 a century
`later. This is a consequence of a number of
`factors, including increased medical know-
`ledge, better technologies and surgical tech-
`niques, better health care, better public health
`and the discovery of drugs such as aspirins,
`antibiotics, the statins, and numerous other
`such innovative and crucial medicines from
`the pharmaceutical industry. However, the
`current challenges facing the pharmaceutical
`industry are unprecedented in its history.
`Perhaps most foremost among these are the
`industry’s lower revenue growth, poor stock
`performance, the lowest number of new
`chemical entities (NCE) approvals and the
`poor late-stage R&D pipelines prevalent
`throughout the industry.
`In 2002, overall top-line revenue growth
`in the pharmaceutical industry was just
`8% and improved only slightly in 2003 to
`
`approximately 9%. Similarly, in 2003 large
`pharma stock prices were among the worst
`performing sector on the New York Stock
`Exchange (NYSE), with an average apprecia-
`tion of 0.3%, compared with the general
`S&P500 market appreciation of 26%. At
`present the average price to earnings (P/E)
`ratio of large pharma stocks is trading at a
`discount to the entire market. By contrast,
`this sector has historically traded at a pre-
`mium to the rest of the market, mainly
`because of pipeline valuations.
`
`Depressing approval rates
`In 2002, the US FDA approvals of NCEs were
`lower than at any other time in the past
`decade, and a total of just 17 NCEs were
`approved; the situation improved marginally
`in 2003 to 21 approvals. Even if biologics and
`NCEs are considered together, the number of
`FDA approvals were at their lowest since
`1994. The situation is even bleaker when the
`number of innovative NCEs approved by
`regulatory authorities are factored into this
`performance. Prous Science1 reported that
`in the eleven-year period 1990–2000 inclusive,
`the year with the lowest number of NCEs
`approved with a novel mechanism of action
`was 2000. These data are further substantiated
`by the number of FDA priority reviews of
`NCEs (an indirect measure of innovativeness
`or addressing true unmet medical need), in
`which 2002 and 2003 showed lower numbers
`of such reviews than any two-year rolling
`period in the preceding ten years2.
`This lower rate of success in the past few
`years could be accounted for, in part at least,
`
`by a number of explanations: the industry is
`currently attacking diseases of great com-
`plexity; the entry bar for new drugs is higher
`because they are often competing with
`enhanced standard of care; and/or the regu-
`latory authorities are more demanding.
`Whatever the case, these features define the
`new playing field on which the industry has
`to compete to produce NCEs that are required
`to achieve necessary growth; an examination
`of the factors that impact R&D success is
`therefore crucial in terms of devising a
`strategy that can build a pipeline needed to
`sustain the business case for large pharma.
`
`Defining the business case
`A recent survey by Accenture3 defined the
`business case for large pharmaceutical com-
`panies in terms of NCEs required to remain a
`growth company on the basis of their current
`revenues and their desired percentage growth
`(TABLE 1). On the basis of this calculation,
`Pfizer, with pharmaceutical revenues in 2003
`of approximately US $45 billion, will need to
`generate approximately nine high-quality
`NCEs per annum. GlaxoSmithKline, with
`revenues in excess of £18 billion (~ US$ 32
`billion), will need to generate about six high-
`quality NCEs per annum, and Merck, with
`US $22.5 billion in revenues, will need
`approximately 4.5 NCEs. The next tier (in
`terms of revenues) would need to deliver
`between three and four NCEs per annum and
`even the smaller companies in the top ten
`would need to deliver approximately two
`NCEs per annum.
`
`Rates of attrition
`FIGURE 1 analyses success rates from first-in-
`man to registration during a ten-year period
`(1991–2000) for ten big pharma companies in
`the United States and Europe. The data
`indicate that the average success rate for all
`therapeutic areas is approximately 11%; or,
`put another way, in aggregate only one in nine
`compounds makes it through development
`and gets approved by the European and/or the
`US regulatory authorities. More interestingly,
`
`NATURE REVIEWS | DRUG DISCOVERY
`
`VOLUME 3 | AUGUST 2004 | 7 1 1
`
`002014
`
`AVENTIS EXHIBIT 2021
`Mylan v. Aventis, IPR2016-00712
`
`
`
`P E R S P E C T I V E S
`
`$45 billion
`$30 billion
`$20 billion
`$15 billion
`
`Table 1 | NCEs required to achieve specific real growth targets as a function of 2002 revenues*
`2002 sales‡
`Anticipated sales
`Annual real
`Sales gap for
`Estimated number
`from current
`growth target
`new products
`of NCEs required to fill
`products in 2012
`to fill in 2012
`gap (over ten years)
`$30 billion
`$43.5 billion
`75–90
`$20 billion
`$29 billion
`50–60
`$13.3 billion
`$19.3 billion
`33–40
`$10 billion
`$22 billion
`40–45
`$17 billion
`30–35
`$14.5 billion
`25–30
`$12 billion
`20–25
`$ 9.67 billion
`16.5–20
`
`5%
`5%
`5%
`8%
`6%
`5%
`4%
`5%
`$6.67 billion
`$10 billion
`*Adapted from REF. 3. ‡All figures in US $. NCE, New Chemical Entity.
`
`Year 2012 required
`NCE output
`
`9.5–11
`6.5–7.5
`4.3–5
`5.5–6.0
`4.0–4.5
`3.25–3.75
`2.5–3.0
`2.15–2.25
`
`59%. Approximately 62% of all compounds
`that enter Phase II trials undergo attrition,
`and again the highest rate of attrition at this
`phase is in the oncology field: more than 70%
`of oncology compounds fail in this phase. It is
`therefore crucial that the industry develop
`and embrace paradigms (such as obtaining
`proof of concept in man early in develop-
`ment) and methodologies to identify risk
`preclinically, and to couple this with experi-
`mental medicine procedures to interrogate
`such risks in man.
`
`Underlying causes of attrition
`An examination of the root causes of why
`compounds undergo attrition in the clinic is
`very instructive and helps in the identification
`of strategies and tactics to reduce these rates
`and thereby improve the efficiency of drug
`development. The data in FIG. 3 show the rea-
`son why compounds undergo attrition and
`how this has changed over time. In 1991,
`adverse pharmacokinetic and bioavailability
`results were the most significant cause of
`attrition, and accounted for ~40% of all attri-
`tion. By 2000, these factors had dramatically
`reduced as a cause of attrition in drug develop-
`ment, and contributed less than 10%. These
`data provide further compelling evidence that
`the industry can identify and remedy the
`causes of attrition. It might also, however, be
`that the solving of this problem has signifi-
`cantly shifted the temporal attrition profiles to
`later stages, because pharmacokinetic/bioavail-
`ability failures would have occurred in Phase I
`mainly and this might now result in com-
`pounds progressing to Phases II and III and
`failing there for other reasons.
`The major causes of attrition in the clinic
`in 2000 were lack of efficacy (accounting for
`approximately 30% of failures) and safety
`(toxicology and clinical safety accounting for a
`further approximately 30%). The lack of effi-
`cacy might be contributing more significantly
`
`the success rates vary considerably between the
`different therapeutic areas: cardiovascular, for
`instance, have a ~20% rate of success, whereas
`oncology and central nervous system (CNS)
`disorders have ~5% and ~8% success, respec-
`tively. Any R&D portfolio, therefore, would
`need to aggregate the percent success based on
`the weight of the various therapeutic areas to
`calculate how many first-in-man studies are
`needed to approximate the requisite business
`case for growth.
`The high rate of attrition in drug develop-
`ment and the need for efficiency, both in
`terms of real and opportunity costs, becomes
`even more compelling when one considers
`where most of the attrition occurs in the
`pipeline. In 2001, the costs of discovering and
`developing a drug were of the order of US
`$804 million4; current estimates are closer to
`about US $900 million; considerably more of
`these costs are incurred later in the pipeline,
`and the vast majority of attrition occurs in
`full clinical development (Phases IIb and III).
`
`FIGURE 2 illustrates the top 10 drug companies’
`success and failure rates from 1991 to 2000
`across different therapeutic areas.
`The failure rate of compounds even at the
`registration stage is 23%; that is, roughly one
`in four compounds fail after all the trials and
`the documentation for submission have been
`completed, thereby incurring the full dis-
`covery and development costs and the oppor-
`tunity costs, which, on average, could be as
`much as 12 years 10 months (the average time
`taken for the development of all the drugs
`that gained approval in 2002)5. In some thera-
`peutic areas, such as woman’s health, the failure
`rate is as high as 42%, and in oncology it is as
`high as 30%. Even the rate of failures in
`Phase III trials — by which stage significant
`amounts of the costs of discovering and
`developing a drug would have been incurred
`— is far too high: approximately 45% of all
`compounds that enter this phase of full devel-
`opment undergo attrition and in some thera-
`peutic areas, such as oncology, it is as high as
`
`11%
`
`20
`
`15
`
`10
`
`5
`
`0
`
`Percentage of success
`
`CNS Infectious
`Cardio-
`Arthriris
`Urology Women's
`Metabolic
`Oncology Opthal-
`disease
`vascular
`and pain
`health
`disease
`mology
`Figure 1 | Success rates from first-in-man to registration. The overall clinical success rate is 11%.
`However, if the analysis is carried out by therapeutic areas, big differences emerge. The data are from the
`ten biggest drug companies during 1991–2000. (The companies are AstraZeneca, Bristol-Myers Squibb,
`Eli Lilly, F. Hoffman-LaRoche, GlaxoWellcome, Johnson & Johnson, Novartis, Pfizer, Pharmacia,
`Schering-Plough and SmithKline Beecham; data were obtained by Datamonitor in the Pharmaceutical
`Benchmarking Study). CNS, central nervous system.
`
`All
`
`712 | AUGUST 2004 | VOLUME 3
`
`www.nature.com/reviews/drugdisc
`
`
`
`P E R S P E C T I V E S
`
`0
`10
`20
`30
`40
`50
`60
`70
`80
`90
`100
`
`Failure rate (%)
`
`a
`
`Success rate (%)
`
`100
`90
`80
`70
`60
`50
`40
`30
`20
`10
`0
`
`Phase I
`
`Phase II
`
`Phase III
`
`Registration
`
`Arthritis and pain management
`Oncology
`Women's health
`
`Cardiovascular
`Opthalmology
`All
`
`Central nervous system
`Metabolic disorders
`
`Infectious disease
`Urology
`
`Figure 2 | Success rate by phase of
`development and by therapeutic area.
`a | Data are shown as percent success or percent
`attrition (second X axes) of compounds entering
`that particular phase of development by certain
`therapeutic areas and by the total aggregate for
`that particular phase of development. The data
`clearly show that different therapeutic areas have
`greatly different success or attrition rates, and
`that significant attrition occurs late in the pipeline.
`b | Shows the percentage rate of success of
`compounds entering first in man that progress to
`subsequent development phase. App, approval;
`Reg, registration.
`
`around the world. ”; Pfizer’s India Homepage
`states that “approximately 1 out of every 15
`drug candidates entering development com-
`pletes phase III evaluation and obtains
`approval, ” both suggesting that their rate of
`attrition might be 93–96%. These five factors
`therefore provide compelling evidence that the
`rate of attrition could be significantly reduced
`and that drug development per se does not
`have this current high attrition rate as an
`inherent constraint. Indeed, it points to the
`idea that a systematic evaluation of the sci-
`ence, strategy and processes currently used
`in drug development merit rigorous evalua-
`tion, critical appraisal and modification to
`fulfil the onerous business case demanded
`by our patients, shareholders, consumers
`and governments worldwide.
`
`How can attrition be reduced?
`Several companies in the industry are now
`beginning to take on this problem and are
`starting to make progress. Below we propose
`some approaches that are likely to be valuable,
`but this is clearly not an exhaustive list. It is
`important that the mindset of reducing attri-
`tion in development should be in place from
`the earliest stages of discovery.
`For instance, building the need to get
`very strong evidence for proof of mecha-
`nism into the discovery paradigm is crucial,
`
`100
`90
`80
`70
`60
`50
`40
`30
`20
`10
`0
`
`b
`
`Success rate (%)
`
`I
`
`III
`II
`Reg.
`Stage of development
`
`App.
`
`many of these companies are indeed working
`on innovative mechanisms of action, then
`clearly this cannot be the whole explanation.
`The rate of attrition of compounds with
`novel mechanisms of action is higher than
`that of those with previously precedented
`mechanisms of action (a precedented mecha-
`nism of action is defined as one hitting a thera-
`peutic target that a drug in the market place
`hits, or which has shown proof of concept in
`late clinical trials).
`Last, even comparable large companies
`with extensive portfolios that would average
`out the differences in success between differ-
`ent therapeutic areas, and therefore portfolio
`success, have different probabilities of success.
`For instance, data from the 2002 Certified
`Medical Representatives Institute survey shows
`that the success rate that Merck enjoyed from
`first human dose to market was approximately
`twofold greater than the aggregate of the six
`companies in the same cohort with R&D
`budgets of >US $2 billion per annum8. On the
`other hand, in a briefing to analysts on 17 June
`2003 Pfizer’s current President of Research
`and Development, John La Mattina, was
`quoted as saying “Right now, only one in 25
`early candidates survives to become a pre-
`scribed medicine. We think we can improve
`those odds to one in ten and greatly enhance
`our ability to bring new medicines to patients
`
`to therapeutic areas in which animal models
`of efficacy are notoriously unpredictive6, such
`as CNS and oncology, both of which have
`relatively higher failure rates in Phase II and III
`trials. In the case of oncology, small Phase II
`trials looking at tumour regression in small
`cohorts of patients with different tumour
`types does not always translate to outcomes
`subsequently obtained in larger Phase III
`trials. Nevertheless, in general, failures due to
`lack of efficacy and safety demonstrate the
`need for the development of more predictive
`animal models where possible and, more
`importantly, the need to develop experimen-
`tal medicine paradigms that are more pre-
`dictive of outcomes and to carry out such
`proof-of-concept clinical trials much earlier
`in development.
`
`Can success be increased?
`Several strong strands of evidence indicate
`that it is possible. First is the fact that different
`therapeutic areas have different rates of success
`and this implies that if we understood the
`inherent factors that make one area successful
`as compared with another, we could then
`attack such factors.
`Second is the finding that biologicals have
`a higher rate of success from first-in-man to
`launch — approximately 24%7. It is true that
`most biologicals have been generated in the
`areas of immunology and cancer, but the aver-
`age rate of these two therapeutic areas should
`even out to ~11% (16% for arthritis and pain
`and 5% for cancer, based on the data in TABLE 1,
`which averages to ~11% if the two were in
`equal parts).
`Third, licensing-in compounds has a con-
`sistently higher probability of success in most
`studies, at approximately 24%7. This is the case
`even if the compounds are categorized by
`the stage that the licensing-out company has
`categorized them. This phenomenon cannot,
`therefore, be attributed purely to the fact that
`the licensing-in companies gather more data
`or because they usually put the compound at
`an earlier stage in the pipeline.
`Fourth, companies with R&D budgets of
`less than US $400 million also have higher
`success rates of approximately 18%7. This
`could partly be explained by the possibility
`that these smaller companies might be more
`inclined to work on me-too drugs (which
`should have a higher rate of success), and that
`their portfolios could be more skewed towards
`one therapeutic area or another with a greater
`probability of success. However, if one con-
`siders that many of the biotech companies fall
`into these categories, that many biotech com-
`panies are working in high-attrition-rate
`therapeutic areas such as cancer, and that
`
`NATURE REVIEWS | DRUG DISCOVERY
`
`VOLUME 3 | AUGUST 2004 | 7 1 3
`
`
`
`much earlier in the drug discovery process,
`and/or by better due diligence with respect
`to competitor development programmes
`and the likelihood of true differentiation
`from such drugs that might be ahead in
`development.
`
`Future perspectives
`The demands on pharmaceutical companies
`to meet their business objectives, as well as
`the demands of consumers for cost contain-
`ment of prescription medicines, is forcing
`the industry to think about ways that effi-
`ciencies can be achieved. A particular
`emphasis is being placed on R&D because of
`the relatively dry late-phase pipelines, the
`spiralling costs of drug discovery and drug
`development, and the patent expirations of
`major blockbusters innovated in the past
`two decades. These pressures inevitably lead
`to a healthy evaluation of the science, strate-
`gies and processes involved in drug develop-
`ment, because the rate of attrition in drug
`development is simply too high, which
`makes the R&D process inefficient; effi-
`ciency and sustained profitability by the
`pharmaceutical industry are important for
`reinvestment in further R&D so that thera-
`pies for debilitating human diseases can
`continue to be developed and the price of
`medications contained.
`This inefficiency becomes even more
`acute when one considers the number of
`compounds that undergo attrition in pre-
`clinical research, and that only three out of
`every ten drugs that makes it to market
`recover the original investment made in
`them. Factors that clearly affect attrition
`rates will lead to a more efficient industry
`and will benefit shareholders, and, more
`importantly, patients and the community.
`The industry will be forced to focus on
`attrition rates to balance the costs of drug
`development, to explore cost containment
`measures while still investing significantly in
`R&D, and to continue to generate share-
`holder value. Scientific and technological
`innovations that affect efficacy and safety
`(factors that most significantly contribute to
`attrition in the clinic) will have to be
`addressed. These include more appropriate
`animal models; biomarkers that can report
`the hitting of the molecular target in dose-
`ranging, efficacy and toxicity studies; and a
`new paradigm for drug development that
`will give early readouts for proof of concept
`and one that will allow attrition to occur
`much earlier.
`We believe that governments and con-
`sumers want to reward truly innovative drugs,
`and/or those that are genuinely differentiated
`
`1991
`2000
`
`P E R S P E C T I V E S
`
`50
`
`40
`
`30
`
`20
`
`10
`
`0
`
`Attrition for each criterion (%)
`
`Clinical
`safety
`
`Efficacy
`
`Formulation
`
`PK/
`bioavailability
`
`Commercial Toxicology
`
`Cost
`of goods
`
`Unknown/
`other
`
`Figure 3 | Reasons for attrition (1991–2000). PK, pharmacokinetics.
`
`and therefore showing that modulation of a
`target in a specific or important disease path-
`way might reduce the attrition of a large
`percentage of compounds that fail because of
`lack of efficacy. The development of imatinib
`(Gleevec; Novartis), for example, was based
`on the targeting of a very specific lesion
`(the BCR-ABL chromosomal translocation
`protein-product or Philadelphia chromo-
`some) that occurs in chronic myelogenous
`leukaemia. We have, in a similar manner,
`provided very strong evidence that inhibition
`of β-secretase inhibits the production of
`amyloid-β in knockout mice9 and that cath-
`epsin K is involved in bone resorption (further
`compelling proof of mechanism is provided
`by humans with pycnodysostosis10,11). How-
`ever, we will have to await approval of thera-
`peutics aimed at these latter two mechanisms
`to see whether drug approvals are eventually
`obtained — for example, cathepsin K is in
`Phase II trials and the impact of this approach
`on attrition is still to early to fully evaluate.
`A second method of reducing attrition is to
`eliminate compounds that have mechanism-
`based toxicity; this risk can be rigorously
`interrogated during discovery using tools
`such as gene knockouts and RNA interfer-
`ence, and, crucially, during preclinical devel-
`opment in toxicity testing. Additional tools
`such as transcriptional profiling can also
`affect attrition due to toxicity by giving specific
`gene-signature readouts that are predictive
`of toxicities obtained by previous compounds
`targeting specific molecular targets that
`have failed, and/or molecular signature algo-
`rithms that have been trained from preclinical
`toxicity studies.
`Third, an important clinical tool that can be
`used is to identify biomarkers that signal cor-
`rect dosing and whether the specific molecular
`target has been hit in early proof-of-concept
`clinical trials.
`
`Fourth, and most important, is the
`design of proof-of-concept clinical trials
`during first-in-man studies. This has the
`distinct advantage of providing evidence in
`man that the molecular target is being hit
`and that hitting such a target gives the antici-
`pated physiological response. Appropriately
`designed proof-of-concept studies (or
`experimental medicine paradigms) could
`reduce attrition due to lack of efficacy
`mostly seen in later development, and also
`have the distinct advantage of allowing
`attrition to occur earlier, which is beneficial
`both in terms of real and opportunity costs.
`This is likely to be important given that lack
`of efficacy accounts for about 30% of attrition
`in this study.
`Fifth, another important tool is the use
`of appropriate animal models for efficacy
`testing in preclinical studies. It is interesting
`that oncology and CNS —two therapeutic
`areas with very high attrition rates in the
`data provided here — are also the areas in
`which animal models are not very predic-
`tive of the true human pathophysiology.
`For example, most pharmaceutical compa-
`nies still use xenograft models for oncology
`testing, in which a tumour cell line that might
`have little relevance to the tumour in vivo is
`injected into a nude mouse (which does not
`resemble the immunology of the host; nor
`does the artificial location of the tumour
`significantly resemble what happens in vivo
`during tumorigenesis). The use of appro-
`priate genetic models (for example, trans-
`genic and gene knockout animals) of
`tumorigenesis might be more pathophysio-
`logically relevant.
`Last, another area in which attrition can
`be reduced is the discontinuation of com-
`pounds for commercial reasons either by
`gaining alignment between the research,
`development and marketing functions
`
`714 | AUGUST 2004 | VOLUME 3
`
`www.nature.com/reviews/drugdisc
`
`
`
`from existing drugs and that address a true
`unmet medical need; this provides a tremen-
`dous incentive for the pharma industry to
`conduct R&D in this arena, and this in itself
`could affect R&D productivity. Drugs that
`target novel mechanisms have higher attrition
`rates12, but a combination of better-validated
`preclinical targets that have significant pre-
`clinical proof of principal, and the scientific
`and technological innovations that posi-
`tively affect efficacy and safety of drugs dis-
`cussed earlier in this article, can mitigate
`such attrition risks. It is clear that in
`the twentieth century the pharmaceutical
`industry has had significant positive impact
`on the health and longevity of humans
`across the globe, but the early twenty-first
`century will demand both great effective-
`ness and efficiency from the industry, and it
`is therefore vital that the industry rapidly
`gears up to meet these demands.
`
`Ismail Kola, Ph.D. (Med), is Senior Vice-President
`of Basic Research at Merck Research Labs, 126 East
`Lincoln Avenue, Rahway, New Jersey 07075, USA.
`John Landis , Ph.D., is Senior Vice-President
`Pharmaceutical Sciences and Compliance Clinical
`Sciences at Schering-Plough Research Institute,
`2000 Galloping Hill Road, Kenilworth New Jersey
`07033, USA. Correspondence to I.K.
`e-mail: ismail_kola@merck.com
`doi:10.1038/nrd1470
`Kola, I. & Rafferty, M. New technologies that may impact
`drug discovery in the 5–10 year timeframe workshop.
`2002 Biomed. Expo Ann Arbor Michigan (data derived
`from Prous Science, Drugs News Prospect (2002).
`Franz, S. 2003 approvals: a year of innovation and upward
`trends. Nature Rev. Drug Discov. 3, 103–105 (2004).
`Accenture Consulting. High performance drug discovery:
`an operating model for a new era. Accenture (2001).
`DiMasi, J. A., Hansen, R. W. & Grabowski, H. G.
`The price of innovation: new estimates of drug
`development costs. J. Health Econ. 22, 151–185 (2003).
`Fearn, C. J. IMS Health. The World Pharmaceutical Market
`Presented at the Strategic Management Review, Australia,
`19–20 June (2002).
`Booth, B., Glassman, R. & Ma, P. Oncology’s trials.
`Nature Rev. Drug Discov. 2, 609–610 (2003).
`Benjamin, G. A. & Lumley, C. E. Industry Success Rates
`2003 Including Trends in Success Rates CMR Report
`Number 03–202 R (CMR International Surrey, UK, 2003).
`
`5.
`
`6.
`
`7.
`
`1.
`
`2.
`
`3.
`
`4.
`
`P E R S P E C T I V E S
`
`8. Van den Haak, M. A., Palachandran, J. & Benjamin, G. A.
`Performance Metrics in Global Pharmaceutical R&D,
`Company specific review 2002 for Merck & Co. (CMR
`International, Surrey, UK, 2003).
`9. Roberds, S. L. et al. BACE knockout mice are healthy
`despite lacking the primary β-secretase activity in brain:
`implications for Alzheimer’s disease therapeutics.
`Hum. Mol. Genet. 10, 1317–1324 (2001).
`10. Gowen, et al. Cathepsin K knockout mice develop
`osteopetrosis due to a deficit in matrix degradation but
`not demineralization. J. Bone Miner. Res. 14,
`1654–1663 (1999).
`11. Lazner, F., Gowen, M. & Kola, I. An animal model for
`pycnodysostosis: the role of cathepsin K in bone
`remodeling. Mol. Med. Today 5, 413–414 (1999).
`12. Ma, P. & Zemmel, R. Value of novelty? Nature Rev. Drug
`Discov. 1, 571–572 (2002).
`
`Acknowledgements
`We wish to acknowledge Datamonitor for the assembly of data
`(Pharmaceutical R&D Benchmarking Forum) used in this study.
`
`Competing interests statement
`The authors declare competing financial interests: see Web version
`for details..
`
`Online links
`
`FURTHER INFORMATION
`PhRMA: http://www.phrma.org/
`Access to this interactive links box is free online.
`
`NATURE REVIEWS | DRUG DISCOVERY
`
`VOLUME 3 | AUGUST 2004 | 7 1 5