[mAbs 1:4, 387-389; July/August 2009]; ©2009 Landes Bioscience
`
`Editor’s Corner
`Probabilities of success for antibody therapeutics
`
`Janice M. Reichert
`
`Tufts Center for the Study of Drug Development; Boston, MA USA
`
`Probabilities of success (POS) play a key role in determining the
`distribution of resources by both investors and the pharmaceutical
`industry. Resources such as time, money and personnel are more
`likely to be directed toward programs in categories with acceptable
`rates of success. What is considered acceptable may, of course, vary
`between companies and other decision-makers. With the increased
`focus on development of antibody therapeutics, it is important
`for stakeholders to understand the utility, and limitations, of POS
`values such as cumulative approval success rates and clinical phase
`transition probabilities. A key point is that cumulative approval
`success rates are derived from data for only those candidates with
`known fates (either approved or terminated), but clinical phase
`transition probability calculations include data on the status of all
`candidates.
`POS values for various cohorts of monoclonal antibody (mAb)
`therapeutics have been reported previously.1-6 For mAb POS, a
`key consideration is the source of the protein sequence. Data for
`humanized and human mAbs must be analyzed separately because,
`overall, these molecules display improved safety and efficacy
`profiles compared to murine and chimeric versions. Humanized
`mAbs comprise the ‘canonical’ cohort because a large number
`(>150) of these candidates have entered clinical study over the last
`two decades (1988–2008), and 12 have been approved (Table 1).
`However, ultimate fates (approval or termination) are known for
`only about half, and the cumulative approval success rate for the
`entire cohort of humanized mAbs will only be an estimate until the
`fates of all the molecules have been decided. The current cumula-
`tive approval success rate estimate for humanized mAbs is 17%.2
`It is important to note that time plays an essential role in POS
`calculations. In general, clinical study and regulatory review periods
`for therapeutics are lengthy, and mAbs are not exceptional in this
`regard. The mean (median) for the combination of the clinical and
`US Food and Drug Administration (FDA) approval phases for 23
`mAbs (Table 1) is currently 8 (7) years. This has important impli-
`cations for POS calculations for mAb cohorts that include high
`percentages of candidates that have entered clinical study within
`the past seven or eight years. Candidates that have entered clinical
`
`Correspondence to: Janice M. Reichert; 75 Kneeland Street; Suite 1100; Boston,
`MA 02111 USA; Email: janice.reichert@tufts.edu
`
`study since 2001 have not had sufficient time, on average, for
`approval, but might have been terminated for a variety of reasons.
`This suggests that there is a downward bias in cumulative success
`rates for cohorts that include candidates that recently entered
`clinical study. Indeed, the cumulative success rate for humanized
`mAbs changes dramatically when the cohort is divided into two
`groups: candidates that entered clinical study during 1988–1996
`(n = 30; eight approved) and 1997–2008 (n = 125; two approved).
`Ultimate fates are known for 87% of the older candidates, and the
`cumulative success rate for the cohort is 31%. However, ultimate
`fates are known for only 33% of the newer candidates, and because
`many have not been in clinical study long enough to accumulate the
`data needed for approval, the cumulative success rate is 5%. This
`value will rise to 9% if the two humanized mAbs in FDA review
`(Table 1) are approved.
`Clinical phase transition probabilities are another important
`measure of the success of a cohort such as humanized mAbs.
`Whereas cumulative approval success rates include data only for
`candidates that are either approved or terminated, clinical phase
`transition probabilities take the status of all candidates into
`account. It is critical to understand the relationship between the
`two parameters in order to interpret POS values appropriately.
`The mathematical product of the phase transition probabilities
`will exactly equal the cumulative success rate only when the fates
`of all the candidates are known. In practice, the two values will
`converge as the percentage with known fates goes to 100%. When
`the fates of fewer than 50% are known, then the values can be
`quite different. One reason for this phenomenon is that candi-
`dates that will ultimately be discontinued remain, technically, at
`Phase 2 for long periods while the company decides whether to
`advance these perhaps marginal candidates into expensive Phase 3
`studies, or attempts to partner or out-license the projects. In these
`cases, the candidates contribute in a positive way to the Phase 1 to
`Phase 2 transition probability, and inflate the mathematical
`product, but are not yet included in the cumulative success rate
`calculation because they have not been officially terminated.
`A comparison of phase transition probabilities for human-
`ized mAbs with the cumulative approval success rates provides
`a good example of the phenomenon. The values for candidates
`that entered clinical study during the three periods 1988–2008,
`1988–1996 and 1997–2008 are quite similar: Phase 1 to 2 tran-
`sition probabilities were 83, 90 and 80%, respectively; Phase 2
`to 3 transition probabilities were 48, 50 and 46%, respectively;
`Phase 3 to FDA review transition probabilities were 75, 73
`Hospira v. Genentech
`IPR2017-00805
`Genentech Exhibit 2003
`
`387
`
`Submitted: 5/18/09; Accepted: 5/18/09
`
`Previously published online as a mAbs E-publication:
`www.landesbioscience.com/journals/mabs/article/9031
`
`www.landesbioscience.com
`
`mAbs
`
`

`

`Table 1 Therapeutic monoclonal antibodies in FDA review or approved
`
`Probabilities of success for antibody therapeutics
`
`Note: Information current as of May 15, 2009. *Proposed trade name; #Voluntarily withdrawn from US market in April 2009. FDA, US Food and
`Drug Administration. Source: Tufts Center for the Study of Drug Development
`
`and 80%, respectively; and the review
`to approval transition probability was
`100% for all three cohorts. The mathe-
`matical products of the phase transition
`probabilities for the three cohorts are
`similar: 30, 33 and 29%, respectively,
`despite the fact that the current cumu-
`lative approval success rates vary (17,
`31 and 5%, respectively). This suggests
`that, so far, the newer candidates are
`proceeding through clinical studies at a
`pace that is similar to the older candi-
`dates. However, the cohort of candidates
`that entered clinical study recently (n =
`125) is much larger compared to the
`cohort of candidates that entered clin-
`ical study during 1988–1996 (n = 30),
`and many are in early clinical studies.
`It remains to be seen whether a similar
`proportion of the newer candidates will
`ultimately be approved.
`POS for human mAbs are affected
`by the same factors. Analysis of this
`cohort is additionally affected by the
`time-frame of clinical entry because of
`technological advances in production
`methods. Early attempts to produce
`human mAbs from hybridomas were
`largely unsuccessful, so human mAbs
`did not start entering clinical study
`in large numbers until after trans-
`genic mice and display technologies
`were developed. As a consequence, the
`majority of candidates are in clinical
`studies, and thus far, only three human
`mAbs, adalimumab, panitumumab and
`golimumab, have been approved in the
`US. However, five additional human
`mAbs (Table 1) are undergoing review
`by FDA (as of May 2009). Approval
`of these candidates would dramatically
`affect the cumulative success rate of the
`cohort.
`Additional complexity arises when
`POS values from various sources are
`compared. Such comparisons should
`be done cautiously because factors such
`as variations in methodology, time-
`frame, and cohort inclusion criteria can
`have dramatic effects on the calculated
`results. End users, including investors
`and strategic planners, should carefully
`consider whether a distinction between
`a cumulative approval success rate and
`the mathematical product of phase
`
`388
`
`mAbs
`
`2009; Vol. 1 Issue 4
`
`

`

`Probabilities of success for antibody therapeutics
`
`transition probabilities has been made, and whether sufficient
`information about the cohort and methodology has been provided
`so that the POS values presented can be clearly understood.
`References
` 1. Nelson AL, Reichert JM. Development trends for therapeutic antibody fragments. Nat
`Biotechnol 2009; 27:331-7.
` 2. Reichert JM. Monoclonal antibodies as innovative therapeutics. Curr Pharma Biotechnol
`2008; 9:423-30.
` 3. Reichert JM, Rosensweig CJ, Faden LB, Dewitz MC. Monoclonal antibody successes in
`the clinic. Nat Biotechnol 2005; 23:1073-8.
` 4. Reichert J, Pavlou A. Monoclonal antibodies market. Nat Rev Drug Disc 2004;
`3:383-4.
` 5. Reichert JM. Therapeutic monoclonal antibodies: trends in development and approval
`in the US. Curr Opin Mole Ther 2002; 4:110-8.
` 6. Reichert JM. Monoclonal antibodies in the clinic. Nat Biotechnol 2001; 19:819-22.
`
`www.landesbioscience.com
`
`mAbs
`
`389
`
`