Of every 100 drug programs that enter Phase I, about eight reach FDA approval. [1] A portfolio team must decide which of the other 92 to stop, and when. This article collects the public numbers behind that decision: success rates, development cost, R&D returns, loss-of-exclusivity exposure, and the pricing rules that now limit how long a drug earns full revenue. It then tests those numbers against five real portfolio decisions made between 2018 and 2025.
Across 9,704 development programs started from 2011 to 2020, the likelihood of approval from Phase I was 7.9%. [1]
Source: BIO, Informa Pharma Intelligence and QLS Advisors, 2021, as summarized by Informa.
The Short Answer: How Do You Make Good Pharma R&D Portfolio Decisions?
Good portfolio decisions use three inputs: a realistic probability of success for the current phase, a value estimate that accounts for how long the drug will earn full revenue, and a stop rule that the team writes down before the data arrive. AstraZeneca reported that success from candidate nomination to Phase III completion rose from 4% in 2005-2010 to 19% in 2012-2016 after it adopted a five-factor decision framework. [20]
The inputs changed in the last three years. Projected returns on late-stage pipelines reached 7.0% in the 2025 cycle, but a small group of obesity and metabolism assets drives most of that gain. [10][11] Medicare price negotiation can start 9 years after approval for small molecules and 13 years after approval for biologics. [29] A large block of branded revenue loses exclusivity before 2030. [24] Each of these facts changes the value of an asset on paper before any trial result arrives.
We propose a simple screen for this problem. We call it the three-clock test: a science clock (will it work), an exclusivity clock (how long is it protected), and a price clock (how long does it earn full price). The framework is our own. It is not an established industry classification. The section titled “The Three-Clock Test” explains it.
What the Data Shows: Ten Findings That Matter
Each finding below stands alone and carries its source. Findings 3, 5, 6 and 9 are calculations made for this article from the cited inputs. They are labeled as analysis.
Programs that entered Phase I from 2011 to 2020 had a 7.9% likelihood of approval. The data cover 9,704 programs and 12,728 phase transitions. [1]
Phase II had the lowest transition rate of any phase: 28.9% of candidates moved on to Phase III. [2][4]
Analysis: Phase II removes about 37 of every 100 Phase I entrants. That is about 40% of all programs lost before approval. (Calculated from the rates in [4].)
The Tufts estimate of average cost per approved new compound is $1,395 million out-of-pocket and $2,558 million when capitalized to the approval date, in 2013 dollars. [8]
Analysis: Capitalization adds $1,163 million, or 45.5%, of the $2,558 million Tufts total. (Calculated from [8].)
Analysis: Without the high-value assets that entered late-stage pipelines in 2024, Deloitte reported that the cohort IRR would have been 3.8% instead of 5.9%. Those assets add 2.1 percentage points. (Calculated from [14].)
The forecast IRR on the top 20 biopharma late-stage pipelines was 7.0% in 2025, up from 5.9% in 2024. The median company stayed at 4.0%. [10][13]
Morgan Stanley estimated that products losing exclusivity through 2030 generate $183.5 billion in annual sales. Exposure ranges from 6% of revenue at Vertex to 67% at Amgen. [24]
Analysis: After a failed Phase III add-on trial, one analyst cut a 2030 sales forecast for Cobenfy from $5.8 billion to $2.6 billion, a 55% reduction. Bristol Myers Squibb had paid $14.0 billion for the company that owned the drug. [38][39]
Under the Inflation Reduction Act, small molecules become eligible for Medicare price negotiation 9 years after approval, and biologics after 13 years. [29]
Methodology: How Were the Derived Figures Built?
What data did this analysis use?
We did not query a private database. We used published figures from the sources in the reference list and recalculated a small number of derived metrics from them. Every derived figure carries the label “analysis” and sits next to its inputs.
Which periods and records are included?
The main success-rate data cover programs tracked from January 2011 to November 2020. [4] The cost estimate uses 2013 dollars. [8] The returns data cover the 2022 to 2025 forecast cycles of the Deloitte series. [10][17] Deal and trial events run from 2018 to November 2025. We include a record only if a regulator, a company filing, a peer-reviewed paper, or a named news report states it. Where two sources give the same number, we cite the source closest to the original. We did not need to remove duplicates, because we used no row-level data.
How were the calculations done?
Cumulative probabilities multiply the phase transition rates. For example, 52.0% x 28.9% x 57.8% x 90.6% equals 7.87%, which matches the 7.9% that BIO reports. [1][4] Percentage changes divide the difference by the starting value. Programs lost at a stage equal the programs that entered the stage minus the programs that left it. The valuation example in the rNPV section uses placeholder cash flows. We label every placeholder as hypothetical.
What are the limits of this analysis?
Average success rates hide large differences between disease areas. Hematology programs had a 23.9% likelihood of approval from Phase I. Urology programs had 3.6%. [2] Cost studies use different methods and dollar years, so their results do not compare directly. Analyst sales forecasts are estimates, not outcomes. Trial and deal events can change after the sources were published. Check current sources before you rely on any status in this article.
Definitions: Which Portfolio Terms Matter Here?
What is likelihood of approval (LOA)?
Likelihood of approval is the share of programs that reach FDA approval from a stated starting phase. BIO uses LOA from Phase I as its headline measure. [1][2]
What is risk-adjusted net present value (rNPV)?
Risk-adjusted net present value multiplies each future cash flow by the probability that the program reaches that point. A full model also discounts the result to today. Teams use it to compare assets at different phases. It is a standard valuation method. The inputs in our worked example are hypothetical.
What is projected IRR in the Deloitte series?
Deloitte calculates the internal rate of return that large biopharma companies can expect from assets in their late-stage pipelines. It uses this figure as a proxy for the industry’s ability to balance R&D outlay against projected cash inflows. [19] It measures forecast returns on a defined pipeline. It does not measure realized profit.
What is loss of exclusivity (LOE)?
Loss of exclusivity is the point at which patents and regulatory exclusivities no longer block competing generic or biosimilar products. FDA exclusivities and patents are separate protections. Pediatric exclusivity, for example, does not stand alone. It attaches to existing patents and exclusivity. [22]
Why Is Pharma R&D Portfolio Management So Hard?
What is Eroom’s Law, and does it still describe R&D efficiency?
Eroom’s Law describes a long fall in output per dollar. Scannell and co-authors found that the number of new drugs approved per billion US dollars of R&D spending halved roughly every 9 years from 1950, a fall of about 80-fold in inflation-adjusted terms. [6] They named four causes: the “better than the Beatles” problem, the “cautious regulator” problem, the “throw money at it” tendency, and the “basic research-brute force” bias. [6]
The paper is from 2012. Returns have moved since then, as the Deloitte data below show. The long-run pattern is still a useful warning: more spending does not guarantee more approvals.
Why do teams keep funding weak programs?
Scannell and co-authors proposed that companies appoint a “Dead Drug Officer” to run a post-mortem on programs that fail, according to a Scientific American summary. [7] The idea targets a known problem. The team that built a program has a strong reason to keep it alive. A stop rule written before the data arrive removes part of that pressure.
What Are Clinical Success Rates by Phase?
What is the success rate at each phase of drug development?
From 2011 to 2020, 52.0% of programs moved from Phase I to Phase II, 28.9% moved from Phase II to Phase III, 57.8% moved from Phase III to filing, and 90.6% moved from filing to approval. [4] The table adds two columns we calculated.
Stage
Transition rate, 2011-2020
Analysis: Phase I entrants still active per 100
Analysis: programs lost at this stage per 100 entrants
Source
Phase I to Phase II
52.0%
52.0
48.0
[4]
Phase II to Phase III
28.9%
15.0
37.0
[4]
Phase III to NDA/BLA filing
57.8%
8.7
6.3
[4]
NDA/BLA filing to approval
90.6%
7.9
0.8
[4]
Why is Phase II the biggest hurdle?
Phase II is where a program first tests efficacy in patients at scale, and it has the lowest transition rate. [2] The 2016 BIO report for 2006-2015 reached the same conclusion: Phase II was lowest of the four phases. [3] A report on that earlier study said about 70% of candidates did not move on to Phase III, and linked this to Phase II being the first test of proof of concept in humans. [3]
Analysis: the table shows that Phase I removes more programs by count (48 per 100) than Phase II (37 per 100). Phase II matters more for cost, because a Phase II failure arrives after a larger spend than a Phase I failure.
How much did Phase I change between BIO editions?
The likelihood of approval from Phase I fell from 9.6% in the 2006-2015 edition to 7.9% in the 2011-2020 edition. [2][3] The Phase I transition rate fell from 63.2% to 52.0%. The Phase II rate moved from 30.7% to 28.9%, and the Phase III rate moved from 58.1% to 57.8%. [3][4]
Measure
2006-2015
2011-2020
Change (percentage points)
Source
Likelihood of approval from Phase I
9.6%
7.9%
-1.7
[2][3]
Phase I transition
63.2%
52.0%
-11.2
[3][4]
Phase II transition
30.7%
28.9%
-1.8
[3][4]
Phase III transition
58.1%
57.8%
-0.3
[3][4]
Analysis: the product of the first three transition rates fell from 11.3% to 8.7%, a drop of 2.6 percentage points. If only the Phase I rate had changed, the product would have been 9.3%. Phase I alone therefore explains about 2.0 of the 2.6 points, or about 77%. This one-at-a-time method is a simplification. The 2016 report also warns that Phase I rates may benefit from delayed reporting bias, so part of the earlier Phase I rate may reflect how programs were reported, not how they performed. [3]
Which disease areas and drug types have the best odds?
Odds vary by a factor of about seven between disease areas. Hematology had the highest likelihood of approval from Phase I at 23.9%, and urology had the lowest at 3.6%. [2] Oncology overall had 5.3%, and immuno-oncology programs had 12.4%. [2] Biologics had 9.1% against 5.7% for new molecular entities, and vaccines had 9.7%. [1] On average, a Phase I asset needed 10.5 years to reach approval. [2]
Do patient-selection biomarkers raise success rates?
Yes. Programs that used patient-selection biomarkers had a 15.9% likelihood of approval, about twice the overall 7.9%, driven by a Phase II success rate of nearly one in two. [2] The 2016 report found the same direction: 25.9% with selection biomarkers against 8.4% without them, in the earlier data set. [3] The 2016 figures come from a different data set and period, so we do not compare them with the 2021 figures.
Do all datasets agree on success rates?
No. A working paper on arXiv that uses firm-level data from a different source reports a Phase II to Phase III transition of 34.5% and a Phase I to approval estimate of 12.1%. [5] The authors note that most earlier studies use clinical trial data to estimate Phase I to approval. [5] A portfolio team should treat the BIO figures as one benchmark, then test how results change with a second source. The rNPV section below does this.
How Much Does It Cost to Develop a New Drug?
What is the Tufts estimate of drug development cost?
DiMasi, Grabowski and Hansen surveyed 10 firms on 106 randomly selected drugs. They estimated an average out-of-pocket cost of $1,395 million per approved new compound, in 2013 dollars. [8] After they capitalized costs to the approval date at a 10.5% real discount rate, the total was $2,558 million. [8] Adding post-approval R&D raised it to $2,870 million. [8] The authors also found that capitalized costs grew 8.5% a year above general price inflation compared with the prior study in the series. [8]
How much of the Tufts figure is the cost of money?
Analysis: the gap between capitalized and out-of-pocket cost is $1,163 million, which is 45.5% of the capitalized total. Time and the cost of capital explain almost half of the headline number. A portfolio that shortens development by one year lowers this part directly. (Calculated from [8].)
How much is spent before a human receives the drug?
A briefing note from Knowledge Ecology International reports that the Tufts study implies pre-human costs of $430 million out-of-pocket and $1,098 million capitalized per approved drug. [9] The note is a critique of the Tufts study, so treat it as an opposing view and not a neutral summary. It shows which part of the estimate depends on the authors’ assumptions about pre-clinical spending.
What does Deloitte estimate for cost per asset?
Deloitte’s 2025 analysis put the cost of bringing a new drug to market at $2.23 billion per asset for the top 20 biopharma companies. [14] Deloitte also estimated that the cohort spent $7.7 billion on clinical trials for assets that were terminated in 2024. [15]
Why can you not compare $2.56 billion and $2.23 billion directly?
The two numbers come from different methods, samples and dollar years. Tufts used a survey of 106 compounds in 2013 dollars. [8] Deloitte uses reported financials from 20 large companies. [14][19] Use each figure to track trends inside its own series. Do not subtract one from the other.
What Are R&D Returns Today? The Deloitte IRR Timeline
How has forecast R&D return changed from 2022 to 2025?
Forecast IRR for the top 20 biopharma late-stage pipelines rose from 1.2% in the 2022 cycle to 7.0% in the 2025 cycle. [10][17] Deloitte called the 2022 figure the lowest point for the cohort since its analysis began. [17]
Forecast cycle
Projected IRR, top 20 cohort
Note
Source
2022
1.2%
Lowest for the cohort since 2010
[17]
2023
4.1%
A later Deloitte release gives 4.3% for 2023
[17][13]
2024
5.9%
Up 1.6 points from the prior year per TechTarget
[13][14]
2025
7.0%
Third year of improvement; median 4.0%
[10][12]
Analysis: IRR rose 5.8 percentage points between the 2022 and 2025 cycles (7.0% minus 1.2%). Total R&D spending by the cohort rose from $139.2 billion in 2022 to $145.5 billion in 2023, about 4.5%. [18] Returns moved much faster than spending.
How concentrated is the improvement?
Deloitte says the gain is concentrated in a narrow set of GLP-1 and GIP assets. [11] The 2025 average IRR is 7.0%, but the median is 4.0%. [10] Analysis: the 3.0 point gap between mean and median shows that a few companies lift the average. In 2024, Deloitte reported that IRR would have been 3.8% without the high-value assets, against the reported 5.9%. [14] Obesity, endocrine and metabolism sit just behind oncology among the areas with the highest forecast revenue. [15]
What is the average peak sales forecast per asset?
The average forecast peak sales per asset rose from $510 million in the 2024 cycle to $598 million in the 2025 cycle. [16][10] Analysis: that is an increase of $88 million, or 17.3%. Deloitte also notes that value is spreading unevenly. [10]
How to Value a Pipeline Asset: A Worked rNPV Scenario
This section is scenario analysis. The cash flows are hypothetical placeholders chosen for clarity. Only the phase probabilities come from published data. We left out discounting to keep the arithmetic visible. A real model must discount every cash flow.
What is the base case for a Phase II asset?
Assume a Phase II asset with a launch value of $2,000 million if it reaches approval, and remaining costs of $60 million for Phase II, $250 million for Phase III and $20 million for filing. Use the BIO transition rates. [4] The probability of approval from Phase II is 28.9% x 57.8% x 90.6%, which equals 15.1%.
Expected launch value is 15.1% x $2,000 million, or $302.7 million. Expected cost is $60 million for Phase II, plus $72.3 million for Phase III (28.9% chance of reaching it), plus $3.3 million for filing (16.7% chance). That totals $135.6 million. The rNPV is $167.1 million.
How much does the choice of success-rate source change the answer?
Change one input: replace the 28.9% Phase II rate with the 34.5% rate from the arXiv working paper. [5] Keep all other inputs. The probability of approval rises to 18.1%, expected launch value rises to $361.3 million, expected cost rises to $150.2 million, and rNPV rises to $211.1 million. That is a $44.0 million gain, or 26.3%.
Why does this matter for a stop decision?
A Phase II probability that differs by 5.6 points changes value by more than a quarter. Teams should record which source set each probability and review the stop rule if the source changes.
What happens if price negotiation cuts the last years of revenue?
Return to the base case. Assume, as a hypothetical, that years 10 to 13 after approval produce 20% of launch value ($400 million), and that price negotiation halves that portion. Launch value falls to $1,800 million. Expected launch value falls to $272.4 million, and rNPV falls to $136.8 million. That is a $30.3 million loss, or 18.1%. The 9-year and 13-year dates in the Inflation Reduction Act decide whether this scenario applies to a small molecule or only to a biologic. [29]
Scenario (hypothetical inputs)
Probability of approval from Phase II
Expected launch value ($M)
Expected remaining cost ($M)
rNPV, undiscounted ($M)
A. Base case, BIO rates [4]
15.1%
302.7
135.6
167.1
B. Phase II rate of 34.5% [5]
18.1%
361.3
150.2
211.1
C. Base case, years 10-13 revenue halved
15.1%
272.4
135.6
136.8
The Three-Clock Test: An Original Screen for Portfolio Decisions
The three-clock test is our own framework. It is not an established industry classification. It asks three questions about each asset and records the evidence for each answer.
What is the science clock?
The science clock asks how likely the asset is to reach approval and how long that takes. Inputs are the phase-based success rates, disease-area rates, and the 10.5-year average from Phase I to approval. [2][4] AstraZeneca’s work shows how a company can raise the quality of this input. Its 5R framework sets five questions: right target, right tissue, right safety, right patient and right commercial potential. [20]
What is the exclusivity clock?
The exclusivity clock asks how long patents and FDA exclusivities block competitors after launch. Inputs are Orange Book patents, FDA exclusivity periods and known litigation. [22][23] The next section lists the main FDA periods.
What is the price clock?
The price clock asks how long the drug earns full price before Medicare negotiation can apply. Under the Inflation Reduction Act, that is 9 years after approval for small molecules and 13 years for biologics. [29] The exclusivity clock and the price clock can end on different dates. A biologic can reach negotiation eligibility after its main patent has expired, and a small molecule can reach it while patents still run.
Clock
Question for each asset
Evidence to collect
Published reference point
Science
What is the probability of reaching approval from the current phase?
Phase, disease area, biomarker use
7.9% from Phase I overall; 15.9% with biomarkers [1][2]
Exclusivity
When can generics or biosimilars enter?
Orange Book patents, FDA exclusivity, Paragraph IV suits
5, 7 or 3 years of FDA exclusivity plus 6 months pediatric [22]
Price
When can Medicare negotiation apply?
NDA or BLA pathway, approval date
9 years small molecule; 13 years biologic [29]
What FDA Exclusivity and Patent Rules Set the Exclusivity Clock?
What are the FDA exclusivity periods?
FDA lists four exclusivity types in its patents and exclusivity guidance: new chemical entity exclusivity (5 years), orphan drug exclusivity (7 years), new clinical investigation exclusivity (3 years), and pediatric exclusivity (6 months added to existing patents and exclusivity). [22] Sponsors must submit patent information for the Orange Book within 30 days after approval of an NDA or supplement. [22]
Protection
Length
Key rule
Source
New chemical entity exclusivity
5 years
Applies to new active moieties only
[22][23]
New clinical investigation exclusivity
3 years
Based on new clinical trials
[22][23]
Orphan drug exclusivity
7 years
Protects against approval of the same active moiety
[22][23]
Pediatric exclusivity
6 months
Attaches to existing patents and exclusivity; needs an FDA written request
[22][23]
How do Paragraph IV challenges and the 30-month stay work?
For each patent listed in the Orange Book, a generic applicant must certify its position. [23] A challenge under Paragraph IV can start as early as four years into the five-year new chemical entity period. [23] If the brand company files suit in time, FDA stays approval of the generic application for 30 months. [23] This is why a portfolio model should list each Orange Book patent, not only the first expiry date.
Does pediatric exclusivity extend the patent?
No. FDA training material states that pediatric exclusivity is not a patent extension, and that it adds 6 months at the end of listed patents and exclusivity. [22] The distinction matters for litigation timing, because the 6 months follows the underlying protection.
Where does DrugPatentWatch fit in this work?
DrugPatentWatch compiles patent and exclusivity records by drug. A portfolio team can use that view to fill the exclusivity-clock row for each asset, then confirm each date against the FDA Orange Book, the Purple Book and the patent documents. A database is a finding aid. The primary record decides the date.
Who Is Exposed? The 2030 Patent Cliff by Company
How much revenue is at risk from loss of exclusivity?
Morgan Stanley calculated that products losing exclusivity through 2030 generate a combined $183.5 billion in annual sales. [24] The industry average exposure is 38% of revenue. [24] EY, in a separate estimate reported by CNBC in January 2024, put sales at risk for the top 20 biopharma companies at $180 billion from patent expirations through 2028. [25] The two figures cover different years and company sets, so we do not treat them as one series.
Company
Share of revenue exposed through 2030
Products named by the source
Source
Amgen
67%
Enbrel ($3.7 billion), Otezla ($2.2 billion)
[24]
Bristol Myers Squibb
63%
Eliquis ($12.2 billion in 2023), Opdivo ($9 billion), Revlimid ($6.1 billion, US exclusivity already lost)
[24]
Merck
56%
Keytruda (US exclusivity expected 2029)
[24]
Industry average
38%
Not applicable
[24]
Johnson & Johnson
33%
Not named in the excerpt
[24]
Pfizer
33%
Not named in the excerpt
[24]
Eli Lilly
31%
Not named in the excerpt
[24]
AbbVie
29%
Not named in the excerpt
[24]
Gilead
24%
Not named in the excerpt
[24]
Vertex
6%
Not named in the excerpt
[24]
What does the exposure spread mean?
Analysis: the spread between the highest and lowest exposure in the table is 61 percentage points (67% minus 6%). Amgen’s exposure is about 1.8 times the industry average, BMS is about 1.7 times, and Merck is about 1.5 times. (Calculated from [24].) Exposure is a figure about the past and present portfolio. It does not say how much a company can replace through its pipeline.
What is the Keytruda exclusivity timeline?
GlobalData reported that Keytruda and Johnson & Johnson’s Darzalex/Faspro will lose US exclusivity by 2029. [26] Keytruda generated over $29 billion in 2024, and GlobalData expects it to stay in the top 10 selling drugs in 2030 with lower revenue. [26] A Merck executive, Davis, said Merck views the expiration as a “hill, not a cliff,” according to CNBC. [25]
Many of the drugs facing expiry are biologics, including Keytruda, Stelara and Opdivo. [25] For biologics, the competitor is a biosimilar. Eliquis is a small-molecule product, and its competitors are generics. [25]
What Does the Inflation Reduction Act Mean for Small-Molecule and Biologic Portfolios?
What is the 9-year versus 13-year rule?
The Inflation Reduction Act makes biologics exempt from Medicare price negotiation for 13 years after approval and small molecules for nine years. [29] Critics call the four-year gap the “pill penalty.” [29][31] A bipartisan bill, the Ensuring Pathways to Innovative Cures (EPIC) Act, would equalize the period. [29] Congress reintroduced it in 2025. [31] Check its current status before you model either date.
What did research find about the effect on investment?
A study in Therapeutic Innovation and Regulatory Science found that the pill penalty reduced investment in small-molecule drugs aimed at the Medicare-aged population. It also estimated that moving small molecules to 13 years would raise FDA approvals for that population by 21%. [30] This is a modeled result from one study, not an observed approval count.
What changes have companies reported?
BioSpace reported that Pfizer was the first big pharma company to cite the Act when it moved its oncology R&D toward antibody-drug conjugates and bispecific antibodies and away from small molecules. [32] An investment tracker from Incubate listed small-molecule cuts at Pfizer, Bristol Myers Squibb and Roche, according to PharmaVoice. [31] The same report notes that companies face other business pressures too, so the Act’s share of each decision is not always clear. [31]
Case Studies: Five Portfolio Decisions Tested Against the Data
Case 1: How did AstraZeneca’s 5R framework change success rates?
AstraZeneca studied 142 of its drug discovery and development projects and found five factors that explained project outcomes. [21] The company started to apply the 5R framework in 2011, after R&D productivity fell below industry averages in 2005-2010. [20] It reported that success from candidate nomination to Phase III completion rose from 4% in 2005-2010 to 19% in 2012-2016. [20] The same paper says the company wanted decisions that used more rigorous and quantitative reasoning, and a culture where truth seeking was encouraged. [20]
Analysis: the AstraZeneca result is a company-reported figure for a defined stage range. It is not the BIO Phase I to approval measure, and the two do not compare directly. The useful part for portfolio teams is the method: define the questions before the data arrive.
Case 2: Why did Pfizer stop danuglipron?
Pfizer stopped the twice-daily form of danuglipron in late 2023 after most patients in a mid-stage trial dropped out because of side effects such as nausea and vomiting. [34] On April 14, 2025, Pfizer stopped the once-daily form after one asymptomatic trial participant had a potential drug-induced liver injury that resolved when the participant stopped the drug. [33]
Pfizer said the frequency of liver enzyme elevations across more than 1,400 participants was in line with approved drugs in the class, and that its dose-optimization studies met their pharmacokinetic objectives. [33] The company said it decided after a review of all clinical data and recent input from regulators. [33] Analysis: this stop did not rest on a poor class-level safety rate. The stated inputs were one serious case, regulator input, and the full data set. A stop rule that names these inputs in advance is easier to defend later.
Case 3: What did Pfizer pay for Metsera, and what did the structure show?
After it stopped danuglipron, Pfizer bought an obesity pipeline. Metsera agreed to Pfizer’s final terms of up to $86.25 per share: $65.60 in cash and a contingent value right (CVR) of up to $20.65. [35] Pfizer’s first offer in September 2025 was nearly $4.9 billion. [35] Metsera chose Pfizer partly because the Federal Trade Commission had raised antitrust risk on the competing Novo Nordisk bid. [35] BioSpace valued the upfront payment plus CVR at $9.8 billion, based on S&P Capital IQ. [36] Pfizer’s chief executive said the therapies could reach the market as soon as 2028. [36]
Analysis: the CVR is 23.9% of the $86.25 headline price ($20.65 divided by $86.25). About one quarter of the price depends on milestones. A Bernstein analyst, as reported by News Arena Network, said Pfizer would need to assume $11 billion of revenue by 2040, nearly double the current projections for the assets. [37] That is one analyst’s view of price against forecast.
Case 4: What did the Karuna and Cobenfy deal teach about add-on trials?
Bristol Myers Squibb closed the Karuna acquisition on March 18, 2024, for $14.0 billion in cash, or $12.9 billion net of cash acquired. It accounted for the deal as an asset acquisition because Cobenfy represented substantially all of the fair value. [38] In April 2025, the Phase III ARISE trial of Cobenfy as an add-on to atypical antipsychotics did not meet its primary endpoint. [39][43] The Cobenfy group improved by 14.3 points on a 210-point symptom scale, about two points better than placebo, and the difference was not statistically significant (p=0.11). [40][43]
Analysts cut forecasts. Leerink Partners lowered its 2030 estimate from $5.8 billion to $2.6 billion, and Citi lowered its estimate by $600 million to $2.8 billion. [39] Raymond James cut its estimate from a little over $3.7 billion to a little under $3 billion. [40] Analysis: Leerink’s cut is 55%, and Citi’s cut is about 18%. The range of reactions shows how much of the value rested on one indication.
Cobenfy had $27 million in net sales in the first quarter of 2025. [41] BMS later said it found irregularities at a small number of sites in the Phase III ADEPT-2 study of Alzheimer’s psychosis, and that results from the ADEPT program were expected by the end of 2026. [42] Check current sources for the latest status.
Case 5: How is Merck managing the Keytruda exclusivity clock?
FDA approved Keytruda Qlex, a subcutaneous form of pembrolizumab with berahyaluronidase alfa-pmph, on September 19, 2025, for solid tumor indications approved for intravenous Keytruda. [27] In the trial that supported approval, the confirmed objective response rate was 45% for the subcutaneous form and 42% for the intravenous form. [27]
Merck’s SEC filing says it holds an exclusive license for Alteogen’s berahyaluronidase alfa. Merck owes regulatory milestones of $25 million and, if the product reaches certain sales levels, up to $680 million of sales-based milestones, and Merck estimated that the related intangible asset has a useful life through December 2030. [28] CNBC reported in January 2024 that a new form, if approved, might give Merck a separate patent and extend market exclusivity by several years. [25] We did not find a source that states the final effective exclusivity date for the combined product. Check the Purple Book and the patent record before you model a date.
Build or Buy? What Does the Evidence Say About M&A and Portfolio Strategy?
What does Deloitte recommend on M&A?
Deloitte recommended that companies prioritize smaller, early-stage deals instead of large megadeals. [15] Its survey found that 77% of surveyed pharma executives expected more M&A activity in 2025 as they looked to offset revenue lost to expiring patents. [14] Deloitte’s partner Colin Terry said that keeping the current approach to pipeline strategy will limit returns on R&D investment, according to a Deloitte release. [13]
How do the case studies compare on cost and risk?
Decision
Type
Stated size or result
Source
AstraZeneca 5R framework
Internal process change
4% to 19% success, nomination to Phase III completion
[20]
Pfizer, danuglipron
Stop
Program ended April 14, 2025; more than 1,400 participants in safety database
[33]
Pfizer, Metsera
Buy (late and early stage)
Up to $86.25 per share, $20.65 as CVR
[35]
BMS, Karuna
Buy (single lead asset)
$14.0 billion; ARISE add-on trial missed
[38][39]
Merck, Keytruda Qlex
Lifecycle management
Approved September 19, 2025; useful life of intangible through December 2030
[27][28]
Analysis: in three of the five cases (Karuna, Metsera and Qlex), one asset or one formulation carries most of the stated value. The cases do not prove that concentration is wrong. They show that the stop rule, the exclusivity date and the price date all need to be written for that asset before the deal closes.
What This Means for Brand Manufacturers, Generic Entrants and Biosimilar Developers
What this means for brand manufacturers
Put the three clocks in one model for every asset. Use a second success-rate source and record the difference in value, as the rNPV scenario does. Write the stop rule before Phase II data arrive. Where a lifecycle product such as a new formulation is part of the plan, find the patent and exclusivity record that supports the date. Do not rely on press statements for the date.
What this means for generic entry
The 2029 dates for Keytruda and Darzalex are reported as expected US exclusivity dates. [26] A generic or biosimilar team should read the Orange Book or Purple Book entry and the patent litigation record for each product. The 30-month stay and Paragraph IV timing can move the real entry date. [23]
What this means for biosimilars
Many of the biggest products with upcoming expiry are biologics. [25] Biosimilar teams should track new formulations and delivery forms, because the Keytruda Qlex approval shows a brand company can launch a new form before the original product loses exclusivity. [27]
What this means for business development and licensing teams
Deloitte’s advice favors smaller, earlier deals. [15] The Metsera price included a CVR of up to $20.65 per share. [35] Contingent payments tie part of the price to milestones. They do not remove the need to test the underlying probabilities.
What Happens Next? Dates and Events to Watch
Which trial readouts and dates are announced?
BMS said ADEPT program results, including ADEPT-2, ADEPT-1 and ADEPT-4, were expected by the end of 2026. [42] Keytruda and Darzalex are expected to lose US exclusivity by 2029. [26] Merck amortizes the Keytruda Qlex intangible through December 2030. [28] These are announced or estimated dates. They are not confirmed outcomes.
Which policy questions are open?
The EPIC Act, which would set the same 13-year timeline for small molecules and biologics, was reintroduced in 2025. [31] Check Congress.gov for its current status. A change would move the price clock for every small-molecule asset in a portfolio.
Key Takeaways
The likelihood of approval from Phase I was 7.9% for programs tracked from 2011 to 2020. [1]
Phase II had the lowest transition rate at 28.9%. [2]
Analysis: Phase II removes about 37 of every 100 Phase I entrants.
Capitalized cost per approved drug was $2,558 million in the Tufts estimate, and 45.5% of that is the cost of capital (analysis). [8]
Forecast IRR for the top 20 biopharma late-stage pipelines was 7.0% in 2025, with a median of 4.0%. [10]
Morgan Stanley estimated $183.5 billion of annual sales exposed to loss of exclusivity through 2030. [24]
Small molecules face Medicare price negotiation eligibility after 9 years and biologics after 13 years. [29]
In our hypothetical model, changing one success-rate input moved rNPV by 26.3%, and halving years 10 to 13 revenue lowered rNPV by 18.1%.
The five case studies show one pattern: concentration on one asset needs a pre-written stop rule, exclusivity date and price date.
FAQ
What is the probability that a Phase I drug gets FDA approval?
It was 7.9% for programs tracked from 2011 to 2020. The 2006-2015 edition reported 9.6%. [1][2]
Which clinical phase has the highest failure rate?
Phase II has the lowest transition rate, at 28.9% in the 2011-2020 data. [2] Phase I loses more programs by count in our calculation (48 per 100 entrants against 37 per 100), but a Phase II failure follows a larger spend. [4]
How long does it take a Phase I asset to reach approval?
The average was 10.5 years in the 2011-2020 BIO data. [2]
How much does it cost to develop a new drug?
The Tufts study estimated $2,558 million per approved new compound, capitalized, in 2013 dollars. [8] Deloitte estimated $2.23 billion per asset for the top 20 companies in its 2025 analysis. [14] The methods differ.
What is a good return on pharma R&D?
The sources do not set a fixed hurdle. The Deloitte cohort had a 7.0% average and a 4.0% median forecast IRR in the 2025 cycle. [10] The Tufts study used a 10.5% real discount rate to capitalize costs. [8] The two measures differ in method, so they are not a like-for-like test.
How much pharma revenue is exposed to patent expiry?
Morgan Stanley put annual sales of products losing exclusivity through 2030 at $183.5 billion, with an industry average exposure of 38% of revenue. [24]
Does a new formulation extend a drug’s exclusivity?
It can, but each case needs patent and regulatory review. CNBC reported that a new Keytruda form might bring a separate patent and extend exclusivity by several years. [25] FDA approved Keytruda Qlex in September 2025. [27] We found no source for the final effective date.
What is the Inflation Reduction Act pill penalty?
It is the four-year difference in the time before Medicare price negotiation: 9 years for small molecules and 13 years for biologics. [29] One study estimated that equalizing the timelines at 13 years would raise approvals for the Medicare-aged population by 21%. [30]
Do patient-selection biomarkers improve odds of approval?
Programs with patient-selection biomarkers had a 15.9% likelihood of approval from Phase I, about twice the overall rate. [2]
What is the 5R framework?
It is AstraZeneca’s decision framework with five questions: right target, right tissue, right safety, right patient and right commercial potential. The company reported success from candidate nomination to Phase III completion rising from 4% to 19% after it adopted it. [20]
References
Informa Pharma Intelligence. (2023, February 7). Clinical development success rates and contributing factors 2011-2020: Download the report. CPHI Online. https://www.cphi-online.com/whitepaper/clinical-development-success-rates-and-contributing-factors-2011-2020-download-the-report/
Biotechnology Innovation Organization, Informa Pharma Intelligence, & QLS Advisors. (2021). Clinical development success rates and contributing factors 2011-2020 [Key findings, as reproduced by Qiuqiu Pharma]. China Healthcare. https://www.cn-healthcare.com/articlewm/20210508/content-1217774.html
Benchmarking R&D success rates. (n.d.). University of St. Gallen Alexandria repository. https://www.alexandria.unisg.ch/server/api/core/bitstreams/1869410b-3d44-4d2d-9127-89da153d1c0f/content
Valuing pharmaceutical drug innovations. (n.d.). arXiv:2212.07384. https://arxiv.org/pdf/2212.07384
Scannell, J. W., Blanckley, A., Boldon, H., & Warrington, B. (2012). Diagnosing the decline in pharmaceutical R&D efficiency. Nature Reviews Drug Discovery, 11(3), 191-200. https://www.nature.com/articles/nrd3681
Horgan, J. (2018). Is science hitting a wall? Part 2. Scientific American. https://www.scientificamerican.com/blog/cross-check/is-science-hitting-a-wall-part-2/
DiMasi, J. A., Grabowski, H. G., & Hansen, R. W. (2016). Innovation in the pharmaceutical industry: New estimates of R&D costs. Journal of Health Economics, 47, 20-33. https://doi.org/10.1016/j.jhealeco.2016.01.012
Knowledge Ecology International. (2016, April 12). KEI briefing note 2015:4: The Tufts estimates of the risk adjusted out-of-pocket costs to develop a new drug. https://www.keionline.org/23054
Deloitte. (2026). Measuring the return from pharmaceutical innovation 2026 (16th edition). Deloitte Italy. https://www.deloitte.com/it/it/Industries/life-sciences-health-care/perspectives/measuring-return-from-pharmaceautical-innovation-2026.html
Deloitte. (2026, May 13). Measuring the return from pharmaceutical innovation [Dbriefs webcast]. https://www.deloitte.com/us/en/dbriefs-webcasts/measuring-return-from-pharmaceutical-innovation.html
Deloitte. (2026, May 4). Sixteenth annual Measuring the Return from Pharmaceutical Innovation report [Press release]. PR Newswire. https://www.medicaldevicedirectory.com/company/938695/news/3806322/weight-loss-drugs-gains-lift-pharma-r-d-returns-but-bubble-risk-grows
Deloitte UK. (2025). Global pharma R&D returns rise as GLP-1 drugs help drive forecast growth [Press release]. https://www.deloitte.com/uk/en/about/press-room/global-pharma-rd-returns-rise-as-one-glp-drugs-help-drive-forecast-growth.html
Kaylor, A. (2025). Pharma R&D returns improve amid rising costs, competition. TechTarget Pharma Life Sciences. https://techtarget.com/pharmalifesciences/news/366621118/Pharma-RD-returns-improve-amid-rising-costs-competition
BioSpace. (2025). Pharma R&D returns grow again, but Deloitte warns progress is ‘fragile’. https://www.biospace.com/business/pharma-r-d-returns-grow-again-but-deloitte-warns-progress-is-fragile
Deloitte. (2025). Measuring the return from pharmaceutical innovation (15th edition). https://www.deloitte.com/us/en/Industries/life-sciences-health-care/articles/measuring-return-from-pharmaceutical-innovation.html
Deloitte. (2024). Unleash AI’s potential: Measuring the return from pharmaceutical innovation (14th edition). https://www.deloitte.com/ch/en/Industries/life-sciences-health-care/analysis/measuring-the-return-from-pharmaceutical-innovation.html
BioSpace. (2024). Pharma R&D ROI spikes in 2023 as ‘breakthroughs’ re-energize industry: Report. https://www.biospace.com/pharma-r-and-d-roi-spikes-in-2023-as-breakthroughs-re-energize-industry-report
Deloitte LLP. (2024). Methodology: Measuring the return from pharmaceutical innovation, 14th edition. https://www2.deloitte.com/content/dam/Deloitte/uk/Documents/life-sciences-health-care/deloitte-uk-measuring-return-from-pharmaceutical-innovation-methodolgy.pdf
Morgan, P., Brown, D. G., Lennard, S., Anderton, M. J., Barrett, J. C., Eriksson, U., … Pangalos, M. N. (2018). Impact of a five-dimensional framework on R&D productivity at AstraZeneca. Nature Reviews Drug Discovery, 17(3), 167-181. https://doi.org/10.1038/nrd.2017.244
Tsui, R. (2014, May 29). Improve R&D pipeline in big pharma with the 5R framework. NIH Center for Regenerative Medicine blog. https://crex.nih.gov/blog/2014/05/29/improve-rd-pipeline-big-pharma-5r-framework
U.S. Food and Drug Administration. (n.d.). Frequently asked questions on patents and exclusivity. https://www.fda.gov/drugs/developmentapprovalprocess/ucm079031.htm
Berkeley Center for Law & Technology. (2025). BPCIA and Hatch-Waxman Act [Presentation]. https://www.law.berkeley.edu/wp-content/uploads/archive/2025/09/BPCIA-and-Hatch-Waxman-Act_Berkely-Center-for-Law-and-Tech_2025.pptx.pdf
Fierce Pharma. (2024). Amgen, BMS and Merck have the most exposure as Big Pharma eyes $183B patent cliff: Analyst. https://www.fiercepharma.com/pharma/firepower-aplenty-and-patent-cliffs-ahead-time-right-ma-activity-report
CNBC. (2024, January 28). Big pharma is at a crossroads, as J&J, Merck and others prepare to lose heaps of revenue from blockbuster drugs. https://www.cnbc.com/2024/01/28/big-pharma-merck-bristol-myers-jj-prepare-to-lose-revenue.html
European Pharmaceutical Review. (2025). Next drug patent cliff to challenge Big Pharma strategy [Reporting GlobalData analysis]. https://www.europeanpharmaceuticalreview.com/?p=261834
U.S. Food and Drug Administration. (2025, September 19). FDA approves pembrolizumab and berahyaluronidase alfa-pmph for subcutaneous injection. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-approves-pembrolizumab-and-berahyaluronidase-alfa-pmph-subcutaneous-injection
Merck & Co., Inc. (2025). Form 10-Q for the quarterly period ended September 30, 2025. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/310158/000031015825000059/mrk-20250930.htm
BioSpace. (n.d.). Stakeholders hope bill will create IRA drug negotiation parity between biologics and small molecules. https://www.biospace.com/stakeholders-hope-bill-will-create-ira-drug-negotiation-parity-between-biologics-and-small-molecules
Therapeutic Innovation & Regulatory Science study on the Inflation Reduction Act and small-molecule investment. (n.d.). Becaris Publishing. https://becarispublishing.com/doi/10.5555/6da9ca39-6f22-44fc-a87e-0da178adb68e
PharmaVoice. (2025). IRA reforms: Small molecule fix penalty Trump. https://www.pharmavoice.com/news/ira-reforms-small-molecule-fix-penalty-trump/742239/
BioSpace. (2024). Drug companies, VCs rethink R&D strategies as IRA stands strong in court. https://www.biospace.com/article/drug-companies-vcs-rethink-r-and-d-strategies-as-ira-stands-strong-in-court
Pfizer Inc. (2025, April 14). Pfizer discontinues development of oral GLP-1 receptor agonist danuglipron [Press release]. Business Wire. https://www.drugs.com/clinical_trials/pfizer-discontinues-development-oral-glp-1-receptor-agonist-danuglipron-22002.html
Reuters. (2025, April 14). Pfizer ends development of weight-loss pill danuglipron. https://fidelity.com/news/article/company-news/202504140645RTRSNEWSCOMBINED_L1N3QR06A_1
Associated Press. (2025, November 8). Pfizer clinches deal for obesity drug developer Metsera after a bidding war with Novo Nordisk. https://www.clickondetroit.com/business/2025/11/08/pfizer-clinches-deal-for-obesity-drug-developer-metsea-after-a-bidding-war-with-novo-nordisk/
News Arena Network. (2025, November 8). Metsera agrees to Pfizer’s $10 bn bid. https://newsarenaindia.com/economy/metsera-agrees-to-pfizer-s-10-bn-bid/61648
Bristol Myers Squibb Co. (2025). Form 10-K for fiscal year 2024: Acquisitions, divestitures, licensing and other arrangements. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/14272/000001427225000039/Financial_Report.xlsx
Fierce Pharma. (2025). Bristol Myers’ schizophrenia drug Cobenfy stumbles as adjunctive treatment, denting blockbuster plan. https://www.fiercepharma.com/pharma/bristol-myers-schizophrenia-drug-cobenfy-stumbles-adjunctive-treatment-denting-blockbuster
BioPharma Dive. (2025). Cobenfy fails in Bristol Myers trial, prompting lower Wall Street forecasts. https://biopharmadive.com/news/cobenfy-bristol-myers-trial-failure-schizophrenia-adjuvant-wall-street/746073
BioSpace. (n.d.). BMS delays Cobenfy Alzheimer’s psychosis data after ‘irregularities’ in Phase III study. https://www.biospace.com/drug-development/bms-delays-cobenfy-alzheimers-psychosis-data-after-irregularities-in-phase-iii-study
BioCentury. (2025, April 22). Key BMS growth driver Cobenfy fails Phase III in adjunctive schizophrenia. https://biocentury.com/article/655714/key-bms-growth-driver-cobenfy-fails-phase-iii-in-adjunctive-schizophrenia
