Keytruda R&D costs

Weitere Informationen

• Methodology

  - Details anzeigen

  Persons in charge: Patrick Durisch / Gabriela Hertig

  Date: February 2026

  Methodology

  The Research & Development (R&D) costs spent out-of-pocket by Merck Sharp & Dohme (MSD) for pembrolizumab (brand name: Keytruda) were estimated by relying on a similar methodology used for a Public Eye report entitled “Properly unhealthy: Big Pharma rakes in 40–90 % profit margins” (2022). 

  The calculation is based on the global cumulative costs of relevant clinical trials (CTs) sponsored by MSD, for all approved indications until marketing approval, with independent and verifiable data. These CT costs represent 70% of R&D costs, according to the literature. 

   

  Selecting the relevant clinical trials sponsored by MSD

  To find the number of clinical trials (CTs) on pembrolizumab for all approved indications, we first broadly searched clinicaltrials.gov as well as all the European Medicines Agency (EMA) and the U.S. Food and Drug Administration (FDA) approval documents (EPAR and Drugs@FDA database). We then applied different selection criteria to end up with the relevant numbers of CTs sponsored by MSD used for the estimation of out-of-pocket R&D costs.

  Following is a summary of the search steps, ending up in our final selection of CTs:

  Clinicaltrials.gov	Keywords and filters	Results
  Keywords	‘pembrolizumab’ (the system also automatically searches for ‘Keytruda’, ‘MK-3475’, ‘lambrolizumab’ and ‘anti-PD-1 monoclonal antibody MK-3475’)	2770
  Applied filter	Sponsor (lead): Merck Sharp & Dohme LLC	268
  Adding	One EMA & FDA-referenced pivotal trial sponsored by the U.S. National Cancer Institute, financially supported by MSD	269
  Excluding	Phase IV trials	267
  Applied filter	Status: ‘completed’, ‘active not recruiting’, ‘suspended’, ‘terminated’	218

   

  Since we wanted to focus on CTs that served for the approval of Keytruda indications until the end of 2025, we excluded the following trials:

  • 2 phase IV CTs (post-approval trials)

  • 49 more recently launched CTs with status ‘not yet recruiting’ & ‘recruiting’, as well as those with status ‘no longer available’ and ‘withdrawn’.

  This resulted in a final selection of 218 MSD-sponsored CTs. Out of these, we distinguished three categories as a basis for different R&D costs estimates, each marked with different colors:

  • Green, 90 trials: These are the CTs that match the approved indications. Most of them are directly referenced in EMA and FDA approval documents. Additionally, it includes CTs selected from the clinicaltrials.gov search that match the approved indication exactly (drug combination, line of treatment, treatment plan (adjuvant or other), patient population). We included also one CT sponsored by the U.S. National Cancer Institute (NCI) for which MSD provided funding and support through a Cooperative Research and Development Agreement (CRADA).

  • Orange, 69 trials: These are additional CTs that match approved indications, but not exactly. They test for an approved indication but (i) with a difference of mono-/combination therapy (ii) with a different drug combination, (iii) a different line of treatment, or (iv) a different treatment plan (adjuvant or others).

  • Red, 59 trials: These are CTs that (i) did not test an already-approved indication, (ii) a specific patient group (often Chinese extension studies) and/or which were not relevant for the EMA/FDA approvals, or (iii) were terminated. 

  • Phase I/II or phase II/III CTs were categorized as the higher phase (i.e. “in favor” of pharma, as CT costs increase by phase).

  Cutoff dates:

  • Approved indications 31.12.2025

  • EMA and FDA approval 31.12.2025

  • Clinicaltrials.gov search 31.12.2025

   

  Average costs per trial and phase

  As in our estimate 2022, we used average costs per oncology CT from Sertkaya et al. 2016. For a characterization and discussion of their data set, see Public Eye 2022, p. 23.

  • Phase I: 4,500,000 US$

  • Phase II: 11,200,000 US$

  • Phase III: 22,100,000 US$

   

  30% additional costs for other R&D expenses

  See Public Eye 2022 p. 14 and footnote 85.

  These 30% additional costs are generous for the case of Keytruda, for several reasons: (1) preclinical costs invested out-of-pocket by MSD were limited since it acquired pembrolizumab (2009) – the discovery and foundational research (e.g. the humanized version of pembrolizumab) was done by others, largely publicly-funded (2) the CT costs estimate is already generous since average cost per trial used was meant for a classical procedure, not for follow-on indications of the same drug. Keytruda benefitted from shorter trial duration and accelerated approvals, which reduced its R&D costs – see e.g. Whitacre 2024.

   

  Risk adjustment

  See Public Eye 2022, p. 22-23 for a more detailed elaboration on the probabilities of success. In our work in 2022, we used the oncology-specific probabilities of success (POS) for different trial phases. After further discussion with experts, the current estimate uses industry-wide averages for the different trial phases. These are higher than oncology-specific POS, see Wong et al. 2019, meaning CTs are on average more successful than oncology trials, hence the risk-adjusted estimated R&D costs are lower. Big companies like MSD do have a diversified pipeline, and we would only use disease-specific POS for a company specialised in cancer therapies that had little else in its pipeline. This methodological choice is also linked with the nature of the CTs (extension/new indications of the same drug rather than new molecular entities).

  For this study, we used the following POS:

  • Phase I: 13.8%

  • Phase II: 21.0%

  • Phase III: 59.0%

  Note, that 13.8% is within the range of what Interpharma (the Swiss pharma trade association) argues in terms of successful drug candidates. “Only 1 out of 10 drugs does make it from phase I to approval”, see Vom langen Weg zum neuen Medikament (only in German).

   

  Cost of capital

  Our estimate is without costs of capital. See Public Eye 2022 p.14 and Annex p. 23 for the argument.

   

  Results

  We did three different calculations which differ in the number of CTs as the basis of the estimates as well as in the use of success rates.

  Estimate 1: Based on all MSD-sponsored trials for EMA and FDA approved indications (color code green). The probabilities of success account for failures (CTs done for unapproved indications and other unsuccessful drug candidates). Even if MSD had a higher success rate (see Estimate 3), they may not get so lucky next time, so in this estimate we reward them at the “average”. Note that we did not distinguish between POS for lead vs. supplemental indications. In Wong et al. study, POS for lead indications are higher, i.e. they are on average more successful than supplemental indications, hence with such distinction the risk-adjusted estimate would be a little lower. However, it is debated, whether it makes sense to apply this distinction, see Wong et al. 2019 Supplementary Data.

  Result non-risk-adjusted: 1,925,300,000 US$

  Result risk-adjusted: 4,789,613,854 US$

  The US$ 1.9 billion (non-risk-adjusted) estimate is very generous for the reasons exposed above (shorter trials, accelerated approvals, higher average phase costs used than for indication extensions). 

  The risk-adjusted figure is only calculated for the sake of demonstration, highlighting that if we applied such a failure ratio, R&D costs would still only represent a tiny fraction of global revenues (just about 3%).

   

  Estimate 2: Based on all MSD-sponsored trials for EMA and FDA approved indications (color code green plus orange). The additional CTs (orange) match the approved cancer type but vary in comparison to the exact approved indication. These costs are not risk-adjusted as this would double-count CTs that do not exactly match the approved indications.

  Result non-risk-adjusted: 3,284,320,000 US$

   

  Estimate 3: Based on all MSD-sponsored trials for pembrolizumab (color code green plus orange plus red), i.e. those trials that can be found through a clinicaltrials.gov search (see above). This estimate does not use success rates as these are the actual CTs done for pembrolizumab, including those that were “unsuccessful” and did not lead to an approved indication. It is therefore no coincidence that Estimate 1 and Estimate 3 are very similar. In fact, MSD was apparently “more successful” than industry-wide averages, i.e. what is estimated with Estimate 1.

  Result non-risk-adjusted: 4,366,960,000.00 US$

   

  For any query, please contact: patrick.durisch@publiceye.ch, +41 21 620 03 06

  Bereich Schliessen