Human and financial consequences of herding in oncology drug development: clinical trials of TIGIT inhibitors

WHAT IS ALREADY KNOWN ON THIS TOPIC Herding in oncology drug development involves five or more biopharmaceutical companies competing to modulate an identical drug target. Because of high rates of drug attrition during development, approval for clinical use of many of the competing drugs is unlikely.WHAT THIS STUDY ADDS For the first time, using an example, the study quantifies the human and financial resources used as a consequence of herding. Thousands of patients were enrolled in clinical trials of experimental drugs that will not be approved for clinical use, at a cost of billions of dollars.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Herding in drug development should be avoided as it concentrates resources at the expense of investment in a broader portfolio of oncology drug development programmes.Introduction Herding in drug development has been defined as the competition between five or more companies each attempting to find modulators of the same drug target. 1 Herding is particularly prevalent in oncology drug development where it is an increasing trend.1 Herding was described in the financial marketplace as ‘an intrinsic preference for conformity’ that is ‘influenced by others’ actions’ with ‘herds that charge into risky ventures without adequate information and appreciation of the risk-rewards trade-offs’.2Herding is especially prevalent in the field of cancer immunotherapy. For example, more than 180 companies are or were developing inhibitors of the PD1/PDL1 immune checkpoints (ICs).3 Upadhaya et al reported in 2022 that there were 4897 active clinical trials of inhibitors of PD1/PDL1 as single agents or in drug combinations.4 The high attrition rate in oncology drug development of around 95%5 6 suggests that only a few PD1/PDL1 inhibitors from the ‘herd’ will be approved for clinical use. Although clinical inactivity or toxicity is a major cause of attrition of the development of drugs, commercial and strategic considerations can also lead to drug withdrawal in a crowded market.7 We previously estimated the overall costs of oncology drug attrition in the pharmaceutical and biotechnology industry to be more than US$50 billion per year.8Although the approved inhibitors of PD1 and PDL1 are effective therapeutics against some cancers, they have had mixed results in others.9 10 Efforts to ameliorate the activity of PD1 and PDL1 inhibitors in combinations with other agents remain intense.11 Publications12 13 of preclinical data describing the role of TIGIT (T-cell immunoreceptor with Ig and immunoreceptor tyrosine-based inhibitory motif domain) and the augmentation of the efficacy of inhibitors of PD1 or PDL1 by additionally inhibiting TIGIT drew the attention of biotechnology and pharmaceutical companies.14–18 Multiple programmes were initiated to discover and develop TIGIT inhibitors, in a clear case of herding. As of October 2025, there are no TIGIT inhibitors approved by the Food and Drug Administration (FDA) or the European Medicines Agency (EMA) for clinical use despite earlier Breakthrough Therapy designations (eg, for tiragolumab in 2021)19 and optimistic projections. TIGIT remains an investigational IC target, with some candidates remaining in late-stage clinical trials, often combined with PD-1/PD-L1 inhibitors. The scale of the resources, human and financial, that were associated with the clinical trials of multiple TIGIT inhibitors has been estimated here.Methods Clinical trial data Two public databases were used to estimate the number of clinical trials and the number of patients enrolled in trials. A first search interrogated the ClinicalTrials.gov 20 database which cites estimated or actual numbers of patient enrolment, followed by a second search of the ICTRP database (International Clinical Trials Registry Platform) of the WHO which contains entries from 19 Registries world-wide, some of which do not appear in the ClinicalTrials.gov database.21 ‘Cancer’ and ‘TIGIT’ were used as the keywords for the searches together with drug names and/or compound numbers of TIGIT inhibitors (table 1). Data available between 23 May 2016 (NCT02794571) and 30 September 2025 were interrogated.Table 1The names or code numbers of the 30 TIGIT inhibitors whose 220 clinical trials are surveyed here, showing the companies developing the inhibitors, the total number of clinical trials and the phase of the trialsTIGIT inhibitorCompanyTotal number of trialsCurrent phase of trialsAK130 and AK 127Akeso7Phase 2Domvanalimab (AB308)Arcus/Gilead24Phase 3ASP8374Astellas Pharma3Phase 1(T)Rilvegostomig (AZD2936/COM902)AstraZeneca/Compugen28Phase 3Ociperlimab (BGB-A1217)BeiGene/Novartis13Phase 3(T)BAT6005/BAT6021Bio-Thera Solutions3Phase 1PM1009, PM1021, PM1022Biotheus/BioNtech4Phase 2BMS-986207Bristol-Myers Squibb4Phase 2BMS-9864422(T, T)JS006 (CHS-006)Coherus Biosciences/Junshi Bioscience2Phase 1Belrestotug (EOS448)GlaxoSmithKline/iTeos10Phase 3(T)HLX301, HLX53Henlius Biotech4Phase 2HB0036/HB0030Huaota Biopharmaceutical2Phase 2IBI939, IBI321Innovent Biologics6Phase 1SIM0348Jiangsu Simcere Pharmaceutical1Phase 1BC008-1ALuzhou Buchang Biopharmaceutical3Phase 1Vibostolimab (MK-7684A)Merck (MSD)18Phase 3(T)Etigilimab (MPH313)Mereo BioPharma2Phase 2SEA-TGTPfizer/Seagen2Phase 2(T)TiragolumabRoche-Genentech68Phase 3(T)Dargistotug (M6223)Serono/Merck2Phase 2ZGGS15/ZG005Zelgen Biopharmaceuticals12Phase 2TOTAL220‘T’ Denotes that inhibitor development has been terminated.TIGIT, T-cell immunoreceptor with Ig and immunoreceptor tyrosine-based inhibitory motif domain.Data on the patient numbers enrolled in individual phases of each trial, the trial location (country/countries), its start date, study status (eg withdrawn, recruiting) and the type of sponsor of a trial (industry or academic) were captured. For studies described as ‘withdrawn’, zero patient enrolment was accredited.For multinational trials, involving the participation of two or more different countries, unless the public databases declared the distribution of patient numbers enrolled in each of the multiple countries, we had to assume an equal distribution of patients between different countries for the purpose of estimating cost.For a ‘basket’ trial (NCT01042379) with multiple treatment arms, comparing a variety of different treatments, with only one arm containing a TIGIT inhibitor, the total number of patients in the trial was arbitrarily divided by the number of treatment arms to give an average number of patients per arm. This calculated average patient number for a single arm was used to estimate the number of patients the arms containing a TIGIT inhibitor together with one comparison arm.Cost estimates of clinical trials The estimated cost of clinical trials of TIGIT inhibitors was based on published data by Sertkaya et al22 of the cost per patient in the different phases of oncology clinical trials in the USA in 2000–2018 performed by the pharmaceutical industry. Their estimates included both the direct costs and indirect costs, as described earlier by Sertkaya et al.23 The costs of oncology clinical trials were then adjusted for inflation to dollar values in 2025 using the US consumer price index for medical care,24 giving costs in 2025 estimated to be US$122 798 (phase 1), US$93 578 (phase 2) and US$110 679 (phase 3). This permitted an estimate of the cost of clinical trials of TIGIT inhibitors performed solely in the USA by industry. The estimated per-patient cost in countries outside of the USA performed by industry was calculated as a percentage of USA costs using data from Qiao et al,25 namely: Africa 49%, Central Europe 50%, Middle East 53%, Latin America 59%, Western Europe 74%, Asia 80% and Oceania 97%. For trials performed in multiple countries, the global mean of 0.7 or median 0.6 of the values cited by Qiao et al25 was used to calculate costs.Trials that were collaboratively performed by academic institutions with pharmaceutical companies were classified as industry funded. Industry-sponsored trials were distinguished from lower-cost, purely academic-sponsored trials such as those performed by the USA’s National Institutes of Health (NIH) or the National Cancer Institute. In the study of Zhou et al, of all classes of drugs, costs of NIH clinical trials were estimated to be approximately 17% of those of industry, between 2010 and 2019.26 As there are no data on the cost of academic trials performed in different academic centres around the world, we chose to estimate a range of costs arbitrarily based on 0.5× or 0.1× the mean cost of industry-sponsored trials in the USA, similarly to estimated costs made by Jentzsch et al.8Results Table 1 shows the names or code numbers of the 30 different TIGIT inhibitors, and the companies developing them, as per clinical trial data found either in ClinicalTrials.gov20 or the ICTRP database.21 The total number of clinical trials was 220, not counting 10 trials that were withdrawn (figure 1A). Development of 8 (27%) of the 30 TIGIT inhibitors has so far been either terminated for failure to show significant clinical activity or for strategic reasons.27–29Figure 1(A) Percentage of patients in different stages of clinical trials of TIGIT inhibitors as of 30 September 2025, calculated from online supplemental tables 1–4. (B) Percentage of patients in clinical trials of TIGIT inhibitors terminated because of insufficient clinical activity indicated in table 1, calculated from online supplemental tables 1–4. (C) Cost of terminated clinical trials that demonstrated insufficient clinical activity taking an average estimated total cost of trials to be US$3.35 billion (US$3.1bn–US$3.6bn), calculated from online supplemental tables 1–4, using 0.7 and 0.5 of USA estimated costs for ex-USA trials and academic trials respectively, as described in Methods. TIGIT, T-cell immunoreceptor with Ig and immunoreceptor tyrosine-based inhibitory motif domain.SP110.1136/bmjonc-2025-001037.supp1Supplementary dataOnline supplemental table 1 shows the details of the 88 industry-sponsored clinical trials of TIGIT inhibitors which were performed in a single country/region, as detailed in ClinicalTrials.gov.20Online supplemental table 2 shows the data for 12 trials obtained from the ICTRP database21 which were not found in ClinicalTrials.gov. These are a mixture of industry-sponsored trials and academic-sponsored trials performed in different countries.Online supplemental table 3 shows 107 industry-sponsored trials that were multinational,20 where costs were estimated as either 0.7 or 0.6 of the costs of USA trials.Online supplemental table 4 shows details of the 22 trials that were sponsored by academic centres alone.20In trials with multiple arms, the number of patients enrolled in each arm containing a TIGIT inhibitor were added to the number of patients in one appropriate control arm. These trials were: online supplemental table 1 NCT04486352, NCT04632992; online supplemental table 3 NCT05116202, NCT03819465 and NCT02964013; online supplemental table 4 NCT01042379.It is estimated that between 23 May 2016 and 30 September 2025, 48 731 patients were enrolled in trials of the TIGIT inhibitors. Where development programmes have been terminated (table 1, T=Terminated) it was estimated from online supplemental tables 1–4 that 14 901 patients had been enrolled (figure 1B).The estimated total costs of purely industry-sponsored trials (n=92) performed in a single country (online supplemental tables 1 and 2) were US$570 million (US$569 588 642). For trials involving multiple countries (n=111) (online supplemental tables 2 and 3) at 0.7 of USA costs was US$2.99 billion (US$2 988 465 642) and at 0.6 of USA costs was US$2.56 billion (US$2 562 065 978). Trials in academic centres (n=27) (online supplemental tables 2 and 4) were estimated at US$58 million (US$57 832 956) at 0.5 of industrial costs and US$12 million (US$11 566 591) at 0.1 of industrial cost.The total minimal clinical trial cost estimate is US$3.1 billion and at maximal cost US$3.6 billion. It was estimated that the costs of terminated trials (to 30 September 2025) due to insufficient clinical activity (table 1) were US$129 million (online supplemental tables 1–4 and figure 1C).Discussion The development and clinical trials of the 30 TIGIT inhibitors ( table 1) constitute a clear case of herding (>5 companies involved1) around a single drug target by 21 competing biotechnology and pharmaceutical companies. According to the two public databases,20 21 220 trials enrolled nearly 49 000 patients (online supplemental tables 1–4), with some TIGIT inhibitors undergoing more than 20 clinical trials (table 1). With an estimated 95% attrition rate for drug development in oncology,5 6 and with attrition (27%) of some TIGIT inhibitors underway27–29 (table 1, figure 1), it is unlikely that many TIGIT inhibitors will ultimately receive approval for use in oncology practice, with thousands of patients having been or being exposed to experimental drugs that will not be or are unlikely to be approved. The risks of drug toxicity in ineffective clinical trials of this magnitude have raised moral questions concerning the role of patients in such trials.30 Herding also risks diminishing the pool of patients entering oncology clinical trials. This pool is limited in the USA because of continuing complex barriers to trial entry.31 However, as seen in online supplemental table 3, clinical trials of TIGIT inhibitors have followed the trend of becoming fully globalised in the search for patients.32 Globalisation of trials also poses ethical questions, especially with respect to trials performed in low-income and middle-income countries33. Although not the goal of these multiple trials, some patients in these countries may benefit from a treatment arm containing an effective PD1 inhibitor, a drug class that, for economic reasons determined by the pharmaceutical industry, is normally inaccessible to them. Finally, herding in oncology drug development, if multiple successful candidates emerge, will amplify regulatory and reimbursement delay by congesting the review process.Addressing the cost of these 220 trials: the overall costs of drug development are controversial.22 34 35 Our estimates of clinical trial costs alone were based on a comprehensive study by Sertkaya et al22 and are intended here to provide a representation of the scale of costs of a project characterised by herding. The estimated cost of the clinical trials of the 30 TIGIT inhibitors lay between US$3.1 billion and US$3.6 billion. In a blog account of an analyst’s report, whose details have not been published in the public domain, the total cost of current TIGIT inhibitor development was claimed to be US$6 billion.36 This greater sum presumably includes the cost of capital, preclinical studies, antibody production and of in-licensing costs incurred by some companies. Our estimate of clinical trial costs is similarly substantial, raising concerns about the cumulative losses if prediction of further attrition is confirmed.36In the world of finance, ‘herds’ were said to ‘charge into risky ventures without adequate information’.2 Was there adequate information to support multiple programmes to develop TIGIT inhibitors? Understanding of the complexity of the immune system’s response to cancer and the factors that control ICs continues to evolve.9 An example of that complexity, and of the uncertainty surrounding the biology of ICs that impacts on TIGIT inhibitors, is illustrated by the active debate as to whether a functional IgG Fc domain in the inhibitor antibody is required15 28 37; the eight inhibitors whose development has been terminated were all Fc-enabled (table 1). There are ongoing trials of inhibitors which are Fc-silent (table 1: Domvanalimab (AB308), BAT6021 and Rilvegostomig (AZD2936/COM902)). Future reports of their clinical activity from phase 3 trials may answer this debate.Currently, limited knowledge of IC function also restricts the choice of predictive biomarkers that can guide the design of the trials of IC inhibitors.10 Critical commentary on the clinical trials of TIGIT inhibitors wrote that, in the absence of those predictive biomarkers, the current approach was to ‘throw spaghetti at the wall and see what sticks’.38 Second, as a potential cause of attrition, the quality of preclinical models which supported the launch of clinical trials of TIGIT inhibitors should be questioned. Seven TIGIT inhibitors (all Fc-enabled) did not recapitulate in clinical trials the preclinical promise seen in murine models, whereas Fc-silent TIGIT inhibitors were observed to be less effective in preclinical models but appear to have clinical promise.28 The fidelity of preclinical models to capture a complex pathology is critical to subsequent clinical success; clinical failure requires that preclinical models are re-examined for their ability to mimic complex human pathologies.39What drives the creation of a bioscience ‘herd’? The publication of ‘breakthrough’ science on the mechanisms of action of ICs will attract widespread interest and stimulate engagement to investigate their therapeutic potential. While the science describes advances in understanding of biology and pathology using laboratory models, their fidelity to capture the complexity of human pathology is uncertain, imposing significant risks to the search for pharmacological intervention. But, by joining a ‘herd’, that risk may be perceived to be diminished, as a project achieves apparent ‘conformity’,2 or mainstream status as the numbers in the herd increase, especially if the herd is populated by actors with long track records of successful drug discovery. Evolution towards project ‘conformity’ may also reassure investors in fledgling biotechnology companies. However, the high risks for attrition remain, and in a herd, the number of failures will be multiplied, as described here, elevating the loss of both human and financial resources.Peck et al40 have analysed the reasons why failing projects are so hard to terminate, writing ‘The association of project terminations with ‘failure’ leads organisations to want to forget them’. Peck et al and Scannell et al39 40 suggest that careful review of ‘failure’ is informative. Important preclinical and clinical data could be shared federally, within a confidentiality framework, to learn lessons, reduce waste and improve productivity, as described in Astra Zeneca’s landmark analysis of their failures.41While the pharmaceutical and biotechnology industries should explore novel avenues to discover innovative therapies in oncology, efforts to reduce high levels of attrition are required with particular attention paid to the fidelity of preclinical models to recapitulate human disease.39 Because of significant attrition, herding around a single drug target multiplies the potential loss of resources, both human and financial. There are ethical questions about the enrolment of thousands of patients in multiple trials due to herding, trials that are unlikely to result in the approval for clinical use of experimental drugs. The significant human and financial resources invested in cases of herding in oncology drug development could better be used in a broader portfolio of projects, spreading risk, to deliver effective ways to reduce cancer mortality.This study has limitations. The number of patients enrolled in clinical trials of TIGIT inhibitors were sourced from publicly available clinical trial databases with global coverage. It is possible that during the course of clinical trials, these numbers varied, although we updated published estimates20 21 in September 2025. It is also possible that data from a regional database was missed.For the costs of clinical trials, several assumptions were made due to the lack of public accessibility to data on individual clinical trial costs. Estimates, and not precise costs of clinical trials, were based solely on the mean cost of each phase of oncology clinical trials performed in the USA by the pharmaceutical industry in 2000–2018, as calculated by Sertkaya et al,22 adjusted for inflation to costs in 2025. The estimates of Sertkaya et al22 23 are open to challenge, although detailed mean costs of each phase of oncology trials have not, to our knowledge, been published since. For trials performed outside of the USA, estimates of fractions of USA costs of industry-sponsored trials were based on the country-wide estimates by Qiao et al25 in 2015–2016; these fractions may have altered by 2025. The study of Qiao et al might overestimate the per-patient cost in Asia in their baseline estimation, reflecting the advantage of scale. In addition, this study applied uniform regional ratios to all countries within a region, which may not reflect country-specific cost differences. When trials were undertaken in multiple countries, the total number of patients enrolled, as enumerated in the public databases,20 21 was arbitrarily divided equally between the different countries unless specific allocations to specific countries were noted. This potentially impacts on cost estimations. There is scant data, in the databases or subsequent publications of clinical trials, on the precise distribution of patient numbers in trials performed in multiple countries and regions. For trials performed purely by academic institutions, as in Jentzsch et al,8 two fractions of the cost of trials performed in the USA by industry were arbitrarily taken (0.5 and 0.1) giving an estimated range of costs as a fraction of those of industry.SP210.1136/bmjonc-2025-001037.supp2Supplementary data