Public Sector Innovation and the Constraints of ‘Platform Thinking’: An Account of Johnson & Johnson's Adenoviral Vector Vaccines

Throughout January 2024 the World Health Organization (WHO) held a series of consultations about its R&D Blueprint for Epidemics, a global research strategy established in 2016 in the wake of the West African Ebola epidemic. Hundreds of international virtual participants were reminded that the goal of the first consultation, "A Scientific Framework for Epidemic and Pandemic Preparedness," was not to discuss "national initiatives" or "global governance," but rather "how […] to better coordinate research and […] the things of research that we need to focus on" (WHO, 2024). Prominent virologist and inventor Barney Graham then described the WHO-endorsed "prototype pathogen approach for pandemic preparedness." Graham described, among other elements of the approach, the need for coordination and communication between different governments and government agencies making decisions in response to the next pandemic. Issues of governance, politics, and economics, initially declared outside of scope, were raised by multiple speakers in addition to Graham, despite the organizers' efforts to partition "science" from "politics" in their initial framing. A struggle over whether to keep discussion of "the science" separate from governance, policy, and practice persisted throughout the consultation.

The second consultation centered on "Critical Research for Priority Pathogens with Epidemic Potential" while the third, one day later, on "Research Response to Pathogen X During a Pandemic" featured a panel of experts discussing the future of a variety of vaccine "platforms." These "platforms" included mRNA, nanoparticles, protein subunits, DNA, live attenuated and codon deoptimized vaccines, and viral vectors. One panel member argued that computationally designed protein subunits should also be considered a "platform." The questions posed to panelists were, "What platforms have potential to be useful in a pandemic and why? How can existing platforms be improved upon?" The panelists included scientists with both public and private funding, working under a variety of different regulatory regimes around the world. The panel skirted foundational questions: What exactly is a vaccine "platform," and what is its role in science and society?

Biomedical research is itself a product of society and its political economy. Intellectual property and other mechanisms turn scientific knowledge into assets through a process of "assetization," and the value attached to these assets accumulates and circulates based on a social appraisal of their future earning potential (i.e., capitalization) (Birch, 2016). Western and especially American research institutions both public and private exert influence globally in the management of this value as it circulates (Kleinman and Moore, 2019). Scientists move between the academic, government, and industrial research as well as clinical institutions. All these institutions are governed as a part of a Western-dominated global world order (Sampat, 2023). These dynamics form the backdrop for the WHO convening, attended by powerful actors in global health and biomedical research.

Work on the political economy of mRNA, embodied in the famed and lucrative National Institutes of Health (NIH)-Moderna and Pfizer-BioNTech COVID-19 vaccines, has been central in investigations of the role of vaccine "platforms" in biomedical research, pandemics, and society more broadly. In The Messenger, Wall Street Journal reporter Peter Loftus describes many of the investors and executives behind the development of the NIH-Moderna vaccine -"the business gamble that changed the world" (Loftus, 2022). Others frame the history of mRNA-based COVID-19 vaccines in terms of the private appropriation of publicly funded knowledge. In the wake of Moderna's steep price increases on the NIH-Moderna vaccine, Sarpatwari (2023) and Morten (2023) argued that Moderna contributed relatively little scientific knowledge and yet went to great lengths to claim the full value of the vaccine. Dosi (2021) extends this analysis to broader dysfunction in public-private dynamics shaping not only the pandemic response but "innovation" more generally. A public option for biomedical research (Ramachandran, 2023) and university tech transfer policy reform (Herder et al., 2022) has been raised as a potential solution. Many scholars have explored injustice embedded in the political economic dimensions of the pandemic response, including the distribution of biochemical infrastructure and manufacturing capacity (Halabi et al., 2022;Jensen et al., 2022;Herder and Benavides, 2024), intellectual property (Benavides, 2023;Moon et al., 2021;Panagopoulos and Sideri, 2023), human rights violations (Joseph and Dore, 2022), and broader dynamics of structural violence (Gleeson et al., 2023). These and many other analyses (Kumar et al., 2022;Lalani et al., 2023;Niazi, 2022;Okuyama, 2022;Paremoer and Pollock, 2022) focus on mRNA.

Science and technology studies offers helpful tools for understanding the "platform" in vaccines and life sciences more broadly. Birch (2016) theorizes a process of assetization in which biological processes and products acquire their market value through life sciences firms acting principally as financial organizations rather than producers of scientific technology. The vaccine "platform" then is not only a biological product or tool; it is a biological product or tool made legible and valuable to lawyers, financiers, and other non-scientist actors through marketing and various legal devices (e.g., patents, procurement contracts, approval by a regulator). The process of assetizing a biological tool into a financially valuable "platform" may necessitate simplification or even obscuration of underlying scientific complexity and uncertainty.

If vaccine manufacturers have a chiefly financial function, no discussion of vaccine development should disregard political economy. However, in describing the scope of the WHO consultations, Dr. Cristina Casetti of the National Institutes of Allergy and Infectious Diseases (NIAID) attempted to separate science and the political economy of science: "We're not going to talk about global governance, this is a scientific discourse."

The WHO consultations underscore the fact that pandemics are political, social, economic, technical, and natural phenomena (Latour, 1993). The controversy over credit for and control over the NIH-Moderna vaccine, for example, is just one opportunity to understand the inseparability of these aspects of drug development. The science of mRNA, nanoparticles, or any vector does not advance or become useful without the social. The need to build bridges between the artificial dichotomy of social and technological responses was evident throughout the COVID-19 pandemic (Sariahmed et al., 2023), and well before (Ryan et al., 2019;Graham et al., 2018;Holmes et al., 2010, Jones andGraham, 2009). Viewed alongside these controversies, the WHO consultation highlights the tension between those who wish to paint science in purely technical terms and those who surface its political, economic, and social dimensions.

The scholarly attention paid to mRNA is warranted, but there was another blockbuster "platform" at the center of the COVID-19 pandemic. That other "platform" was "AdVac," a viral vector based on recombinant adenovirus type 26 (Ad26). Patented and made "proprietary" by Johnson & Johnson (J&J) (Johnson and Johnson, 2011), the vector has been used to develop vaccine candidates for many different pathogens, but its best-known application was in COVID-19 (Sadoff et al., 2021). The J&J COVID-19 vaccine was originally approved as a single-shot, a relative advantage in achieving wide distribution (Hardt et al., 2022). Use of the J&J COVID-19 vaccine was eventually limited by rare but deadly cases of thrombosis with thrombocytopenia (TTS) (FDA, 2021), but it was still commercially significant, with tens of millions of doses administered (Johnson and Johnson, 2022) and hundreds of thousands of lives saved (The Economist Newspaper, 2022).

Besides the COVID-19 context, J&J's AdVac has a longer history of application against HIV, Ebola, RSV, and Zika. J&J is one of the world's largest pharmaceutical manufacturers by revenue (Dunleavy, 2024), giving it influence in the conception and function of vaccine "platforms" more generally. This confluence of technological potential and capital makes J&J's AdVac a potentially instructive case study for understanding the role of the "platform" in modern vaccine development. Yet we are aware of only one other political economic analysis that analyzes the J&J COVID vaccine in any detail, and it focused only on manufacturing (Bown and Bollyky, 2022).

In this paper, we examine the scientific and political-economic history of J&J's AdVac "platform" from the early 2000s to 2023. We base our examination on a series of interviews with key scientists in industry, government, and academia involved in its development. We show that public laboratories-the NIH especially-made essential yet overlooked contributions to the science supporting the "platform" and to viral vector vaccines more broadly. Our examination suggests that the specifics of J&J's patent portfolio and its mastery in manufacturing one particular viral vector-Ad26-may have led it and NIH to overinvest in that specific vector. The story of AdVac and its corresponding vector-Ad26-is one of great ups and downs: Early promise, mixed success driven by enormous public investment, and ultimately highprofile disappointment. Ad26-based vaccine candidates ultimately proved suboptimal against a range of pathogens, contributing to J&J's decision in 2023 to end most of its vaccine research. Ad26 seemed to have all the necessary characteristics of a vaccine "platform" and yet failed as such.

Through our analysis, we develop the notion of "platform thinking." To do so, maintaining a distinction between "vectors" and "platforms" is necessary. Viral vectors describe a variety of modified viruses used to transmit genetic material. We use "vector" in the technical, scientific sense, to refer to the tool that delivers biological material to cells, and not to the tool's context. By contrast, "platforms" emerge from the social, political, and economic context in which vectors operate. If the vector is part of the economic base in a Marxist sense, along with the supply chain that produces it and the people who developed and manufacture it, the "platform" is superstructure; the "platform" is the vector as understood, characterized, assetized, and reshaped by lawyers and capitalists.

The Ad26 vector, once branded as AdVac, became the object of marketing, financial speculation, legal control through patenting, and certain totalizing, all-or-nothing aspirations-in short, platform thinking. J&J's AdVac "platform" vied for a time with mRNA and other vaccine "platforms" for investment, regulatory approval, media attention, patient and prescriber demand, and so on. The story of AdVac outlines the contours of platform thinking and its potential consequences for science and public health.

In the broad science and technology studies literature on the production of value by life sciences firms, our theoretical perspective is closest to Birch's work on "assetization" (Birch, 2016), and we share a focus on political economy. However, our subject of focus is a large, powerful, productive firm in a field where most biotechnology firms are more thoroughly speculative, capitalizing assets without any commodity production. We approach the scientists and scientific organizations in and around J&J as actors with political economic agency-with agency to translate science into medical commodities marketed to and used by millions of people-although they are of course constrained by a wider milieu in which value is managed by the "assetization" and subsequent capitalization of knowledge, where "speculative" assets abound but are only occasionally realized as commodities (Birch, 2016).

We conducted a critical, qualitative study of the recent history of adenoviral (Ad) vector vaccines, with a focus on J&J's AdVac platform. Our inquiry relied on the triangulation of key patents, scientific literature, press releases, and other documentation, as well as qualitative interviews with key scientists. Study authors have expertise in qualitative methodologies with particular experience in clinical medicine, patent analysis, and social science and law. We report our methods using the Standards for Reporting Qualitative Research (O'Brien et al., 2014) (Appendix 1).

We centered patents in our approach because they capture an important intersection of the social, legal, and technical aspects of vaccine development, and patents are filed by inventors spanning industrial, academic, and government laboratories. Patents are legal documents that confer important, sometimes highly valuable exclusive rights; their attribution of credit by naming owners, also make them social documents. (Balconi et al., 2004;Fleming et al., 2007, Fleming andFrenken, 2007). Timestamped and describing developments in science and engineering contextualized with literature citations, they are scientific and technical documents (Fromer, 2009).

Our patent search focused on U.S. government-owned patents relevant to viral vector vaccine technology using a data set published by the Government Accountability Office (GAO) in 2020 of 4446 patents owned by the Department of Health & Human Services (HHS) and its constituent agencies (GAO, 2020a,b).

We searched the titles and abstracts of these patents for terms related to viral vector vaccines such as "Ad5," "rAd5," etc. The abbreviations "Ad5" and "rAd5" refer to recombinant adenovirus type 5, a viral vector used in vaccine development that was important to the early scientific history of Ad26-based vaccines (Travieso et al., 2022). We use the abbreviations Ad5, Ad26, etc. to refer to the recombinant viral vectors based on the particular adenovirus serotype, not to the wild-type viruses themselves. "ChAd" refers to chimpanzee adenovirus, and "RhAd" to rhesus adenovirus, both also used as viral vectors in vaccines. Vesicular stomatitis virus (VSV) and poxvirus are an additional virus and virus family used to develop viral vector vaccine candidates (Travieso et al., 2022). More search details are in Appendix 2.

From the above patent search, we constructed a preliminary timeline of key U.S. government inventors-focusing on those who appeared on numerous relevant patents-and their key inventions (Appendix 2). Using PubMed, we searched for scientific publications by these scientists.

We added to our timeline of the most relevant patents and scientific papers, informed by ongoing interviews described below. To complete the timeline (Fig. 1), we searched the websites of Johnson & Johnson and NIH for press releases and other materials describing key patents, clinical trials, regulatory submissions, and other significant events. We also conducted a second round of patent searches related to adenovirusbased viral vector vaccines owned by J&J (Appendix 2).

Outreach to expert scientists in academia, government, and industry was conducted with an email describing the purpose of the project, reviewed and exempted by the [institution redacted for review] IRB. Interview respondents provided informed consent and interviews were conducted via Zoom between December 2022 and October 2023. Interviews were recorded for transcription and coding, securely stored, and de-identified. Interviews focused on the key elements of the AdVac development timeline (Appendix 3). In our interviews, we used a form of "snowball sampling"; we asked interview respondents to recommend additional scientists with relevant expertise that we should seek to interview.

This project applied a qualitative grounded theory framework analytical approach informed by Clark's "situational analysis," which is particularly well-suited to the thematic analysis of complex social situations (Clarke et al., 2018). Transcripts were analyzed using a three-round team-based approach, with codes, coding and emergent themes discussed by all co-authors after coding by one author. Group meetings were held to discuss and appraise the analysis alongside our patent review and scientific development timeline to develop a chronological and thematic analysis of AdVac. The patents, scientific literature, consolidated timeline (Fig. 1), the list of relevant scientists to contact for interviews, and our analysis co-evolved in an iterative triangulation process as interviews progressed.

The audit trail includes scientific timeline drafts and documentation of each round of thematic transcript analysis. To ensure anonymity of our respondents, we share the minimum amount of direct quotations necessary to communicate themes emerging from our analysis, sometimes removing non-essential details which could lead to respondents' unwanted identification. Assurance of anonymity was important to some respondents.

We conducted selective searches of databases of U.S. government spending to gather data on federal spending on research, development, and purchases of adenoviral vector vaccines, including J&J's vaccines. The complete search protocol is in Appendix 4.

Our initial patent search yielded seven patents specifically relevant to Ad26 and the term "AdVac," all filed either by NIH or jointly filed by NIH and Crucell/J&J. See Fig. 1 and Appendix 2. Our second-round search yielded seven additional J&J-owned or co-owned patents and patent applications covering Ad26-based vaccines and vaccine manufacturing processes. See Fig. 1 and Appendix 2.

Sixteen scientists were contacted. Interviews were successfully conducted with 8 scientists during a total of 12 h of interviews. Most of the scientists interviewed spent the majority of their careers in government and academic laboratories. We sought interviews with more private sector interviewees, but our outreach to private sector scientists suffered a low response rate, and our "snowball" approach yielded connections to more government and academic scientists than scientists employed by industry.

Our respondents and the scientific literature place the selection and development of the AdVac platform and the Ad26 vector in the context of a longer story of viral vectors in general, and their many applications across the field of molecular biology. To understand the "life" of the Ad26 vector, one must first understand the earlier development of the related recombinant vector based on adenovirus type 5 (Ad5), which was tested in the 1990s and 2000s in vaccine candidates for Ebola, HIV, and other pathogens. Some of this history is needed to understand the Ad5 vector. It begins in the 1970s and early 1980s with recombinant DNA genetic engineering, and so we start our timeline then (Fig. 1).

Viral vectors are modified viruses that deliver genetic material into cells. Their development follows a general formula: Engineer a viral vector that readily enters human cells but cannot itself replicate or cause serious illness, and that contains a pathogen-specific transgene. Once delivered into the cell, the transgene is expressed and marshals the host cell's molecular machinery to manufacture an antigen that induces an immune response, conferring immunity. In other words, the viral vector delivers the pathogenic virus's antigenic "cargo" to the host immune system (Zhao et al., 2021).

The first viral vectors were created early in the history of genetic engineering, with the first viral vector vaccines emerging in the 1980s. The first recombinant viral vector vaccine candidate was for hepatitis B and was invented by scientists at the National Institute of Allergy and Infectious Diseases (NIAID) at NIH using a poxvirus as the vector (Smith et al., 1983).

The complex and sometimes unpredictable relationships and interactions between a vaccine vector, the pathogenic antigen it carries, the human immune response, and the pathogen itself reappeared as a theme across interviews. Technical challenges emerge, prompting solutions that themselves present new technical challenges.

Adenoviruses (best known as one of the virus families that cause the "common cold") emerged as promising vectors in the 1990s because of their high efficiency of transduction and relative safety (Natuk et al., 1992;Xiang et al., 1996;Hitt et al., 1997). Recombinant Ad5 became an early favorite of researchers because of its potency (Gabitzsch et al., 2009) and ease and efficiency of manufacturing (Chen et al., 2018).

Recombinant Ad5 was used as a vector in foundational research on both Ebola and HIV vaccines conducted in the early 2000s at the NIH Vaccine Research Center (VRC) headed by Gary Nabel and John Mascola at the time, and the NIH-funded Barouch Laboratory led by Dan Barouch (Barouch and Nabel, 2005). NIH researchers showed that a combination of an Ad5-delivered Ebola glycoprotein and an Ebola DNA vaccine protected rhesus macaques from an otherwise lethal Ebola challenge, proving that a preventative vaccine for Ebola was possible (Sullivan et al., 2000). Our review of government-owned patents revealed that, beginning in 2001, NIH VRC filed multiple patent applications on Ad5-based vaccine candidates for Ebola, Marburg, and related viruses (U.S. Patent Nos. 7,635,688,8,106,026,8,106,027 & 8,124,592;U.S. Patent Nos. 7,635,485 & 8,017,130) (Sullivan et al., 2003). The NIH-funded Step trial of an Ad5-based HIV-1 vaccine was also underway during the same period (Buchbinder et al., 2008).

The Ad5 vector suffered a major and unexpected setback in 2007: The Step trial was terminated due to lack of efficacy and safety problems. Trial investigators observed increased HIV acquisition in a subgroup of vaccinees (Barouch, 2010). Increased acquisition of the pathogenic virus had implications for any scientists using Ad5 in their vaccine research, including Ebola researchers. Researchers quickly surmised that preexisting immunity to wild-type Ad5-a naturally circulating and common cold virus-may have been responsible.

One researcher working on Ebola during this period described how the ineffectiveness of an Ad5-based HIV vaccine and the potential connection with pre-existing immunity to wild-type Ad5 led them to do additional pre-clinical experiments before investing further in that vector. In those experiments, efficacy decreased in the Ad-immune animals compared to the control:

"That was a disappointment. NIH had planned at the time to advance an Ad5-based Ebola vaccine as a lead clinical candidate but halted that approach in light of the preclinical data."

The Step trial results ignited a search for alternative Ad vectors with less pre-existing immunity and less vector-induced immunogenicity. Research on such vectors was already underway, much of it led by two centers: the Barouch Laboratory, located at Beth Israel Deaconess Medical Center and affiliated with Harvard Medical School, and a small Dutch pharmaceutical company called Crucell Holland, B.V.

Crucell began in 1993 as a company called IntroGene, affiliated with the University of Leiden (InvestEurope; US Securities Exchange Commission, 2005). In 2000 IntroGene merged with another company to form Crucell (InvestEurope; US Securities Exchange Commission, 2005).

By the mid-2000s, Crucell and the Barouch Laboratory had established themselves as leaders in the search for recombinant adenoviruses. According to academic and industry respondents, Crucell had expertise in the development and manufacture of recombinant vectors from different wild-type adenoviruses, complementing the HIV research program at the Barouch Laboratory. In 2007, scientists from Crucell and the Barouch Laboratory jointly published a now widely-cited paper describing the performance of six non-Ad5 recombinant adenovirus vectors with comparatively low wild-type seroprevalence (Abbink et al., 2007). One of these vectors rose to the top: Ad26. The paper concluded, "[o]ur data indicate that rAd26 vectors have several characteristics that are attractive for vaccine development" (Abbink et al., 2007).

Crucell, Ad26, and collaboration with the Barouch Laboratory would soon become the core of J&J's international vaccine program.

J&J purchased a minority stake in Crucell in 2009 (Gray-Block, 2009). J&J purchased remaining shares in Crucell in 2010 and fully absorbed the company in 2011 (Dow Jones, 2011). J&J managed Crucell within Janssen, J&J's main pharmaceutical and vaccine division (Gray-Block, 2011;Pagliarulo, 2023). Some scientists we spoke with used the names J&J, Janssen, and Crucell interchangeably.

Crucell filed patent applications on Ad vectors in the late 2000s and early 2010s. Crucell filed patent applications on specific processes for making Ad35 and Ad35-based vaccines in 2008 (U.S. Patent No. 10,041,049), on general processes for making recombinant adenovirusbased vectors and vaccines in 2009 (U.S. Patent No. 8,460,920), and on processes for making Ad26 in 2010 (European Patent No. EP2536829B1). Ad35 was another vector considered viable alongside Ad26 in this early period; J&J initially branded Ad35 alongside Ad26 under the AdVac label (Fierce Biotech, 2010). Crucell began describing AdVac as "proprietary" at least as early as 2010, presumably based on these then-still-pending patent applications (Fierce Biotech, 2010). J&J would ultimately obtain patents in Europe, Canada, China, Japan, and other countries on its processes for making Ad26-based viral vectors and vaccines (European Patent No. EP2536829B1). Interestingly, J&J's patent application on its specific process for making Ad26 met resistance from the U.S. Patent & Trademark Office and was eventually abandoned by J&J (US20180080010A1) (Prosecution History of U.S. Patent Application No. 15/823,174). Yet J&J's patent protection on Ad26 in other large markets was presumably enough for the company to consider Ad26 effectively "proprietary." In 2021, one independent expert pointed to J&J's European patent, EP2536829B1, as crucial to J&J's proprietary position on its Ad26-based COVID vaccine (Storz, 2021).

The relationship between NIH and J&J quickly became the key public-private interface in the development of Ad26. Crucell's expertise in manufacturing Ad26 made it NIH's preferred partner, and NIH money flowed to J&J (Appendix 4). As one academic scientist explained: "The negative results from the Step study gave further impetus to develop alternative Ad serotypes, including Ad26. And our program with first Crucell, then J&J was entirely funded through government dollars." Even though an Ad26-based HIV vaccine candidate would later fail for lack of efficacy, this scientist described how growing experience with Ad26 made rapid development of a COVID-19 vaccine possible, as well as Ad26's application to other pathogens including Ebola (Milligan et al., 2016) and RSV (Widjojoatmodjo et al., 2015). Crucell became the preferred partner of NIH-especially the VRC and Division of AIDS (DAIDS), both divisions of NIAID-early on, largely because of its expertise in making high-quality Ad vectors on scales large enough for use in clinical trials. In this telling, it is clear that the NIH de-risked the development of the Ad26 vector by funding the early-stage collaborations between the Barouch Laboratory and Crucell, when there was no promise of any return on investment (Beam Alliance, 2019) (Appendix 4). Many scientists we spoke with see industry partnership as a necessity and expressed reverence for Crucell's and J&J's specialized manufacturing prowess.

To understand how AdVac-branded vaccines are products of the relationship between the NIH and industry, and how Ad26 became not just a vaccine vector but marketed as a multipurpose vaccine "platform," we turn next to the application of Ad26 in the context of ebolavirus. The storyline connecting the failure of Ad5 in HIV to the success of Ad26 in COVID-19 must be read alongside the response of Ebola researchers to the Step trial, including those who shifted to chimpanzee Ads (ChAds) with no natural immunity among humans. Even prior to the failure of the Step trial, it was widely postulated that human immunity against naturally occurring adenoviruses could limit their usefulness as vaccine vectors, and various groups actively sought alternatives (Hofmann et al., 1999;Yang et al., 2003;Barouch et al., 2004a,b).

Ebola scientists in government and academia tested various humanadenovirus-based vaccine candidates, including candidates based on Ad26 and Ad35, often in collaboration with Crucell. In 2010, Crucell and NIH VRC jointly filed a first patent application on a two-shot Ad26based + Ad35-based vaccine for Ebola (U.S. Patent No. 9,701,718). They continued to collaborate on Ad26-based Ebola vaccine research until at least 2014, when they jointly filed a first patent application on an Ad26and MVA-based Ebola vaccine candidate (U.S. Patent No. 10,561,721). Different perspectives on the clinical viability of Ad26 appear to be at the center of different directions later taken by Crucell and NIH VRC.

Despite joint efforts on prime-boost regimens (evidenced by the patents above), single-shot regimens would prove key for candidate Ebola vaccines to be useful during an outbreak. Early single-shot Ebola vaccine candidates based on Ad26 and Ad35 did not protect monkeys from a lethal Ebola challenge. This finding was described in a 2011 paper published by scientists at NIH VRC, the United States Army Medical Research Institute of Infectious Diseases, and Crucell (Geisbert et al., 2011). It showed that a single shot of the Ad26-based Ebola vaccine candidate provided some efficacy, but not the level provided by the Ad5-based vaccine candidate. Ad26 bypassed problems with preexisting immunity but failed to provide potency on par with the Ad5-based candidate. One scientist elaborated on this and related findings:

"Even at the highest dose, which probably couldn't be used in humans, 10 12 [viral particles], the [Ad26-based] vaccine achieved only 75 % protection compared to 100 % for Ad5. But it was a pivotal finding when the immune data were analyzed because it showed that immunologically, it wasn't just the antibody, or total T cells, but it was the actual quality of the T cells … the one variable that was key for protection. Ad35 and Ad26 did not generate that [quality] of T cell."

In the years following the disappointing Step trial results, the VRC also started to explore ChAd vectors, which promised both potency and an absence of preexisting immunity. There appears to have been a consequential divergence of approaches between VRC and various product developers including Crucell, as demonstrated by the scientific record of the mid 2010s and one scientist's commentary:

"The VRC had begun looking at ChAds as vectors. They worked like Ad5, in terms of efficacy. The VRC hoped that the two product developers for human and chimp-derived vectors could collaborate. However, this was not feasible and NIH ended up partnering with a smaller biotech startup to produce ChAd."

After Crucell and J&J declined to work on a ChAd3-based vaccine candidate, NIH VRC ultimately partnered with GSK to study ChAd3 as a vector for an Ebola vaccine (NIH, 2014).

Other NIH scientists separately worked with J&J to complete development of an Ad26-based prime-boost regimen, which would eventually be approved and marketed by J&J as Zabdeno; Bavarian Nordic provided the booster, Mvabea, which uses the modified vaccinia Ankara (MVA) viral vector (NIAID, 2020). By the mid-2010s, J&J seems to have become more specifically focused on Ad26 and Ad26 alone. As we've narrated above, by the mid-2010s, J&J had filed numerous patent applications on processes for making Ad26 and had begun using the AdVac brand name to refer specifically to Ad26-based vaccines.

A 2021 review of Ebola vaccines describes the ChAd3-based NIH-GSK candidate Ebola vaccine as similar to J&J's Ad26-based Zabdeno; both are boosted by Mvabea, with the only difference being the vector used in the first dose of the prime-boost series (Woolsey and Geisbert, 2021). The review downplays the potential importance of vector choice for the first shot, ChAd3 or Ad26. One scientist responded to this framing by stressing, to the contrary, the importance of the vector: "If [the author] says it's similar, it's similar only in the fact that they're both adenovirus vectors. But beyond that, in terms of efficacy, they're dissimilar."

We did not find any direct empiric comparison of Ad26 and ChAd3 vaccines in the Ebola literature. There are clinical studies showing strong efficacy of a ChAd3-based Ebola vaccine in monkeys (Stanley et al., 2014). A meta-analysis of pre-clinical and clinical variation in antibody responses to different Ebola vaccine candidates (Gross et al., 2018) expressed concern about the focus on antibody responses, noting the lack of evidence that this is a good correlate of protection against the virus. In ultimately granting marketing approval for Zabdeno/Mvabea, the European Medicines Agency (EMA) noted the need for a 2-month interval between prime and boost to attain maximal protection in nonhuman primates; shorter intervals produced lower survival rates (EMA, 2020). This explains the WHO recommendation to limit use of Zabdeno/Mvabea during an Ebola outbreak to lower risk populations in areas neighboring the outbreak area, not for those directly exposed (WHO, 2021). Zabdeno/Mvabea was administered to over 20,000 people during the 2018 to 2020 outbreak (Woolsey and Geisbert, 2021).

At the time, J&J said in a press release that the approval of their Ebola vaccine

it is also the first approved vaccine to be developed using Janssen's AdVac® technology. The same technology is being used to develop vaccines candidates to protect against SARS-CoV-2, as well as Zika, RSV and HIV" (Johnson and Johnson, 2020). This press release emphasized Ad26's claimed applicability against a wide range of pathogens-Ebola, SARS-CoV-2, Zika, RSV, and HIV. The press release positioned Ad26-based "AdVac® technology" as "transformational" against "deadly infectious diseases" plural, not just Ebola.

One scientist interviewed noted that a different, single-shot, VSVbased viral vector Ebola vaccine developed by Canada's National Microbiology Laboratory and manufactured by Merck (Herder et al., 2020) would be preferred for an acute outbreak. The scientist echoed the EMA recommendation that the Ad26-based Ebola vaccine could be reasonably used for lower risk populations and acknowledged J&J's role:

"To Janssen's great credit, they persisted with development [of Zabdeno/Mvabea]. They were able to deploy it in the Eastern part of Congo, not in an outbreak area, but around there. And you know, they did a good job. However, in an outbreak setting, it's likely that the single-shot VSV vaccine will be the first choice."

Multiple NIH-funded scientists with close working relationships with multiple manufacturers emphasized the manufacturing benefits of Ad26 made the vector particularly commercially viable, even if it produced a vaccine that was less effective in some outbreaks. In discussing HIV, Ebola, and other vaccine targets, they emphasized technical manufacturing know-how as J&J's principle contribution. The proprietary cell line used by Crucell and J&J to manufacture Ad26, Per.C6, was hailed by one NIH scientist as J&J's "special sauce" and even its "genius." According to these scientists, this unique advantage was one of several technical features which helped Ad26's rise attract more and more NIH support following a period of NIH investment in many candidate vector technologies-a sort of renaissance of vector development funded by NIH between 2005 and 2013: "[J&J] were also zealously protective of that IP [on Per.C6], as you can understand. And it made it so that they could take a thousandliter bioreactor and make millions of doses of their Ad26. And that, that was that was the genius that they had." When asked about the need for different vectors to address different pathogens, another important aspect of all vaccine development, one academic scientist minimized the distinctness and variation between vectors. (Note that many of our interviewees used the terms "vector" and "platform" interchangeably.): "I mean, the platforms do the same thing. So you can encode almost any antigen, and then you get a reasonably good immune response humoral and cellular, with different nuances, whether you're dealing with mRNA, or protein or viral vectors … What is different is how relevant those immune responses are for protection, and that part is pathogen specific."

In this scientist's view, J&J's work on the Ad26 vector-supported by science at the NIH VRC, Barouch Laboratory, and other labs and fueled with NIH money-had successfully created a multipurpose vaccine "platform." But another scientist emphasized how, in some pathogen contexts, the vector could be the primary difference between an effective and an ineffective vaccine, and switching vectors might be necessary for progress in some cases. When asked whether vaccine candidates based on Ad26 or other vectors generate immune responses which are reasonably predictable, the scientist's answer was clear: "Ad35 doesn't protect against Ebola in a single shot as Ad5 does. So clearly, efficacy by all Ad vectors as a platform is not predictable if one adenovirus protects and the other one doesn't. There's something fundamentally different about the immune response that is not predictable. There may have been some sense that adenoviral vectors should behave more or less the same way, apart from the pre-existing immunity issue. Like, "all Ad vectors are interchangeable." It was a shock and a disappointment to the field, that Ad26 and Ad35 didn't work the same as Ad5, didn't produce the same potent efficacy against Ebola."

The story of Ad26 and the AdVac-branded "platform" from the period from 2020 to the present is one of highs and lows, culminating in J&J's decision to shut down most of its vaccine research.

A boom began with the 2020 outbreak of SARS-CoV-2. A team at NIH VRC sent the Barouch Laboratory the precise sequences of engineered, antigenic coronavirus proteins, to be used in an Ad26-based vaccine developed with J&J; ultimately, Barouch and J&J, Moderna, and BioNTech-Pfizer would all use this same protein sequence from NIH in the three COVID-19 vaccines authorized by the U.S. Food & Drug Administration (FDA) in 2020 and 2021 (Fauci, 2021;Rutten et al., 2024).

The Barouch Laboratory led early work on what would become "Ad26.COV2.S," the J&J COVID-19 vaccine, with support from both J&J and NIH VRC. To quote one government scientist, "Dan [Barouch] is the mother and the father and the older brother of all of this [ …] He had the funding from the NIAID to do a lot of this. He did much of the formative non-human primate studies. … [H]e's the one that saw, 'Can I take the 1273 insert for SARS-CoV-2 [i.e., the sequence designed by NIH VRC] as glycoprotein and put it into Ad26, and very quickly immunize non-human primates and demonstrate a strong immune response?'"

In March 2020, the U.S. government's Operation Warp Speed-the massive cross-government initiative to speed development, manufacturing, and distribution of anti-COVID-19 technologies, including vaccines-began support to J&J for accelerated clinical trials and other development of its COVID-19 vaccine (Erman, 2020) (Appendix 4). In August 2020, before the vaccine had been proven safe and effective in clinical trials, let alone approved, the Center for the Biomedical Advanced Research and Development Authority (BARDA), a center within HHS, cut a $1 billion deal with J&J, guaranteeing the U.S. government 100 million doses of the vaccine (Erman, 2020) (Appendix 4). Around the same time, perhaps as part of its deal with Operation Warp Speed, J&J agreed to sell the vaccine in the United States on a nonprofit basis throughout the "emergency pandemic" (Kuchler, 2020;Mitchell, 2020). In January 2021, J&J released interim results from a Phase 3 trial of its COVID-19 vaccine, ENSEMBLE, showing safety and efficacy (NIH, 2021).

The U.S. government would support the J&J COVID-19 vaccine in other ways. For example, BARDA spent additional hundreds of millions of dollars to support a Baltimore facility that manufactured the vaccine (Baker and Koons, 2020). NIH and BARDA partially funded the first Phase 3 trial of the vaccine (NIH, 2020). Furthermore, in the words of President Biden, the White House "invoked the Defense Production Act to equip two Merck facilities to the standards necessary to safely manufacture the J&J vaccine" (The White HouseOffice of the Press Secretary, 2021). In 2020 and 2021, inventors at the NIH-funded Barouch Laboratory and J&J jointly filed patent applications on the J&J COVID-19 vaccine, which issued as enforceable patents in late 2022 (U. S. Patent No. 11,384,122;U.S. Patent No. 11,498,944). These patents acknowledge U.S. government support via a contract with HHS: "This invention was made with Government support under Agreement HHSO100201700018C, awarded by HHS. The Government has certain rights in the invention."

The USFDA granted an emergency use authorization (EUA) to J&J's COVID-19 vaccine in February 2021 (Johnson and Johnson, 2021a). As with Zabdeno, J&J again used the authorization to tout its "proprietary AdVac® technology" (Johnson and Johnson, 2021b). In 2021, J&J earned $2.4 billion in global COVID-19 vaccine sales, (Johnson & Johnson, 2022) a blockbuster even when sold at nonprofit prices. In January 2022, the company projected COVID-19 vaccine sales of $3 to $3.5 billion in 2022 (Johnson & Johnson, 2022). Nineteen million Americans received the J&J vaccine by May 2023 (McPhillips, 2023).

Sales and vaccinations declined in 2022. The rare but deadly side effect of thrombosis with thrombocytopenia (TTS) was reported as an adverse event with the vaccine (FDA, 2021) and a growing perception that mRNA-based vaccines were superior challenged its commercial success. In December 2021, the CDC Advisory Committee on Immunization Practices declared that mRNA-based vaccines were preferred in most cases (Oliver et al., 2021).

Yet as late as April 2023, J&J's website still touted the "AdVac platform," a means to "combat life-threatening infectious diseases" (Johnson and Johnson, 2023). A 2021 review article by J&J scientists (and others) reported no fewer than eleven Ad26-based vaccine candidates then in development against Ebola, HIV, malaria, RSV, filovirus, Zika, and HPV (Custers et al., 2021). Remarkably, in August 2023 J&J decided to abandon the majority of its infectious disease and vaccines research (Bayer, 2023). The reason for the sudden pivot is uncertain, but it likely relates to disappointing results in trials of Ad26-based, AdVac-branded vaccine candidates for multiple pathogens. In January 2023, a major trial (Mosaico) of J&J's Ad26-based HIV vaccine candidate was stopped due to lack of efficacy (NIAID, 2023). Also in January 2023, J&J announced it was "slashing" production of its COVID-19 vaccine due to declining demand (Hopkins and Loftus, 2023). In May 2023, J&J asked the FDA to revoke the EUA for its COVID-19 vaccine; the FDA did so in June (FDA, 2021). In February 2023, the first trial results of J&J's Ad26-based RSV vaccine candidate were released (Falsey et al., 2023), and while these were positive, the candidate's efficacy was lower than that reported around the same time for protein-based RSV vaccines manufactured by GSK and Pfizer (Papi et al., 2023;Walsh et al., 2023). These competitor vaccines were granted FDA approval in May 2023 (Jewett, 2023). In August 2023, media reported that the J&J Leiden vaccine unit-the facility first spun out of the University of Leiden in the 1990s-was closing (Bayer, 2023).

This whipsaw-from blockbuster to bust-begs the question: Why did AdVac fail? Why did J&J decide to halt R&D on Ad26-based vaccines?

J&J's dramatic decision to divest from Ad26-based vaccines is notable because Ad26 has held a privileged position in recent scientific history of viral vector vaccines, receiving unparalleled levels of financial and technical support from the U.S. government (at compared to other viral vectors). Despite the trial setbacks in some pathogen contexts, several things remained true about their potential: Ad26-based vectors had proven safety; they could overcome the issue of Ad5 seroprevalence in some contexts; and scientists across sectors trusted J&J's manufacturing know-how, even if they didn't see Ad26 as a "silver bullet."

By our estimate, vaccines based on Ad26 and Ad35 and J&J's capacity to manufacture such vaccines received at least $1.5 billion in direct support from the U.S. government-at least $285 million from NIH in support for clinical trials, preclinical research, manufacturing, and other R&D, and over $1.2 billion from BARDA and DOD to manufacture, study, and procure J&J's Ebola vaccine and COVID-19 vaccine (Appendix 4). Two notable BARDA and BARDA/DOD awards to J&J provided the company $132 million and $1 billion in 2017 and 2020, respectively (Appendix 4). J&J also received additional, valuable taxpayer-funded, non-monetary support from leading scientists at the NIH VRC and the NIH-supported Barouch Laboratory, as summarized above.

Beginning in 2005, numerous NIH awards to the Barouch Laboratory directed tens of millions of dollars directly to J&J to build out Ad vector manufacturing capacity, with J&J scientists Jaap Goudsmit and Hanneke Schuitemaker listed as "Contact PI/Project Leader" of various "sub projects" within the awards (Appendix 4). We document NIH's direct support for J&J's manufacturing totaling over $65 million-this in addition to money received from BARDA and DOD in connection with Operation Warp Speed (Appendix 4).

This history of NIH funding underscores the Barouch Laboratory's role as J&J's and NIH VRC's main collaborator on adenoviral vector research. Dan Barouch is named as Principal Investigator (PI) on 33 separate NIH awards related to Ad26 and Ad35 that total over $155 million, and other investigators at Beth Israel Deaconess Medical Center are named as PI on other relevant awards (Appendix 4).

The privileged position of Ad26 is only partially explained by the promise of the technology itself. Scientists identified other influences on its trajectory. One was already mentioned: J&J's enormous success in manufacturing Ad26. Another influence was intellectual property (IP). Multiple scientists suggested that patents, once obtained, created incentives to invest further in the vectors and "platforms" they protected (and dissuaded investment in other directions). A scientist with a career spanning multiple sectors of drug development describes the incentive: "There also is a tendency I think, for anybody, I don't want to single out J&J. But you know, when you know something, and you have a hammer and a nail, you know how to hit the hammer with the nail [sic] … What often happens though is they -not just them, but others-don't look at the fine details."

This scientist argues that ignoring these "fine details" is not about carelessness. Rather, patent incentives make it easy to focus on encouraging results like high antibody titers, and neglect details which could undermine support for the patented vector's continued use. Another scientist also noted that it may be easier to make patent claims on a vector compared to other equally important vaccine components, like the antigen, making the vector a particularly appealing basis on which to build a proprietary brand:

"Yeah, it's probably easier to get your intellectual protection around the vector platform [than around other components of viral vector vaccines] … And there's also a lot of steps that you have to take in that manufacturing process, all of which can be proprietary. But it probably is easier to build your proprietary base around your manufacturing approach than around your design approach."

As J&J built out manufacturing experience and capacity with Ad26 and obtained patents on the vector, it began applying the AdVac brand name specifically to Ad26. And as J&J found regulatory and some commercial success with Ebola and COVID-19 vaccines based on Ad26, it began marketing Ad26 and AdVac as a multipurpose vaccine "platform" applicable against a wide range of pathogens, perhaps then driving further NIH funding-until the company pulled the plug.

Importantly, multiple scientists believed that TTS was not a good reason to abandon Ad26 completely, and certainly not all research into adenovirus-based vaccines. Clinical data published in late 2022 indicated that J&J's COVID-19 vaccine offers significantly more durable immunity than the mRNA-based Moderna or Pfizer-BioNTech vaccines (Zheutlin et al., 2022), an overlooked fact in analysis that describes mRNA-based vaccines as the undisputed "winners" of the COVID-19 vaccine race. Two scientist respondents argued that the regulatory abandonment of the J&J vaccine may have been out of proportion to the side effect, which only occurs measurably in the context of mass vaccination campaigns. They further agreed that there would likely be clinical scenarios in which Ad26 is a good choice of vector, and also that TTS is likely a problem with a technical solution (Favaloro et al., 2022;Baker et al., 2021). One lamented J&J's decision to cease its research program: "I think that [TTS] could probably be solved by doing more genetic testing and more detailed molecular analysis [ …] I would hate to lose Ad-vectors from our armamentarium. I don't think that it would be good to lose them."

Another scientist, working in industry, elaborated on this possibility. Platelet factor 4 (PF4) activity, thought to be an important aspect of TTS pathophysiology, was not observed in some preclinical studies of alternative rhesus adenoviral (RhAd) vectors:

"We have shown actually, in vitro, that there is no binding to PF4 … And all of our preclinical studies, show equal and durable immunogenicity [of rhesus adenoviruses] as compared to human and chimp adenoviral vectors. [ …] I expect that we would be able to translate that and use the RhAd vectors that retain all the immunogenicity, characteristics, benefits of the currently used Ad vectors, but then with the improvements of eliminating TTS. I think that is the way forward, for me at least."

The future of viral vector vaccine research is uncertain. NIH's disinvestment in mRNA-based vaccines during the second Trump Administration and mysterious investment in a new project dubbed "Generation Gold Standard" clouds the funding and R&D outlook for vaccines generally (Cohen, 2025).

One bright spot: There is continued work on Ebola with ChAd3 (Happe et al., 2024). It is notable that, in this active area of Ebola research, BARDA recently committed at least $35 million to unsolved problems in the field using ChAd3 (Sabin Vaccine Institute, 2023). The Ad26-based and VSV-based J&J and Bavarian Nordic vaccines for Ebola-Zabdeno/Mvabea-have marketing approval, but only for Zaire ebolavirus (Woolsey and Geisbert, 2021). There was no approved vaccine for an outbreak of Sudan ebolavirus in Uganda in 2022 (WHO, 2023). BARDA therefore is supporting research and development of a single-dose ChAd3-based vaccine for Sudan ebolavirus through the nonprofit Sabin Vaccine Institute (Sabin Vaccine Institute, 2023. At this point, there is no apparent industry partner for this work, perhaps because ebolaviruses, despite being an important public health problem, are a less attractive investment for private manufacturers. Work on a separate Sudan ebolavirus vaccine candidate based on VSV is also apparently proceeding without industry support (Marzi et al., 2023). Disinterest from industry is unlikely due to lack of potential viabilitya 2024 preprint shows that a single dose of the ChAd3-based vaccine provided 100 % protection from Sudan ebolavirus challenge in macaques (Honko et al., 2024).

HIV, on the other hand, is at a crossroads. Some scientists at NIAID and the CEO of Moderna have expressed hope that trials of mRNA-based HIV vaccines will push the field forward (Stulpin, 2023;Lusso, 2023). However, HIV experts among our respondents did not express great confidence in their potential, because they do not see mRNA or any other new vector as likely to solve the complex problems presented by HIV. J&J was the largest pharmaceutical company in the HIV vaccine field as of early 2023. J&J's divestment from vaccine research leaves many questions about potential uses of viral vectors unanswered.

Platform thinking sometimes drove the development of AdVac rather than the best science. J&J's interest in building and exploiting a "platform" based on Ad26 shaped not only its own scientific choices but those of NIH and other key actors. This concept helps us to organize and understand how Ad26 went from being first one promising vector among many, to being J&J's preferred vector for every pathogen and the centerpiece of an international brand name, to being all but abandoned when J&J decided to cut its losses. It also helps us relate the social organization of vaccine science and its protagonists to the fields and structures which govern science and capital.

In Ebola, J&J assiduously pursued an Ad26 vaccine candidate while NIH collaborators with pathogen expertise pursued a seemingly empirically superior alternative (ChAd3). J&J's research, development, manufacturing, patent, marketing, and sales positions made Ad26 the only vector worthy of consideration for J&J for more than a decade, and J&J's strong relationships with NIH, the Barouch Laboratory, and other actors helped keep momentum behind Ad26. When conditions shifted in 2022 and 2023, J&J apparently came to see the same vector as completely unviable.

This sharp change does not reflect any scientific truth or consensus, but rather what we call "platform thinking." Platform thinking is a capital-driven approach to vaccine development that strives to use a single vector "platform" for every known and unknown pathogen, because patent rights, manufacturing and distribution capacity, branding, and other incumbency advantages guarantee immense profits when "platform" products are administered at a population level. We propose platform thinking as a way to conceptualize the tendencies that this singular focus produces in individuals and in organizations, and the shared political economic context of those tendencies. Platform thinking can marshal resources to ensure that the idealized "platform" supplants potentially viable (even superior) alternatives, underestimate complexity, and overstate certainty. Platform thinking can attribute "genius" to one component of a vaccine and overstate the component's promise, while overlooking other components and downplaying disappointments as mere details. Of course, some research organized around vectors is sensible. However, a research ecosystem overly influenced by this approach, as seen with AdVac, leaves important scientific questions unanswered, and relevant technical insight and resources unshared.

The concept of platform thinking adds to existing literature on the political economy of drug development, and literature on the management of value within the bioeconomy more broadly. Financialization and assetization explain how capital assigns value to products and product components in the drug development pipeline (Birch, 2016;Bruch et al., 2024;Roy, 2023). The concept of "biomedical platforms" conceived by Keating and Cambrosio (2003) and taken up in studies of the relationship between the laboratory and the clinic (Levy, 2022, Chiapperino et al., 2025) has been applied to studies of the regulatory regimes to which biomedical innovations are subject (Hogarth and Martin, 2021;Gardner and Webster, 2016). Also relevant to our work are Haraway's (1997) analysis of the branding, patenting, and aggressive marketing of the "OncoMouse™" and other genetic engineering technologies, and their impact on the "technoscientific" process itself, and Graham and Ritchie's (2006) analysis of how classification of cognitive impairment and dementia has both shaped and been shaped by the market for pharmaceutical treatments. However, these important references do not articulate precisely the way that technoscience produces value. This gap is filled by Birch's work on assetization, which elucidates "the political economy of technoscience" (Birch, 2016). This is possible with an emphasis on the life science firms who principally manage value by monetizing IP (Pisano, 2006). Birch (2016) confronts the contradiction between, on one hand, ever-increasing valuations on the speculative assets produced in the process of monetizing IP and, on the other hand, the fact that most firms in the field will never realize their value in the form of treatments or other marketed commodities. A venture capitalist investor described how knowledge either disappears or is sold very cheaply when firms "go bust" in this environment, as so many do (Birch, 2016).

To theorize the social practices not just of small firms but of the large, dominant firms in this milieu and their implications for knowledge production and public health, we have focused this paper on one of the scientific organizations whose size, profits, and manufacturing capacity permit them to exert political economic power through the scientific process itself (alongside other means). J&J and other dominant pharmaceutical companies are at less risk of dissolution than so many other firms are and are relied upon by public agencies, including NIH, to translate and commercialize the products of the scientific process. AdVac and platform thinking elucidate how political economy is entangled within the scientific process, and not a distant or strictly "superstructural" influence.

Platform thinking attends to the interface between capital and the scientists, wherever they may be operating; our respondents had experience in public, for-profit, and non-profit spheres. These perspectives help to outline a vaccine pipeline which is at once technical and political. Latour argues that a technology is inseparable from the social and political context which produced it (Latour, 1993). Intellectual property's influence on AdVac is a vivid instance of this. While branding and the accrual of value are absolutely necessary to give a politically inert vector the life of a "platform," our findings suggest that science and scientists are part of this political economic life. This is true even for academic and government scientists who are not directly part of the companies branding and profiting from various "platforms"; most of the academic and government scientists we interviewed used the words "platform" and "vector" more or less interchangeably. Some of these scientists mentioned that vaccines need an industry partner to get to patients, that would-be industry partners need profits to agree to partner, and that industry partners need patents to make profits-further suggesting how platform thinking may be common not just among industry scientists. Leadership at NIH may have accepted that, by the mid-2010s, J&J had made a bet on Ad26 and AdVac and was less interested in developing alternative vectors. In the absence of other large American pharmaceutical companies conducting viral vector vaccine research, J&J may then have been able to dictate some of the terms of public-private partnership. J&J seems to have been able to concentrate NIH's viral vector resources on Ad26 and the AdVac "platform" even though NIH was itself contributing much of the science and footing most of the bill.

One may even view J&J's influence on NIH's research and funding priorities as a kind of "deep capture," as theorized by Hanson & Yosifon and Cohen-that is, capture of a regulatory agency at the level of regulators' ideology (Hanson and Yosifon, 2003;Cohen, 2019). J&J did not corrupt NIH, but J&J's influence (and, of course, the influence of other companies, and the broader political economy of the United States) may have limited the ability of NIH leadership to imagine making alternative policy choices less favorable to J&J. This phenomenon resembles in some ways the phenomenon documented by Graham and Ritchie of independent researchers, clinicians, and patients coming to share the vocabulary, practices, and outlook of pharmaceutical companies selling treatments for cognitive impairment (Graham and Ritchie, 2006).

When we asked scientists whether a "platform" is a scientific or business entity, all said that it is both things at once. The branding of AdVac made promises (as branding is meant to do) beyond the empiric reality of Ad26, and J&J's 21st century development strategy did everything it could to actualize these promises. When the vector's totalizing potential as an all-purpose "platform" lost credibility, J&J shut down the vast majority of its vaccine development. Platform thinking explains the gap between the many aspirational functions of AdVac and the science of Ad26, which could not keep up.

J&J's decision to shut down the AdVac brand might be viewed, in sympathetic light, as proof that vaccines are financially unappealing compared to other kinds of pharmaceutical products and indication that even bigger grants or other incentives are necessary to induce companies to partner with NIH to develop vaccines. On the other hand, in less sympathetic light, we might see the "AdVac affair" as tragic. NIH and BARDA together invested at least $1.5 billion, by our estimate, in Ad26 and Ad35, and yet J&J abandoned the vectors and its AdVac "platform" after perceiving them as insufficiently profitable. J&J retains patent protection on its processes for manufacturing Ad26 and other adenoviral vectors, meaning that there may be legal barriers to entry for other pharmaceutical companies should they wish to develop Ad26-based vaccines of their own. NIH did not manage in the meantime to cultivate any other major manufacturers capable of bringing a viral vector vaccine to market, nor did it foster late-stage research into other promising viral vectors (aside from ChAd3).

"Pathogen thinking," or, as the WHO proposes, the prototype pathogen approach, could be a more promising alternative way to structure vaccine science (WHO, 2024). When WHO assembled experts on "platforms" in early 2024, tensions were on display, and some scientists were openly critical of "platforms." Many discussants repeated and endorsed the facilitator's framing, rejecting any "one size fits all," "one vector fits all" approach to a hypothetical "Pathogen X." Pathogen thinking instead promotes exploring a variety of technologies as potentially useful when orienting towards a field of known and unknown pathogenic threats. David Weiner, an expert on DNA vectors, was asked by the facilitator Philip Krause about their commercial viability given their potential use in prime-boost combinations with other vectors owned by other companies. In response, Weiner described in broad terms how combinations of different vectors could improve immune responses and that the field should take advantage of their unique properties. Also present was Rajeev Dhere, Executive Director of the Serum Institute of India, now the largest global vaccine manufacturer thanks to early tech-transfer agreements and financialization through the WHO and Gates Foundation (Graham and Jones, 2016). Because the Serum Institute has wide-ranging, adaptable manufacturing infrastructure, he de-emphasized the vector-centric approach even more strongly. While this more pluralistic manufacturing approach and the resources invested in it do not cure drug development of every pressure created by global pharmaceutical capital, it goes a long way in challenging the assumptions of platform thinking, in which other scientists and organizations are more thoroughly embedded.

However, some participants in the 2024 WHO consultation endorsed platform thinking. Chris Parks, associate Vice President of Vaccine Design and development at the International Aids Vaccine Initiative (IAVI), explicitly called out recombinant VSV as having "plug and play" potential, referring to its application in Sudan ebolavirus and other pathogens. Clearly, the prototype pathogen approach has not completely supplanted platform thinking.

Measures are needed in research and regulation to ensure that platform thinking does not unduly influence the trajectory of future pandemic preparedness or emergency responses. At the discursive level, scientists need to be able to name platform thinking as it is occurring so that these tendencies do not undermine open science and collaboration (Torreele et al., 2023). Due consideration of all viable medical countermeasures through an investment in basic science could turn medical countermeasures for pandemics into a finite "engineering problem," as Barney Graham put it, referencing his work with Nancy Sullivan on prototype pathogens (Graham and Sullivan, 2018). The state of flux in vaccine development following the acute phase of the COVID-19 pandemic could create more space for this approach to be taken up across more research institutions.

We also need to better understand the social, political, and economic dimensions of pandemics and build capacity to address these challenges. At its worst, platform thinking helps to prop up R&D capital at the expense of both scientific advancement and public health.

A variety of interventions can challenge this regime of knowledge and power. These include supporting the autonomy of scientists in lower-and middle-income countries, sharing life-saving IP and knowhow, and including affected stakeholders in decision-making about pandemics and other threats to our health (Graham and Jones, 2016). The WHO consultations on pandemic preparedness were a site where the relationship between the "platform" and the society that makes it were clearly visible, along with its constitutive power dynamics. Though these dynamics are rarely visible or explicit, the proposed interventions must act on and through the fields and institutions which together produce them.

One small but promising step in the right direction has already been undertaken by NIH: In January 2025, NIH announced a new Intramural Research Program Access Planning Policy that requires all licensees of NIH-owned patents to submit Access Plans to NIH "outlining steps they intend to take to promote patient access to those licensed products" (NIH, 2025a,b). This policy was announced near the end of the Biden administration and has been reissued by the second Trump administration (NIH, 2025b). The policy will require greater levels of IP-and know-how-sharing by companies that themselves license NIH-owned patents and will encourage those companies to take other steps to ensure "equitable access for underserved communities in the U.S. and for populations in low-and lower-middle-income countries" (NIH, 2025b). In effect, the policy commits NIH to taking a stronger hand in its negotiations with companies of public-private partnerships that commercialize certain NIH-created and NIH-patented technologies. But the policy has many limitations; for example, it will not apply to companies that license patents that are NIH-funded but held by other institutions (e.g., NIH-funded universities), nor will it apply to companies that receive direct grants from the NIH (Love, 2025).

As long as NIH and other public agencies remain dependent on industry not just to commercialize, manufacture, and distribute products but to co-create scientific research agendas, they may remain beholden to platform thinking. The financialized pharmaceutical industry will continue to dictate what is "pragmatic," even possible, and will likely lead to overinvestment in profit-making platforms at the expense of more diverse and potentially more valuable science. In our view, what we need is not a doubling down on incentives for industry or moderate reform of existing regulations but a fundamentally different paradigm.

One fundamentally different and, to us, appealing paradigm is truly public pharma-end-to-end nonprofit, public-sector, and publicly accountable research, development, validation, manufacturing, and distribution of medicines. Public sector agencies empowered to take promising technologies all the way from laboratory bench through regulatory approval and manufacturing to patients without the need for a profit-seeking private-sector partner may be able to focus more fully on the research most important to public health, avoid platform thinking, and negotiate more favorable public-private partnerships when they do choose to work with industry. Vaccines may be a particularly sensible focus of such initiatives (Sarpatwari, 2023;Ramachandran, 2023). Public drug companies once thrived in Canada, the U.S., the Netherlands, and elsewhere in the Global North and continue to endure, if not thrive, in Cuba, Brazil, India, and elsewhere in the Global South, despite strong and persistent opposition (Sarpatwari et al., 2019;Costa Chaves et al., 2021;Blume and Baylac-Paouly, 2022;Brown and Latkowski, 2022;Paranhos et al., 2023;Barber, 2024). Early stage research within these public initiatives must be publicly funded, transparent, and cooperative, rather than competitive. It must also, consistent with pathogen thinking, be focused on the pathogens of greatest concern to human health, and attentive to complexity, rather than bound by the reductive fantasy of one-size-fits-all "plug-and-play" "platforms." The most promising technologies should ideally be transferred not to industry but to new public bodies capable of clinical testing, management of knowledge and any IP rights, manufacturing on scale, and wide and equitable distribution. This may seem impossible today, but the crisis presented by the Trump administration and widespread distrust of both government and pharmaceutical industry may both invite and require transformative change.

Our study faced several limitations, namely that many major stakeholder institutions in AdVac and Ad vectors could not be reached or did not agree to an interview. Our synthesis of interview data from interview respondents was also limited by the need to maintain their anonymity. Where these perspectives limited us, patents, scientific literature, publicly available government documents, and other media filled in gaps.

This project reviewed and exempted by the Columbia University IRB (IRB-AAAU2507). Interview respondents provided informed consent and interviews were conducted via Zoom between December 2022 and October 2023.