The FDA has now authorized three vaccines and several treatments (including both monoclonal antibodies and small-molecule drugs) for the prevention and treatment of COVID-19. But the initial evidence supporting these products’ introduction into the market did not include information about how they might work together. Nevertheless, information about mixing-and-matching COVID-19 vaccines and therapies would be highly valuable not only to physicians and their patients, who must already make decisions about what treatment options to pursue under conditions of uncertainty (if the treatments are available), but also for policymakers, who want to know what products to prioritize for investment. Why is it so difficult to obtain this information? How can policymakers encourage its development?
What, if anything, is known about mixing-and-matching COVID-19 vaccines and therapies?
A growing body of evidence suggests that mixing and matching different COVID-19 vaccines is safe and generates an effective immune response, but evidence is more uncertain on whether switching vaccines mid-course leads to a more or less effective response, particularly when the initial dose was of an mRNA vaccine. This information seems especially important given that global vaccine distribution remains patchwork and viral variants continue to arise. What evidence is there for vaccines’ and boosters’ interchangeability?
The CDC currently advises that vaccines are not interchangeable for the initial two doses of the Pfizer-BioNTech and Moderna mRNA vaccines, but that mix-and-match dosing (more formally known as “heterologous” dosing) is allowed for booster shots. People who received non-FDA-authorized vaccines can also receive a heterologous primary dose and booster. The agency cites preliminary results from a pre-Omicron study by the NIH-funded Mix and Match Team, which found that heterologous boosters resulted in similar or higher antibody responses in the first month after boosting. But short-term antibody responses may not indicate clinical outcomes, and the study authors caution that the study was not designed to compare different booster regimes (given the sample size and lack of controls for relevant variables). Last week, the Mix and Match Team posted very preliminary results that most booster combinations (heterologous or homologous) increase antibody response to the Omicron variant, but these results have similar limitations.
The European equivalents of the FDA and CDC—the European Medicines Agency (EMA) and the European Centre for Disease Prevention and Control (ECDC)—issued updated guidance on heterologous vaccination in December, concluding that the mix-and-match approach may be used not just for boosters, but also for initial courses. This guidance was supported by a literature review of studies available by December 3, including both immunogenicity studies like those from the Mix and Match Team and longer-term studies of vaccine effectiveness. Most of these studies involve the AstraZeneca vaccine that has not yet been authorized in the United States, and while a few describe the benefits of boosting J&J recipients with mRNA vaccines (like the COV-BOOST and SWITCH trials), the EMA emphasizes that there is “limited data on interchangeability of mRNA vaccines.”
The WHO’s recommendations, also published in December along with a literature review, state that although homologous dosing is “considered standard practice,” the WHO “supports a flexible approach to homologous and heterologous vaccination schedules” and says countries may consider administering mRNA vaccines or vectored vaccines (like J&J or AstraZeneca) after any other vaccines, but it doesn’t recommend switching to inactivated vaccines (like Sinovac-CoronoVac) after initial dosing with other vaccines. Recommendations for the global context may be more lenient toward mix-and-match dosing in part because of the inequitable global vaccine rollout. As the WHO notes, “[a] common reason for considering heterologous COVID-19 vaccine schedules is lack of availability of the same vaccine product in settings with limited or unpredictable supply.”
As incomplete as the evidence base is for heterologous vaccines, there is even less information like this about combining different COVID-19 therapies like monoclonal antibodies and the antivirals molnupiravir and Paxlovid. The information we want is in part comparative effectiveness (which therapy is best?) and in part combinatory (is it better to have two therapies rather than one?), which would be valuable for patients, providers, and healthcare payers. But as discussed in the following section, firms have limited incentives to develop this kind of information about their products.
Why is it difficult to obtain information about the comparative benefits of different health care technologies?
Despite the benefits of comparative and combinatory information for health care technologies, data is hard to come by. Companies generally don’t have good incentives to generate such information. Two “lingering questions” concerning the antivirals molnupiravir and Paxlovid, for example, are whether they work better in combination and how they compare with monoclonal antibodies. In an assessment of these questions in STAT News, Dr. Céline Gounder at NYU’s Grossman School of Medicine bluntly noted that “neither Merck nor Pfizer is incentivized to run a combination therapy trial.”
To the contrary, such trials are prone to yield “negative information,” i.e., information that’s harmful to the market prospects of either drug. In a 2013 article in the Yale Law Journal, Professors Amy Kapczynski and Talha Syed presented the case study of Norvasc for cardiovascular disease and older, similar cardiovascular drugs. At the time of approval, Norvasc’s manufacturer (Pfizer) failed to conduct any study of its drug compared to the standard treatment at the time—a study which, when conducted thirteen years later by NIH, demonstrated Pfizer’s product was not clinically superior—and in some respects, was inferior. This, Professors Kapczynski and Syed argue, is an example of how the downside to developers in producing such comparative data far outweighs any potential upside. Professors Kapczynski and Syed further categorize this negative information into two groups: information that the studied drug is not as safe and efficacious as previously demonstrated (like Vioxx); and information that the studied drug is comparatively less safe or efficacious than its competitors. In either case, the only entity currently incentivized to produce such information seems to be an independent third-party—like NIH, the sponsor for the COVID vaccine Mix and Match trials.
Incentives aside, there are practical barriers to conducting comparative trials as well. Developers may have a hard time negotiating with a competitor to obtain enough of their drug to do a comparative analysis. If the competitor’s drug is already approved or authorized, it can typically be obtained from the market. But garnering enough—at reasonable prices—is often difficult for even the simplest analyses, and a point of antitrust scrutiny. Further, the mere asking for excess product for a clinical trial is likely a signal of the developer’s confidence in its product’s superiority—further discouraging sharing. And, for comparative trials at least, differing dosing regimens may complicate any analysis. For the COVID-19 mRNA vaccines, for example, how should researchers compare mixing and matching second shots when Pfizer’s original data spaced doses three weeks apart to Moderna’s four?
There are practical barriers to combinatory trials, too. Combinatory trials generally require more treatment arms, which require more participants to be robustly statistically powered—and thus more time, and more money. This is likely exacerbated by the need for participants who fulfill each arm’s criteria. Fielding enough participants to demonstrate the combinatory efficacy of mixing-and-matching two-dose COVID-19 vaccines, for example, requires a subset of participants who have received one dose but not the second. As the number of permutations of combinations increase, there may simply not be enough participants out there to enroll—and especially to include participants from traditionally underrepresented communities. And, for observational studies as opposed to controlled trials, combinatory information relies on both products being approved for heterologous use, available, and covered by patients’ insurance—none of which are a guarantee.
Lastly, the desire for comparative and combinatory data is ultimately about updating information against constraints of financial resources, patients, and time. The pandemic continues to evolve as new viral variants and new therapies come online; running comparative trials each time a new therapy is authorized or a new variant comes to the fore is impractical and, of course, costly. Even with good incentives for developers, there may simply not be enough money or patients to go around—or enough time to conduct a study before new variants continue their march down the Greek alphabet.
What tools can policymakers use to encourage the development of this information?
Despite these difficulties, policymakers have multiple tools to help develop this sort of combinatorial information, including incentivizing or mandating drug companies to come up with the information, directly generating the information, or facilitating its collection in some other way.
One possibility is using agency incentives to encourage drug companies to develop the information. Authorization or approval of a therapy or reimbursement for its use by federal payers like CMS could be contingent on evidence of not just effectiveness versus a placebo, but also effectiveness versus the current standard of care. If one product is developed later than another, the second product’s developer could be required to include a comparison arm or a mix-and-match arm in the clinical trial. In the Paxlovid trials, for instance, the treatment arm was compared to placebo, and no patients had received monoclonal antibody treatment. To develop comparative information, the FDA could have considered requiring Pfizer to include a comparator arm for mAb therapy or one where patients got both to see what the added benefit would have been (assuming medical feasibility). This approach is, of course, more complicated where both products are being developed simultaneously, as with the mRNA vaccines. In the simultaneous development context, requirements or guidelines for more standardized clinical trials could help make the data at least more directly comparative, though, as we have discussed, such standardization efforts bring their own complications.
A more direct and thus far substantially more important pathway is direct government funding. The government can either run the relevant trials directly or provide grant funding for others (for instance, academics) to run the trials themselves. This results in higher on-book government expenses, of course, but can enable the government to design a relatively neutral trial rather than one designed to favor one product over another. It also allows the enrollment of multiple manufacturers, and the possibility to conduct adaptive trials that examine the relevant questions in a more efficient way, as the RECOVERY trial has done in the UK.
Policymakers can also act to facilitate the development of such information from both private-sector and public-sector clinical trials. As noted above, those wishing to run comparative or mix-and-match trials may face problems acquiring enough of the relevant products to run their trials; policy mandates could require as a condition of authorization or approval that drug companies make sufficient amounts of their products available for later trials (the CREATES Act, enacted in December 2019, goes some way along this path). Policymakers could consider other facilitating actions, such as immunizing drugmakers for liability arising from side effects in clinical trials run by anyone other than the drugmaker. (In the COVID-19 context, for example, liability for injuries related to vaccines has been a persistent sticking point, generally).
Finally, policymakers can facilitate the development of knowledge about comparative effectiveness and mix-and-match treatments outside the context of clinical trials. Health care providers are already treating patients under conditions of uncertainty, sometimes mixing-and-matching vaccines, combining treatments, and otherwise acting outside the scope of clinical trials (including treating pregnant people even though they are often excluded from trials). Policymakers should encourage the collection of high-quality data about these ongoing practices so that this real-world evidence can be used to increase knowledge of what works, how, and when. This might mean the provision of infrastructure for better data collection, pathways for this data to be considered by regulators (already an area of significant and contestable interest), or incentives for private actors to share information. Clinical trials will remain the gold standard, but knowing something is far better than knowing nothing. Policymakers can move forward on all fronts to augment information about mixing and matching the COVID-19 vaccines and therapies we already have.
This post is part of a series on COVID-19 innovation law and policy. Author order is rotated with each post.