How ARPA-H can master the subtle art of risk-taking

Despite remarkable progress in health and medicine, biology’s final frontiers remain beyond our reach. Humanity has not yet cured Alzheimer’s or ended cancer, and was unable to prevent Covid-19 from growing into a pandemic.

As physicians and scientists, we understand the frustration of patients who still feel left behind despite the marvels of modern medicine. And we wonder: What would it take to cure Alzheimer’s? End cancer? Prevent pandemics?

advertisement

We are not the only ones. In line with President Biden’s reignited Cancer Moonshot, Congress recently created the Advanced Research Projects Agency for Health (ARPA-H) to fast-track breakthroughs in disease prevention, detection, and treatment. ARPA-H’s “special forces” model of innovation emulates the legendary Defense Advanced Research Projects Agency (DARPA). Founded in 1958 as a response to the shocking Soviet launch of the first satellite, Sputnik, DARPA’s original mission was to prevent future technological surprise to the U.S. It ended up creating surprise instead. DARPA has played a crucial role in the development of technologies like the internet, GPS, self-driving cars, and mRNA vaccines, to name but a few.

We are excited to see ARPA-H catalyze similar seismic changes in health. But ARPA-H has also acknowledged that disruptive innovation carries an inherent element of risk. Those at the forefront of breakthrough technologies will also be the first to encounter risks associated with those technologies. Agency leadership has said they want to bring their technologies to fruition safely, in addition to successfully.

DARPA’s history underscores that technological breakthroughs can be a double-edged sword. While innovations have the power to fundamentally change the world for the better, that transformative power can also be hazardous. Many technologies are “dual use,” meaning that they can have both benevolent and malicious applications. For instance, drones (another DARPA invention) grant millions access to aerial photography and surveying but can also be weaponized by those same individuals. Or take DARPA’s introduction of Agent Orange as a defoliation compound during the Vietnam War, despite many experts at the time warning that it would harm humans. Ultimately, the widespread use of Agent Orange exposed millions to toxic, cancer-causing chemicals.

advertisement

Conversely, there have been instances where the U.S. government has prevented misuse and unintended consequences of DARPA technology: When DARPA-developed GPS became available for civilian use, the government implemented speed and altitude limits and export controls to prevent the new navigation system from being used in weapons systems. Recently, DARPA issued a Request for Information on ethical, legal, and societal implications of emerging technologies, with the goal of collecting information on risk assessments and frameworks for responsible innovation.

As ARPA-H sets out to emulate DARPA’s successes in health, many of its programs will harness the power of cutting-edge biotechnology, like next-generation sequencing or genome editing. And frankly, we believe this is cause for optimism. Biology’s accelerating pace may enable ARPA-H to drive the breakthroughs it hopes to achieve.

But while genome editing techniques like CRISPR offer the promise to heal once-incurable diseases, they have also sparked concerns about misuse, as evidenced by the case of the CRISPR babies. It is clear that our ability to address misuse risks isn’t as adept as it needs to be.

Consider this: In 2002, it took world-class experts and $300,000 to first synthesize an infectious virus (polio) in the lab. The scientists involved received their funding from DARPA, arguing their work would help protect the U.S. against bioterrorism because it would identify if “would-be bioterrorists might be able to synthesize viruses.”

When their work was published in the journal Science, other experts expressed concern that their methods could indeed be used by others to build dangerous pathogens like Ebola virus or smallpox. Two decades on, Ebola virus now has been created in the lab, and so has a close relative of smallpox, horsepox. Today, it doesn’t take several hundreds of thousands of dollars to manufacture these viruses. Instead, the step-by-step protocols and required synthetic DNA are available online, and not every purchase of synthetic DNA is screened for potential hazards such as sequences corresponding to harmful pathogens.

This lack of sufficient safeguards allows an increasing number of skilled individuals to edit and engineer dangerous pathogens, exposing us not only to the risk of deliberate misuse (biosecurity risks) but also to the risk of accidents (biosafety risks). The U.S. government has taken important steps to institutionalize safety, security, and risk-mitigation practices, but much work remains.

It is worth noting that the risk landscape extends beyond biosafety and biosecurity risks. The term “high risk” in ARPA-H’s high-risk, high-reward approach to innovation traditionally refers to the willingness to take high technical risk: The agency acknowledges that ground-breaking technological approaches might fail due to unforeseen technical hurdles. In fact, failure is expected and demonstrates that the agency is being sufficiently ambitious. But although ARPA-H is equipped with a sizeable budget ($1.5 billion in 2023), its resources are ultimately finite. So how can ARPA-H determine which technical risks are worth taking?

This is where the secret sauce of every ARPA — the Heilmeier Questions — come in. These questions are designed to help program managers clarify their goals, strip away jargon, and substantiate the broader implications of their work. Program managers are subject-matter experts, borrowed from industry or academia for a term-limited period, who spearhead the agency’s technical programs. When pitching a research program, they must answer these questions:

  • What are you trying to do? Articulate your objectives using absolutely no jargon.
  • How is it done today, and what are the limits of current practice?
  • What is new in your approach and why do you think it will be successful?
  • Who cares? If you are successful, what difference will it make?
  • What are the risks?
  • How much will it cost?
  • How long will it take?
  • What are the midterm and final “exams” to check for success?

ARPA-H leadership appears to have learned from DARPA’s past of both regrettable errors and remarkable foresight. They even drafted their own version of the Heilmeier Questions. The classic question, “What are the risks?” received an addendum, now stating, “Identify any risks that may prevent you from reaching your objectives, as well as any risks the program itself may present.” ARPA-H also added an entirely new question, asking, “How might this program be misperceived or misused (and how can we prevent that from happening)?”

Misperception risks can be limited by proactively engaging in public outreach and combating misinformation. Yet biosafety and biosecurity are a different challenge altogether. Although misuse events or accidents associated with ARPA-H products might be rare, a single incident could result in catastrophic outcomes. Thus, it will be essential to consider and mitigate safety and security risks upfront, rather than after-the-fact. We believe the most ambitious technologies are often also the safest ones: We should be aiming to create the health equivalent of a comfortable and reliable passenger jet built with safety in mind, not simply a bare-bones aircraft prone to failure.

Many health technologies can be harmful in their early stages of development, but are wonderful tools once they can be applied safely. X-ray technicians who used their bare hands to calibrate machines at the turn of the century often developed cancer requiring amputation. Now, no hospital is complete without radiographic imaging capabilities. In fact, many aspects of the U.S. healthcare system have been built with safety in mind: Regulatory bodies like the FDA are the manifestation of our collective commitment to protecting patients from adverse outcomes. Before the FDA was established in 1906, many drugs were completely useless, or worse, actively harmful. Today, novel medicines must prove their safety and efficacy in clinical trials, and ARPA-H’s products are no exception. The agency will fund clinical trials and closely collaborate with the FDA, ensuring that its cutting-edge products reach patients in a timely and safe manner.

In contrast to the FDA approval process — which ensures the safety of individual patients — accidents or the misuse of biotechnology could pose a significant public and even global health risk. Therefore, ARPA-H should take a proactive stance on safety and security risks throughout the entire R&D lifecycle. Promisingly, agency leadership has thus far exhibited a prudent awareness of the need for anticipating and mitigating these risks — it is hard to overstate the importance of their decision to update the near-canonical Heilmeier Questions. However, to fully actualize a forward-thinking biosafety and biosecurity framework, a few more steps should be taken. The Heilmeier Questions, while valuable, focus only on risks apparent before a project’s inception. And they miss some important aspects, such as accident risk. To lead the world in both transformative innovation and pioneering safety, ARPA-H should:

Establish a Research and Technology Protection office. Like the intelligence ARPA (IARPA), ARPA-H could create a Research and Technology Protection (RTP) office tasked to identify biosecurity risks. This office, potentially staffed by just one or two full-time experts, could help safeguard ARPA-H’s innovations throughout each project’s lifecycle and beyond.

Foster collaboration with security agencies. As a civilian agency, ARPA-H may be more vulnerable to espionage or cyberattacks. By granting staff security clearances and promoting collaboration between the RTP office and security agencies, ARPA-H can enhance its security posture — especially as biotechnology emerges as a key area of national security.

Include an addendum to the ARPA-(H)eilmeier Questions. ARPA-H published a number of “hidden questions” to help their PMs answer the Heilmeier Questions. Yet this guide could be improved further. First, there is no mention of accident risk, despite laboratory-acquired infections being surprisingly common. Second, we believe program managers deserve more support to answer the risk-related Heilmeier Questions. While they are technical subject-matter experts, few of them will have received formal training in security risk assessment and mitigation. Support from biosafety and security experts is something that other government agencies like the NIH provide.

Prior instances of successfully mitigating risks from technology show us it is achievable and worth the effort. For example, scientists substituted ozone-depleting CFCs with less harmful alternatives. Experts are developing post-quantum cryptography resistant to attacks by quantum computers before building actual quantum computers, as they anticipate the risks posed to our information security systems. And automakers and policymakers have equipped cars with a number of vehicle safety technologies, including seatbelts, airbags, and front headrests, making them substantially safer.

Thus, ARPA-H could support its program managers by incorporating the following informal questions into its guide:

  1. How can we prevent, mitigate and monitor for accidents?
  2. Can we use one or more of the following strategies to prevent, mitigate and monitor for misuse and unintended consequences? There are three subquestions here: First, can we substitute this technology with a lower-risk alternative? (For example, substituting ozone-depleting CFCs with less harmful alternatives.) Second, can we advance a risk-decreasing technology before a risk-increasing one? (For example, developing post-quantum cryptography before building quantum computers.) Finally, can we integrate safety mechanisms? (For example, equipping cars with safety mechanisms like airbags.)

It’s worth emphasizing that awareness of safety and security risks and ambitious technical risk-taking are not mutually exclusive. DARPA innovations like disaster response robots or airbag sensors have advanced the frontiers of both technology and safety. ARPA-H can redefine what it means to be a leader in responsible progress while maintaining the technical aspirations woven into its very essence. In doing so, ARPA-H has the potential to transform health, ultimately shaping a brighter and more equitable future for all.

Janika Schmitt is a biosecurity fellow with the Institute for Progress. Jacob Swett is an advisor to Open Philanthropy on biosecurity and pandemic preparedness. Jassi Pannu is a resident physician at the Stanford School of Medicine and a fellow at the Johns Hopkins Center for Health Security.