Uncontrolled spread, p.40
Uncontrolled Spread, page 40
We need to look across all of our capabilities to make sure that such an event cannot happen again. Other nations are looking to their national security instruments to identify and reduce pandemic risks. We will need to make the same adjustments. Yet, as we go down this same path, we need to make sure we engage our tools of national security in a way that preserves our multilateral institutions and the work of scientific agencies. We need to continue to invest in capacity building in nations with whom we collaborate. We need to make sure that scientific exchange doesn’t become encumbered. Agencies like the NIH don’t want to be burdened with having to handle reams of classified information. Researchers want to know the provenance of any samples they receive and the environment in which it was collected so they can properly assess its usefulness. The rules of engagement will need to be carefully worked out.
The flip side is that embedding public health information in our intelligence reporting could elevate the work of our scientific agencies, allowing this analysis to get in front of policymakers in a form and function where it’s more actionable. For policymakers, making good use of intelligence requires repetition. You need to be able to read intelligence estimates over time to develop a cadence for the information and how threats evolve. When it comes to threats from emerging diseases, policymakers aren’t reading academic papers. It’s not how they digest information that’s outside their domain. But they are reading the CIA’s World Intelligence Review and other national security estimates. If more health reporting were regularly included in these assessments and treated on par with other threats, it would keep it in front of policymakers, so they would be in a more informed position to secure the nation against biological risks.81
The components of HHS have access to a lot of knowledge that fits into the national security dialogue, but most of these operating units don’t view this information through that prism. It’s not the world they live in. The national security agencies could help fit this reporting into a broader mosaic that would help identify risks. At the same time, intelligence agencies could better hone their collection and reporting through collaboration with health experts who could provide context to help guide intelligence collection. The National Security Agency scoops up an enormous amount of signals intelligence but may not know precisely what to look for without the help of health experts, or how to interpret the information they gather. If the NSA found itself in possession of laboratory data, for example, it might not have a clear sense for how to pull this information out of the signals intelligence or how to make an initial analysis to determine its importance; but collecting these data would fall feasibly into their existing mission given that most of this information is now shared digitally.
Leveraging our intelligence assets to inform us better on risks would also complement our diplomatic tools. More information could be shared with our diplomats to let them target their own efforts around the US public health mission. If intelligence estimates identified certain weaknesses in a foreign country’s scientific institutions, our diplomats would be in a better position to know where to focus their efforts on capacity building. Perhaps it’s a foreign lab with poor internal controls or certain regions where surveillance is especially weak. These sorts of missions, where diplomatic initiatives are being fed by intelligence agencies, has been a hallmark of arms control efforts, nuclear nonproliferation, and nuclear inspection activities and a similar approach can help improve our biosecurity and surveillance as well.
When it has been proposed to public health officials that they should work more closely with intelligence services, and they express reluctance, the unwillingness often turns on their contempt of an episode where the CIA was reported to have administered a fake hepatitis B vaccine campaign in an effort to collect DNA from Osama bin Laden’s children as a way to target his location in Pakistan in 2011.82 A local physician, Shakil Afridi, was reported to have organized a vaccine drive, first in a poorer neighborhood, to appear “more authentic.” Public health leaders denounced the scheme and its potential impact on legitimate vaccination campaigns, including the ongoing effort to eradicate polio.83 Far from enabling these kinds of potential incidents, if public health experts collaborated more closely with our intelligence community it might help forestall such an approach. In such a case, the public health perspective may have been woven into operational planning early and closed down the idea of using a vaccination campaign as cover for intelligence gathering because of the impact that the tactics would have on overarching public health goals. The public health imperative could prevail, and intelligence operators might find a different way to pursue a mission that didn’t risk undermining a legitimate vaccination effort.84 Had public health officials been at the table, they might have discouraged such an effort at the outset. The more engaged intelligence services are with public health experts, the more they’ll be cognizant of the lines that shouldn’t be crossed in trying to balance security interests with the fundamental goals of public health. In May 2014, the Obama White House announced that the CIA would no longer use vaccination programs as a cover for espionage.85
COVID showed the importance of timely, reliable reporting. Consider two hypothetical scenarios that might have unfolded in the early days of the outbreak, when the virus was still localized within China. In the first scenario, US intelligence officials are briefing the president sometime around December 20, 2019. They tell the president that intelligence agencies are tracking a mysterious outbreak of pneumonia in Wuhan. The intelligence officials are relying largely on open-source material—information they’ve gleaned from WeChat and other social media. They don’t know the characteristics of the virus and haven’t established whether it’s spreading person-to-person. However, they’ve found some troubling postings on social media. Doctors in China are worried. A few local sources have told them that hospitals are seeing a higher number of flu cases than is normal for this time of year, and emergency rooms are full. Taken together, US agencies believe it could be an outbreak with a novel virus.
Is that enough to instigate any decisive action in the US? Probably not.
Now consider the second scenario. Intelligence officials tell the president that they were made aware of an outbreak in Wuhan based on the same open-source material. Digging deeper, they’ve been told by local assets, who work inside Chinese medical facilities, that ten healthcare workers have already become infected with the virus. This is a crucial piece of information that strongly suggests the pathogen is capable of spreading between people. Additionally, they’ve obtained samples of the virus and sequenced it. The genetic diversity between the different specimens suggests that the virus has been spreading for more than a month. Moreover, comparing the novel virus to a database of known pathogens, they’ve assessed that the virus bears disturbing similarities to SARS-1 and is likely deadly and highly contagious.
If US officials had had access to that second briefing rather than the first, it could have instigated actions that would have given us a monthlong head start on COVID. We squandered a lot of opportunities to get an edge on the epidemic, so it’s not clear that, at least in this instance, more time would have made a significant difference in our preparations. But in the hands of federal officials who are astute to these threats, a month’s head start can be enormous.
The other piece of this puzzle, in addition to the clandestine capabilities, is the sequencing data. The COVID pandemic was the first time that scientists used sequencing at a massive scale to evaluate the evolution of a virus’s pathogenicity and trace its spread. In the future, this practice will be an essential tool for doing routine global surveillance.86 Leveraging these data as a way to get a head start on gauging a pathogen’s pandemic potential will require a more reliable means to gain access to samples, and then to link the unique characteristics of a virus’s sequence with clinical information that can predict how these features correlate with its transmissibility and lethality.
With the current technology it’s hard to deduce from just a sequence alone the behavior of a virus. Still, some things can be inferred. For example, we might be able to surmise from the RNA sequence, combined with a knowledge of the structural biology, that a novel pathogen has the potential to bind well to human tissue. That can be a key insight that tips us off to its ability to cause disease. In the case of SARS-CoV-2, there was early evidence that it bound well to the ACE2 receptor in the human respiratory tract. Its sequence also resembled SARS-1. Those two findings gave some key insights that the novel strain might pose a serious danger. These insights could also help with the early development of treatment options or the creation of ways to identify spread through the monitoring of symptoms. And as we expand the quality and quantity of sequence data that we have—correlating different genetic features with how viruses behave, how contagious they are, and how likely they are to cause disease—then the usefulness of sequence information as a predictive tool will greatly expand. Given the enormous potential of this approach, we need to build on these capabilities.
Achieving this goal will require a massive amount of data to better connect key genetic features found in sequencing data with their clinical significance. More information will be needed on how viruses behave in animals and humans.87 This is the biggest obstacle right now to using sequencing data to deduce a pathogen’s severity and infectivity. We don’t know precisely how small changes in a virus’s genetic sequence will translate into differences in how it behaves once it spreads among people. But as sequencing data get more robust, as we collect and store more information to correlate genetic changes with their clinical significance, and as the ability to use computational methods to make these predictions continues to improve, so will the practical application of these tools. We need to look toward this future and invest in these capabilities.
The first time that sequencing was used to help guide the investigation and response to an epidemic was with Ebola in 2014. Owing largely to improvements in the technology for sequencing, scientists at the Broad Institute were able to use sequencing information derived from Ebola samples collected in the field in West Africa to trace patterns of spread. This information, in turn, helped public health workers better understand how the virus was being passed from person to person, and gave them new ways to intervene to break off the chains of transmission.88 But COVID was a dramatic step forward. It was the first time scientists used sequencing at a massive scale, in a near real-time fashion, to trace the spread of individual cases, characterize an epidemic, and monitor its expansion.89 The biology and the informatics will improve to the point that we’ll be able to make better predictions based on sequence information alone. Even partially predictive assessments are valuable if they can be made early, when new strains and variants first emerge.
There are other advantages offered by these capabilities.
When a new cluster of infections first emerges, we’ll be able to deduce whether the infections are the result of transmission between people or repeated introductions from a single animal source to different people. The latter scenario was the first theory advanced by China to explain the spread of COVID. It was wrong. For weeks the prevailing belief was that the initial infections were each the result of people interacting with an infected animal in a local food market and not the result of person-to-person spread. As we’ve seen, in the right hands, even a few weeks’ head start on recognizing that the virus was spreading between people can give us a crucial edge at containing it.
The ability to use sequencing to monitor for emerging pathogens, and evaluate their potential harm, is not a new idea. It was envisioned by the National Strategy for Pandemic Influenza released in May 2006, but no federal agency has ever fully implemented this vision as a strategic priority.90 Back then, the technology didn’t exist to make such a framework practical. The advent of fast, accurate, and less-costly sequencing, along with better experimental evidence, finally makes it more obtainable. In 2007, the cost of sequencing a single human genome was about $10 million, according to the National Human Genome Research Institute; today it’s under $1,000, and most predict the cost will soon fall below $100.91 For sequencing a virus’s genome, the costs are typically about one-tenth the cost for sequencing a human genome. In recent years, public health agencies have been incorporating pathogen genome sequencing into their infectious disease surveillance with support from the Advanced Molecular Detection program, which Congress established at the CDC in 2014.92 The FDA also relies on its GenomeTrakr network, the first distributed network of labs that use whole genome sequencing for pathogen identification in instances of outbreaks of foodborne illness. However, most of the prior uses for sequencing were retrospective. The tools were used to conduct deeper analysis of outbreaks that had already been identified or resolved. Such work can be valuable because it allows researchers to uncover outbreaks that they may not have previously recognized, which can yield important insights into the risks posed by pathogens and how they’re transmitted. However, until COVID, sequencing was never deployed in a large scale to identify the origins of a burgeoning outbreak and help contain its present spread. Information on viral genomes was never widely exchanged outside the setting of influenza. During COVID, the framework that enabled viral sequence data to be shared around the world was a system originally built for flu surveillance, the Global Initiative on Sharing Avian Influenza Data (GISAID). We now know that such a collaborative network can be used to exchange information on other viruses.
At proper scale, sequencing can be used in a near real-time fashion to identify emerging threats and to predict their activity.93 At Harvard, a project is under way to create a sentinel surveillance system that could detect novel strains of viruses before they evolve into the next pandemic. Started in response to COVID, the idea is to deploy sequencing equipment in hot zones to allow novel pathogens to be evaluated when they make their first jump into man. By uncovering these dangers early, and sharing the information quickly, the hope is that such a system can serve as a tripwire that will give us the capability to head off the next pandemic.94 The CDC also committed to sharply expanding its use of sequencing for sentinel surveillance in the US. The agency is looking for ways to integrate its surveillance into municipal and national planning, such as through sequencing wastewater. The CDC launched a project called the National Wastewater Surveillance System (NWSS) that is monitoring for variants of SARS-CoV-2, but also starting to look for new viruses that may be circulating in the population.95 Because we shed many viruses into our feces, city wastewater can serve as a “liquid biopsy” to help inform us of emerging outbreaks.
It now appears that COVID may have been spreading widely in China, and had broken out of the country to regions with close ties to Wuhan, much earlier than we first suspected.96 In Italy, the first known COVID case was reported in the town of Codogno in Lombardy on February 21, 2020, but since then, a few studies have suggested it may have been spreading earlier, including one study that found a positive sewage sample in northern Italy in mid-December 2019 and another that detected the novel coronavirus in a banked patient sample that was taken in the same region in early December. It was from the throat culture of a child first suspected of having measles.97
In France, another banked throat swab, this one from a patient with pneumonia who was admitted to a hospital on December 27, 2019, was later found to have SARS-CoV-2 RNA in it.98
In Brazil, testing of sewage found SARS-CoV-2-positive results in samples collected on November 27, 2019, much earlier than the first reported case in the Americas.99
If these and similar findings are true, the window to detect the virus’s initial jump into people may have been open much longer, along with the opportunity to intervene to avert a global pandemic. Better surveillance, using sequencing tools, might have uncovered its spread earlier. In the future, using these approaches, we can build a system that alerts us sooner to when a new virus has gained the capacity for human transmission, so we can start expanding our surveillance, expanding our testing, and developing therapies and vaccines, much earlier.
The ability to turn raw sequence data into actionable, digital information is a complex and computationally intensive endeavor. It will turn on the availability of a well-curated and up-to-date database of sequences that can serve as a reference for comparing novel strains to known viruses, along with information on how genomic features correlate with the clinical behavior of different pathogens. COVID proved the importance of having these proficiencies. In addition to being able to identify new viruses, we’ll need these capabilities to monitor known viruses and identify when they mutate and undergo a dangerous evolutionary change. We’ve already seen the appearance of variants that are more contagious and less susceptible to our antibody drugs and vaccines. The UK had in place a systematized approach for sequencing samples and was able to detect these variants in late 2020. The United States had no similar framework, and so we became heavily seeded by the time we started to look for and find these new strains.100 It was an echo of what had happened in February and March 2020, when we lacked the PCR testing to detect the first wave of infection until it was too late.
These efforts should be expanded to focus on all RNA respiratory pathogens. As a category of pathogens, these viruses share a plethora of the basic features that give them pandemic potential.101 RNA viruses mutate relatively quickly, and many, like influenza, are able to undergo a process known as antigenic drift, by which the virus is able to alter the surface antigens that are the targets of our antibodies—thus evading our existing immunity. Some viruses, like measles, cannot change their genomic sequence in ways that substantially alter enough of their surface proteins, so measles remains susceptible to our vaccines or the immunity that we get from prior infection. However, for viruses like influenza, as their surface proteins undergo change, the virus is able to dodge the protective antibodies that we’ve developed from past infection or vaccination.102 This is how flu slips past our immunity every season and why we need to constantly update our vaccines. Being transmitted through respiratory secretions means that these viruses can spread widely. By having a global enterprise for sampling, sequencing, and describing the clinical features of RNA respiratory pathogens and the disease they cause, we can detect troubling variants when there’s still time to target them with new antibody drugs or vaccines.
