Supporting Discovery and Development of Medical Countermeasures for Chemical Injury to Eye and Skin

Abstract

Vesicants, from vesica (Latin for blister), can cause local and systemic toxicity. They include the chemotherapy drug nitrogen mustard and chemical warfare agents sulfur mustard, Lewisite, and phosgene oxime. These agents are commonly released in vapor form and consequently, eyes and skin are the most vulnerable. The ocular and cutaneous injuries can be acute, subacute, or chronic, and can predispose casualties to secondary deleterious effects. Underlying these broad organ responses are shared and tissue-specific cellular and molecular biological cascades that attempt to counteract such chemical injuries. Depending on the severity of the chemical insult, biological responses often lead to inadequate wound healing and result in long-term pathology instead. Exposure to other toxic industrial chemicals such as acrolein, chloropicrin, and hydrogen fluoride, can also cause prominent eye and skin damage. There are currently no FDA-approved drugs to counteract these injuries. Hence, the possibility of a mass casualty emergency involving these chemicals is a major public health concern. Recognizing this critical challenge, the United States Department of Health and Human Services (HHS) is committed to the development of medical countermeasures to advance national health and medical preparedness against these highly toxic chemicals. Here, we provide an overview of various HHS funding and scientific opportunities available in this space, emphasizing parallels between eye and skin response to chemical injury. We also discuss a main limitation of existing data and suggest ways to overcome it.

Introduction:

In the book, “Weapons of Mass Destruction”, Robert Hutchinson describes “man’s inhumanity to man” with the first use of sulfur mustard on the battlefields of Ypres, Belgium, on July 12, 1917 (Hutchinson, 2011). Called mustard gas given its garlic or mustard smell, it can be toxic even at concentrations undetectable by smell. Pure sulfur mustard (SM) is actually an odorless liquid but its impurities give it its name (Rafati-Rahimzadeh et al., 2019). Given SM’s lipophilic nature, it does not dissolve in water, but it adheres to epithelial tissues with their lipid bilayers. SM can also freeze, and once thawed, remains potent. Hutchinson quotes Major Harry Gilchrist, a medical director in the American Expeditionary Force who witnessed the attack, “At first the troops didn’t notice the gas but in the course of an hour or so, there was marked inflammation of the eyes. They vomited and there was erythema of the skin. Later, there was severe blistering and by the time the gassed cases reached the casualty clearing stations, the men were virtually blind”. Predictably, both SM production and the number of vesicating agents rapidly proliferated. By 1919, output of mustard gas in American factories reached 19 tons a day (Mansour Razavi et al., 2012). Lewisite (L), named after U.S. Army Captain Winford Lewis, is a similar but faster-acting war agent which Hutchinson describes as being fatal with just 30 drops on the skin and, “if the eyes are affected, total blindness occurs within one minute”. A blend of nitrogen mustard (NM) and L – called ‘Winterlost’ – had a low freezing point ensuring effectiveness even at frigid Russian fronts (Ledford, 2020). Phosgene oxime (CX), a highly toxic urticant commonly described as a nettle/corrosive/vesicating agent, was subsequently produced solely as a chemical warfare agent (Singh et al., 2021). In 1925, the Geneva Protocol banned use of chemical warfare agents, yet nearly a century later, the threat remains (Sezigen and Kenar, 2020).

Therapeutic options, also known as medical countermeasures (MCMs), to treat vesicant injuries are scarce with only one U.S. Food and Drug Administration (FDA)-approved MCM product to date, in the form of a silver-plated nylon dressing (Public Health Emergency, 2019), but no FDA-approved MCM drugs. Vesicating agents can cause moderate to debilitating injuries and pain to the eye, skin, and mucous membranes. The primary modes of exposure are through contact, inhalation and ingestion. Depending on the dose, route, and duration of exposure, toxic symptoms can range in varying degrees of ocular and dermal burns, blister formation, bronchospasm, dyspnea, pulmonary edema, bronchitis, and immune- and bone marrow suppression (National Institutes of Health, 2017). Due to the potential manifestation of these devastating health effects after exposure, vesicating chemicals have been identified by the U.S. Department of Homeland Security (DHS) and the Department of Health and Human Services (HHS) as highly toxic chemicals of concern to public health security. As such, developing MCMs to address the toxic acute and chronic effects of SM, NM, L, and CX is a critical health security and public preparedness need.

Threats involving the deliberate release of chemical, biological, radiation, and nuclear (CBRN) agents (Tin et al., 2021), spurred the HHS to establish the Public Health Emergency Medical Countermeasures Enterprise (PHEMCE). Launched in July 2006, PHEMCE coordinates the Federal efforts to enhance CBRN and emerging infectious diseases (EID) preparedness from a MCM perspective to advance the nation’s medical response to, and recovery from disasters and public health emergencies. PHEMCE, in turn, engages various U.S. governmental agencies and their components including the Department of Defense, Biomedical Advanced Research and Development Authority (BARDA), the Centers for Disease Control and Prevention (CDC), the FDA, and the National Institutes of Health (NIH) to fulfill its overall mission. On behalf of the NIH, the National Institute of Allergy and Infectious Diseases (NIAID) was tasked with leading the HHS civilian-focused biodefense research initiative supporting discovery and early development of MCMs against CBRN threats.

Complementing the NIAID Biodefense and EID and Radiation and Nuclear Countermeasures programs, the Chemical Countermeasures Research Program (CCRP) was implemented to foster a trans-NIH extramural effort focusing on discovery of novel, safe, and effective MCMs to prevent mortality and serious morbidity after acute exposure to DHS-designated highly toxic chemicals (HTCs) (Yeung et al., 2020a). These HTCs cover multiple toxidromes including ocular and dermal injuries caused by vesicants. Recognizing the need for highly specialized and relevant subject matter expertise, the NIAID/CCRP established collaborative partnerships with the National Eye Institute (NEI) and the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) to administer the overall NIH chemical MCM initiative directed at counteracting eye and skin chemical injury, respectively, two prime target organs for chemical exposure. The lung is also a prime target for vesicant injury but it is outside the scope of this commentary.

Funding Opportunity Announcements:

How does one get to an FDA-approved drug against vesicant injury? Drug development is highly arduous and uncertain. It is estimated that there are 10⁶⁰ molecules in chemical space that are drug-like (Bohacek et al., 1996). In their 2019 review of 185,994 clinical trials between 2000 and 2015, and corresponding to 21,143 compounds, it was estimated that only 13.8% of drug development programs eventually lead to approval (Wong et al., 2019). The probability of success ranged from a low of 3.4% for oncology drugs to a high of 33.4% for vaccines (Wong et al., 2019). Besides time and uncertainty, there is also cost. The drug development process is expensive with the average R&D investment estimated at $1.3 billion per drug (DiMasi et al., 2016). Recognizing these major financial hurdles, U.S. Government agencies provide various funding opportunities and incentives towards de-risking the investment necessary for MCM research and development. NIH and BARDA are the two primary funding agencies within the HHS to advance chemical MCM public health preparedness.

The overall NIH chemical MCM funding is administered by the CCRP together with the trans-NIH partner overseeing the research award. CCRP-supported research awards are typically focused on one or both of the following broad objectives: 1) To understand the immediate and/or long-term pathophysiology associated with acute exposure to HTCs and 2) Apply fundamental knowledge of toxicities to discover MCMs that are safe and therapeutically effective when administered 30 minutes or more after exposure, i.e., post-exposure efficacy. Research progress (or degree of maturity) towards accomplishing the stated objectives is gauged by Technology Readiness Levels 1 through 9 (TRL), with TRL 1 representing a review of scientific knowledge base while TRL 9 representing post-licensure and post-approval activities (Medical Countermeasures, 2022a). TRL scale was developed by PHEMCE partners to objectively evaluate and communicate the status of MCMs and research tools such as assays, models, and reagents. In general, NIH chemical MCM funding supports early-stage research activities – specifically TRL 3 and below.

TRL 3 research activities are primarily hypothesis driven incorporating the current state of the science with investigator-initiated innovations to establish novel research tools and models towards discovery of therapeutic targets and candidates against chemical threats. TRL 3 research is also where the tension between serendipity and rational drug discovery typically comes in. A recent review lists a large number of drugs that were discovered by serendipity including metformin, methotrexate, rapamycin, statin, and of course aspirin (Prasad et al., 2016). The book, “Serendipity: Accidental discoveries in science”, discusses even more surprising examples, including the accidental discovery of the anti-cancer property of nitrogen mustard following the World War II air raid on Bari harbor, Italy (Roberts, 1989).

Chance of course favors the prepared mind and early-stage drug development is likely to benefit when grounded in basic research directed at understanding fundamental mechanisms and natural history of the disease (Kamil and Mohan, 2021; Fuchs et al., 2021). The eyes and skin are usually the first and most frequent route of toxic exposure, making them especially vulnerable to chemically induced injuries. However, compared to skin, ocular chemical injury typically manifests earlier and at lower toxicant concentrations. Dissecting and counteracting the eye’s vulnerability to chemical injury requires traction in at least four domains: (1) Responses to corneal toxicity (acute and chronic); (2) Models of corneal toxic injury; (3) Mechanisms of corneal pathologies and wound healing, and (4) Therapies for corneal pathology. Building on this premise, the NIH recently issued a Request for Applications titled “CCRP Initiative: Chemical Threat Agent-induced Pulmonary and Ocular Pathophysiological Mechanisms” (National Institutes of Health, 2021a). The central objective of the Funding Opportunity Announcement (FOA) is to support basic research directed at understanding mechanisms of chemical toxicity through characterization of fundamental molecular, cellular, and physiological pathways involved in acute and delayed toxicity/pathology and identifying potential targets that can mitigate the ocular injuries caused by chemical threat agents, particularly vesicants. This FOA involves funds set-aside by NIH with per application budget of up to $300,000 direct costs per year for up to 3 years. Ocular chemical injury in general, and ocular vesicant injury in particular, involves a pleiotropic response. As such, projects that attempt to address more than one aspect of the ocular chemical injury cascade spanning pain, photophobia, corneal epithelial defects, keratitis, endothelial cell loss, edema, inflammatory response, conjunctivitis, tear disruption, neovascularization, corneal scarring, opacity, through blindness would be particularly responsive.

Not surprisingly, the dermal chemical MCM community is similarly limited and primarily consists of toxicologists rather than skin biologists, dermatologists, and other subject matter experts in cutaneous regenerative medicine, wound healing, or skin disease management. While the current toxicologist-driven applied/translational research approach has been effective in evaluating candidate chemical MCMs, much of the obvious and immediate approaches, i.e., the low-hanging fruits, may have been largely exhausted and new therapeutic targets, ideas, and approaches are needed to innovate and sustain the dermal injury portfolio. Consequently, the NIH recently issued PAS-21–245, titled, “Limited Competition: Promoting a Basic Understanding of Chemical Threats to Skin” (National Institutes of Health, 2021b) to engage the skin research community to: 1) Contribute to the basic understanding of vesicant chemical injuries, 2) Explore the local and systemic mechanisms as they relate to chemical wound development, healing and long-term consequence of the injury, and 3) Understand skin biology in the context of whole-body physiology. This FOA is intended for investigators who are not currently supported by the NIH chemical MCM initiatives but whose research interests are relevant to dermal injuries and to skin wound healing in general. It is assumed that the proposed project will represent a new direction into dermal vesicant injury for the applicant, which requires the research lab to acquire new skills (e.g., handling toxic chemicals), gather new resources (e.g., establishing contractual relationships with facilities certified to utilize restricted/controlled chemicals), develop multidisciplinary collaborations, and generate new preliminary data. Therefore, the overall goal of the 3-years R34 Planning Grant is to support the preparatory steps toward entry into this new line of research. Applicants do not need to present preliminary data supporting the proposed study in concept nor publication records and specific skills and proficiency in experimental execution in the chemical MCM field. However, applicants should demonstrate prior experience in wound healing research and general knowledge relating to skin injury and repair.

In addition to the previously mentioned FOAs, the CCRP is also soliciting applied/translational research projects with pre-identified MCM candidates ready for exploratory studies. Such exploratory research includes, but not limited to, refining mechanisms of action for MCM intervention based on the potential clinical indications of interest and demonstration of proof-of-principle animal efficacy. Ideally, these projects should have an animal model(s) relevant to the human disease or chemical injury identified and tentatively characterized to assess the ability of the candidate MCM to induce a biological response with qualitative endpoints such as enhanced survival and/or reduced major morbidity. Applicants interested in this FOA should review “Countermeasures Against Chemical Threats (CounterACT) Exploratory/Developmental Projects” (National Institutes of Health, 2020). Projects supported by this FOA are expected to generate preliminary data that would facilitate the development of competitive applications for more extensive support upon completion. Complementing this FOA, the CCRP is also leveraging the NIH Small Business Education and Entrepreneurial Development (SEED) program to solicit and support the innovator community to advance healthcare products that could be directed towards the chemical MCM field (SEED, 2022a). To learn more about this public-private partnership opportunity, see “PHS 2021–2 Omnibus Solicitation of the NIH and CDC for Small Business Innovation Research Grant Applications” (National Institutes of Health, 2021c; 2021d) and the NIAID section of the “NIH, CDC, and FDA Program Descriptions and Research Topics” (SEED, 2022b).

Accompanying the NIH funding strategies to support discovery and early development of chemical MCMs, BARDA offers several public-private partnership opportunities with industry as well. More specifically, BARDA uses its advanced research & development authorities to help innovators develop promising candidate MCMs from preclinical development to clinical trials and manufacturing scale-up to FDA approval. Based on the current BARDA Broad Agency Announcement (BAA) solicitation, the agency is particularly interested in development of MCMs that limit harmful aspects of exposure to vesicating agents such as SM and L (Medical Countermeasures, 2022b). Preference is given to therapies with the potential to prevent or ameliorate the chronic effects of vesicant exposure (Medical Countermeasures, 2022c). In general, BARDA prioritizes MCM candidates in more advanced stages of development over those in earlier stages. The minimum TRL for Chemical MCM candidates should be at TRL 4 or higher for the relevant indication. Hence, in vivo efficacy consistent with the product’s intended use as an MCM against a threat agent (i.e., dose, schedule, duration, and route of administration) must already be demonstrated. More advanced candidates that have progressed into and completed some clinical development studies (i.e., Phase 1 or 2) and have achieved manufacturing at a scale greater than benchtop are preferred. Strong preference will be given to drug candidates that are already approved or are in late-stage clinical development for a conventional indication which has similar symptomology to that arising from exposure to a chemical agent (Medical Countermeasures, 2022b). In addition to the BARDA BAA, applicants seeking to identify new uses for FDA approved or late-stage investigational therapeutics that are outside of their original clinical indication, i.e., repurposing or label expansion, should consider opportunities within the BARDA Division of Research Innovation, and Ventures ReDIRECT program (HHS). ReDIRECT is focused on identifying “treat the symptom” MCM approaches that is agnostic to the cause of the injury/identity of the chemical itself. Repurposing MCMs holds the potential to greatly enhance national health preparedness and medical response capabilities to chemical threats as the therapies would be easily accessible in the community during an emergency while limiting commercial and inventory risks typically borne by the product sponsor and vendor.

Other Support Mechanisms:

Chemical vesicant research generally involves highly toxic substances. Due to special biosafety concerns, access to chemicals such as SM, L, and CX are highly regulated and restricted to only a few biosurety laboratories across the nation. This in turn leads to a significant obstacle in obtaining adequate preliminary data that would enable submission of a competitive application to the NIH peer review system. With this in mind, the NIH created the CounterACT Ocular Therapeutics Screening (COTS) program. The COTS program employs a screening model in rabbits to determine the efficacy of investigational compounds in mitigating SM-induced ocular damage. The primary purpose of the COTS program is to provide investigators with pre-application, pilot proof-of-principle efficacy data in support of potential follow-on research efforts. The application process for the COTS program is straightforward and involves submission of a brief application and supporting data to justify the study investment. This is reviewed administratively by NIH staff. Once approved, NIH will cover the costs associated with the study while the participant will retain custody and have primary rights to the data developed. The applicant is only responsible for provision of the candidate MCM and proposed treatment paradigms. Thus, an investigator with a promising vesicant MCM can have that candidate tested in vivo under two different treatment paradigms at no cost. Three groups, each composed of up to ten New Zealand rabbits, are used to evaluate the efficacy of the MCM candidate for its ability to ameliorate SM injury to the eye (up to 56 days post-exposure). Outcome measures include effects on corneal stromal injury; neovascularization; eyelid notching/damage; and corneal thickness/ulceration. Fixed tissues, such as the eyes with optic nerve, may be collected and provided to the applicant when requested. For more information, please visit: https://www.nei.nih.gov/grants-and-training/funding-opportunities/programs-and-research-priorities/counteract-ocular-therapeutics-program

Eye and Skin – CounterACT Centers of Excellence:

Rubor, tumor, dolor. Redness, swelling, pain. Classic signs of inflammatory response. Not surprisingly, being the first tissues to encounter chemical exposure, both the ocular surface and dermal surface share overlapping responses. Collaborative team projects that simultaneously tackle the multifaceted problem of chemical injuries in both eye and skin are therefore likely to fertilize and cross-inform each other. NIH recently funded two teams with that aim in mind. The Northwestern University CounterACT Center of Excellence (NUCCX) (NIH Reporter, 2022a) was established to test various interventions including PLGA-IMPs (poly lactic acid-glycolic acid immune modifying nanoparticles) with vitamin D3 against nitrogen- and sulfur-mustard injury to skin and eye. High-density lipoprotein nanoparticles (HDL NPs) will also be tested (Lavker et al., 2021). The Rutgers University CounterACT Center of Excellence (NIH Reporter, 2022b; Laskin et al., 2020) was established to test FDA-approved drugs, known to enhance DNA repair, against vesicant-induced skin injury, while other FDA approved drugs, known to suppress immune response, being tested against vesicant-induced eye injury. A significant advantage is that repurposing an FDA-approved drug may considerably shorten drug approval time.

Eye and Skin – Parallels and Intersections:

There is clear need for improved therapeutic approaches against chemical injuries (McNutt and Mohan, 2020). Novel therapies as well as repurposed drugs are promising. Nonetheless, it is abundantly clear that tissue responses to chemical lesions are complex, non-linear and can involve both acute and chronic stages that vary for many known reasons, such as concentration and duration of exposure, and unknown reasons. Redundancy is a known feature in biological systems. It is estimated that ~26% of human genes are duplicates originating from ancient whole-genome duplication events (Kuzmin et al., 2022). Andreas Wagner discusses in his book, “Arrival of the Fittest”, the concept of robustness (Wagner, 2014). As an example, he points out that the bacterium E. coli is capable of using more than 80 different molecules as its only source of energy. This redundancy, pleiotropy, penetrance, robustness and inherent biological noise make the common reductive approach of knocking out one gene at a time to understand the gene’s function, or testing one intervention at a time to rescue some phenotype of interest, liable for generating an insufficient response or even confusion. Case in point: despite examining various anti-inflammatory, anti-fibrotic, anti-oxidant, anti-neovascularization, antibiotic, and even cell-based therapies, there is still no FDA-approved drug against vesicant injury whether to the eye or the skin (Sun et al., 2018; Timperley et al., 2021).

Given the challenge in understanding the complexity of tissue responses to chemotoxic injury – pursuing high-throughput screening approaches becomes important (for example, see Ruff et al., 2016). Similarly, microphysiological tissue devices, including microfluidic organ-on-a-chip, provide powerful screening technologies (Kwak et al., 2020). These allow the extension of analysis beyond vesicant exposure into other toxic chemicals such as acrolein, chloropicrin, and hydrogen fluoride, among other threats (Gupta et al., 2021; Singh et al., 2021). Several in vitro corneal models are being developed under the principle of the 3Rs with its Refine, Reduce, Replace foci that can reduce the need for animal experimentation (Citi et al., 2020). In silico models also help in this aim and have been applied to the skin (Ta et al., 2021; Harding et al., 2021).

Optimizing screening technologies is a key step in tackling the biological complexity of response to chemical injury. Yet beyond this practical step, there is a conceptual obstacle that can impede progress. Namely, settling for the expedience of labeling rather than the prudence of longer-term investment in true understanding. To quote Derek Leebaert, “We live in a world to be labelled not understood” (Leebaert, 2010). So how can we move from labeling to understanding? From knowing fragmented detail of biological responses to chemical injury, some of which specific to the model system being used, to understanding the true multifaceted response to chemical injury and wound healing. A cohesive view is needed.

Labeling vs. Understanding:

We discussed elsewhere the attractiveness of looking for linear cause-effect relationships in biological responses to ocular chemical injury (Araj et al., 2020; Yeung et al., 2020b). But with the majority of the low-hanging fruit already harvested, especially in pinpointing various signaling pathways and their effectors in tissue responses to injury – other approaches become vital. The last decade or two witnessed a near revolution in machine learning algorithms especially those that involve deep learning. The rapid accumulation of big data, essential for training these models, combined with powerful graphics processing units, essential for massive underlying algebraic computations, led to remarkable advances using classification, clustering and regression applications in multiple domains including image, voice, text and even natural language processing. A recent spectacular example was the 2021 Breakthrough of the Year: DeepMind’s structure determination for 350,000 proteins found in the human body representing 44% of the known total (Service, 2021).

But of course, deep learning is not deep understanding. To give an analogy, a middle schooler may learn how to accurately solve a quadratic equation by applying an algorithm, say “completing the square”. But that same middle schooler may still not understand why that algorithm works or why its generalization in the quadratic formula is doing the exact same thing. This brings up a major challenge faced in understanding and counteracting chemical injury: the flood of scientific information that needs to be continually condensed and incorporated into a faithful representation of the biology and pathobiology. The NSF reports that from 2008 to 2018, publications in the Scopus database grew from 1.8 million to 2.6 million articles per year (National Science Foundation, 2022). This information overload does not even account for the explosion of preprints (Xie et al., 2021) nor the reproducibility crisis (Turkyilmaz et al., 2020).

Faced with this deluge of scientific reports, it becomes easier to label rather than to understand. The temptation is to put neat, seemingly well-circumscribed labels, on biological responses to chemotoxic injury: vesication, inflammation, fibrosis, etc. Certainly, vesication can be seen in both eye and skin injury. SM can cause the corneal epithelium to separate from the underlying stroma, just like it can cause the dermal epithelium/epidermis to separate from the underlying dermis. Inflammation is there too manifesting as edema and likely involving both innate and adaptive immune reactions as seen in the lung (Boskabady et al., 2011). Fibrosis is also a common outcome of chronic injury or inadequate tissue repair and wound-healing. Beneath such potentially misleading precise labels – there is a potentially equally misleading view of the underlying signaling pathways be it cell-adhesion molecules, pro-inflammatory chemokines and effectors, remodeled extracellular matrix (ECM) proteins, or myriad signal transducing ligands, receptors and second messengers. It is easy to assume a direct causality from a molecular pathway to an injury phenotype/label. The Nobelist Daniel Kahneman calls such bias the “automatic search for causality” and points out that, as with other cognitive fallacies, it remains ingrained even when one recognizes it for what it is (Kahneman, 2011).

Between the seeming clarity of gross morphological responses to ocular and dermal chemical injury (e.g., vesication, inflammation, fibrosis), and the seeming precision of histological and biomarker readouts and assays (e.g., TGFβ, FGF, other growth factors, interleukins, MMPs, etc.), there appears to be an “explanatory gap”, to borrow the words from the book, “Possible Minds: 25 ways of looking at AI” (Brockman, 2019). Explanatory gaps abound: How do the molecular changes explain the latency that is characteristic of mustard exposure? Why does mustard gas keratopathy (MGK) sometimes develop decades after initial exposure? How do angiogenic factors like VEGF, thought to drive corneal neovascularization, cross-talk with the inflammatory response? How do the differential regenerative abilities of the various corneal and dermal layers affect the acute and chronic phases of the chemical injury? Why do promising monotherapies frequently yield only partial efficacy or even vanishing efficacy? How do the cellular networks and feedback-loops interplay with both the mechanism(s) of injury and mechanism(s) of therapy?

In their paper, “What does it mean to understand a neural network?”, AI researcher, Timothy Lillicrap, and neuroscientist Konrad Kording argue for the need of an intermediate level of understanding (Lillicrap and Kording, 2019). In a sense, a mesoscale. To give an example of a very popular deep learning algorithm used for image classification, convolutional neural networks (CNN), one can understand at a high level the concept of an artificial neural network extracting features with layers, nodes, weights, backpropagation and such, as well as abstract concepts like matrix multiplication, convolution, activation functions and the like. Unfortunately, there is an effective gap between the languages used for the two levels that precludes intuitive understanding. Similarly, there is a gap between the language used to describe pathophysiology of chemical lesions and that used to describe gene/protein effectors (especially when the latter are analyzed in reductive isolation). In sum, apparent knowledge exists; real understanding on the other hand is more elusive.

How does one proceed around such obstacle? Generating more and more data is certainly attractive and indeed a key part of the solution. But besides data generation – data mining should also be highlighted especially given the impressive recent accomplishments of deep learning. A recent example from Russell Poldrack and colleagues applied machine learning and natural language processing (NLP) techniques on nearly 20,000 human fMRI neuroimaging articles published over the past 25 years (Beam et al., 2021). The team mapped a data-driven framework for neurobiological domains. This involved clustering brain structures into circuits, based on the similarity of their co-occurring mental functions, and then mapping functions back onto the circuits. k-means clustering values at 6, 8 and 22 levels generated the best metrics.

Performing similar NLP analysis on chemical injury publications, especially when done across silos, may yield similar compression of existing data and may lead to deep understanding of the vexing problem of chemical injury. Maybe by the 2025 centenary of the Geneva Protocol that banned use of chemical warfare agents, an integrated approach to counteracting chemical weapons would be named a Breakthrough of the Year.

List of Acronyms:

BAA

Broad Agency Announcement solicitation

BARDA

Biomedical Advanced Research and Development Authority

CBRN

Chemical, biological, radiation, and nuclear agents

CCRP

Chemical Countermeasures Research Program

CDC

Centers for Disease Control and Prevention

COTS

CounterACT Ocular Therapeutics Screening

CounterACT

Countermeasures Against Chemical Threats

CX

Phosgene oxime

DHS

United States Department of Homeland Security

ECM

Extracellular matrix

EID

Emerging infectious diseases

FDA

Food and Drug Administration

FOA

Funding Opportunity Announcement

HHS

United States Department of Health and Human Services

HTCs

Highly toxic chemicals

L

Lewisite

MCM

Medical countermeasures

MGK

Mustard gas keratopathy

NEI

National Eye Institute

NIAID

National Institute of Allergy and Infectious Diseases

NIAMS

National Institute of Arthritis and Musculoskeletal and Skin Diseases

NIH

National Institutes of Health

NLP

Natural language processing

NM

Nitrogen mustard

PHEMCE

Public Health Emergency Medical Countermeasures Enterprise

RFA

Request for Applications

SEED

NIH Small Business Education and Entrepreneurial Development

SM

Sulfur mustard

TRL

Technology Readiness Level