4

Innovative Approaches to Accelerating Lyme IACI Research

Many people living with Lyme infection-associated chronic illnesses (IACI) suffer from debilitating symptoms. There is an urgent need for effective treatments to improve symptoms, function, and quality of life for those affected. Innovative research approaches can accelerate progress toward safe and effective treatments and address the key research questions outlined at the end of Chapter 2. Many of the approaches discussed in this chapter are already being used in Lyme IACI research, but their implementation has been limited thus far. Greater coordination can increase the likelihood that these approaches will be used without duplicated or competing efforts and can also maximize the efficiency of financial resources. To promote such coordination, this chapter offers a common framework for the numerous funder, research, and patient organizations to identify shared research priorities for Lyme IACI treatment studies.

A COMMON FRAMEWORK TO PRIORITIZE LYME IACI RESEARCH

Clinical case reports, small or uncontrolled trials, and anecdotal evidence from people living with Lyme IACI and their care providers have identified therapeutic interventions that may be worthy of further investigation in the effort to understand these interventions’ safety, efficacy, and potential for development into approved treatments. The finite nature of funding and other resource constraints generally necessitates prioritization of the areas of research with the greatest potential to improve symptoms, daily function, and quality of life.

Many factors may contribute to or influence what evidence is assessed, how the evidence is weighed, and how these considerations translate into research or funding decisions. Various priorities may be pursued by different research groups or funders in parallel; nonetheless, it is helpful for the field to agree on a general approach to identify research priorities to pursue and to understand the rationale in arriving at these decisions. The committee proposes a framework for research prioritization that outlines a systematic approach to evaluating the available evidence and making decisions (Figure 4-1).

Specifically, this proposed framework is intended to provide a balanced strategy to identify and evaluate potential therapeutic agents to inform prioritization decisions for continued or new research and build toward a portfolio approach to optimally allocate available resources. (1) Source candidates may be identified from established or emerging science documented in published or unpublished (e.g., proprietary) data as well as from lived experience or clinical anecdotes (Figure 4-1A). From these sources, these potential therapeutic agents may be grouped into broad (2) intervention categories based on their mode of action (e.g., acting on pathogen, host, symptoms, or other processes). This can then inform the relevant criteria for evaluating efficacy, clinical readiness, and safety, as well as other considerations for success in development in a systematic series of (3) prioritization activities to assess how a candidate fits into a coordinated portfolio of clinical research.

The series of prioritization activities can include a triage process to rule out candidate agents if they are already being investigated in other robust clinical trials for Lyme IACI, if no preclinical or clinical data on safety exist yet, or if there are insufficient preclinical or clinical data that support treatment effects. This triage reduces redundancies in the research portfolio and redirects research efforts to obtain the necessary data on safety and preliminary indications of efficacy to support consideration in the research portfolio (Figure 4-1B). Therapeutic candidates need to be screened to ensure sufficient toxicology and human data are available to guarantee that they can safely be tested in larger human clinical studies. In some instances, the safety profile of candidates may be well known, such as with approved antimicrobials, but there may be differences between the labeled and proposed use that need to be considered in clinical trials for Lyme IACI (e.g., duration of use). In other cases, the candidate may be a new molecular entity for which comprehensive preclinical development is needed prior to human studies, including toxicology.

Candidates that have been triaged may be evaluated based on four major domains: biological plausibility, availability of supportive preclinical and clinical data, and potential impact (individual and population level), as well as other considerations for the feasibility of clinical development

(Figure 4-1C). For each domain, the evaluation considers whether sufficient data exist, whether the data are from rigorously conducted studies, and whether any elements listed within the domain are not required for the assessment of a particular candidate (e.g., real-world data may be applicable to repurposed products but are generally not available for new entities). In addition to a close examination of the available data, prioritization decisions may also consider factors for a viable clinical development pathway of the potential therapeutic product. Some factors include:

• Are the available data sufficient for an investigational new drug application (IND)?¹
• Are there sufficient supply or manufacturing logistics to enable testing in large-scale clinical trials (and market access if the therapeutic agent is approved)?
• Is it supported or likely to be supported by adequate capital throughout the clinical development pathway?

Overall, potential therapeutic candidates are evaluated to ensure that those included in the prioritized portfolio have a high probability of success, while balancing the inherent risks of failure in clinical development with the envisioned benefit (e.g., high-risk–high-reward candidates may be included in the portfolio).

It is important for the assessment criteria to be clearly defined at the start of this prioritization exercise as part of a rational and systematic review of potential new treatments to avoid post hoc decisions or other sources of bias, even if the specific criteria for each of these domains may differ among the candidate types (e.g., preclinical data may derive from observational studies with human participants for some and from animal models for others). This framework thus serves as a common ground to systematically evaluate, document, and understand the prioritization process for funding, particularly for challenging situations where different researchers or funders may ultimately arrive at different decisions. For instance, consider two promising therapeutic candidates that each have robust biological plausibility and strong preclinical and clinical data that are sufficient for an IND: Candidate A is likely to affect only a subset of patients (e.g., targeting a specific symptom or symptom cluster) and thus have a small population size, while Candidate B is believed to be broadly applicable to a majority of patients who report a common symptom. However, Candidate A has a

___________________

¹ An investigational new drug application (IND) is “a request for authorization from the Food and Drug Administration (FDA) to administer an investigational drug or biological product to humans.” It is a necessary authorization for new potential therapeutic products or uses that have not been previously approved for humans. See: NIH CC (n.d.).

more favorable clinical development pathway (e.g., clear product supply, logistics, and clinical studies design), while the many components in the clinical development pathway for Candidate B are unclear (Table 4-1). One funder may decide to prioritize Candidate A if the targeted symptom does not have any other promising treatment options and the funder perceives this as a high-need area. Given the same analysis, another funder might prioritize Candidate B if they value the larger population size for the potential treatment more, and the decisions align with the funder’s “high-risk/high-reward” ethos in supporting new research. Given the dynamic nature of biomedical research, a landscape review of potential therapeutic agents, including those recently reported and those already in the portfolio, must be conducted regularly, rather than being a one-time assessment exercise. This is particularly important in light of the opportunities for applying modern technologies and methods and drawing from similar health conditions to investigate the pathogenesis and treatment of Lyme IACI that may reveal significant findings (see Box 3-2).

TABLE 4-1 Example of Applying Prioritization Assessment of Lyme IACI Treatment Interventions

NOTE: IACI = infection-associated chronic illness; IND = investigational new drug; PD = pharmacodynamics; PK = pharmacokinetics.

Prioritization among potential treatments inevitably means some candidates will be excluded from the prioritized portfolio. However, a lower-priority ranking from one round of assessments does not imply that the particular research direction or proposed candidate will not produce results that lead to substantial discoveries and treatments in the future. It is important that the prioritization process be designed to incorporate new evidence as it is generated and allow for ongoing refinement of research priorities.

Given the heterogeneity in people living with Lyme IACI, the committee recognizes the need to consider a variety of interventions, sometimes in combinations, potentially including nonpharmaceutical interventions. Combination therapies have been successfully developed to treat conditions such as cancer (Jin et al., 2023), cardiovascular disease (Agarwal et al., 2024), and immune conditions (Colombel et al., 2010); but there are several challenges in evaluating them through clinical research. These challenges include identifying patients likely to benefit from combination therapy, determining the correct dosing, and designing studies that can provide informative results on the safety and efficacy of the combination. Importantly, enhanced toxicity is a major concern when multiple drugs with varying safety profiles are combined (IOM, 2012).

Ultimately, the committee decided not to apply this framework to propose a specific portfolio of priority candidates for additional study for several reasons. First, the committee acknowledges that individual funders and research organizations may have different perspectives on prioritization. Second, input from people living with Lyme IACI is built into the framework, but the robust community engagement needed to faithfully implement this framework was beyond the capabilities of this committee, given its task and timeline. Finally, the committee did not have all of the relevant expertise or time to evaluate the diversity of potential candidates that are likely to be identified for such an approach.

RESEARCH INFRASTRUCTURE

Careful consideration of needed research infrastructure is important to ensure that studies can answer their proposed research questions. Insufficient or inappropriate research infrastructure can make it challenging, if not impossible, to achieve the goals of a study, delaying treatment discovery and adoption. In this section, the committee evaluated factors related to research infrastructure that may challenge or enable effective research and how these factors can be addressed.

Participant Recruitment and Retention

Recruiting interested individuals who are willing to participate in clinical research is a challenge that is not unique to Lyme IACI research (Houghton et al., 2020). In particular, clinical researchers often fail to recruit representative samples of the intended patient population, commonly underrepresenting racial and ethnic minority groups, children, pregnant and lactating women, and members of rural communities, among others (NASEM, 2022b). The factors that impede people from participating in research are multiple, including mistrust of health care professionals or institutions and the burden of participating in research (Natale et al., 2021).

Mistrust and stigmatization as barriers to research participant recruitment are particularly relevant for Lyme IACI research. Many people with Lyme IACI symptoms have encountered stigmatization and delegitimization when seeking care from medical providers (Dumes, 2020). As a result, they may be suspicious of clinical researchers and the institutions that they represent; conversely, receiving information about clinical research from someone who is perceived as trustworthy by patients is a key facilitator to participant recruitment (Houghton et al., 2020). It will, therefore, be incumbent upon Lyme IACI researchers and clinicians to build trusting relationships with their patients and the Lyme IACI patient community. To improve recruitment, it is also important for researchers to develop relationships with the broad range of health care providers who treat people living with Lyme IACI, since many receive care from primary care physicians, specialized clinics, and complementary and alternative medicine providers (e.g., naturopaths, chiropractors, herbalists) (Ali et al., 2014).

Given the geographic distribution of Lyme IACI without nearby specialized centers for everyone, decentralized clinical trials may be a useful method to allow broader participation. Decentralized trials may be particularly helpful for those with severe symptoms, who may find participating in standard, site-based trials with frequent visits a difficult burden. Decentralized trials also provide an opportunity for community-based clinicians to engage in the conduct of research (Goodson et al., 2022).

Another barrier to participant recruitment in Lyme IACI research is the strict inclusion criteria required in some studies.² On the one hand, using strict inclusion criteria promotes a homogeneous study population and ensures that research participants are comparable to one another and facilitates validity the interpretation of study results. On the other hand, certain criteria may prevent many people who want to enroll in a clinical study from doing so and can slow study enrollment, complicate subgroup analyses, and limit the generalizability of research findings to a relatively

___________________

² As presented to the committee in open session by John Aucott on July 11, 2024.

small subset of the overall patient population. Expanding the study population to be more representative of the broader Lyme IACI population can improve access to trials and help produce more generalizable results but will likely require clear definition and analysis of distinct subgroups, increasing the size and costs of trials. For example, designing studies to recruit participants with and without two-tier test confirmation would enable researchers to capture a greater heterogeneity of people with Lyme IACI. Stratifying data collection and analysis based on these or other relevant patient characteristics might help researchers identify subpopulations who respond to treatment.

Ultimately, improving trial recruitment and retention will depend on people living with Lyme IACI being able to see the value in participating in clinical research. To ensure that participation is meaningful, individuals with lived experience need to be involved in the planning and design of Lyme IACI research (NASEM, 2022a). Similarly, providing mechanisms for clinicians to inform research priorities can assist recruitment, given their role in referring patients to research studies.

Study Designs

The design of a study determines the research infrastructure required. Study designs can be divided into two broad categories: interventional, in which study participants are assigned to a new intervention and typically compared with controls, and observational, in which study participants receive no research intervention and instead are observed. Interventional studies are particularly well suited to evaluate the safety and efficacy of treatments. Observational studies can be useful for understanding the natural history and outcomes of a disease, the risk factors for developing a disease or disease progression, and the long-term safety of treatments. While biomedical research has often focused on randomized controlled trials or prospective observational trials as the preferred methods to generate data in clinical research, there is robust literature on how other experimental designs can be relevant in answering some research questions or navigating limitations to the available resources (Brown et al., 2023).

Ultimately, the most appropriate study design will depend on the research question. While many different interventional and observational study designs exist, this section highlights two examples of efficient designs that are well suited to answer pressing research questions for Lyme IACI and that provide long-term value despite high start-up costs.

Interventional Adaptive Platform Trials

Traditional clinical trials are often slow and costly and do not achieve economies of scale. When multiple trials are needed for a given disease area,

as is likely the case for Lyme IACI, then the serial or parallel process of setting up and conducting trials grows redundant and offers limited flexibility in adapting as learning unfolds. In contrast, platform trials, which operate off a single master protocol (Woodcock and LaVange, 2017), and adaptive trial designs, which enable flexibility in how the study is conducted by responding to evidence as it is generated (Pallmann et al., 2018), offer a learning system that can accelerate evidence generation and adapt to the most important priorities for the development of effective treatments.

Adaptive platform trials can change over time in response to emerging data, allowing new treatments to be added as they become available and ineffective treatments to be dropped in order to optimize the use of resources and time. For example, if a specific subset of participants, such as individuals with a particular condition or several subpopulations across different conditions, respond more favorably to the intervention, the randomization could be adapted to increase the number of participants within that subset who are assigned to that intervention. Master protocols used within platform trials may allow the study of multiple interventions and populations, increasing the efficiency of evidence generation. By using a common placebo or control group and shared infrastructure, the time and resources required to test new treatments can be reduced. Moreover, having a single overarching trial structure synchronizes administrative, operational, and analytical approaches to efficiently test multiple drugs or devices simultaneously or in an accelerated, semi-sequential process.

Although more resources may be needed up front to establish a flexible platform, once it is operational the potential benefits can exceed the marginal costs for each additional intervention tested when the adaptive trial is thoughtfully designed and executed. Alternatively, the serial process of one trial at a time generates operational delays and opportunity costs that could be avoided (Mahlich et al., 2021). Examples of well-designed, successful platform trial platforms include the RECOVERY (RECOVERY Collaborative Group, 2021), REMAP-CAP (REMAP-CAP Investigators, 2021), and ACTIV-6 platforms (Naggie et al., 2022). All of these were performed in the setting of acute COVID-19 in which multiple interventions needed to be tested efficiently with common data collection and endpoints.

The advantages of adaptive platform trials may be particularly relevant for Lyme IACI research, since it is complex condition likely involving multiple subpopulations of interest and mechanisms to target. However, it should be noted that adaptive designs often rely on interim analyses that depend on surrogate markers to assess benefit. In many disease areas, such surrogates do not necessarily correlate with outcomes and may not allow for a full evaluation of safety if the trials are shortened. Notably, there are no known biomarkers for Lyme IACI that correlate to treatment response, though several have been suggested in the literature (see Chapter 2).

Observational Prospective Studies

Prospective longitudinal studies are important for developing a fundamental understanding of the characteristics and mechanisms of Lyme IACI. Longitudinal study designs permit researchers to observe change over time. When longitudinal studies are conducted prospectively, research participants are observed starting before the outcome of interest occurs (e.g., disease development or symptom resolution). As described in Chapter 2, between 5 and 36 percent of people diagnosed with Lyme disease experience persistent symptoms 6 months or more after antibiotic treatment, but the causes and pathogenesis of these persistent symptoms remain unknown. A major benefit of obtaining prospective longitudinal data is that it enables researchers to study the development and progression of Lyme IACI in a group of individuals starting from the time they receive a Lyme disease diagnosis. Understanding how Lyme IACI develops and progresses can inform the design of targeted treatments. Several prospective, longitudinal studies designed to provide insights on Lyme IACI mechanisms and disease progression are in progress at the time of this report’s publication (Box 4-1).

For the evidence generated by prospective observational studies to be informative, the studies must be well designed. Ideally, participants would be enrolled soon after Borrelia spp. infection and followed regularly for several years. Timely enrollment requires early detection of Lyme disease, which is often complicated by the inconsistent display of symptoms following infection and imperfect diagnostics (Branda and Steere, 2021; Schwartz et al., 2017). Notwithstanding, this study design would enable examination of the disease course and help identify factors that predict which individuals will return to health versus those who will develop Lyme IACI. Moreover, such studies could be designed to incorporate comparison groups, such as individuals with IACI symptoms of unknown origin, or people without Borrelia spp. infection, to evaluate how the comparison groups change over time relative to the participants with Borrelia spp. infection, given the similarities between Lyme IACI and other IACI, and the high prevalence of IACI-like symptoms in the general population (Wormser et al., 2020). Recruiting and following sufficiently large cohorts of study participants is important for several reasons. First, Lyme IACI is highly heterogeneous, and studies must have large enough sample sizes to capture that heterogeneity and help to understand its relevance to disease progression and treatment. Furthermore, large cohorts whose exposures have been well characterized over time provide greater opportunities to identify risk factors and novel therapeutic targets. For example, sex differences that are influenced by reproductive stage could be masked in a small study.

Because prospective longitudinal studies are expensive and time-intensive, they often are designed to gather data on many different exposures

that may be of interest. Collecting and storing well-characterized biological samples (e.g., serum, cells, tissues) during a prospective longitudinal study can increase the study’s value and potential for use in future research as new knowledge and assays are developed. Furthermore, the application of machine learning (ML), which is discussed in more detail later in this chapter, to large datasets generated through these types of studies may help detect new and relevant factors, including genetics, co-infections, or prior infections that may result in a confluence of events that lead to Lyme IACI. The complexity of these contributions may be masked in the results of smaller studies, which is why large-scale, prospective longitudinal studies are critical to understanding the etiology and pathogenesis of Lyme IACI.

Trial Networks

Multicenter trials are a feasible approach to conducting studies that are intended to enroll large numbers of participants. Such projects are harmonized for study participant recruitment and data-collection methods and can help speed enrollment, allowing large datasets to be acquired in a relatively short time frame. However, multicenter trials require centralized oversight (e.g., a principal investigator or oversight committee), with significant administrative burden, and are expensive to execute. Trial networks can help facilitate multicenter trials, coordinate or facilitate single-center studies, and can help address the costliness of multicenter studies by reducing the administrative costs incurred by each site. One such network already operates in the Lyme disease area, the Clinical Trials Network (CTN) for Lyme and Other Tickborne Diseases, based at Columbia University. The CTN model provides a means of collaboration and general research support to Lyme studies at other sites within the network.

As explained in Chapter 3, coordinating centers can help promote harmonization and decrease administrative burden on individual sites within a trial network. For example, coordinating centers can facilitate the development or selection of uniform data content and formats used by the sites participating in the network. Moreover, the use of common data elements (CDEs) can improve data interoperability among sites by aligning common case definitions, metrics, outcomes, and data formats. Data interoperability is critical within a trial network, especially when the network is conducting a multisite trial, to allow data generated from distinct sites to be aggregated and analyzed together.

The use of a single institutional review board (sIRB) can also help ease administrative burden within a multisite trial by streamlining the IRB process through a single institution (Wolinetz and Collins, 2017). In fact, certain National Institutes of Health (NIH)–funded research that is

conducted at multiple sites may require use of a sIRB (NIH, 2016).³ A platform protocol, a type of master protocol, is another tool that trial networks can use to reduce the burden on individual sites and promote coordination within the network. Similar to platform trials, platform protocols enable the study of multiple interventions for a specific disease over time (Woodcock and LaVange, 2017). A platform protocol offers the benefit of allowing all sites within the network to operate off the same protocol while conducting research on different interventions. Ideally, the existing trial network infrastructure could be used to launch large-scale multi-site clinical trials with identical protocols across study sites. Each site could, in addition, include specialized research arms to explore additional avenues of research.

Conclusion 4-1: Clinical trial networks can foster collaboration across disparate sites that conduct Lyme IACI research.

Data Sources

Advancing Lyme IACI research will require the integration and gathering of diverse data sources to address scientific questions accurately and efficiently. The previously mentioned study designs highlight the importance of tailoring research infrastructure to the specific types of data needed to address issues. In addition to different study designs, it will also be necessary to use data sources whose contents extend beyond what clinical trials traditionally collect. The sections below explore these data sources and provide an overview of their unique benefits, limitations, and background for their use in research.

Real-World Data

The Food and Drug Administration defines real-world data (RWD) as “data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources” (FDA, 2024). Real-world evidence is “the clinical evidence about the usage and potential benefits or risks of a medical product derived from analysis of RWD” (FDA, 2024). Sources of RWD are diverse and can include electronic health records, health insurance claims, product and disease registries, and wearable and other mobile technology. Accordingly, some sources of RWD are also sources of patient-generated data. In addition to the heterogeneity in data sources used to generate RWD, the fact that the data are captured in real-world environments increases the variability within and among the individuals from whom data are collected compared with the carefully defined subjects included in

___________________

³ Cooperative Research, 45 CFR Part 46.114 (July 19, 2018).

clinical trials. For example, patients may have more variable adherence to the intervention in a study using RWD, compared with a randomized trial. Despite these limitations, RWD can be useful for a number of applications in clinical research and may in some instances, typically through including patients more fully reflective of the heterogeneity of the actual treated population, be better suited than randomized trials to draw conclusions regarding real-world safety, compliance, and effectiveness and to efficiently produce informative results (Simon et al., 2022).

There are a multitude of ways in which RWD can be used in research and decision making, including identifying potential research hypotheses and priorities, conducting long-term safety monitoring, and even making causal inferences about an intervention’s effectiveness (Duke-Margolis Center for Health Policy, 2024). Regardless of the application, RWD must be fit for use, meaning that the data can address the relevant research or regulatory questions. Multiple sources of RWD—or a combination of RWD and research data—may ultimately be necessary to improve the fitness of the data (Duke-Margolis Center for Health Policy, 2019).

Assessing the quality of data obtained from an RWD source is fundamental to predicting its likely utility in generating evidence. RWD is generally not produced with future research applications in mind and is often unstructured and heterogeneous and includes errors and missing information (Liu and Panagiotakos, 2022). To address these challenges, it is important to validate that the data meet defined minimum quality principles. While the data-quality principles for RWD may vary across organizations, principles include:

• Conformance: Do the structure and format of the data conform with prespecified standards?
• Completeness: Are data values reported frequently enough for variables of interest (i.e., what is the rate of missingness in the data)?
• Plausibility: Are the data believable and conform with expectations for potential values and relationships? (Duke-Margolis Center for Health Policy, 2019)

ML also presents an opportunity to derive useful information from unstructured RWD, though researchers and clinicians must exercise caution in interpreting results from ML software (Liu and Panagiotakos, 2022).

A particular challenge to the broader adoption of RWD in Lyme IACI is the lack of both generally accepted diagnostic criteria for Lyme IACI and an International Classification of Diseases (ICD) code for the disorder. ICD codes provide a standardized approach to tracking a patient’s diagnosis, and are therefore ubiquitous in electronic health records and billing claims often used in research employing RWD. However, without clear diagnostic

criteria or an ICD code for Lyme IACI, it is challenging to define or identify individuals with Lyme IACI within these RWD sources. However, CDEs may help to identify people with possible Lyme IACI within certain sources of RWD. For example, a Lyme IACI symptom and disease history questionnaire that can both identify potential individuals with Lyme IACI and distinguish patient subgroups based on the likelihood that symptoms are related to Lyme disease would be a useful tool to adopt as a CDE. While it would be infeasible to adopt such a questionnaire as a CDE throughout the entire health system, clinics and registries that frequently encounter patients with Lyme disease and Lyme IACI could increase the research value of the data they collect by incorporating a similar tool into their routine processes. And even clinics without routine Lyme patient encounters could improve the utility of their records for RWD purposes by clearly reporting the method and date of Lyme disease diagnosis and documenting any symptoms that develop after diagnosis.

Patient-Generated Data

One type of real-world data that has been incorporated into existing Lyme IACI research is patient-generated data. Patient-generated data can come in many different forms, such as outcome data collected through a validated self-reporting instrument, data from wearable technologies, and data entered by individuals in disease registries or personal health trackers. This section focuses on patient registries. Patient-reported outcome (PRO) measures are discussed later in this chapter. Existing patient registries for Lyme IACI contain patient-generated and real-world data and are developed and maintained by patients. These include, but are not limited to, self-reported clinical data, such as diagnostic results, symptoms experienced and their severity over time, and experiences or results from a variety of interventions. Some registries are also tied to biological samples. Analysis of such datasets could help researchers identify and understand the characteristics of symptoms that matter most to people living with Lyme IACI, informing the planning and conduct of research. Common symptoms of those with Lyme IACI include fatigue, pain, and cognitive dysfunction (Rebman et al., 2017; Touradji et al., 2019). However, there appears to be a lack of quantitative or qualitative research on the symptoms people living with Lyme IACI deem most necessary to address to “return to functionality.” Similarly, data from patient registries could be interrogated to derive the relevance of existing PROs or to help generate new ones. If the data do not already exist, registries can often be used to send out health surveys and questionnaires to derive the exact information desired from large sample sizes relatively quickly.

For patient registry data to be useful for research applications, the data must be representative of the heterogeneous Lyme IACI population. This standard may be difficult to meet for registries in which patients self-enroll due to the high potential for selection bias. Use of representative samples from within the registry or statistical methods such as weighting could help address this limitation. Further, to understand whether registry data are representative of the Lyme IACI population, the registry must collect and store data on the sociodemographic characteristics (e.g., age, sex, race and ethnicity) and health histories (e.g., diagnostic information, relevant exposures) of participants in sufficient detail.

There are inherent risks to using retrospective data contained in many patient registries without fully understanding and considering the data’s quality. The quality of retrospective data relies upon accurate self-reporting based on an individual’s memory, but in reality, they may be highly variable. Thus researchers must be mindful of learning about the registry’s data-collection procedures and what filters are available for the data recording and extraction process in order to support the scientific rigor and reproducibility of the research using these data. Similarly, it is important for patient registries to implement standardized data collection procedures to minimize measurement bias. For example, the Patient-Centered Outcomes Research Institute outlines standards to improve the quality and transparency of patient registries (PCORI, 2024). Despite such risks, patient registries have the potential to provide important information about the symptoms and course of disease and to identify and elevate patient priorities, potentially accelerating research.

Another benefit to using patient registries is the opportunity they offer for people living with Lyme IACI and researchers to work collaboratively. For decades the Lyme IACI patient community has been raising awareness among researchers, clinicians, and policymakers about the toll of Lyme IACI. Yet, stigmatization of individuals with Lyme IACI has often persisted, in part due to a lack of understanding of both the disorder and patient experiences. Not only does stigmatization marginalize the Lyme IACI community, but it may also limit access to health care, potentially contributing to disability and death (e.g., suicide is a major cause of death in Lyme IACI) (Bransfield, 2017). Rebuilding the bonds of trust and communication, including offering acknowledgments of past harm, will be necessary for effective future collaborations. Working alongside patient communities requires and can help build respect for patients’ perspectives and their decades of investments into the development of registries and the information these registries contain. Therefore, using patient registries in Lyme IACI research not only accelerates the search for symptom-based treatments that are most meaningful to the people living with Lyme IACI,

but also can forge bonds between the medical, academic, and patient communities that strengthen trust and communication.

Patient registries can also be used as a recruitment platform for clinical trials, allowing researchers to meet enrollment targets more effectively. Developing and maintaining a patient registry can be expensive (Gliklich et al., 2020). Therefore, using the participants, infrastructure, and data that already exist within a patient registry to design and conduct clinical trials can make good use of the investment in the registry and save money compared to traditional trials (Anderson et al., 2020).

Biobanks

Biobanks are entities that collect and store human biological samples, such as tissue or blood, along with data associated with those samples for use in research. Biobanks are also often characterized by the presence of established governance structures that permit outside researchers access to the samples and data contained within them (Coppola et al., 2019). The biological specimens and data contained in biobanks are fundamental in biomedical research, enabling discoveries of the etiology, pathological mechanisms, risk factors, and molecular signatures of disease and treatment response (Annaratone et al., 2021; Harris et al., 2012). Researchers increasingly turn to biobanks to support a variety of applications, particularly for studies using human tissues, such as the creation of diagnostic tools, and the identification of biomarkers to guide treatment and diagnosis (Mackenzie, 2014). While their value to research is increasingly recognized, biobanks face financial strains from short-term funding and insufficient cost recovery that imperil their sustainability (Annaratone et al., 2021).

There are several active biobanks dedicated to Lyme disease. While this list is likely not comprehensive, these include:

• Columbia Lyme and Tick-Borne Diseases Research Center specimen bank
• Johns Hopkins SLICE biobank
• Lyme Disease Biobank
• Lyme serum repository
• Pediatric Lyme Biobank (Bay Area Lyme Research Foundation, n.d.; Boston Children’s Hospital, n.d.; CUMC, 2025; Johns Hopkins Lyme Disease Research Center, 2025; Molins et al., 2014).

Still, the samples and data contained within these various biobanks could be used more consistently in Lyme IACI research. Out of 68 articles on Lyme IACI mechanism and diagnosis research included in the scoping review,

26 (38 percent) did not use any human biological samples. Of the other 42 articles that did report use of human biospecimens, three (7 percent) included samples from biobanks, while the other 39 articles reported using specimens that were collected during a previous study but not necessarily stored in a biobank. Most studies reviewed did not use biobanks despite their potential to optimize research for Lyme IACI by cutting down on enrollment times and improving data comparability.

To ensure the continued and expanded access to biobank samples in Lyme IACI research, it is imperative to address some of the challenges commonly encountered in biobanking. First, the quality of biobanks’ biological samples and associated data, as well as the collection and storage methods are fundamental to their utility in clinical research (Dagher, 2022). Since many biobanks develop independently of one another, harmonizing collection and storage procedures can be difficult, affecting the validity and utility of collected samples and complicating research that aggregates or compares samples from multiple biobanks (Coppola et al., 2019). Similarly, standardizing metadata annotations, such as data provenance and experimental protocols, is important to promoting quality and interoperability of a biobank’s data (Alkhatib and Gaede, 2024). Characterizing samples to an appropriate level of detail and specificity is another challenge that can threaten the utility of samples for research,⁴ especially given the lack of established diagnostic criteria for Lyme IACI. Including well-characterized controls—such as healthy participants, samples collected before COVID, and participants with other IACI—within Lyme IACI biobanks will also be important for conducting comparative analyses. Several organizations have published best practices and standards in an effort to promote harmonization and quality in biobanking, including the International Society for Biological and Environmental Repositories and the International Organization for Standardization (International Standard, 2018; Snapes et al., 2023). Adherence to validated quality standards, such as those mentioned, is essential for Lyme IACI biobanks to be useful in research.

Biobank sustainability is yet another concern, given the open-endedness of their operation. Beyond ongoing funding, sustainability in terms of continuing to provide value to participants and researchers is a key consideration (Abdaljaleel et al., 2019; Coppola et al., 2019). To continue to provide value, it is critical for biobanks to adopt a reciprocal relationship with specimen donors, requiring collaboration with donors and relevant patient and advocacy groups. Moreover, it is important for biobanks and the researchers who use their samples to have a plan for returning aggregated research results to the individuals whose samples contributed to the research. There are also considerations unique to Lyme IACI for promoting

___________________

⁴ As presented to the committee in open session by Liz Horn on July 11, 2024.

good stewardship of donor samples, including tiered informed consent approaches that enable donors to opt in to having samples used for research on other IACI and making accommodations during the informed consent process for prospective participants with neurocognitive symptoms as a result of Lyme IACI.⁵ These considerations are important for building trust with the patient community.

Finally, access, governance, and transparency need to be considered in the development and operations of a biobank (Langhof et al., 2018). It is important that the governance structures of biobanks clearly outline the process for granting researchers access to samples and for participants to be informed of how their donated samples and data may be used. The development of a virtual catalog of biological samples and the data elements associated with those samples has been a successful strategy for improving awareness and accessibility of biological samples in research. Examples of where this strategy has been employed include the Stanford Biobank Laboratory Inventory Management System (Stanford Medicine, n.d.), OneDukebio Integrated Biospecimen Network (Duke University, n.d.), and National Heart, Lung, and Blood Institute Biologic Specimen and Data Repository Information Coordinating Center (NIH, n.d.). A similar approach could simplify and expand access to the patchwork network of Lyme IACI biobank samples.

Conclusion 4-2: Data from biobanks and registries can be highly informative and foundational for hypothesis generation and designing and carrying out treatment and mechanistic studies, but is currently underused due to a lack of awareness, coordination, governance, accessibility, sustainability, and standardization across biobanks and patient registries.

Conclusion 4-3: There is inconsistency in the provenance and standardization of registry data, biological samples, awareness of existing data domains, and quality of data generated through Lyme disease registries as well as heterogeneity in patients, which has limited the data’s usefulness in research studies.

Diagnostic Tools

There is currently no diagnostic for Lyme IACI. However, currently available approaches to diagnose Lyme disease and co-infections, despite their limitations, could help designate relevant subpopulations for research. Additionally, emerging approaches to characterizing unique biomarker

___________________

⁵ As presented to the committee in open session by Liz Horn on July 11, 2024.

profiles in Lyme IACI could, given additional research, assist in the diagnosis of Lyme IACI.

Diagnostics to Facilitate Stratification in Research

To investigate treatments for Lyme IACI, it is essential that patients be selected using accepted criteria of clinical diagnosis and it may be necessary to allocate patients to subgroups by scientifically valid criteria, which may include diagnostic testing. Ideally, the diagnostic tools would be widely available so that patients for prospective studies can be recruited widely from geographic areas beyond those located near major academic research centers.

To advance clinical trials of interventions treating Lyme IACI, it is important to use laboratory results from diagnostic assays for which the accuracy is known to stratify research participants in terms of those with and those without clear evidence of prior Borrelia spp. infection. Importantly, there are limitations to current Lyme disease diagnostics, given that standard two-tier serologic tests have a sensitivity of approximately 30 percent at the time of initial diagnosis of early Lyme disease (e.g., erythema migrans) and ranging from 34 percent to 61 percent in the 1–4 weeks after clinical diagnosis and initiation of antimicrobial therapy (Branda et al., 2011; Molins et al., 2014). The Tick-Borne Disease Working Group at the Department of Health and Human Services has submitted reports to Congress with recommendations regarding the improvement of both direct detection and indirect methods to provide laboratory evidence of Lyme disease, babesiosis, anaplasmosis, and deer tick virus. These reports—released in 2018, 2020, and 2022—are fundamental resources that identify the consensus for diagnosis of acute and early-stage tick-borne disease, topics that are out of scope for this committee (TBDWG, 2018, 2020, 2022).

In addition to classifying Lyme disease diagnosis history based on serologic evidence of B. burgdorferi infection and according to prior infection or coinfection with other tick-borne pathogens (e.g. babesiosis, anaplasmosis, Powassan virus), diagnostic tests and biomarkers for IACI, if they are identified, could be used to stratify patients into subgroups such as Lyme IACI only, Lyme IACI plus evidence of prior co-infections, Lyme IACI plus other IACI, and other relevant diagnoses

As described in Chapter 1, several diagnostic strategies are currently used to diagnose B. burgdorferi infection. Serological tools with multiplex capacity are currently available and can be designed to measure evidence of past exposure to the agents of multiple tick-borne infections, including Lyme disease, babesiosis, anaplasmosis, and deer tick virus, simultaneously. Microsphere immunoassay on the Luminex analyzer and line immunoblots with recombinant proteins of the target pathogens are currently available

tools for both research and diagnostic testing (Porwancher et al., 2011; Shah et al., 2023). While these assays have not been reviewed by the FDA, there are proposed means to define accepted criteria for these assays. For example, the Clinical Laboratory Evaluation Program of the Wadsworth Center, New York State Department of Health, publishes their criteria for review and approval of laboratory-developed tests that can then be used for diagnosis and research purposes.⁶ Laboratory-developed tests (i.e., not FDA-approved) that meet validation criteria for research purposes and as clinical diagnostics may in some cases be useful in allocating patients to Lyme IACI subgroups in the design of clinical research.

Diagnostics to Identify Co-Infections

Possible tick-borne coinfections that might be present from an Ixodes tick bite include babesiosis, anaplasmosis, and the deer tick lineage of Powassan viral encephalitis (Project Lyme, 2021). To date there is no strong evidence that bartonellosis results from tick bites (CDC, 2024). Simultaneous co-infections sometimes occur and may cause more severe disease than a single infection (Cutler et al., 2021). On the other hand, immune activation or immune priming may hasten the clearance of the pathogens. In terms of tick-transmitted illnesses, while simultaneous infection may occur, at other times a person is infected first by one pathogen and then weeks or months later by a second pathogen from another tick bite. If a person is at risk of one tick bite, it is possible that their environment or activities put them at risk of another tick bite.

While a review of diagnostics for various tick-borne infections is beyond the scope of this committee’s task, the proper diagnosis of co-infections is important in Lyme IACI research. First, it is important to rule out an untreated co-infection as a cause of, or contributor to, persistent symptoms following Lyme disease. Second, characterizing whether people with Lyme IACI have also had certain tick-borne co-infections could be informative in better understanding Lyme IACI (NIAID, 2025), and may be important in considering design or interpretation of clinical studies. New technologies, such as multiplex diagnostics, may be able to advance accurate and timely diagnosis of co-infections (Nigrovic et al., 2023), but attention will need to be paid to the accuracy feasibility of these technologies before they can be more broadly adopted.

Importantly, if a person has Lyme IACI, it is likely that at least several months have passed since the original tick encounter, meaning that testing

___________________

⁶ Relevant criteria published by the Wadsworth Center include the Diagnostic Immunology Checklist and the Microbiology Molecular Checklist, particularly the Validation of Next Generation Sequencing Methods for Identification and/or characterization of Infectious Agents. See: New York State, n.d.

for the presence of Borrelia infection by direct detection methods is highly likely to be negative. Serology is more likely to be positive, but antibody levels may have waned below a significant cutoff level for positivity. When several months have passed since the infected tick bite, infections may be at a point where no live organisms are present in the individual, and a positive result not only is unlikely (with the exception of joint fluid in active Lyme arthritis) but may not indicate the presence of active infection, but rather persistence of B. burgdorferi DNA (Marques, 2015).

If an individual with Lyme IACI symptoms is diagnosed with an ongoing tick-borne co-infection, it is important that the person receive treatment for the co-infection, if indicated. Fortunately, the generally recommended first-line therapy for Lyme disease, doxycycline, is also highly effective against Anaplasma. Doxycycline is not effective for babesiosis, for which the recommended therapy, if indicated, is atovaquone plus azithromycin or clindamycin plus quinine. There is only supportive therapy for infection with Powassan virus (Yale Medicine, n.d.).

Prospective observational studies are needed to ascertain the possible clinical impact of other concurrent or sequential tick-borne infections on the development, diagnosis, treatment, and prognosis of Lyme IACI. In parallel, trials for new treatments can determine and document participant status on prior infections (i.e., Anaplasma, babesiosis, and deer tick virus infection) at the time of enrollment and throughout the study so that this information can inform the categorization of subgroups and related data analyses for Lyme IACI research.

Diagnostic assays for the most common human disease-causing pathogens transmitted by Ixodes scapularis and I. pacificus are available and reviewed in depth elsewhere (Rowan et al., 2023; TBDWG, 2022). The development and use of diagnostic tests need to be fit for purpose, such as tracking disease epidemiology, guiding treatment, or research and discovery. Immunoassays are appropriate for prospective observational studies designed to gain understanding of the prevalence and influence of coinfections on Lyme IACI without seeking to provide treatment for acute infections and can be developed as multiplex tests for broad use.

Multiplex diagnostic tests can be used to identify multiple pathogens from the same sample within one reaction (e.g., molecular or immunoassay). One key technical challenge in multiplex diagnostic tests is the potential for cross-reactivity between the different antigens covered in the panel. The higher complexity of the reaction may also affect sensitivity and specificity for a multiplex panel (Otoo and Schlappi, 2022). There are no FDA-approved tick-borne diseases multiplex panels for use in humans, although research-use-only panels for common tick-borne pathogens have been published (Buchan et al., 2019; Nigrovic et al., 2023; Tokarz et al., 2018), and some laboratory-developed tests are in use (Mayo Clinic

Laboratories, n.d.). Future development of multiplex diagnostic panels for human use will need to clearly define the intended application of the test and balance benefits from the test with potential drawbacks such as costliness or inappropriate distress that could stem from false-positive results (Box 4-2). Metagenomic next-generation sequencing (NGS) is an emerging diagnostic approach that could enable the simultaneous analysis of genetic material from multiple pathogens (Chiu and Miller, 2019). While in early development, there is emerging evidence that metagenomic NGS could be a tool to conduct host transcriptomic analyses to detect altered gene expression within humans as an indicator of host response (Omura et al., 2025). Without a clear rationale, including use cases and market strategy, it may be difficult to justify the cost-effectiveness of development, manufacturing, and use of the diagnostic test with device developers and health care payors.

Emerging Approaches in Diagnostics for Lyme IACI

As noted in Chapter 2, no current diagnostic is capable of identifying individuals with Lyme IACI. However, novel technologies are being studied for the potential capability to detect changes in biomarkers that may be characteristic of Lyme IACI or other IACI. For example, one study using proteomics on cerebrospinal fluid (CSF) reported increases in several markers of complement activation that appeared to discriminate myalgic encephalitis/chronic fatigue syndrome (ME/CFS) from neurological posttreatment Lyme disease (PTLDS) (Schutzer et al, 2011). Similarly, another study suggested the ability of an ML algorithm to distinguish between individuals with Long COVID and Lyme IACI based on cytokine profiles (Patterson et al., 2024). In other studies, increases in cytokines such as IL-23 and interferon-alpha, and chemokines including CCL19, were reported that, if confirmed, may help discriminate individuals with Lyme IACI from persons who respond to therapy for Lyme disease (Aucott et al., 2016; Hernández et al., 2023; Strle et al., 2014). While in the early stages, the application of ML can be implemented to aid in developing new approaches to diagnosis. Significant advances have been made for cancer and rare disease genetic diagnostics and treatments using ML, but there has been little research conducted on its application in IACI. One study examining the proteomic profiles of individuals with Long COVID used ML to develop a random forest model, which was reported to accurately distinguish those with Long COVID from healthy controls (Gu et al., 2023). These and other promising approaches, particularly if confirmed, may also be tested in Lyme IACI.

Metabolomics analysis may also be useful in developing a diagnostic for Lyme IACI. One study observed notable differences between a group of Lyme disease patients who went on to develop PTLDS and those who returned to health after treatment, measured at the initial time point of diagnosis, at completion of antibiotics treatment (2 to 3 weeks after diagnosis), and one year after completion of treatment. Researchers identified and measured over 100 small molecule metabolites that formed the basis of biosignatures for the PTLDS and non-PTLDS groups. A smaller set of 72 small molecule metabolites demonstrated the strongest discriminatory effect between the two groups. The researchers validated their findings with a second cohort of people with and without PTLDS and observed similar longitudinal patterns (Fitzgerald et al., 2021).

Transcriptomics, which measures the expression of cellular mRNA to observe the effects of an infectious agent on host cell gene expression, is another method that may be promising for understanding and diagnosing Lyme or other IACI (Bouquet et al., 2016). Using RNA sequencing, a study compared the transcriptome profiles of 15 individuals with Lyme disease who returned to health with the profiles of 13 individuals with persistent

symptoms after having Lyme disease. While no differentially expressed genes were observed between the two groups at individual time points (time of diagnosis, post-treatment, and 6-months post-treatment), combining the three time points revealed genes whose expression differed between those with resolved Lyme disease and those with persistent symptoms (Bouquet et al., 2016). Further examination in a larger study could help determine whether transcriptomic-based diagnostics may be capable of identifying individuals with Lyme IACI.

While promising, these tools are in the preliminary stages of validation for Lyme IACI diagnosis, are expensive, and are not widely accessible for primary care sites. Additional research into the candidate biomarkers identified through these emerging diagnostic approaches is needed to validate these tools. However, there are numerous opportunities to further evaluate these tools using prospective observational studies, specimens and data from biobanks, and other samples collected and stored in clinical trials.

DATA INTERPRETATION

Data can only be useful for solving scientific problems if they can be meaningfully interpreted. As more data are generated through improvements in research, it will be essential to develop novel approaches to interpret these larger quantities of data. One of the fundamental aspects of data interpretation is ensuring that the data are standardized with an emphasis on uniformity across different datasets.

Data Standardization

The use of common research tools and outcome measures across studies would facilitate direct study comparisons and data aggregation. Most data must be collected using identical and consistent methods and measures for comparisons across studies to be valid. Moreover, understanding the source of data is fundamental to accurately interpret and compare findings. Therefore, it is critical that certain data domains be routinely included in clinical research. These domains must be guided by patient-reported priorities to target the most important symptoms and quality-of-life issues and by the data quality needed for research. For example, standardized approaches to collecting and reporting disease and sociodemographic characteristics will be particularly important to defining the data’s source population and promoting data quality in future analyses. CDEs, introduced in Chapter 3, encourage data standardization within a disease area across discrete research studies.

The development of CDEs for Lyme IACI may be guided by similar efforts in ME/CFS research. Defined CDEs for ME/CFS are encouraged to

be used for studies on the condition and are classified into four categories: (1) Core, general information typically required for all studies funded by the National Institute of Neurological Disorders and Stroke, (2) Supplemental—Highly Recommended, specific to the disease, (3) Supplemental, common to clinical research but depends upon the protocol, and (4) Exploratory, which still requires validation. Due to the large degree of symptom overlap between Lyme IACI and ME/CFS, some elements may transfer directly (Bai and Richardson, 2023). In developing such CDEs, however, it will be important to develop Lyme-centric, patient-driven symptom measures for treatment endpoints.

CDEs will be particularly valuable if they are able to be tied into health care practice with the use of consistent coding conditions (e.g., to build accurate patient cohorts). As with the ME/CFS example, a committee that includes the research community and Lyme IACI patient representation may be helpful for determining CDEs for future Lyme IACI clinical research that will prioritize outcomes important to the Lyme IACI patient community.

The use of CDEs in Lyme IACI research would strengthen and accelerate research efforts in multiple ways, including data sharing, data mining, comparisons with other IACIs with overlapping CDEs, patient stratification (e.g., based on symptoms), identification of core data characteristics, and uniform classification of clinical markers. While the standardization offered by CDEs confers many benefits, it is important to allow for flexibility across study protocols, especially as exploratory treatments may not align precisely with CDE specifications. Separate CDEs for various study approaches would be helpful, enabling researchers to choose instruments best suited to their research question. For researchers new to the field, or those seeking guidance in a new area of Lyme IACI research, the publication of Lyme IACI CDEs could help identify which instruments to use and which variables to collect, expediting start-up time. Simple CDE sets that are not overly prescriptive would be easiest to implement, promoting early adoption.

The inclusion of standardized and validated PROs that are already known to be acceptable to regulators (e.g., the FDA) within CDEs could expedite approvals of new treatments. Of the 68 articles included in the treatment or mechanisms category of the scoping review, 34 did not report using any PRO tools in their study, 22 used structured PRO tools (e.g., SF-36, Fatigue Severity Scale), eight used unstructured methods (e.g., direct patient-reported symptoms), and four used data from MyLymeData, for which it was not possible to ascertain the types of tools used to collect data from individuals (Figure 4-2). Of the 22 articles that used structured PRO tools, six articles used tools specifically related to Lyme disease or Lyme IACI, such as the General Symptom Questionnaire-30 (GSQ-30) or

Post-Lyme Questionnaire of Symptoms (PLQS). From the evidence reviewed by the committee, it appears the use of standardized PRO tools, especially tools designed for Lyme disease or Lyme IACI, is limited within Lyme IACI research.

Currently, many Lyme IACI studies use the SF-36 as a measure of study outcomes. As a survey that assesses patient-reported quality-of-life measures related to physical health and mental health, the SF-36 is a generic tool used across many different disease areas and has not been specifically tailored to address multisystemic Lyme IACI symptoms. Several outcome measurement tools have been developed specifically to measure outcomes in people with Lyme IACI, including the GSQ-30 (Fallon et al., 2019) and the Multiple Systemic Infectious Disease Syndrome (MSIDS) model (Horowitz and Freeman, 2018). The GSQ-30 was specifically developed to assess patient-reported multisystem symptom burden in a brief format (Fallon et al., 2019), focusing on four core domains: pain/fatigue, neuropsychiatric, neurologic, and viral-like symptoms. However, these scoring domains have not been compared directly with other standard measuring tools. For instance, the GSQ-30 has not been compared with the SF-36 to address whether one may be more effective than the other for measuring treatment outcomes in a way that is meaningful to people living with Lyme IACI. Importantly, the GSQ-30 was modeled on existing scales and clinical experience rather than originating directly from patient input.

A validated outcome measurement tool that captures the symptoms most relevant to people with Lyme IACI is needed to improve comparisons across future studies and to steer those studies toward measuring the most pressing outcomes. The involvement of people with lived experience in the development of PROs is necessary to improve the relevance and comprehensibility of these tools, and to enhance the validity of studies that use the tools (Wiering et al., 2017). It is also essential for outcome measures and tools to be validated and acceptable to regulators so that their use can inform approval decisions. Moreover, outcome measurement tools have not generally been tailored to capture symptoms as experienced by pediatric patients. In developing such surveys, it will be vital to incorporate developmentally appropriate language or to create versions that could be completed by caregivers, or both.

Conclusion 4-4: There is a lack of standardized, validated measures for Lyme IACI study outcomes. It is unclear if existing measures adequately reflect patient well-being and clinical responses to treatment interventions. In particular, there are no validated tools developed for the pediatric population.

Outcome Evaluation

Determining the efficacy of a treatment is one of the core functions of clinical trials. This requires a comparison between the treatment’s effects on a particular outcome or set of outcomes, compared with the effects of no active treatment or an alternative treatment. There are numerous statistical calculations that can determine the effect of an intervention on an outcome (e.g., relative risk, hazard ratio). Likewise, there are several statistical approaches to comparing the effects in different study arms to assess the likelihood that an intervention is associated with a specific outcome. However, the statistical significance of a result is not the same as clinical relevance—a large enough trial can produce a statistically significant result from just a small difference in outcomes that may not be clinically meaningful (Kieser et al., 2013). Conversely, due to issues in study design or size, some clinically meaningful effects may not reach commonly used levels of statistical significance. Therefore, measuring the effect size—or the magnitude of the treatment’s effect—is important to understanding the clinical relevance of a treatment’s effect (Aarts et al., 2014).

The minimal clinically important difference (MCID) is a concept that can help improve understanding of the clinical relevance of a treatment’s effect, especially when PROs are used to assess outcomes. The MCID, which represents the smallest amount an outcome must change to be meaningful to patients, can contextualize an effect size within the values and priorities of

patients (McGlothin and Lewis, 2014; Mouelhi et al., 2020). For example, a relatively small change in fatigue scores may represent meaningful relief for patients, or a large numerical change in an outcome measure may not affect how a patient feels. Ideally, evaluation studies would be conducted to determine the MCID for the particular outcome measure and patient population for which the MCID will be used, but they can be estimated using effect sizes or approximated by anchoring the outcome measure to an external measure—such as the global rating of change score—if a study is not feasible (Angst et al., 2017; Wright et al., 2012).

Artificial Intelligence⁷

Advances in AI models, which include ML approaches, can be applied to create tools that provide unique opportunities to advance Lyme IACI research. These models are particularly well suited to analyze disparate sources and types of data, including large and complex datasets, to identify new insights or generate hypotheses. Research tools that incorporate AI are relatively new technologies and will require significant human oversight, including from individuals with expertise in the development and use of these tools as well as in the underlying biological and clinical science.

Identifying Commonalities and Differences Between IACI

The need for expanding collection of data in research for complex IACI in order to distinguish commonalities and differences, discussed in Chapter 3, could involve deep phenotyping studies on defined disease (e.g., ME/CFS) or broadened inclusion for studies that allow enrollment from more than one disease (e.g., a clinical trial on fatigue that can include Lyme IACI and ME/CFS). Either approach will inevitably generate large amounts of data. AI tools could be used to make qualitative and quantitative comparisons and identify areas of similarity between Lyme IACI and other IACI. For example, AI tools could parse specific research findings, compare biomarkers and biometrics of interest, and generate testable hypotheses that predict relationships among IACI symptoms, pathogenesis, and etiology or risk factors. Immune profiling studies at single timepoints show common inflammatory markers across ages, including for individuals with autoimmune and neurological disorders (Chen et al., 2023; Nature Immunology, 2025; Wang et al., 2025; Zaslavsky et al., 2025). Time-based AI analyses that could be relevant to Lyme IACI also show early promise. Computational models predicting the dynamics of antibody memory formation from omics screens can help understand an individual’s

___________________

⁷ This section draws largely from a commissioned paper authored by Dr. Amina Qutub, a consultant to the committee.

immune cell exposures—potentially elucidating the environmental trigger of IACI symptoms (Cvijović et al., 2025). These tools may have applications in research on diagnostics that can distinguish between various IACI and identify potential treatment targets that several IACI have in common.

One specific application to facilitate usability for researchers is development of an integrative, searchable, AI-friendly database or federated data ecosystem that links IACI symptomatology, pathophysiology, and research categories and empowers researchers to identify connections. If a comprehensive Lyme IACI dataset is available, AI algorithms could quickly learn from it, drawing connections among the data that might otherwise be missed in manual evaluation. Among the benefits of using a well-established big data and an AI framework to host Lyme IACI data are the time saved in using robust database tools and the ability to quickly cross-reference new data for Lyme IACI with external public data resources. Connecting the Lyme IACI dataset with additional data from Long COVID, ME/CFS, and other conditions could allow identification of pathogenesis mechanisms that are shared across these diseases. There are existing examples of relevant, open public databases, including Google’s COVID Symptoms dataset and the National Library of Medicine’s COVID sequencing datasets.

Large language models (LLMs) are generative AI models that pre-train on massive amounts of textual or image data, or both (e.g., books, scientific literature, programming code) and then use statistical methods to predict relationships among letters, words, and sentences in the sequence of text or pixel relationships in an image. Since LLMs are adept at analyzing data and probabilistically inferring connections based on prior knowledge, they can generate hypotheses for Lyme IACI research if curated, high-quality data were available. Each of such AI-generated mechanistic hypothesis would still need to be tested, and any predictive models will need further validation from new patient datasets or clinical studies.

Challenges for the application of AI to Lyme IACI and other IACI include heterogeneity of the chronic symptoms; the combinatorial complexity of environmental, social, and psychological effects on the disease course; and lack of a single defining molecular feature (e.g., a single mutation present in some rare genetic diseases). One approach that has been taken to account for analogous challenges in other areas of biology research, such as experiments that measure a limited signal per molecule or drug development research with small effect sizes is the incorporation of weights for specific biomarkers and mathematical rules based on known immunological significance of a protein in predictive models. For example, a study on Long COVID developed the “long-hauler index,” a formula that defines an algebraic relationship among the cytokines IFN-α, CCL4, and IL-2 (Patterson et al., 2024).

However, rule-based AI approaches will need to address the lack of a unified vetting process for data quality, inconsistency across research methods, and heterogeneity across patients and within patients over time. Similarly, without defined “gold standards” to delineate Lyme IACI and other IACI by molecular or phenotypic characteristics, the ability to apply supervised learning that predicts categories of IACI remains limited. Unsupervised approaches, including clustering, LLMs, and dimensional reduction methods that find patterns across datasets without needing labeled input data, offer alternatives that do not require known delineations of IACI, which may be particularly useful when the cause of IACI symptoms is unknown.

Subgroup Stratification

The heterogeneity of symptoms that people with Lyme IACI experience suggests that there may be distinct patient subgroups whose symptoms stem from different mechanisms or combinations of mechanisms. However, attempts to pinpoint specific biomarkers that classify such patient subgroups have so far been unsuccessful. Unsupervised ML approaches can be used within a well-curated IACI dataset to identify subgroups of patients based on one or many attributes of interest (e.g., symptoms, proteomics, genomics, brain-imaging abnormalities) (Hu et al., 2019; Petukhova et al., 2024; Zaslavsky et al., 2025). Such methods could limit the noise created by the heterogeneity within the Lyme IACI population, enabling more rapid discoveries of biomarkers and mechanisms for subgroups. This unbiased approach also allows the potential identification of individuals with Lyme IACI who group more closely with other IACI and may benefit from a different course of treatment.

In an example of this approach applied to Long COVID, semantic clustering was employed to identify six distinct clusters of people with Long COVID based on profiles of phenotypic abnormalities. Clusters identified people who closely shared phenotypic changes. One cluster had notably increased mortality, and others had distinct pulmonary, neuropsychiatric, and cardiovascular abnormalities (Reese et al., 2023). Another study applied a clustering-based method to a proteomic screen to conduct a molecular-based stratification and classify subsets of people with acute myeloid leukemia, which is also a highly heterogeneous disease. The ML method analyzed >300 protein expression levels plus clinical characteristics of 205 patients and identified expression patterns in protein functional groups and relationships among functional groups to discover global proteomic patterns (Hu et al., 2019). Neither of these examples was a completely unsupervised method, as both used publicly available reference databases to weight or

structure the data. However, no labels were given to the patients a priori. These illustrative examples of ML applications demonstrate options that could be used to infer patient subgroups.

Increased availability of reference data could strengthen the capabilities of AI-driven analyses of multi-omics and mixed biological data in Lyme IACI. For example, interoperability of data gathered across various research studies in Lyme IACI, promoted through CDE adoption, would facilitate aggregation and comprehensive analysis of the Lyme IACI evidence base. Collection of relevant metadata in registries, biobanks, and other research repositories could also help inform AI analyses. Yet the potential for ML to amplify existing biases in the data necessitates the implementation of careful data-quality controls to minimize biases in the source data.

Enhancing Drug Development

AI could provide a deeper understanding of the relationships between therapeutic interventions and biomarkers or symptoms through computational methods and guide the design of new therapies or choice of FDA-approved drugs to test in a clinical trial. Examples include AI used to identify therapeutic targets in cancer from multi-omics biomarkers and drug screening efforts underway in many biotech companies—some with direct relevance to IACI. Unsupervised pattern-recognition methods may be a useful approach to finding relationships among symptoms, therapies, and molecular biomarkers. For example, clustering could be used to identify relationships between different symptoms or potential biomarkers against multiple therapeutic candidates. Temporal changes in symptoms and biomarkers in response to therapies could also be tracked by AI. AI methods including diffusion models and generative adversarial networks can predict changes in tissues as a function of a negative stress (e.g., inflammatory cytokines in the blood, spirochete bacteria) or therapeutic compound, and produce images and graphs that replicate what a medical test might produce. Overall, using these approaches could not only help identify therapeutic candidates, but also help inform disease mechanisms and identify quantitative endpoints for clinical trials.

Another route to identifying effective Lyme IACI treatments is through assessment of repurposing of existing therapies. Linking molecular targets and patient symptoms to known targets and clinical endpoints of FDA-approved drugs can help identify whether a drug can be repurposed for use in a subset of patients with Lyme IACI. A convergence of symptoms alone can help narrow the computational space needed to predict molecular targets and repurpose FDA-approved drugs for Lyme IACI. However, validation will require knowledge of target proteins or pathways that are not currently understood in Lyme IACI.

CONSIDERATIONS FOR IMPLEMENTATION

It is important that the evidence generated through clinical research informs future research and clinical practice. Even when clinical research demonstrates that an intervention is effective, a failure to consider how the intervention will be used in real-world settings can limit its adoption (Bauer and Kirchner, 2020). Careful consideration for the implementation of research findings while initial effectiveness data are being generated can increase the likelihood of a new intervention being adopted.

Dissemination

A robust dissemination strategy is fundamental to the successful and timely translation of research findings into clinical practice that is accessible to patients. This is underscored by the long lag times between medical discoveries and their implementation in the real world (Morris et al., 2011). Similarly, it is important to disseminate negative study results to prevent patients being given ineffective or unsafe treatments. Individuals with Lyme IACI and Lyme disease have organized a vibrant patient community and various advocacy groups (McPhail, 2017; Morrison et al., 2021). These existing networks of individuals living with Lyme IACI and their advocates are likely to champion the dissemination of promising research results on Lyme IACI treatments as well as to make others aware of negative findings. Yet dissemination cannot rest solely on the patient community.

There are several challenges that need to be addressed to improve the dissemination of results through the research ecosystem to the end users. First, many researchers lack the resources and capacity—or sometimes experience—to effectively convey their findings beyond traditional academic forums to lay audiences. Furthermore, clinicians may not be aware of the emerging evidence without targeted dissemination.

Many systematic challenges exist to the accurate dissemination of research to the public, but on an individual level researchers have a role to play in actively engaging the public, including through news and social media, and in scientific discourse, resisting pressure to hyperbolize results, and clearly describing the nuances and limitations of their findings (West and Bergstrom, 2021). Adopting better science communication practices can help researchers equip people with Lyme IACI and advocacy groups with accurate evidence to readily disseminate through their networks. Successful science communication strategies include transparently conveying uncertainty in scientific understanding and how scientific conclusions are drawn and tailoring messages to specific segments of the target audience based on values, beliefs, and information sources (NASEM, 2017). Researchers can access resources such as the American Association for the Advancement of

Sciences Communication Toolkit (AAAS, 2025) or the CDC Clear Communication Index (CDC, 2023) to help guide their communication strategies and messages.

Given that individuals with Lyme IACI can encounter dismissal and delegitimization of their symptoms and experiences from clinicians, it is also important to consider how best to disseminate Lyme IACI research results to treating clinicians. Lessons from HIV/AIDS advocacy may be particularly illuminating on how to bridge the gap between research and clinical practice. Similar to the case with Lyme IACI, people living with HIV were, and continue to be, vocal self-advocates who have enmeshed themselves in the scientific process, as exemplified through the creation of transnational advocacy networks that connected patients and researchers across the world (Colvin, 2014). Likewise, there are many other patient-centered groups that have advanced science and addressed unmet needs through dissemination, including groups dedicated to rare (e.g., cystic fibrosis) and common (e.g., breast cancer) diseases. In the case of HIV, for instance, scientific meetings and conferences offer an opportunity for advocates, researchers, and clinicians to share knowledge with one another and serve to keep clinicians and people living with HIV apprised of the latest research. HIV Source is another resource, maintained by NIH, which centrally collects resources on HIV research and treatment for clinicians, researchers, and the public (NIH OAR, n.d.). Within the realm of IACI, Project ECHO Global Health Initiatives has developed a course to share best practices in the care and treatment of Long COVID and ME/CFS (Project ECHO, n.d.). Clinician training is a key component of research dissemination, and programs that can help practitioners recognize IACI generally and distinguish between individual IACI will be valuable as more evidence is generated. Similar models of dissemination may be effective in aligning clinical practice with emerging research in Lyme IACI treatment. Additionally, implementation science could help further the understanding of dissemination processes that can facilitate the adoption of effective treatments for people with Lyme IACI.

Implementation Strategy

Engagement with the populations who will use an intervention is foundational to implementation. For this reason, patient engagement in clinical research must occur early and often and continue after research is completed. Even though this report highlights patient engagement as part of a research implementation strategy, patient engagement that starts after the research is completed is too late and often lacks authenticity. Intentional and sustained patient engagement enables lived experience to be embedded in research and can improve the responsiveness of research to

patients’ needs and values (Cavaller-Bellaubi et al., 2021). There are numerous opportunities to build in patient engagement across the research, development, and product life cycle, conferring benefits to patients, researchers, and developers throughout. Engagement in the early phases of research can involve identifying the research questions that matter most to patients and having patients provide input on the study design to improve accessibility and the capture of relevant data (Sacristán et al., 2016). Most relevant to the implementation phase is the potential for the patient engagement process to inform regulatory and coverage decisions about the product and to improve patient satisfaction with and use of the intervention (Puerini et al., 2024). Empowering expert patients through education on the clinical research process can further advance the goals of patient-centered and quality research throughout the product development life cycle (Sacristán et al., 2016). Examples of good practice in patient engagement published by Patient Focused Medicines Development can help guide patient engagement strategies in Lyme IACI research (PFMD, 2020).

A successful implementation strategy will also include engagement with industry and regulators. For interventions to be widely adopted, they need to achieve market viability (i.e., meet a demand in the market, be sold at an acceptable price, offer a return on investment, etc.). Collaborating with industry partners can thus strengthen the economic argument for implementation (Proctor et al., 2021). Early coordination with regulators, such as the FDA, is critical to the eventual adoption of Lyme IACI innovations. Discussions with FDA can help ensure that appropriate data of adequate quality are being collected in well-designed trials to inform and support regulatory decisions.

As evidence-based Lyme IACI treatments become available, it will become increasingly important to understand how they are used in clinical practice settings, especially given the heterogeneity of people living with Lyme IACI and the challenges that many individuals may have in receiving a timely and accurate diagnosis. The Agency for Healthcare Research and Quality could be well suited to conduct this research and to develop associated educational resources for clinicians.⁸ Moreover, should various treatments be found to be effective, comparative effectiveness trials may be appropriate to determine the most effective Lyme IACI treatments and whether different patient subgroups respond better to certain treatments. These comparative effectiveness trials would align well with the research being funded by the Patient-Centered Outcomes Research Institute.

___________________

⁸ As part of restructuring the Department of Health and Human Services announced on March 27, 2025, the Agency for Healthcare Research and Quality will be merged with the Assistant Secretary for Planning and Evaluation under a new Office of Strategy. See: HHS (2025).

REFERENCES

Branda, J. A., K. Linskey, Y. A. Kim, A. C. Steere, and M. J. Ferraro. 2011. Two-tiered antibody testing for Lyme disease with use of 2 enzyme immunoassays, a whole-cell sonicate enzyme immunoassay followed by a VlsE C6 peptide enzyme immunoassay. Clinical Infectious Diseases 53(6):541–547.

Branda, J. A., and A. C. Steere. 2021. Laboratory diagnosis of lyme borreliosis. Clinical Microbiology Reviews 34(2).

Bransfield, R. C. 2017. Suicide and Lyme and associated diseases. Neuropsychiatric Disease and Treatment 13:1575–1587.

Brown, A. W., S. Aslibekyan, D. Bier, R. Ferreira da Silva, A. Hoover, D. M. Klurfeld, E. Loken, E. Mayo-Wilson, N. Menachemi, G. Pavela, P. D. Quinn, D. Schoeller, C. Tekwe, D. Valdez, C. J. Vorland, L. D. Whigham, and D. B. Allison. 2023. Toward more rigorous and informative nutritional epidemiology: The rational space between dismissal and defense of the status quo. Crit Rev Food Sci Nutr 63(18):3150–3167.

Buchan, B. W., D. A. Jobe, M. Mashock, D. Gerstbrein, M. L. Faron, N. A. Ledeboer, and S. M. Callister. 2019. Evaluation of a novel multiplex high-definition PCR assay for detection of tick-borne pathogens in whole-blood specimens. Journal of Clinical Microbiology 57(11):e00513–e00519.

Buchman, T. G., R. Draghia-Akli, S. J. Adam, N. R. Aggarwal, J. P. Fessel, E. S. Higgs, J. P. Menetski, S. W. Read, and E. A. Hughes. 2021. Accelerating coronavirus disease 2019 therapeutic interventions and vaccines-selecting compounds for clinical evaluation in coronavirus disease 2019 clinical trials. Critical Care Medicine 49(11):1963–1973.

Cavaller-Bellaubi, M., S. D. Faulkner, B. Teixeira, M. Boudes, E. Molero, N. Brooke, L. McKeaveney, J. Southerton, M. J. Vicente, N. Bertelsen, J. García-Burgos, V. Pirard, K. Reid, and E. Ferrer. 2021. Sustaining meaningful patient engagement across the lifecycle of medicines: A roadmap for action. Therapeutic Innovation & Regulatory Science 55(5):936–953.

CDC (Centers for Disease Control and Prevention). 2023. The CDC Clear Communication Index. https://www.cdc.gov/ccindex/index.html (accessed January 28, 2025).

CDC. 2024. Bartonella infection. https://www.cdc.gov/bartonella/about/index.html (accessed February 13, 2025).

Chen, Y., J. Dai, L. Tang, T. Mikhailova, Q. Liang, M. Li, J. Zhou, R. F. Kopp, C. Weickert, C. Chen, and C. Liu. 2023. Neuroimmune transcriptome changes in patient brains of psychiatric and neurological disorders. Molecular Psychiatry 28(2):710–721.

Chiu, C. Y., and S. A. Miller. 2019. Clinical metagenomics. Nature Reviews Genetics 20(6): 341–355.

Colombel, J. F., W. J. Sandborn, W. Reinisch, G. J. Mantzaris, A. Kornbluth, D. Rachmilewitz, S. Lichtiger, G. D’Haens, R. H. Diamond, D. L. Broussard, K. L. Tang, C. J. van der Woude, and P. Rutgeerts. 2010. Infliximab, azathioprine, or combination therapy for Crohn’s disease. New England Journal of Medicine 362(15):1383–1395.

Colvin, C. J. 2014. Evidence and AIDS activism: HIV scale-up and the contemporary politics of knowledge in global public health. Global Public Health 9(1–2):57–72.

Coppola, L., A. Cianflone, A. M. Grimaldi, M. Incoronato, P. Bevilacqua, F. Messina, S. Base-lice, A. Soricelli, P. Mirabelli, and M. Salvatore. 2019. Biobanking in health care: Evolution and future directions. Journal of Translational Medicine 17(1):172.

CUMC (Columbia University Irving Medical Center). 2025. Columbia specimen bank. https://www.columbia-lyme.org/columbia-specimen-bank (accessed January 30, 2025).

Cutler, S. J., M. Vayssier-Taussat, A. Estrada-Peña, A. Potkonjak, A. D. Mihalca, and H. Zeller. 2021. Tick-borne diseases and co-infection: Current considerations. Ticks and Tick-borne Diseases 12(1):101607.

Cvijović, I., M. Swift, and S. R. Quake. 2025. Long-term B cell memory emerges at uniform relative rates in the human immune response. Proceedings of the National Academy of Sciences 122(9):e2406474122.

Dagher, G. 2022. Quality matters: International standards for biobanking. Cell Proliferation 55(8):e13282.

Duke University. n.d. OneDukeBio Integrated Biospecimen Network (ODIN). https://medschool.duke.edu/research/research-support/core-facilities-service-centers/onedukebio-integrated-biospecimen-network (accessed December 11, 2024).

Duke-Margolis Center for Health Policy. 2019. Determining real-world data’s fitness for use and the role of reliability. https://healthpolicy.duke.edu/sites/default/files/2019-11/rwd_reliability.pdf (accessed December 11, 2024).

Duke-Margolis Center for Health Policy. 2024. Real-world evidence to support causal inference: Methodological considerations for non-interventional studies. https://healthpolicy.duke.edu/publications/real-world-evidence-support-causal-inference (accessed January 28, 2025).

Dumes, A. A. 2020. Lyme disease and the epistemic tensions of “medically unexplained illnesses.” Medical Anthropology 39(6):441–456.

Fallon, B. A., N. Zubcevik, C. Bennett, S. Doshi, A. W. Rebman, R. Kishon, J. R. Moeller, N. R. Octavien, and J. N. Aucott. 2019. The General Symptom Questionnaire-30 (GSQ-30): A brief measure of multi-system symptom burden in Lyme disease. Frontiers in Medicine (Lausanne) 6:283.

Fitzgerald, B. L., B. Graham, M. J. Delorey, A. Pegalajar-Jurado, M. N. Islam, G. P. Wormser, J. N. Aucott, A. W. Rebman, M. J. Soloski, J. T. Belisle, and C. R. Molins. 2021. Metabolic response in patients with post-treatment Lyme disease symptoms/syndrome. Clinical Infectious Diseases 73(7):e2342–e2349.

FDA (Food and Drug Administration). 2024. Real-world evidence. https://www.fda.gov/science-research/science-and-research-special-topics/real-world-evidence (accessed December 11, 2024).

Gu, X., S. Wang, W. Zhang, C. Li, L. Guo, Z. Wang, H. Li, H. Zhang, Y. Zhou, W. Liang, H. Li, Y. Liu, Y. Wang, L. Huang, T. Dong, D. Zhang, C. C. L. Wong, and B. Cao. 2023. Probing long COVID through a proteomic lens: A comprehensive two-year longitudinal cohort study of hospitalised survivors. eBioMedicine 98:104851.

Gliklich, R., M. Leavy, and N. Dreyer. 2020. Registries for evaluating patient outcomes: A user’s guide [internet]: Agency for Healthcare Research and Quality. https://www.ncbi.nlm.nih.gov/books/NBK562567/ (accessed June 6, 2025).

Goodson, N., P. Wicks, J. Morgan, L. Hashem, S. Callinan, and J. Reites. 2022. Opportunities and counterintuitive challenges for decentralized clinical trials to broaden participant inclusion. npj Digital Medicine 5(1):58.

Hanson, K. E., and M. R. Couturier. 2016. Multiplexed molecular diagnostics for respiratory, gastrointestinal, and central nervous system infections. Clinical Infectious Diseases 63(10):1361–1367.

Harris, J. R., P. Burton, B. M. Knoppers, K. Lindpaintner, M. Bledsoe, A. J. Brookes, I. Budin-Ljøsne, R. Chisholm, D. Cox, M. Deschênes, I. Fortier, P. Hainaut, R. Hewitt, J. Kaye, J. E. Litton, A. Metspalu, B. Ollier, L. J. Palmer, A. Palotie, M. Pasterk, M. Perola, P. H. Riegman, G. J. van Ommen, M. Yuille, and K. Zatloukal. 2012. Toward a roadmap in global biobanking for health. European Journal of Human Genetics 20(11):1105–1111.

HHS (Department of Health and Human Services). 2025. HHS announces transformation to make america healthy again. https://www.hhs.gov/about/news/hhs-restructuring-doge.html (accessed March 27, 2025).

Hernández, S. A., K. Ogrinc, M. Korva, A. Kastrin, P. Bogovič, T. Rojko, K. W. Kelley, J. J. Weis, F. Strle, and K. Strle. 2023. Association of persistent symptoms after Lyme neuroborreliosis and increased levels of interferon-α in blood. Emerging Infectious Diseases 6(29):1091–1101.

Horowitz, R. I., and P. R. Freeman. 2018. Precision medicine: The role of the MSIDS model in defining, diagnosing, and treating chronic Lyme disease/post treatment Lyme disease syndrome and other chronic illness: Part 2. Healthcare (Basel) 6(4):129.

Houghton, C., M. Dowling, P. Meskell, A. Hunter, H. Gardner, A. Conway, S. Treweek, K. Sutcliffe, J. Noyes, D. Devane, and et al. 2020. Factors that impact on recruitment to randomised trials in health care: A qualitative evidence synthesis. Cochrane Database of Systematic Reviews 10(10):MR000045.

Hu, C. W., Y. Qiu, A. Ligeralde, A. Y. Raybon, S. Y. Yoo, K. R. Coombes, A. A. Qutub, and S. M. Kornblau. 2019. A quantitative analysis of heterogeneities and hallmarks in acute myelogenous leukaemia. Nature Biomedical Engineering 3(11):889–901.

International Standard. 2018. Biotechnology-biobanking-general requirements for biobanking. https://www.iso.org/standard/67888.html (accessed December 11, 2024).

IOM (Institute of Medicine). 2012. Facilitating collaborations to develop combination investigational cancer therapies: Workshop summary. Washington, DC: The National Academies Press. https://doi.org/10.17226/13262.

Jin, H., L. Wang, and R. Bernards. 2023. Rational combinations of targeted cancer therapies: Background, advances and challenges. Nature Reviews Drug Discovery 22(3):213–234.

Johns Hopkins Medicine Lyme Disease Research Center. 2025. Research at the Lyme disease center. https://www.hopkinslyme.org/research-at-the-lyme-disease-center/ (accessed February 26, 2025).

Kieser, M., T. Friede, and M. Gondan. 2013. Assessment of statistical significance and clinical relevance. Statistics in Medicine 32(10):1707–1719.

Langhof, H., H. Kahrass, T. Illig, R. Jahns, and D. Strech. 2018. Current practices for access, compensation, and prioritization in biobanks. Results from an interview study. European Journal of Human Genetics 26(11):1572–1581.

LaPoint. 2024. Large-scale study will seek to unearth causes of persistent symptoms of Lyme disease. https://medicine.tufts.edu/news-events/news/large-scale-study-will-seek-unearth-causes-persistent-symptoms-lyme-disease (accessed December 11, 2024)

Liu, F., and D. Panagiotakos. 2022. Real-world data: A brief review of the methods, applications, challenges and opportunities. BMC Medical Research Methodology 22(1):287.

Mackenzie, F. 2014. Biobanking trends, challenges, and opportunities. Pathobiology 81(5–6): 245–251.

Mahlich, J., A. Bartol, and S. Dheban. 2021. Can adaptive clinical trials help to solve the productivity crisis of the pharmaceutical industry? A scenario analysis. Health Economics Review 11(1):4.

Marques, A. R. 2015. Laboratory diagnosis of Lyme disease: Advances and challenges. Infectious Disease Clinics of North America 29(2):295–307.

Mayo Clinic Laboratories. n.d. Tick-borne coinfection—Honing in on multiple infections. https://news.mayocliniclabs.com/infectious-disease/vector-borne-diseases/tick-borne-coin-fection/ (accessed January 30, 2025).

McGlothlin, A. E., and R. J. Lewis. 2014. Minimal clinically important difference: Defining what really matters to patients. JAMA 312(13):1342–1343.

McPhail, M. 2017. Making Lyme disease law: The role of patient advocacy in health law and policy. Queen’s Policy Review 8(1):105–120.

Molins, C. R., C. Sexton, J. W. Young, L. V. Ashton, R. Pappert, C. B. Beard, and M. E. Schriefer. 2014. Collection and characterization of samples for establishment of a serum repository for Lyme disease diagnostic test development and evaluation. Journal of Clinical Microbiology 52(10):3755–3762.

Morris, Z. S., S. Wooding, and J. Grant. 2011. The answer is 17 years, what is the question: Understanding time lags in translational research. Journal of the Royal Society of Medicine 104(12):510–520.

Morrison, T., S. Madaras, C. Larson, and R. Harrison. 2021. Personal agency and community resilience: Narratives of women navigating health care with chronic Lyme disease. Qualitative Health Research 31(14):2706–2714.

Mouelhi, Y., E. Jouve, C. Castelli, and S. Gentile. 2020. How is the minimal clinically important difference established in health-related quality of life instruments? Review of anchors and methods. Health and Quality of Life Outcomes 18(1):136.

Naggie, S., D. R. Boulware, C. J. Lindsell, T. G. Stewart, N. Gentile, S. Collins, M. W. McCarthy, D. Jayaweera, M. Castro, M. Sulkowski, K. McTigue, F. Thicklin, G. M. Felker, A. A. Ginde, C. T. Bramante, A. J. Slandzicki, A. Gabriel, N. S. Shah, L. A. Lenert, S. E. Dunsmore, S. J. Adam, A. DeLong, G. Hanna, A. Remaly, R. Wilder, S. Wilson, E. Shenkman, A. F. Hernandez, and the Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV-6) Study Group and Investigators. 2022. Effect of ivermectin vs. placebo on time to sustained recovery in outpatients with mild to moderate COVID-19: A randomized clinical trial. JAMA 328(16):1595–1603.

NASEM (National Academies of Sciences, Engineering, and Medicine). 2017. Communicating science effectively: A research agenda. Washington, DC: The National Academies Press.

NASEM. 2022a. Envisioning a transformed clinical trials enterprise for 2030: Proceedings of a workshop. Edited by T. Wizemann, A. W. Gee and C. Shore. Washington, DC: The National Academies Press.

NASEM. 2022b. Improving representation in clinical trials and research: Building research equity for women and underrepresented groups. Washington, DC: The National Academies Press.

Natale, P., V. Saglimbene, M. Ruospo, A. M. Gonzalez, G. F. M. Strippoli, N. Scholes-Robertson, C. Guha, J. C. Craig, A. Teixeira-Pinto, T. Snelling, and A. Tong. 2021. Transparency, trust and minimizing burden to increase recruitment and retention in trials: A systematic review. Journal of Clinical Epidemiology 134:35–51.

Nature Immunology. 2025. Single-cell profiling of the immune landscape across the human lifespan. Nature Immunology 26(2):172–173.

New York State. n.d. Department of Health Wadsworth center test approval. https://www.wadsworth.org/regulatory/clep/clinical-labs/obtain-permit/test-approval (accessed March 13, 2025).

NIAID (National Institute of Allergy and Infectious Diseases). 2025. Lyme disease co-infection. https://www.niaid.nih.gov/diseases-conditions/lyme-disease-co-infection (accessed February 13, 2025).

Nigrovic, L. E., D. N. Neville, L. Chapman, F. Balamuth, M. N. Levas, A. D. Thompson, A. B. Kharbanda, D. Gerstbrein, J. A. Branda, and B. W. Buchan. 2023. Multiplex high-definition polymerase chain reaction assay for the diagnosis of tick-borne infections in children. Open Forum Infectious Diseases 10(4):ofad121.

NIH (National Institutes of Health). n.d. Biologic Specimen and Data Repository Information Coordinating Center. https://biolincc.nhlbi.nih.gov/about/ (accessed December 11, 2024).

NIH. 2016. Final NIH policy on the use of a single institutional review board for multi-site research. https://grants.nih.gov/grants/guide/notice-files/NOT-OD-16-094.html (accessed. December 11, 2024).

NIH CC (National Institute of Health Clinical Center. n.d. What is an IND? https://www.cc.nih.gov/orcs/ind/what-is-an-ind (accessed January 20, 2025).

NIH OAR (National Institute of Health Office of AIDS Research). n.d. HIV source. https://hivinfo.nih.gov/hiv-source (accessed December 11, 2024).

Omura, C., M. de Lorenzi-Tognon, P. Benoit, V. Servellita, K. Foresythe, N. Brazer, M. Oseguera, D. Ingebrigtsen, J. Streithorst, D. Stryke, K. Zorn, M. Karalius, M. R. Wilson, and C. Chiu. 2025. Host response profiling from clinical metagenomic sequencing data for diagnosis of central nervous system infections. Open Forum Infectious Disease 12(Suppl 1).

Otoo, J. A., and T. S. Schlappi. 2022. REASSURED multiplex diagnostics: A critical review and forecast. Biosensors (Basel) 12(2):124.

Pallmann, P., A. W. Bedding, B. Choodari-Oskooei, M. Dimairo, L. Flight, L. V. Hampson, J. Holmes, A. P. Mander, L. Odondi, M. R. Sydes, S. S. Villar, J. M. S. Wason, C. J. Weir, G. M. Wheeler, C. Yap, and T. Jaki. 2018. Adaptive designs in clinical trials: Why use them, and how to run and report them. BMC Medicine 16(1):29.

Patterson, B. K., J. Guevara-Coto, J. Mora, E. B. Francisco, R. Yogendra, R. A. Mora-Rodríguez, C. Beaty, G. Lemaster, G. Kaplan Do, A. Katz, and J. A. Bellanti. 2024. Long COVID diagnostic with differentiation from chronic Lyme disease using machine learning and cytokine hubs. Scientific Reports 14(1):19743.

PCORI (Patient-Centered Outcomes Research Institute). 2024. PCORI methodology standards. https://www.pcori.org/research-related-projects/about-our-research/research-methodology/pcori-methodology-standards#Data%20Registries (accessed February 24, 2025).

Petukhova, A., J. P. Matos-Carcalho, and N. Fachada. 2024. Text clustering with large language model embeddings. International Journal of Cognitive Computing in Engineering 6:100–108.

PFMD (Patient Focused Medicines Development). 2020. The PFMD book of good practices. https://patientfocusedmedicine.org/bogp/2020/the-book-of-good-practices.pdf (accessed January 28, 2025).

Porwancher, R. B., C. G. Hagerty, J. Fan, L. Landsberg, B. J. Johnson, M. Kopnitsky, A. C. Steere, K. Kulas, and S. J. Wong. 2011. Multiplex immunoassay for Lyme disease using vlse1-igg and pepc10-igm antibodies: Improving test performance through bioinformatics. Clinical Vaccine Immunology 18(5):851–859.

Proctor, E. K., E. Toker, R. Tabak, V. R. McKay, C. Hooley, and B. Evanoff. 2021. Market viability: A neglected concept in implementation science. Implementation Science 16(1):98.

Project ECHO. n.d. Long COVID and fatiguing illness recovery program echo. https://iecho.org/public/program/PRGM1699044218879IERCAXHJ8Y (accessed February 25, 2025).

Project Lyme. 2021. Introducing co-infections. https://projectlyme.org/resource/introducing-co-infections/ (accessed February 13, 2025).

Puerini, R., A. Suthar, K. Forth, and K. Schneeman. 2024. Defining and demonstrating the value of patient engagement in medtech research and product development. https://milkeninstitute.org/sites/default/files/2024-10/PatientEngagementMedtech.pdf (accessed February 6, 2025).

Rebman, A. W., K. T. Bechtold, T. Yang, E. A. Mihm, M. J. Soloski, C. B. Novak, and J. N. Aucott. 2017. The clinical, symptom, and quality-of-life characterization of a well-defined group of patients with posttreatment Lyme disease syndrome. Frontiers in Medicine (Lausanne) 4:224.

RECOVERY Collaborative Group. 2021. Dexamethasone in hospitalized patients with COVID-19. New England Journal of Medicine 384(8):693–704.

Reese, J. T., H. Blau, E. Casiraghi, T. Bergquist, J. J. Loomba, T. J. Callahan, B. Laraway, C. Antonescu, B. Coleman, M. Gargano, K. J. Wilkins, L. Cappelletti, T. Fontana, N. Ammar, B. Antony, T. M. Murali, J. H. Caufield, G. Karlebach, J. A. McMurry, A. Williams, R. Moffitt, J. Banerjee, A. E. Solomonides, H. Davis, K. Kostka, G. Valentini, D. Sahner, C. G. Chute, C. Madlock-Brown, M. A. Haendel, P. N. Robinson, H. Spratt, S. Visweswaran, J. E. I. V. Flack, Y. J. Yoo, D. Gabriel, G. C. Alexander, H. B. Mehta, F. Liu, R. T. Miller, R. Wong, E. L. Hill, L. E. Thorpe, and J. Divers. 2023. Generalisable long COVID subtypes: Findings from the NIH N3C and RECOVER programmes. eBioMedicine 87:104413.

REMAP-CAP Investigators. 2021. Interleukin-6 receptor antagonists in critically ill patients with COVID-19. New England Journal of Medicine 384(16):1491–1502.

Rowan, S., N. Mohseni, M. Chang, H. Burger, M. Peters, and S. Mir. 2023. From tick to test: A comprehensive review of tick-borne disease diagnostics and surveillance methods in the United States. Life (Basel) 13(10).

Sacristán, J. A., A. Aguarón, C. Avendaño-Solá, P. Garrido, J. Carrión, A. Gutiérrez, R. Kroes, and A. Flores. 2016. Patient involvement in clinical research: Why, when, and how. Patient Preference and Adherence 10:631–640.

Schutzer, S. E., T. E. Angel, T. Liu, A. A. Schepmoes, T. R. Clauss, J. N. Adkins, D. G. Camp, B. K. Holland, J. Bergquist, P. K. Coyle, R. D. Smith, B. A. Fallon, and B. H. Natelson. 2011. Distinct cerebrospinal fluid proteomes differentiate post-treatment Lyme disease from chronic fatigue syndrome. PLOS One 6(2):e17287.

Schwartz, A. M., A. F. Hinckley, P. S. Mead, S. A. Hook, and K. J. Kugeler. 2017. Surveillance for Lyme disease—United States, 2008-2015. MMWR Surveillance Summaries 66(22):1–12.

Shah, J. S., J. J. Burrascano, and R. Ramasamy. 2023. Recombinant protein immunoblots for differential diagnosis of tick-borne relapsing fever and Lyme disease. Journal of Vector Borne Diseases 60(4):353-364.

Simon, G. E., R. Platt, J. H. Watanabe, A. B. Bindman, A. John London, M. Horberg, A. Hernandez, and R. M. Califf. 2022. When can we rely on real-world evidence to evaluate new medical treatments? Clinical Pharmacology & Therapeutics 111(1):30–34.

SLICE (Study of Lyme Disease Immunology and Clinical Events) Study. n.d. The SLICE study. http://www.slicestudies.org/ (accessed November 27, 2025).

Snapes, E., J. J. Astrin, N. Bertheussen Krüger, G. H. Grossman, E. Hendrickson, N. Miller, and C. Seiler. 2023. Updating International Society for Biological and Environmental Repositories Best Practices, fifth edition: A new process for relevance in an evolving landscape. Biopreservation and Biobanking 21(6):537–546.

Stanford Medicine. n.d. Biobank services. https://med.stanford.edu/biobank/services.html (accessed December 11, 2024).

Strle, K., D. Stupica, E. E. Drouin, A. C. Steere, and F. Strle. 2014. Elevated levels of IL-23 in a subset of patients with post-Lyme disease symptoms following erythema migrans. Clinical Infectious Diseases 58(3):372–380.

TBDWG (Tick-Borne Disease Working Group). 2018. Tick-Borne Disease Working Group 2018 report to Congress. https://www.hhs.gov/sites/default/files/tbdwg-report-to-congress-2018.pdf (accessed November 27, 2024).

TBDWG. 2020. Tick-Borne Disease Working Group 2020 report to Congress. https://www.hhs.gov/sites/default/files/tbdwg-2020-report_to-ongress-final.pdf (accessed November 27, 2024).

TBDWG. 2022. Tick-Borne Disease Working Group 2022 report to Congress. https://www.hhs.gov/sites/default/files/tbdwg-2022-report-to-congress.pdf (accessed November 27, 2024).

Tokarz, R., N. Mishra, T. Tagliafierro, S. Sameroff, A. Caciula, L. Chauhan, J. Patel, E. Sullivan, A. Gucwa, B. Fallon, M. Golightly, C. Molins, M. Schriefer, A. Marques, T. Briese, and W. I. Lipkin. 2018. A multiplex serologic platform for diagnosis of tick-borne diseases. Scientific Reports 8(1):3158.

Touradji, P., J. N. Aucott, T. Yang, A. W. Rebman, and K. T. Bechtold. 2019. Cognitive decline in post-treatment Lyme disease syndrome. Archives of Clinical Neuropsychology 34(4):455–465.

Tufts University. 2024. Large-scale study will seek to unearth causes of persistent symptoms of lyme disease https://medicine.tufts.edu/news-events/news/large-scale-study-will-seek-unearth-causes-persistent-symptoms-lyme-disease (accessed March 17, 2025).

Wang, Y., R. Li, R. Tong, T. Chen, M. Sun, L. Luo, Z. Li, Y. Chen, Y. Zhao, C. Zhang, L. Wei, W. Lin, H. Chen, K. Qian, A. F. Chen, J. Liu, L. Chen, B. Li, F. Wang, L. Wang, B. Su, and J. Pu. 2025. Integrating single-cell RNA and T cell/B cell receptor sequencing with mass cytometry reveals dynamic trajectories of human peripheral immune cells from birth to old age. Nature Immunology 26(2):308–322.

West, J. D., and C. T. Bergstrom. 2021. Misinformation in and about science. Proceedings of the National Academy of Sciences 118(15):e1912444117.

Wiering, B., D. de Boer, and D. Delnoij. 2017. Patient involvement in the development of patient-reported outcome measures: The developers’ perspective. BMC Health Services Research 17(1):635.

Wolinetz, C. D., and F. S. Collins. 2017. Single-minded research review: The common rule and single IRB policy. American Journal of Bioethics 17(7):34–36.

Woodcock, J., and L. M. LaVange. 2017. Master protocols to study multiple therapies, multiple diseases, or both. New England Journal of Medicine 377(1):62–70.

Wormser, G. P., D. McKenna, C. L. Karmen, K. D. Shaffer, J. H. Silverman, J. Nowakowski, C. Scavarda, E. D. Shapiro, and P. Visintainer. 2020. Prospective evaluation of the frequency and severity of symptoms in Lyme disease patients with erythema migrans compared with matched controls at baseline, 6 months, and 12 months. Clinical Infectious Diseases 71(12):3118–3124.

Wright, W. F., D. J. Riedel, R. Talwani, and B. L. Gilliam. 2012. Diagnosis and management of Lyme disease. American Family Physician 85(11):1086–1093.

Yale Medicine. n.d. Tick-borne illnesses. https://www.yalemedicine.org/conditions/tick-borne-illnesses (accessed February 13, 2025).

Zaslavsky, M. E., E. Craig, J. K. Michuda, N. Sehgal, N. Ram-Mohan, J. Y. Lee, K. D. Nguyen, R. A. Hoh, T. D. Pham, K. Röltgen, B. Lam, E. S. Parsons, S. R. Macwana, W. DeJager, E. M. Drapeau, K. M. Roskin, C. Cunningham-Rundles, M. A. Moody, B. F. Haynes, J. D. Goldman, J. R. Heath, R. S. Chinthrajah, K. C. Nadeau, B. A. Pinsky, C. A. Blish, S. E. Hensley, K. Jensen, E. Meyer, I. Balboni, P. J. Utz, J. T. Merrill, J. M. Guthridge, J. A. James, S. Yang, R. Tibshirani, A. Kundaje, and S. D. Boyd. 2025. Disease diagnostics using machine learning of B cell and T cell receptor sequences. Science 387(6736):eadp2407.

This page intentionally left blank.