4

Building an Equitable Precision Health Care System

Building on the personal perspectives of equity in genomic medicine that were shared by the first panel, the second session speakers discussed the concerns of diverse communities and what can be done to help achieve equitable precision care at the health system level. The session was moderated by Gabriel Lázaro-Muñoz, assistant professor of psychiatry and member of the Harvard Medical School Center for Bioethics at Harvard Medical School.

ADDRESSING THE NEEDS AND CONCERNS OF DIVERSE COMMUNITIES

A Focus on Sexual and Gender Minorities

“For precision medicine to be successful, we need to make sure that we are thinking about it not just in the clinic, but also from the perspective of the social determinants of health, the things that happen outside of the clinical setting that really affect how well our medical technologies and our treatments work,” said Kellan Baker, executive director and chief learning officer at the Whitman-Walker Institute. Not taking social determinants of health—as well as patient demographics such as race, ethnicity, disability, sexual orientation, and gender—into account when implementing precision medicine tools and technologies will exacerbate existing disparities and further disadvantage already underserved populations, he said.

Barriers to the Uptake of Genomics and Precision Care

To benefit from precision medicine, patients must have access to it, Baker said. Some of the barriers to access that can disproportionately affect diverse and underserved populations include costs of care, insurance coverage of evaluations and treatments, availability of providers with expertise in precision medicine, and transportation to the provider’s office, Baker said. Health literacy also affects access, and Baker noted that information about the availability and benefits of new medical breakthroughs often does not reach diverse populations, especially those who do not receive routine

clinical care. Discrimination also affects access to precision medicine, both directly, and because “Previous and anticipated experiences of discrimination in health care settings create medical mistrust,” Baker said.

Many individuals and communities are skeptical about advances, such as within precision medicine, and harbor concerns that they could be hurt, or that they will once again give of themselves and receive nothing in return. For example, there is great mistrust of genomic research and precision medicine among many lesbian, gay, bisexual, transgender, and queer (LGBTQ) persons, Baker noted. This stems from a history of mistreatment of LGBTQ individuals in medical care: a recent nationwide study showed that LGBTQ respondents were more than three times as likely than non-LGBTQ individuals to “postpone or avoid getting medical care because they were afraid of mistreatment,” Baker said. LGBTQ patients may also mistrust genomics because of misguided research efforts to identify a so-called “gay gene” or a “trans gene,” which raise concerns of eugenics should such genes ever be found, he added.

Making Genomics and Precision Medicine Acceptable to Patients

“Medical mistrust is not a condition to be cured of,” Baker said. “It is an adaptive response to historical and ongoing mistreatment.” The onus is on the medical establishment to become trustworthy. One approach is to develop a workforce that better reflects the communities it serves, meaning patients are served by clinicians and researchers who have similar racial or other backgrounds. This requires creating career paths that attract and retain people of color, women, people with disabilities, and people who identify as LGBTQ, Baker said. Evidence suggests that these populations drop out of science, technology, engineering, and math (STEM) and health career pathways early on. For example, it is estimated that the representation of LGBTQ people in STEM fields is 20 percent less than expected, he said. Often when students do not see health or research professionals who look like them, they have difficulty envisioning themselves in those roles.

Another approach is to ensure that clinicians and researchers receive continuing education, training, and resources on working with diverse populations. Baker observed that there is much attention on advancing the technical aspects of science and medicine, but these tools can be of limited use if they are not implemented or administered. Genomic medicine tools must be acceptable to, and understandable by, patients and broader communities, and trainings as well as other resources that incorporate the experiences and concerns of diverse patient populations need to be readily available to, and taken up by, clinicians and researchers, he said. For example, Baker said that resources for working effectively with LGBTQ patients exist but are often viewed as difficult to find.

Incorporating Patient Identities and Experiences

Precision medicine’s focus on the genome and specific therapeutic targets mean that the focus on the person at the center of this approach can be lost, Baker observed. Data on people’s identities and day-to-day experiences could help inform the effective use of these advances, he added. The field is lacking data on patient sex, gender identity, and sexual orientation; precise definitions of what is meant by these terms; and methods for how best to collect these data. In 2022, a consensus study committee at the National Academies of Sciences, Engineering, and Medicine¹ released recommended standards for the collection of data on sex, gender identity, and sexual orientation, Baker noted (NASEM, 2022). These are all “complex multidimensional elements of human identity” that are being incorrectly broken down into a simple binary, he said. For example, sex includes components of genetics, anatomy, and hormones, and one’s sex as assigned on a birth certificate dictates how a child is raised and the way the world interacts with them but may not reflect the individual’s true gender. Baker asked, “Does the truth of who we are really lie in our genes, or is it instead a really complex relationship between the genome [and] so many other elements of our biology, of our surroundings, of our upbringings?”

A Focus on Racial and Ethnic Minority Groups

Health equity is “the absence of avoidable, unfair, or remediable differences in health outcomes among socially disadvantaged populations,”² said Consuelo Wilkins, chief equity officer, professor of medicine, senior vice president, and senior associate dean for Health Equity and Inclusive Excellence, and engagement core director for the All of Us research program at Vanderbilt University Medical Center. “Precision medicine has been built on a foundation of inequity,” she said. Inequitable practices in health care, precision medicine, and genomics have resulted in flawed science. Nearly 80 percent of the individuals in the genome-wide association studies (GWAS) catalog are of European ancestry. Although inclusion of non-European ancestry individuals in GWAS has increased over the last decade, they are still woefully underrepresented (Popejoy and Fullerton, 2016; Quansah and McGregor, 2017; Sirugo et al., 2019). As such, the benefits of the treatments and tools, such as diagnostics, screening tests, and algorithms, that were developed based on these studies do not translate equally to all people, she said. The population of the world continues to change, and 80 percent of the global population is expected to be of Asian

___________________

¹ Herein referred to as “the National Academies.”

² Adapted from the World Health Organization. See https://www.who.int/health-topics/social-determinants-of-health#tab=tab_3 (accessed December 14, 2022).

or African descent by 2050, Wilkins added. Therefore, reference data sets need to change, she emphasized.

Race, Ethnicity, and Genetic Testing

“Race is a sociopolitical construct,” Wilkins said. Although there can be some overlap with ancestry, people are grouped into races based on physical characteristics, and race-based discrimination and oppression persist in the United States. Although there is no biological basis for race, it cannot be disassociated from health, Wilkins said.

There is much discussion of how to use race in science going forward, whether or how genetic ancestry might be used in place of race, and how categorizing people affects genomics and precision health (Lewis et al., 2022). Wilkins discussed research being done by the Vanderbilt-Miami-Meharry Precision Medicine Disparities Collaborative using samples from Vanderbilt’s DNA biorepository which are linked to medical records. This research found that when race is controlled for, the disorders that are most associated with African genetic ancestry, determined by GWAS, are kidney disease and sickle cell anemia. However, when instead of race, genetic ancestry is controlled for, the conditions most associated with those who identify by race as Black or African American are hypertension and fatigue, conditions that are often associated with allostatic load and with one’s environment (e.g., experiencing racism and discrimination, living in disinvested communities), Wilkins said.

Current genetic screening and diagnostic tests do not provide equal benefit for all populations, Wilkins said. Studies have shown that rates of variants of uncertain or unknown significance (VUSs) are higher in racial or ethnic minority individuals compared to individuals of European ancestry (Kwon et al., 2020; Ndugga-Kabuye and Issaka, 2019; Roberts et al., 2020). Yet, there are few discussions of revisiting these VUSs to determine their significance, she said.

Improving Equity in Precision Health Care

In general, the delivery of health care is inequitable, and this is not unique to genomics and precision medicine. The blame is often placed on implicit bias, but Wilkins said there is evidence that the root cause of racial and ethnic disparities in health care is racism (IOM, 2003), adding:

Further, many of the inequities in health outcomes could be attributable not to genetics, but to the body’s response to the physiological stress of conditions such as living in poverty or in a disinvested community, experiencing racism, or having an increased allostatic load, she suggested. These stressors “are increasing [the] risk of disease, including risk of epigenetic changes that could lead to disease,” Wilkins said (Carlos et al., 2022). “We can talk about access. We can talk about all sorts of things. But if we’re not going to address the underlying racism, then we’re not going to get very far,” she added.

Wilkins concluded her talk by highlighting opportunities for building equitable precision health care:

• Committing funding to closing the large gap in quality reference data for diverse populations. These new data could identify novel genomic variants, establish clinical relevance, and interpret VUSs. Incremental steps will not close the data gap; a large injection of funding to support increased activity is needed.
• Acknowledging that some racial and ethnic groups are benefiting less from precision medicine and precision health and committing to remedying that.
• Recognizing that delivering on the promise of precision medicine involves more than genetics and genomics and incorporating information on patients’ social and structural determinants of health in precision health care.
• Increasing diversity in the health care and research workforce, and reaching out to, and engaging with, marginalized communities.
• Understanding and incorporating the needs and priorities of minority groups in precision health care by involving communities in study design.
• Working to build trust in health care and clinical research. “Acknowledge again that people have real reasons to distrust us based on our past and current activities and the way we deliver care and provide and perform research,” she concluded.

CASE EXAMPLE: REDUCING BIAS BY IMPLEMENTING PANEL-BASED PREEMPTIVE PHARMACOGENETIC TESTING FOR EVERYONE

“Most adults and many children receive at least one high-risk actionable pharmacogenetic drug in their lifetime,”³ said Mary Relling, coinvestigator

___________________

³ An example of a high-risk actionable pharmacogenetic drug is clopidogrel or warfarin (Schildcrout et al., 2012).

and cofounder of the Clinical Pharmacogenetics Implementation Consortium (CPIC)⁴ and member of the Pharmaceutical Sciences Department at St. Jude Children’s Research Hospital (Chan et al., 2021; Chanfreau-Coffinier et al., 2019; Kimpton et al., 2019; Kuch et al., 2016; Samwald et al., 2016). In addition, more than 95 percent of the U.S. population has a high-risk diplotype for at least one of the original 12 genes identified by CPIC as pharmacogenetically actionable (Dunnenberger et al., 2015).

In any given year, about half of the patients at St. Jude Children’s Research Hospital will receive at least one high-risk drug, Relling said. The cost of multigene panel testing is comparable to that of single gene testing and so, in 2011, St. Jude Children’s Research Hospital launched the PG4KDS protocol to implement panel-based preemptive pharmacogenetic testing as the standard of care for all consenting patients (Haidar et al., 2022).⁵ This approach supports the mission of St. Jude to “advance cures, and means of prevention, for pediatric catastrophic diseases through research and treatment,” as well as the hospital’s culture of evidence-based prescribing, she said.

Implementing the PG4KDS Protocol

Pharmacogenetic Testing in Children

St. Jude Children’s Research Hospital believes “There is no ethical reason to exclude children from preemptive pharmacogenetic testing,” Relling said. Germline genomic variants do not change as a person ages, and CPIC gene–drug prescribing guidelines are applicable to children. She highlighted some of the special considerations for pharmacogenomic testing in pediatric patients at St. Jude. Parents provide consent on behalf of underage patients, which includes specific consent for the possibility of finding sex chromosome abnormalities. Consent is obtained again directly from the patient when they reach 18 years of age, and Relling noted that most pediatric practices have patients who have reached adulthood. At St. Jude, “Parents are not allowed to refuse incidental findings on behalf of their child,” Relling said. The incidental findings that must be reported to parents and patients are those that have “a very strong association with disease risk, [are] not likely to be discovered via other means, and [are] actionable before the child reaches 18 years of age,” she explained.

___________________

⁴ CPIC was established in 2009 to “create, curate, update, [and] make freely available specific peer-reviewed, evidence-based, updatable clinical guidelines for actionable gene/drug pairs,” Relling said. See https://cpicpgx.org (accessed December 14, 2022).

⁵ See also www.stjude.org/pg4kds/implement (accessed December 14, 2022).

The Process

The PG4KDS process begins with research nurses identifying patients for potential enrollment in the protocol from the daily list of new patients, and Relling said that nearly all are approached to participate (Hoffman et al., 2014). Of patients that are asked, more than 95 percent consent to participate and more than 96 percent of participants reconsent at age 18. Pharmacogenetic testing to inform prescribing is considered to be part of the patient’s medical care, “similar to how we would treat administration of flu vaccine,” Relling said. Testing is approved by an oversight committee, implemented institution-wide, and each enrollment in the protocol does not require consultation with the primary physician. Once enrolled, a blood sample is collected and more than 1,100 genes are genotyped. All results are entered into the research database, and results for select genes are entered into the patient’s electronic health record (those for which there is clinical decision support available for at least one drug affected by the gene), she explained.

Because genomic findings have life-long implications, genotype results are entered into a dedicated pharmacogenetics tab in the EHR so they are not encounter specific, Relling explained. The EHR contains the diplotype, the interpretation, and patient letter and is uploaded to the patient portal. High-risk diplotypes are also automatically populated into the patient’s problem list as phenotypes and are linked to clinical decision support that guides prescribing through interruptive alerts, she said. There are also pretest alerts that are triggered for select gene–drug pairs when the relevant genetic test result is absent from the patient record. Clinical decision support grows more complex when more than one gene affects prescribing for a drug, Relling said. Standardization of terminology is important for sharing data across EHR systems, she noted.

Progress

More than 6,600 patients have been enrolled in PG4KDS and genotyped, and around 53 percent are White, Relling said. Thus far the protocol has been implemented for 14 genes that affect a total of 66 drugs, and 95 percent of patients at St. Jude have been found to have at least one high-risk phenotype for those 14 genes. Adherence to the prescribing advice in the clinical decision support alerts among practitioners has been very high (92 percent) over the past decade (Nguyen et al., 2022).

Relling reviewed the process for implementing a new gene–drug pair beginning with results interpretation, clinical consultations, creation of problem list entries, and development of clinical decision support, which is done using the resources of CPIC. This is a time-consuming process, and only one or two new genes are added per year. St. Jude then updates its

formulary, drug policies, patient and provider educational materials, and clinical competencies; approval is obtained from the pharmacogenetics oversight committee; and information about the new gene–drug pair is shared publicly, she said.

The expansion of precision medicine, including clinical genomics and pharmacogenomic testing, is now part of the St. Jude Children’s Research Hospital strategic plan and is supported by institutional leadership, Relling said. Pharmacogenomic testing for thiopurines is also now an institutional patient safety metric (e.g., the goal is to have 100 percent of patients tested for the relevant genes before receiving their first dose of the drug).⁶

DISCUSSION

Protecting Against the Misuse of Information from the Genome

A topic of much discussion was preventing the misuse of genetic information and ensuring that the application of genomics in health care does not further exacerbate health disparities and discrimination. One participant asked, how can “the systemic extinction of neuroatypical persons and other marginalized groups” be prevented?

Greater legal and regulatory attention on how to prevent “the misuse of information from the genome to selectively apply our social or cultural preferences for particular groups of people or different elements of human diversity” may be essential, Baker said. As an example, the selective abortion of fetuses based on genetic screening results has essentially eliminated Down syndrome in some places around the world, he said. The Genetic Information Nondiscrimination Act was designed to help protect people from discrimination based on their genetic information. However, it does not consider how genomic information might be used to make decisions about an unborn fetus, particularly from a eugenics standpoint, he added.

An underlying issue is the flawed determination of the value of the quality of life of a member of a particular population, especially as assessed by those who are not part of that population, Baker suggested. For example, studies have found that people without disabilities rate the quality of life of people with a given disability much lower than those with the disability rate their own quality of life. It is this perception (or misperception) of the value of life that drives “what ultimately would be eugenics practices of trying to ensure that children with particular disabilities or particular elements of their identity are simply not born,” he said. Cost-benefit analyses of preci-

___________________

⁶ Thiopurines are commonly used to treat autoimmune disorders and acute lymphoblastic leukemia; however, there are various genetic interactions that can affect drug response (Fridley et al., 2010).

sion medicine interventions can take the firsthand perspectives and experiences of people living with genetic conditions into account, Baker said.

Wilkins added that current policies and practices are focused on protecting individuals, not groups of people, from misuse of genetic information. Any policy or practice solutions must start with acknowledging the country’s history of eugenics and of creating structures of oppression directed at groups of people, Wilkins said, and individuals from affected groups should be engaged in developing policies and oversight mechanisms. Genetic data or presumed genetic data can be misused “to reinforce stereotypes, systems of oppression, [and] inferiority.” The fears and concerns that marginalized and minoritized people have about the misuse of genetic information are valid and must not be dismissed, she said. The smaller the minority group, the greater the risk of the loss of individual data privacy, she added. Baker added that there is a belief that disparities in health are associated with “something fundamental” in the genome. A focus on the individual is the core of precision medicine. However, “a focus on the individual in terms of disparities is simply shifting the blame from the structural forces that are creating and maintaining the disparities…to the individual,” emphasizing genetics and behavior and taking the focus off addressing these societal structures, he said.

The perspective that genomics is objective scientifically supports precision medicine, but Baker observed that genomics has also been “weaponized” in policies relating to individual social and cultural identity. For example, people with intersex variations and those who are transgender have been targeted, and it has become “the province of the government or a sporting body…to tell you who you are and how you fit into society, and what kinds of benefits, protections, and access are available to you,” he said.

Increased efforts may be essential to educate the public and professionals about both the benefits and the limitations of genomic information, Lázaro-Muñoz said.

Capturing Racial, Ethnic, and Ancestral Identity Information

Panelists also discussed the need for new approaches for capturing data on an individual’s identity. Relling said St. Jude uses the National Institutes of Health (NIH) categories for race and ethnicity and includes self-declared race and genomic ancestry data when publishing studies. Individuals who participate in the All of Us research program are asked to select which race and ethnicity categories they feel best identifies them from a list of many options, Wilkins said, which offers an alternative to the constraints of the Office of Management and Budget (OMB) categories, which NIH uses. Participants can choose as many as they want. They can also answer branching questions and select any ethnicity or nationality subgroups that correspond

with how they identify. Data are compiled and still reported according to the OMB categories as required, she noted. For pharmacogenetics, focusing on the genomic variants instead of racial or ancestral identity results in clinical prescribing guidelines that are broadly applicable regardless of race or ancestry in almost all cases, Relling noted.

How the Genomics Community Can Promote Equity

As an example of the challenges of equitable use of genomic data, Relling said that the human leukocyte antigen (HLA) genotyping algorithms being used are largely based on data from individuals of European ancestry and do not provide accurate results for persons who are not of European ancestry. Although St. Jude could deliver accurate results to its patients of European ancestry, it was decided to not provide preemptive HLA genotype results to any patient at this time. The situation with HLA genotyping is a good example of how some groups currently benefit less from genomics and precision care than others, Wilkins said. Funding can help to close these types of gaps, in this case, to develop HLA genotyping algorithms in other ancestry groups so accurate results can be provided to all patients, she noted. There is a need to better integrate data on social and structural determinants of health with other health-related data, Wilkins highlighted. Precision medicine is a powerful tool for health, and the genomics community should not only develop and apply these types of tools and technologies but also participate in ensuring they reach all those who can benefit from them, Baker said.

Drawing on the St. Jude experience, Relling supported a model in which genomic testing would be embedded in institutional policies that direct best practices in care. This model reduces provider-based variability, and testing becomes routine for all patients, regardless of background, sex, race, or ancestry, she said. For example, weight, height, age, and liver and renal function are always considered when prescribing, she noted. If practice evolved such that prescribing was not done without knowledge of the patient’s pharmacogenetic testing results, “then all patients would benefit from pharmacogenomics…not just the select few at academic medical centers where the physicians are trained to order pharmacogenetic tests.”

This page intentionally left blank.