Assessing the psychological impact of genetic susceptibility testing: Where have we been, where do we go from here?

Introduction

Genetic susceptibility testing for various medical conditions is becoming more commonplace across health care, research, and direct-to-consumer settings. Such expanded use of genetic testing raises key ethical and policy questions regarding the likelihood of benefits and harms among those receiving disease risk information. Concerns about such harms have often focused on the potential for psychological distress in response to testing. These concerns, rooted in the core bioethics principle of nonmaleficence, have long provided a rationale for either withholding or strictly regulating access to genetic risk information, particularly for more severe and untreatable diseases such as Huntington’s and Alzheimer’s. Yet many commentators (e.g., Angrist, 2015)¹ assert that fears of significant negative psychosocial impacts of testing are overblown, and that a paternalistic approach to genetic susceptibility testing undermines both patient and consumer autonomy. Research on the extent and likelihood of psychological harms of genetic testing is therefore a key information source in current ethical and policy debates on the provision and regulation of genetic testing. This paper briefly reviews what we have learned so far about the psychological impact of disclosing disease risk information, highlights several methodological limitations of the work to date, and offers some suggestions to guide a future research agenda in this area.

The early days: Testing for Huntington’s disease

Breakthroughs in research on the genetic causes of Huntington’s disease (HD) in the 1980s spurred some of the earliest studies of the psychological impact of genetic testing. As predictive testing for HD was initiated in a clinical setting, survey research posing hypothetical test scenarios suggested that the vast majority of at-risk relatives would like to know whether or not they carried a disease-causing mutation.² However, only a small minority (under 20%) ultimately availed themselves of this opportunity, in part because many at-risk individuals concluded that a positive test result for this feared, incurable condition would be too psychologically overwhelming for themselves and their family members.³ HD testing protocols have long paid significant attention to the potential for psychological harm by mandating pre-test education and counseling and routinely incorporating patient assessment by a mental health professional prior to disclosure of test results.⁴ Yet even with these safeguards in place, there was still concern among the medical community that individuals might react poorly to a positive test result, with suicide as the most feared outcome. A landmark study by Wiggins et al (1992),⁵ published in the New England Journal of Medicine, which followed people who had requested and undergone predictive testing for HD, helped allay some of these fears. The researchers administered multiple psychiatric symptom checklists both before and after testing, and found that learning positive test results (i.e., finding out there was a high likelihood that one would ultimately develop HD later in life) was not associated with elevated levels of depression or reduced levels of psychological well-being after testing. This pattern of findings—that psychological distress in response to genetic susceptibility testing is generally mild and transient for those who choose to undergo testing—has been replicated in numerous HD studies⁶ and observed across multiple other disease contexts, from hereditary cancer syndromes to Alzheimer’s disease.⁷ In fact, participants’ pre-test emotional state has been shown to predict subsequent distress to a much greater degree than test results themselves.⁸

Why the null findings?

The broader psychology literature on affective forecasting offers potential explanations not only for this pattern of findings, but also for the common presumption that emotional reactions to genetic test results will be stronger than they ultimately turn out to be. For example, a substantial literature suggests that individuals are often more resilient than anticipated in coping with many different health-related stresses.⁹ Humans are prone to what cognitive psychologists call an impact bias: a tendency to overestimate the length and/or intensity of future emotional states in response to a particular life event.¹⁰ Several cognitive processes may underlie this bias, particularly in regard to response to negative life events. As Halpern and Arnold (2008)¹¹ nicely summarized, “people focus more on what they will lose than what will stay the same (focalism), they fail to envision how their own coping skills will lessen their unhappiness (immune neglect), and they fail to envision how they might develop new values (adaptation).” It should be pointed out that such biases may not only operate in those undergoing genetic testing, but also in clinicians and policymakers who overrate the psychological harms of testing!

Yet those contending that studies to date have underestimated the potential for psychological harm in response to genetic testing have several valid critiques to offer. First, several notable selection biases limit the generalizability of findings from this line of research. These studies have typically been conducted at major academic centers in urban locations, with patient populations lacking in diversity in terms of race, ethnicity, and socioeconomic status. Furthermore, psychologically vulnerable individuals are unlikely to volunteer for such studies in the first place and/or might be screened out based on concerns over protection of human subjects. Second, genetic testing in these studies is generally provided in well-controlled research settings that may represent a best-case scenario for disclosure of test results. This typically means that highly trained genetic counselors are disclosing test results within standardized protocols including intensive, in-person, pre-test education and counseling. In today’s rapidly changing genetic testing landscape, these conditions may well not apply—particularly in the case of direct-to-consumer genetic testing, where pre-test education and interaction with genetic counselors are typically absent. A third reason for skepticism about null findings of psychological distress in response to genetic testing concerns the timing of study measurements. Some studies have not administered their initial assessments of post-test responses until several weeks after disclosure of results, which may be too late to detect significant distress responses. Moreover, most studies in this area have not followed participants beyond a year after disclosure of results. These study designs would fail to capture the potential emergence of psychological distress as individuals near the age where their parent developed the genetic disease in question. This phenomenon was observed in an interesting longer-term study of HD test recipients 7–10 years after receiving results, which showed increases in hopelessness and intrusive thoughts among carriers as they approached potential age of disease onset.¹²

A closer look at measurement

The issue of measurement is worth considering in greater detail. Most major studies in this area have used measures of depression and anxiety as their primary outcome. The advantages of this approach are that these outcomes are widely recognized as clinically important, and existing depression and anxiety scales are well-validated, with strong psychometric properties and established clinical cut-off scores. Depression and anxiety scales are easily administered and can be used across multiple disease contexts. Yet responses to such measures are often influenced by many factors and life events beyond genetic testing, and these scales are often not sensitive enough in detecting psychological effects specific to learning test results. For example, the primary outcome in the aforementioned Wiggins et al (1992)⁵ HD study was the Symptom Checklist 90-Revised (SCL-R),¹³ a multidimensional scale whose items ask respondents to indicate how much they have recently been bothered by a wide variety of problems and complaints, including “feeling shy or uneasy with the opposite sex” and “others not giving you proper credit for your achievements.” It is not hard to see how such measures may be assessing psychological phenomena unrelated to the effects of genetic testing.

To help address this concern, studies in the field have increasingly incorporated measures of test-specific distress. A systematic review of outcome measurement in clinical genetic services¹⁴ found that the most commonly used distress measure, by far, has been the Impact of Events scale (IES).¹⁵ Unlike the general psychiatric scales mentioned above, the IES poses survey items that are anchored to a particular life event (in this case, learning one’s genetic risk for disease) and measures distress responses specific to that stressor. This approach provides greater sensitivity in detecting psychological impact of testing, but the IES is not without its own limitations. The measure was not developed with genetic testing in mind; rather, it was designed to assess common distress reactions consistent with post-traumatic stress disorder (PTSD). As such, it includes subscales capturing both intrusive symptoms (e.g., nightmares about the stressful event) and avoidance symptoms (e.g., trying not to think about the event). This design is highly appropriate when measuring impact of severely traumatic events such as military combat, rape, and assault, but less so when applied to genetic susceptibility testing.

More recently, instruments specifically designed to gauge psychological effects of genetic susceptibility testing have been developed. Measures such as the Psychological Adaptation to Genetic Information Scale (PAGIS),¹⁶ the Multidimensional Impact of Cancer Risk Assessment (MICRA),¹⁷ and the FACToR scale¹⁸ have gone beyond merely measuring distress responses to testing. These scales capture additional domains including the following: feelings of uncertainty; privacy concerns; confidence in using test results; and psychological benefits (e.g., feelings of relief, empowerment). Such instruments have shown good reliability and validity in their use, and they address many of the limitations of measures mentioned above. Yet the field continues to struggle with inconsistent use of measures across studies. This is true not only for the outcome of psychological impact, but also for related domains including understanding of results, disease risk perceptions, and test expectations and decision-making.¹⁴ The lack of “gold standard” measures across these domains makes it difficult to compare findings across studies and to conduct pooled or meta-analyses that would be useful in summarizing knowledge gained beyond an individual study.

REVEAL: A case study in study design

Much of my own work in the field has been through the Risk Evaluation & Education for Alzheimer’s Disease (REVEAL; Robert Green, PI) study: a series of randomized, clinical trials (RCTs) assessing the impact of genetic susceptibility testing on asymptomatic individuals at-risk for Alzheimer’s disease.¹⁹ The study has focused on the disclosure of individuals’ APOE genotype status, as certain variants of this gene have been shown across hundreds of studies to be associated with elevated risk of Alzheimer’s disease.²⁰ Individuals with the ε4 variant in their APOE genotype (who represent approximately a quarter of the general population) are at significantly increased risk for AD, but this risk allele is neither necessary nor sufficient to cause the disease. These limitations in the predictive value of testing, coupled with a lack of proven options to reduce AD risk, prompted several consensus statements to conclude that the use of APOE for predictive testing purposes was generally inappropriate [e.g., Post et al, 1997].²¹ Although such conclusions seemed reasonable, they were made with a striking lack of empirical data to guide practice. Parties on both sides of the debate made speculative claims about the potential benefits and harms of risk disclosure in this context, but until the REVEAL Study no empirical studies had directly assessed the psychological impact of APOE testing.

Given the focus of this paper, my emphasis in describing the REVEAL Study here will be more on research methodology than on its specific findings. And in fact, many of the aforementioned methodological critiques could be applied to REVEAL. Our recruitment strategies have resulted in many of the same sample selection biases mentioned above, and despite some success in enhancing involvement of African American participants,²² our study populations have been disproportionately white and well-educated. In addition, our primary outcomes have typically been general measures of depression and anxiety [e.g., Green et al, 2009],²³ as opposed to more sensitive measures of test-specific impact. Nevertheless, our experience in developing and implementing four successive, multisite trials does provide some potentially useful lessons for the field.

First, each of our studies has used an RCT design. A major limitation of the field to date has been a lack of well-controlled studies. Most examinations of the psychological impact of genetic testing have been through observational studies without a comparison group, and some of these have even lacked pre-test data against which to compare post-test psychological states. Our use of an RCT design has allowed us a greater opportunity to isolate the impact of genetic risk disclosure from other potential influences on psychological functioning. Second, we have assessed psychological responses to testing across several different methods of genetic risk disclosure. Genetic services have traditionally been provided in a relatively small number of specialty clinics, with time-intensive case preparation and counseling procedures. This model of care offers only limited access to services for the growing number of individuals who will need them, suggesting a need for more condensed protocols and greater use of telemedicine approaches.²⁴ Thus, in the second REVEAL trial, we examined the impact of a more clinically feasible genetic risk communication protocol for first-degree relatives. The extended protocol from our first trial was compared against a condensed protocol where an in-person pre-test genetic counseling session was replaced with provision of an educational brochure; this modification resulted in fewer sessions and 57% less face-to-face time with the study clinician.²⁵ Our results suggested that genotype disclosure via the condensed protocol was generally as safe and effective as the original extended protocol, as participants did not differ by protocol in terms of post-test depression, anxiety, or risk recall and comprehension.²⁶ We continued in our third trial to examine differences between methods of service delivery in genetic counseling and testing, again with an eye toward approaches that could expand the reach of services. We found that a telephone disclosure protocol for communicating APOE test results achieved comparable results to an in-person disclosure protocol in terms of psychological distress.²⁷

We have also paid significant attention to the selection and creation of measures designed to assess psychological impact of testing. The results from our initial trial were informative in determining whether genetic risk information resulted in clinically significant depression and anxiety symptoms; however, available measures lacked the sensitivity to detect subtler psychological responses specific to genetic testing for a late-onset condition. We therefore developed a new measure called the Impact of Genetic Testing for Alzheimer’s Disease (IGT-AD) scale. This 16-item scale was found to have excellent internal reliability, and construct validity was established by observing correlation with other standardized self-reported psychological measures.²⁸ This measure can assist in the identification of patients susceptible to the negative psychological effects of genetic testing for AD, as well as the monitoring of patients who have already received such information. To address limitations of survey measures, another approach we have taken is to conduct or facilitate in-depth qualitative interviews to supplement our quantitative findings. For example, the medical anthropologist Margaret Lock led an ancillary study examining how REVEAL participants made sense of their test results, and how these findings were integrated (or not) into their existing mental models of dementia risk based on their family experience of the disease.²⁹

Finally, over time we have expanded our measures of psychological impact beyond a narrow focus on psychological distress. Knowledge of biomarker status for a feared condition like Alzheimer’s disease has the potential to change how individuals feel about themselves and how they believe others may be judging them. We are thus now drawing upon the literature on disease-related stigma to inform study assessments.³⁰ Genetic testing for AD also has the potential to alter individuals’ perceptions of their future (e.g., how many “good years” they have left). Indeed, the gerontology literature suggests that differences in future time perspective may influence key life decisions and prioritization of social goals, with well-validated scales available to measure this construct.³¹ Finally, preliminary work outside of REVEAL suggests that knowledge of APOE status could bias how individuals rate their own memory functioning and perform on neuropsychological tests.³² We are currently exploring whether this effect, potentially explained by the stereotype threat phenomenon observed in other test-taking contexts, is influential among individuals learning their AD risk status.

Conclusions and directions for future research

In this paper I have highlighted many needs for future research on the psychological impact of genetic susceptibility testing. In terms of research methods, there is a strong need for prospective, controlled, theoretically informed designs. Several health behavior and health services utilization frameworks could inform studies in this area, but these have generally been underutilized. Such frameworks include the adaptation of stress and coping theory for a genetic testing context³³ and family systems models specifically focused on genetic disease, such as the Family Systems Genetic Illness model.³⁴ This latter model could help guide assessments that go beyond the level of the individual; despite the clear implications of genetic susceptibility test results for family members, relatively few studies have directly involved them as research participants. There also clearly needs to be greater attention to ensuring diversity in study populations, especially with regard to race, ethnicity, and socioeconomic status. It is unclear whether and to what extent findings from studies on predominantly white, well-educated participants extend to the population at large. Greater diversity is not only needed in studies of the impact of genetic testing, but also in the clinical and genetic epidemiology studies on which our disease risk estimates are based.³⁵ Such efforts could assist in the development of more appropriate and precise genetic risk disclosure techniques.

As noted above, another challenge in the field is harmonizing measures used across studies. Key constructs in the domain of psychological impact have been operationalized in different ways, measured inconsistently across studies by a wide range of instruments. Greater consensus on use of measures would allow for better comparison of findings and pooling of data for meta-analyses. A positive recent development in this area has been work within the federally funded Clinical Sequencing Exploratory Research (CSER) consortium, which has sponsored multiple major projects across the US examining the impact of clinical use of whole genome sequencing in contexts including oncology, pediatrics, cardiology, and primary care.³⁶ An Outcomes and Measures workgroup within CSER has developed a framework to guide research in this area, highlighting not only psychological impact of sequencing, but related domains including preferences for disclosure of sequencing findings, understanding of test results, behavioral impact, health care utilization, and decisional satisfaction / regret.³⁷ Assessments of these domains should consider mixed-methods approaches incorporating qualitative interviews that examine individual differences in how people make meaning of genetic risk information explore contextual factors (e.g., cultural influences, lived experience of family history of disease) that affect individuals’ psychological response to testing.

Finally, research on the psychological impact of genetic susceptibility testing should keep pace with new developments in how genetic information is being offered. The rapid emergence of direct-to-consumer offerings of genetic susceptibility testing poses many important questions for psychosocial research.³⁸ This model of testing generally does not include either pre-test or post-test education and counseling with a genetic counselor or other expert clinician. The potential for being “blindsided” by unexpected test results therefore seems greater in this context, particularly as consumers are beginning to submit their raw genomic data to third-party websites for interpretation.³⁹ In addition, genetic susceptibility testing will increasingly be combined with other biomarker information to identify high-risk populations. Returning to the example of Alzheimer’s disease, new amyloid neuroimaging technologies now make it possible to detect amyloid plaques in the brain more than a decade before any clinical dementia symptoms are manifest. The Food & Drug Administration (FDA) has approved the use of amyloid imaging in symptomatic patients, and the technique is being used—along with genetic susceptibility testing—to identify asymptomatic at-risk individuals for prevention drug trials.⁴⁰ Studies will therefore be needed to better understand the differential and additive psychological effects of disclosing genetic vs. biomarker information to at-risk populations.

In conclusion, research on the psychological impact of genetic susceptibility testing has made significant strides over the past few decades, but many areas for improvement remain. Such work will become increasingly important given current trends of expanded use of testing not only within clinical medicine, but also consumer and research contexts. More people will be asking for their own personal genetic information, and even some expert groups have suggested that greater access to genetic information should be built into the research enterprise.⁴¹ Better understanding of the psychological impact of genetic testing will help us decide when such access is appropriate, and how best to disclose results in a manner that supports adjustment to test findings and promotes use of genetic information to improve human health.