Returning Individual Genetic Research Results to Research Participants: Uptake and Outcomes Among Patients With Breast Cancer

Abstract

Purpose

Understanding the outcomes of returning individual genetic research results to participants is critical because some genetic variants are found to be associated with health outcomes and have become available for clinical testing.

Materials and Methods

BRCA1/2-negative women with early-onset breast cancer, multiple primary cancers, or a family history of breast cancer who participated in a gene discovery cancer registry were offered the opportunity to learn their individual genetic research results of 24 breast cancer susceptibility genes with a genetic counselor after predisclosure genetic counseling. Outcomes included uptake of research results, knowledge, informed choice, psychosocial adjustment, uncertainty, satisfaction, and uptake of clinical confirmation testing.

Results

Four hundred two potential participants were contacted. One hundred ninety-four participants (48%) did not respond despite multiple attempts, and 85 participants (21%) actively or passively declined. One hundred seven participants (27%) elected for predisclosure counseling and were more likely to be younger, married, and white. Ninety percent of participants who had predisclosure counseling elected to receive their genetic research results, and 89% made an informed choice. Knowledge increased significantly after predisclosure counseling, and anxiety, intrusive cancer-specific distress, uncertainty, and depression declined significantly after receipt of results. General anxiety and intrusive cancer-specific distress declined significantly for both participants with a positive result and those with a negative result. Sixty-four percent of participants had clinical confirmation testing when recommended, including all participants with a mutation in a high-penetrance gene.

Conclusion

Uptake of genetic research results may be lower than anticipated by hypothetical reports and small select studies. Participants who elected to receive research results with genetic providers did not experience increases in distress or uncertainty, but not all patients return for confirmation testing.

INTRODUCTION

As large studies with banked DNA are increasingly used in genetic sequencing studies, there has been increasing debate over the obligation to share individual research results with research participants.^1-7 Over time, genetic variants identified in research may be found to be associated with health outcomes and may become available in clinical practice.^8,9 Arguments against returning research results include a blurring of the distinction between research and clinical care, misunderstanding, the right not to know, and costs.^3,7,10-17 Others argue that results should be returned based on the principles of beneficence, autonomy, reciprocity, and respect for persons.^6,18-25 There remains no consensus on how and which results should be returned.^1-7,26-28 Numerous studies have reported high interest in receiving research results, even if the patients knew there was nothing they could do about the results.^29-35 Although some argue that only actionable results from a laboratory certified by the Clinical Laboratory Improvement Amendments (CLIA) should be returned,^3,36,37 others argue that research results should be offered to patients more broadly, allowing patients to make decisions about which results they would like to receive, with recommendations for CLIA confirmation for results that could change clinical care.^{6,20,21,38-41}

In the few studies reporting uptake of research results (as opposed to hypothetical interest), uptake varied considerably (17% to > 80%) in cohorts with cancer.^42-48 Most studies returned results with recommendations to have confirmatory CLIA testing, but none report on uptake of confirmatory testing,^42-44,48,49 and the majority of these studies were small, with select populations and limited outcomes.^42-45,50-52

In the Returning Genetic Research Panel Results for Breast Cancer Susceptibility (RESPECT) study, we sought to evaluate uptake of research results in patients with breast cancer who participated in the research registry, factors associated with uptake of results, and participant cognitive and affective outcomes. Equally important, how frequently patients obtain confirmatory testing is critical to understanding the risks and benefits of returning research results. Thus, we sought to evaluate how frequently participants obtained clinical confirmation testing, the concordance of clinical testing with research results, and the impact on medical management.

MATERIALS AND METHODS

Participants

Potential participants were individuals who previously consented to a research registry at the University of Pennsylvania (Philadelphia, PA). The original informed consent stated that results would not be returned unless they would significantly affect cancer risk. Five hundred two samples from BRCA1/2-negative women with breast cancer and with a diagnosis at < 40 years of age, multiple primary cancers, or at least three first- or second-degree relatives with breast cancer were selected for sequencing of 24 breast cancer susceptibility genes (CDH1, CDKN2A, MLH2, MSH2, EPCAM, MSH6, MUTYH homozygous, PMS2, PTEN, STK11, TP53, ATM, BAP1, BARD1, BMPR1A, BRIP1, CHEK2, MRE11A, MUTYH heterozygous, NBN, PALB2, RAD50, RAD51C, and RAD51D) for related research.⁵³ Sequencing was by either targeted or whole-exome hybrid capture methodologies, as described.^53-55 All potential participants were checked against a death index and were confirmed to have permission for recontact. As shown in Figure 1, 402 women from 378 families were eligible for contact.

Fig 1.

Study flow diagram. (*) Seven patients were men, 12 had clinical multigene panel testing (MGPT), five had return of results (ROR) before Returning Genetic Research Panel Results for Breast Cancer Susceptibility (RESPECT) study, and research charts were not available for 11. (†) Eight patients died, six already had clinical MGPT, and two were disclosed off-study. (‡) Reasons for declining included concerns about research or time burdens, not being interested in genetic information, only wanting actionable results, preferring clinical testing, and concerns about uncertainty or distress. (§) Predisclosure sessions lasted, on average, 51 minutes (range, 16 to 84 minutes). (‖) Three participants reported concerns about distress, three reported concerns about uncertainty, three reported a preference for clinical testing, three reported that they were not ready, and one was not ready at the time because of relative in hospice. Some participants provided more than one reason for declining. (¶) Disclosure sessions were, on average, 20 minutes (range, 6 to 66 minutes). (#) Three high-penetrance genes (MSH2, MSH6, and TP53) and 10 moderate-penetrance genes (ATM, n = 3; CHEK2, n = 3; MUTYH, n = 2; and RAD51D, n = 2). T0, baseline; T1, after predisclosure counseling; T2, after disclosure counseling; VUS, variant of uncertain significance.

Institutional review board approval was obtained. Potential participants were contacted between May 2014 and October 2015. Study invitation letters explained that research testing had been completed and that they could enroll onto a study evaluating the outcomes of returning research results. The letter explained that everyone whose research sample had been tested was being invited to participate in the study (ie, that the invitation was not an indication of a positive result). This was the only method provided to participants to receive their research results.

Predisclosure and Result Disclosure Genetic Counseling

Predisclosure genetic counseling sessions using the tiered-binned model were conducted by genetic counselors (n = 5) via phone or in person.^56,57 Genetic counselors were blinded to the results at predisclosure and completed counseling checklists for all sessions. Participants received results from a genetic counselor by phone or in person. Genetic counselors shared the need for clinical confirmation when applicable and any clinical genetic testing consistent with current clinical practice.

Confirmatory Testing in a CLIA-Certified Laboratory

Participants with a deleterious or likely deleterious research result in any gene and a variant of uncertain significance (VUS) in a high-penetrance gene were recommended to have clinical confirmation of the research finding. Genetic counselors assisted participants with clinical confirmation testing, which was provided through a single CLIA-certified clinical laboratory. Research funds covered any testing costs not covered by insurance.

Measures

We used a novel conceptual model⁵⁸ grounded in the Self-Regulation Theory of Health Behavior to inform outcomes and hypotheses.^59,60 Our model proposes that uptake of research results and response to (eg, psychosocial adjustment) and use of (performance of risk reduction behaviors) result disclosure are products of an individual’s understanding and perception of the disease threat.^58,60 Participants completed surveys at baseline (T0) and after predisclosure counseling (T1) and disclosure counseling (T2).

Knowledge of genetic disease was evaluated (at T0, T1, and T2) using an adapted version of the Cancer Genetics Knowledge scale and ClinSeq knowledge scale,^61-63 including knowledge of inheritance and test interpretation (eight items), the benefits (three items) and limitations (six items) of multigene testing, and the differences between research and clinical testing (five items; Cronbach’s α = .78 to .84).

Psychosocial adjustment was evaluated (at T0, T1, and T2) using three measures. State anxiety was measured using the 20-item State Inventory of the State-Trait Anxiety Inventory^64,65 (Cronbach’s α = .95 to .97). General anxiety and depression were assessed using the seven-item Hospital Anxiety and Depression Scale⁶⁶ (Cronbach’s α = .85 to .90 for anxiety and .79 to .88 for depression). Cancer-specific distress was evaluated using the 15-item Impact of Events Scale⁶⁷ (Cronbach’s α = .81 to .93).

Satisfaction with genetic services was measured (at T1 and T2) with a 16-item scale used in related research and adapted for return of research results^68,69 (Cronbach’s α = .81 to .82). Uncertainty was assessed (at T0, T1, and T2) using a three-item scale adapted from the Multidimensional Impact of Cancer Risk Assessment Question⁷⁰ (Cronbach’s α = .84 to .89).

Perceived utility was assessed (at T1 and T2) with a novel scale developed to evaluate patient perceptions of the utility of genetic results, including two 12-item subscales evaluating medical and personal utility now and in the future (Cronbach’s α = .94). Attitudes about genetic testing were assessed at T1 using an eight-item scale evaluating attitudes about testing (Cronbach’s α = .76).^71,72

Statistical Analyses

On the basis of prior small studies reporting high potential interest in receiving research results, we hypothesized that 70% of invited participants would agree to receive research results. We selected a sample size to have a narrow 95% CI in estimating the uptake percentage. With at least 350 participants contacted and 70% agreeing to receive their results, we estimated that the CI width would be < 10%. As a result of symmetry, an uptake rate of 30% would also give similar width.

We used multiple logistic regression models to evaluate baseline characteristics associated with uptake. We used forward stepwise regression to increase power but still control for confounders. We adjusted for individuals within the same family using cluster-corrected standard errors.⁷³ In regression models, we excluded 26 of 372 participants who had missing data for continuous variables, race, or proband status.

We used paired t tests for psychosocial responses between the first and second survey rounds. For longitudinal analyses across the three waves, we used multiple linear regressions estimated by generalized estimating equations with robust SEs to account for within-individual temporal correlation. Because only seven pairs of individuals within a same family had longitudinal data (7 × 2 = 14 individuals), we did not account for familial correlation in longitudinal analyses. We assumed exchangeable correlation matrices for generalized estimating equations.

We did not find significant differences when comparing baseline response means between those who completed the first two surveys and those who completed all three surveys (P > .28 in all cases). Similarly, we did not find significant differences in the second set of survey responses when comparing the same two groups (P > .075 in all cases). Given the lack of differences, we determined that missing data as a result of loss to follow-up was likely not a major confounder. Thus, we performed complete case analyses. P < .05 (two-sided) was considered statistically significant.

Informed decision making was defined using the multidimensional measure of informed choice (MMIC), which incorporates three dimensions of choice, including knowledge, individual values and attitudes toward choice options, and the extent that choice reflects attitude.^72,74 Consistent with prior applications of the MMIC, adequate knowledge was defined as more than half of the items correct with an average item score > 3.5 on a 5-point Likert scale. A positive attitude about genetic testing was defined by a sum score at or greater than the midpoint of the scale.^71,72 A more informed choice is defined by adequate knowledge and a testing decision concordant with attitudes about genetic testing. All other categories are considered less informed.

RESULTS

Participant Characteristics

All of the 402 potential participants contacted had a history of breast cancer; the mean age at diagnosis was 40.2 years old (Table 1). All participants had previously tested negative for BRCA1/2 mutations.

Table 1.

Characteristics of All Potential Research Participants Contacted and by Uptake Outcome

Uptake of Genetic Research Results

Despite multiple mailings and phone calls, we were not able to directly speak with 196 potential participants (49%). For only 14 potential participants was the mailing returned to sender (Fig 1). In a multiple logistic regression analysis, nonresponders were more likely to be older at the time of their first cancer diagnosis, to not have been seen in the clinical program (eg, participated in research only), and to be nonwhite (Table 2). Twenty-three potential participants (6%) actively declined participation. Sixty-two potential participants (15%) spoke with research staff but did not respond to follow-up inquiries. In a multiple logistic regression analysis, those who actively declined participation (eg, communicated that they did not want to participate) or passively declined participation (eg, never responded after multiple contacts) after learning about the study were more likely to have been younger age at cancer diagnosis and to have had a greater number of relatives with cancer (Table 2). One hundred seven participants (27%) consented to the study. In a multiple logistic regression analysis, those who consented were more likely to be younger at the time of contact, married, and white (Table 2). Time since sample donation was not associated with actively or passively declining participation in multiple logistic regression analysis. Nine participants who had predisclosure counseling chose not to receive their results.

Table 2.

Factors Associated With Uptake or Denial of Individual Genetic Research Results in Multivariable Models

Informed Decision Making

Eighty-nine percent of participants (74 of 83 participants) who had predisclosure counseling and completed relevant measures were classified as making an informed choice as defined by the MMIC (Table 3). Making an informed choice was not associated with any demographic characteristics but was associated with a 2.0-point change-score increase in research knowledge (t test, P = .048) and marginally statistically significant 3.0-point change score in total knowledge (t test, P = .095) compared with participants who did not make an informed choice.

Table 3.

Frequency of Making an Informed Choice Using the Tiered-Binned Predisclosure Genetic Counseling Model

Patient-Reported Outcomes With Receipt of Genetic Research Results

Knowledge increased significantly after predisclosure counseling and receipt of results (Table 4). General anxiety and uncertainty declined significantly after predisclosure counseling. State anxiety, intrusive cancer-specific distress, uncertainty, general anxiety, depression, and perceived utility declined significantly after receipt of results. In exploratory analyses, we evaluated difference in outcomes by test result (Appendix Table A1). Knowledge increased significantly among those receiving a negative or VUS result. Anxiety declined significantly for participants with a positive and negative result, but cancer worry only declined significantly for those with a negative result.

Table 4.

Patient-Reported Outcomes With Receipt of Genetic Research Results

Clinical Confirmation Testing and Changes in Medical Management

Sixty-four percent of participants (14 of 22 participants) had the recommended clinical confirmation testing (Fig 2), and all recommended confirmation testing was covered by insurance, although one patient had a small out-of-pocket cost. Clinical testing (14 samples with variants where confirmation testing was recommended and three samples where participants elected to have clinical testing to confirm a VUS in a moderate-penetrance gene) identified 16 of 17 variants identified by the research laboratory (94%; Appendix Table A2). The discordant MSH2 mutation not found on clinical testing was a large genomic rearrangement identified by the research bioinformatic pipeline in place at the time. Research-based confirmatory testing was not performed, and clinical confirmatory testing was used for patient management. The variant interpretation was concordant for 12 (75%) of 16 of the research-identified variants. For two samples, the research classification was VUS and the clinical result by the commercial laboratory was benign, although other clinical laboratories also classify the variant as VUS (thus, discordant when compared with the call by the specific clinical laboratory used but not discordant with results in ClinVar).⁷⁵ For two samples, the research classification was likely deleterious, and the clinical result was reported as a VUS. Finally, clinical testing identified three additional variants that were not reported on research testing because they were considered likely benign by the research laboratory.

Fig 2.

Percentage of participants who completed recommended clinical confirmation testing (n = 22). D, deleterious; LD, likely deleterious; VUS, variant of uncertain significance.

DISCUSSION

To our knowledge, this is the largest study offering a range of research genetic cancer susceptibility results (eg, actionable mutations as well as no identified mutations, genes of variable clinical utility) to research participants and collecting comprehensive outcomes as well as uptake of confirmatory testing in a commercial CLIA laboratory. Although hypothetical query and some published studies have suggested high interest in receiving genetic research results,^29,30,48,76 in this study, we contacted > 400 research participants and only a small subset elected to receive research results. Nonetheless, patient-reported outcomes for those who elected to receive results were favorable, with no significant increase in distress or uncertainty, even among subgroups receiving a positive or VUS result. We found increases in understanding of the limitations of research testing and that the majority of participants made informed decisions about receiving individual genetic research results.

Of note, in our study, we were not able to reach 50% of research participants. Although some studies have reported higher contact rates,^43,48 our rate is consistent with that in the Kathleen Cuningham Foundation Consortium for Research Into Familial Breast Cancer cohort, where 62% of participants did not acknowledge receipt of an invitation to receive results.⁴⁶ These data highlight that it may be more difficult than anticipated to reach research participants in some cohorts. It is possible that some participants did not understand what was being offered^77-79 or that they had already had updated clinical testing and therefore did not respond. Similar to at least one other study,⁷⁶ uptake was lower among nonwhite participants, raising important questions about the potential for disparities in uptake and highlighting the need to better understand how to clearly communicate the opportunity to receive research results among diverse populations.

In addition, 22% of research participants contacted actively or passively declined. These data are consistent with other studies reporting that some participants are not interested in receiving research results.^49,76,79,80 Until further data are available, researchers and institutional review boards considering return of results should anticipate that there will be a subset of participants who will not be interested in receiving results.

Although uptake was lower than anticipated in our study, the majority of participants who spoke with a genetic counselor elected to receive their research results, and most met criteria for making an informed choice with favorable patient-reported outcomes. Our findings are consistent with the favorable patient-reported outcomes reported in the Colon Cancer Family Registry,^47,48 and our study included additional knowledge measures, patients with a wide range of results, and greater sociodemographic diversity. Many studies have suggested that research participants and the lay public report interest in receiving genetic research results even when there is unclear utility or uncertainty.^29-32 Our findings support the premise that providing participants with a choice to receive a wide range of results may not be associated with negative psychosocial outcomes. This work was performed in a population of patients with breast cancer. Outcomes could be different when the research results are unrelated to the initial purpose of the study. In addition, we used a tiered-binned genetic counseling approach, and all patients spoke with a genetic counselor. Outcomes could differ with different approaches, and providing results with a genetic provider for all research participants will be challenge given the limited number of genetic counselors.⁸¹

Importantly, all participants with a deleterious or likely deleterious finding in a high-penetrance gene proceeded with confirmation testing, and for the most part, this testing was covered by insurance. Thus, for studies returning only actionable findings, uptake of confirmation testing may be high and cost may not be a barrier for insured patients. We did observe lower uptake of confirmation testing in other subgroups. Determining why participants did not proceed with confirmation testing will be important to understanding the risks of returning research results. However, returning only results obtained in a CLIA-certified research laboratory would deny some individuals critical medically actionable information that could be relevant to their health if subsequently confirmed in a clinical laboratory. For a small number of participants, the return of results and confirmation in a clinical laboratory did change their medical management and, even more frequently, could have medical implications for their relatives. Equally important, all 92 participants who had predisclosure counseling were candidates for clinical testing (eg, they were all reasonable candidates for clinical multigene panel testing), but only 32% proceeded with some clinical testing. Longitudinal data and further data will be needed to understand whether declining available clinical testing is based on informed preferences or potential misunderstandings or reliance on research results.

Although this study is, to our knowledge, the largest report of patient-reported outcomes in a sociodemographically diverse population receiving a wide range of genetic panel research results, we acknowledge several limitations. We do not know the reasons some participants never responded after receiving the study letter. Some may not have responded because they had already had testing. We expect this to be rare because panel testing had only recently become available (approximately 1 year) at the time we contacted participants for this study. It is possible that some never received our invitation. Our population of affected survivors of breast cancer may not be representative of other cohorts. In addition, participants who did not respond could have had higher baseline distress, and these results may not reflect outcomes for all individuals who are contacted to receive research results. The number of individuals who received positive and VUS results was small.

In conclusion, uptake of genetic research results may be lower than reported by hypothetical and small select studies. Participants who received results with genetic providers did not experience increases in distress or uncertainty, but not all patients return for confirmation testing. Understanding barriers to returning genetic research results and longitudinal outcomes will be important to informing policies regarding obligations to return genetic research results.

Appendix

Table A1.

Patient-Reported Outcomes With Receipt of Genetic Research Results by Test Result

Table A2.

Characteristics of Participants, Outcomes of Clinical Confirmation, and Other Clinical Testing and Changes to Medical Management for Positive Results

AUTHOR CONTRIBUTIONS

Conception and design: Angela R. Bradbury, Linda Patrick-Miller, Brian L. Egleston, Katherine L. Nathanson, Susan M. Domchek

Financial support: Angela R. Bradbury, Brian L. Egleston, Katherine L. Nathanson, Susan M. Domchek

Administrative support: Laura DiGiovanni

Provision of study material or patients: Kara N. Maxwell, Katherine L. Nathanson

Collection and assembly of data: Angela R. Bradbury, Kara N. Maxwell, Laura DiGiovanni, Jamie Brower, Dominique Fetzer, Jill Bennett Gaieski, Amanda Brandt, Danielle McKenna, Jessica Long, Jacquelyn Powers, Jill E. Stopfer, Katherine L. Nathanson, Susan M. Domchek

Data analysis and interpretation: Angela R. Bradbury, Linda Patrick-Miller, Brian L. Egleston, Kara N. Maxwell, Laura DiGiovanni, Jill Bennett Gaieski, Jessica Long, Katherine L. Nathanson, Susan M. Domchek

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Angela R. Bradbury

No relationship to disclose

Linda Patrick-Miller

No relationship to disclose

Brian L. Egleston

No relationship to disclose

Kara N. Maxwell

No relationship to disclose

Laura DiGiovanni

Employment: Carevive

Research Funding: Carevive (Inst)

Jamie Brower

No relationship to disclose

Dominique Fetzer

No relationship to disclose

Jill Bennett Gaieski

No relationship to disclose

Amanda Brandt

No relationship to disclose

Danielle McKenna

No relationship to disclose

Jessica Long

Employment: DePuy Synthes Companies (I)

Stock and Other Ownership Interests: DePuy Synthes Companies (I)

Travel, Accommodations, Expenses: DePuy Synthes Companies (I)

Jacquelyn Powers

Employment: Carevive Systems

Honoraria: CureConnect

Consulting or Advisory Role: Carevive Systems

Travel, Accommodations, Expenses: Hospital of the University of Pennsylvania

Jill E. Stopfer

No relationship to disclose

Katherine L. Nathanson

No relationship to disclose

Susan M. Domchek

Honoraria: AstraZeneca, Clovis Oncology

Research Funding: AstraZeneca (Inst), Clovis Oncology (Inst), Abbvie (Inst), Pharmamar (Inst)