Personalized & Precision Medicine: September 2018

Can Learning Genomic Risk Really Affect Behavior?

Tara Haelle

August 16, 2018 (Medscape Medical News) – For years, researchers have explored whether telling patients about their genetic risk for chronic disease might inspire those patients to reduce their risk through lifestyle changes, such as eating more healthfully, quitting smoking, or increasing their physical activity.

But the evidence hasn’t panned out: two systematic reviews, published in 2010 and 2016, found no evidence that giving people DNA­ based risk estimates for chronic disease changes behavior, and more recent research has shown the same.

Now, however, preliminary findings from a large Finnish study challenge the status quo — and the intervention is just different enough from past ones that this research may have “the potential to kick off the new generation of research and understanding on when genomic information can influence health behavior,” according to Susan Persky, PhD, head of the Immersive Virtual Environment Test Unit at the National Human Genome Research Institute in Bethesda, Maryland.

The study found that high percentages of participants had lost weight and quit smoking a year and a half after learning their risk for ischemic heart disease, based on both traditional and genomic risk factors.

“It is encouraging that any sort of personalized intervention is achieving these results,” Persky, who was not involved in the study, told Medscape Medical News. “Behavior is incredibly difficult to change over the long­term, using most sorts of interventions.”

But the findings, presented at the European Society of Human Genetics meeting in Milan, Italy, in June, are preliminary, and the study itself has substantial limitations, including no control group. Still, the study suggests a new way forward in exploring the complexities of human behavioral change.

“Our study shows it is possible to motivate and support individuals to successfully reduce their disease risk and achieve sustained lifestyle change by applying a web­based tool to communicate and interpret complex disease risk data,” lead author Elisabeth Widen, MD, a senior scientist at the University of Helsinki Institute for Molecular Medicine in Finland, told Medscape Medical News.

GeneRISK is a prospective longitudinal study in which 7328 southern Finnish participants, ages 45 to 65 years at baseline, learned their composite 10­year risk score for cardiovascular disease from the web­based tool KardioKompassi. The tool analyzed their traditional anthropomorphic and lifestyle risk factors, along with their polygenic risk, based on 49,000 single nucleotide polymorphisms.

The researchers focused on cardiovascular disease for several reasons, Widen told Medscape Medical News: it’s common and severe, genetic factors explain half its risk variation in a population, and polygenic risk scores have been shown to improve risk prediction.

“Moreover, there are efficient means to reduce the risk, both through lifestyle changes and medical intervention, but nonetheless, the clinical use of polygenic risk information for disease prevention has remained minimal so far,” Widen said.

The score’s traditional factors included age, sex, cholesterol levels, smoking status, and blood pressure. When participants’ scores were ready, they received a text message to log onto the web portal, where they could manage their information. KardioKompassi provided their results and advised patients with at least a 10% increased risk for heart disease to speak to their doctor about ways to reduce risk.

“KardioKompassi is interactive and displays personal disease risk in multiple different ways, showing the individual risk factors and the combined personal disease risk,” Widen said.

Dramatic Lifestyle Changes 18 Months Later

At baseline, a quarter of the study population had increased cardiovascular disease risk, 40% of whom smoked and a third of whom were obese (body mass index > 30 kg/m2). Only 17% were taking medication to manage cholesterol. In the overall cohort, 20% of the men and 15% of the women smoked.

The follow­up data the researchers presented came from 3278 participants assessed 18 months later. Overall, 15% of average­risk smokers and 17% of high­risk smokers had quit smoking compared with 4% of smokers in the general population. Further, 12% of average­risk participants and 15% of high­risk participants had lost weight and kept it off.

More than a third (36.4%) of the high­risk participants had taken action to reduce their risk by losing weight, quitting smoking, and/or visiting their physician. Among those with lower risk, 20.5% took action.

The researchers then compared high­risk participants who took action vs those who didn’t. “We found there wasn’t any big difference in their risk profiles regarding age, sex, [body mass index], or smoking status,” Widen told Medscape Medical News. “However, we did find that individuals who had undertaken action to lower their disease risk had a higher polygenic load than the others.”

Participants’ questionnaires revealed that 90% believed the risk information was useful and easy to interpret.

“The participants also said that receiving personal risk information encouraged them to take better care of their health, and a majority believed that clinical doctors are capable of interpreting and using genomic information in their work,” Widen continued.

The extent to which participants had made lifestyle changes surprised the researchers, as did the participants’ particular attention to their genomic data. Widen speculated that the genetic profile knowledge alone may have been particularly motivating or that the information’s novelty played a role in lifestyle change decisions.

“Many of the study participants were aware that they had elevated cholesterol levels from before, whereas this was the first time their genetic risk profile was measured,” Widen said.

Other Behavior Change Factors to Consider

But the research base suggests other possibilities that may explain participants’ behavior change, said Joyanna Hansen, PhD, RD, an assistant professor in the Oregon Health & Science University School of Medicine’s Department of Medical & Molecular Genetics in Portland.

“There’s a lot of research showing that a personalized intervention in general can be helpful in inspiring people to make lifestyle changes,” Hansen told Medscape Medical News. The more personalized an intervention is, the more likely it is to motivate change, past research suggests, which highlights this study’s weakness in not having a control group. Hansen would like to have seen a three­group study, with two groups getting personalized information — one with genomic data and one without — and one group receiving only generic risk information.

“I would like to really understand if it’s the genetic aspect or the personalization,” she said. She also pointed to a limitation of existing research that is potentially relevant here: selection bias.

“Most of the other research that I’ve read [concerns] consumers who have opted into direct­to­consumer genetic testing and those who opt into research, and that’s a very self­selective group,” Hansen said. “We really need more studies looking at a wider range of ethnicities and socioeconomic groups.”

This new study may suffer the same limitation, according to Sarah Knerr, PhD, an acting assistant professor of health services at the University of Washington School of Public Health in Seattle.

“Knowing the way study recruitment works, especially with really healthy, motivated people, there’s the likelihood that they got a bunch of a motivated people who think this is cool and are more likely to be persuaded to do these things anyway,” Knerr told Medscape Medical News. “Without a control group, you can’t tell.” She said study recruitment likely brought in “super early adopter people who are not the people the doctor is really trying to get to change.”

Or, she pointed out, KardioKompassi may be a much better designed tool, in terms of its formatting, modality, and interface, than the clunkier ones often used in research.

“[T]he degree that they actually put some thought and effort into making this user­friendly could also have had an impact,” she said. Echoing Hansen, Knerr also speculated that the degree to which the intervention was tailored to each person — independent of including genomic data — may be more influential.

Genomic Data as the Next Frontier?

Widen acknowledged that the current evidence base shows genetic risk estimates not having a significant effect on smoking cessation, diet, or physical activity.

“However, in most of these studies, the genetic tests used are outdated, since they typically were based on only one few genetic variants,” she said. In contrast, her team relied on 49,000 variants and presented the results in an interactive web­based tool.

And, as Knerr hinted, delivery method does appear to matter, explained Rebecca Ferrer, PhD, a health scientist and program director of the National Cancer Institute’s Division of Cancer Control and Population Sciences in Bethesda, Maryland. She said the literature overall is mixed, but that the medium can be more influential than the message.

“As with other types of risk communication, the framing and creation of the message matters as much or more than the content,” Ferrer told Medscape Medical News.

“Some research suggests that people tend to be defensive against information suggesting that they volitionally engage in behaviors that may cause them harm because this is a threat to self­competence,” she said. “Other research suggests people tend to be more sensitive to information about risks, harm, or loss than information about benefits and gains. So, risk communication researchers think really carefully about how to craft messages about health risk, including messages about genomic risk for disease.”

The use of an overall genomic score also makes this study fairly unique in the field since major literature reviews have focused on narrower genetic data, Persky said.

“That said, many researchers have long been pointing out that the level of risk genetic information can convey is very small,” Persky told Medscape Medical News. “Indeed, we shouldn’t really expect people to change their behavior when their risk for disease increases by 1 percentage point due to a risk­associated genetic variant.”

That’s why this study may represent the next phase of research in this area, moving from single nucleotide polymorphism genetic risks to broader genomic risk, Persky said.

“For me, it would be unsurprising to see the outcomes shift such that risk information derived from genomics (including nongenetic elements) is more influential than what we’ve seen in the past,” Persky said. “Many of us have had the expectation that when we become better able to explain a larger proportion of individuals’ risk through genomic testing, we may start to see different outcomes with respect to behavior change.”

Social and Cultural Factors Play a Role Too

At the same time, this study may be most useful for hypothesis generation, especially given that its homogenous population may limit generalizability. The findings may apply to populations with similar characteristics, such as those with good access to primary care, Persky said, but smoking cessation rates, for example, may not be as dramatic in populations with a lower baseline smoking rate.

It’s also unclear whether cultural factors in general may mediate behavior change, but it’s possible, said Richard L. Street Jr, PhD, a health communications professor at Texas A&M University in College Station.

He and Persky collaborated on research exploring how incorporating genomic information into discussions about body weight influences women’s perceptions and beliefs. They expected it could go two ways: patients might be more forgiving with themselves but more willing to make healthy changes, or they could become more fatalistic, not bothering with changes because they figured there was nothing they could do. Fortunately, results pointed to the former.

“People who got a genetic explanation perceived less blame and less weight stigma than people getting a behavioral explanation,” Street told Medscape Medical News. The genomic information did not affect participants’ health behavior­related attitudes and beliefs, but interestingly, the effects were not moderated by physicians’ supportive vs direct communication styles — suggesting some leeway in delivery methods.

But now, Street said, they are working on a follow­up study looking at differences in effects between white women and black women in the study. They are starting to see some variations between the two groups, including whether and how communication style mediates effects. What they ultimately find will be relevant to future studies testing a sophisticated tool such as KardioKompassi.

Street believes the bigger question about using risk calculation tools relates to their effect on self­perception and self­efficacy.

“I think the important part of this is not the genetic information itself, because quite frankly, when we start talking about risk factors, it’s a moving target,” Street told Medscape Medical News. He speculates that the perception of a “genetic anchor” may give a subgroup of people a reason to be proactive behaviorally.

“Let’s suppose having the genetic explanation had some worrying less that they failed and realizing it’s not their fault,” Street said. “Maybe that is empowering in a way that lets them feel better about themselves so they can go do the things they can do. Instead of focusing on, ‘I can’t run 6 miles,’ they may shift to, ‘I can get 1 mile in today.’ ”

In other words, perhaps alleviating self­blame in patients is key to motivating them. But with multiple, uncontrolled parts to the intervention in the Finnish study, it’s not possible to parse out all the different effects, he said.

“The part that’s novel is the genomic part, but we can’t identify for sure if that’s what’s leading to the change,” Street said. He also wondered whether using the tool as a stand­alone tool at home vs during a clinical encounter makes a difference. But regardless, he said, the length of time the participants maintained their lifestyle changes was remarkable.

“Now [the change is] so routine it’s become a part of their everyday practices,” Street said. “The best predictor of future behavior is current behavior.”

But there’s more to it than an individual person making changes, explained William McCarthy, PhD, an adjunct professor of health policy and management at the University of California, Los Angeles, Fielding School of Public Health. McCarthy researches tobacco cessation, where public health has had great success, but that success involved multiple factors. Lifestyle change never occurs in a vacuum and doesn’t rely solely on knowledge, McCarthy told Medscape Medical News.

“We have tons of research showing that knowledge is good, but knowledge is not enough,” McCarthy said. “It takes a lot more than a single piece of information to adopt a lifelong pattern of healthier living.”

Educational attainment has been shown to correlate with adherence, he said, and social support matters at least as much as a person’s own motivation.

“The extent to which genetic information has a long­standing impact [on lifestyle] would be mediated by relationships with coworkers, loved ones, and other significant people in their life,” McCarthy said. “You’ll take steps in the short run, but the immediate rewards of going along with your social support group is likely to overwhelm any cognition that you may be at a higher risk. But if you shared with others the information that you’re really at risk and really need their help, then you can turn that around.”

He pointed out that having a heart attack is one of the best predictors of permanently quitting smoking — but not entirely because of the person having a heart attack: It’s mostly others’ reactions to it.

“Everybody now sees you with a scarlet letter,” McCarthy said. “No one is going to allow you to smoke anymore because you’ve gone through this major cardiac event, and they don’t want you to go through that again.”

Regardless of what Widen’s final analysis and future research shows, the broader discussion of genomic risk and behavior change underscores a much bigger shift occurring in medicine at large — the effect of personalization and patient self­efficacy.

“A general overall trend in medicine is that the individual’s role is increasing; that is, the development towards participatory medicine where individuals are empowered to play a more active role,” Widen said. “Because disease prevention in particular is dependent on the individual or patient being motivated and committed, it is quite important that we develop methods and procedures to support them.”

Widen reports receiving speaker fees from Amgen. Hansen reports having received a one­time donation for educational purposes from 23andMe. The other interviewees have disclosed no relevant financial relationships.

Medscape Medical News © 2018 WebMD, LLC