Tai chi, Yoga ‘reprogram DNA’? Claims that mind-body practices improve health via epigenetics unfounded
David Gorski| Epigenetics Literacy Project | July 10, 2017

One of the most persistent narratives latched on to by advocates of “integrative medicine” is that the “mind” can somehow “heal” the body. Sometimes, the claim is that such interventions work through “powerful placebo” effects. Sometimes it involves the abuse of emerging science, such overblown claims about what can be accomplished through epigenetic modifications of DNA and gene expression. While my wife and I were away on vacation, there was another example of this claim popping up in the media under headlines like these:

An example of the reporting:

Meditation and tai chi don’t just calm the mind – they seem to affect our DNA too. There’s evidence that such “mind-body practices” dampen the activity of genes associated with inflammation – essentially reversing molecular damage caused by stress.

Mind-body practices such as mindfulness meditation are widely claimed to protect against stress-related diseases from arthritis to dementia. But although there’s plenty of evidence that they can relieve stress, the scientific case for physical health benefits has not yet been proven.

Recent advances mean it’s now easier to study patterns of gene activity inside cells, and there has been growing interest in using this approach to investigate how nurturing inner peace might influence the immune system and disease risk.

Ivana Buric, a psychologist at the Coventry University’s Brain, Belief and Behaviour lab, and her colleagues have now conducted the first systematic review of such studies. The team analysed 18 trials including 846 participants, ranging from a 2005 study of Qigong to a 2014 trial that tested whether tai chi influenced gene activity in people with insomnia.

You get the idea, and there’s more where that came from. (I picked that particular quote because Jo Marchant wrote it, and we’ve met her before.) The idea is that “mind-body” practices change our genes. Some reports even went so far as to imply that such practices can “reprogram” your DNA. Of course, I always find the category of “mind-body” practices to be problematic because it lumps together practices with actual science demonstrating a plethora of health benefits with other modalities that are probably not. Inevitably, the “body” part of “mind-body” emphasizes “exotic” forms of exercise, like yoga, tai chi, or quigong, over more mundane forms of exercise like running, bicycling, or more gentle forms of exercise like walking. The “mind” forms of “mind-body medicine” also tend to mix potentially science-based with more questionable claims. Of course, the whole name “mind-body” is problematic in itself as well. After all, it rather assumes Cartesian mind-body dualism, when in fact the mind is a product of the activity the body, specifically the brain.

Buric is a graduate student in a joint program between Coventry University and Donders Institute for Brain, Cognition and Behaviour, and the senior author of the paper is Inti Brazil at the Collaborative Antwerp Psychiatric Research Institute in Belgium. The study was published in Frontiers in Immunology and is entitled “What Is the Molecular Signature of Mind–Body Interventions? A Systematic Review of Gene Expression Changes Induced by Meditation and Related Practices.” Let’s take a look.

A systematic review of “mind-body” interventions, or: GIGO

I’ve read many systematic reviews and meta-analyses in my time, and if there’s one thing that I’ve learned it’s that they are very, very dependent upon the quality of the input. Computer programmers have a term that is very applicable here: GIGO, which stands for “garbage in, garbage out.” It’s a term that is applicable to virtually any form of data analysis or synthesis. If the data used to do the analysis is not of high quality (i.e., it is garbage), then the results of analyzing the data will also not be of high quality (i.e., it will be garbage too). When it comes to meta-analyses, what I like to say is that aggregating a bunch of low quality studies will not miraculously result in high quality, highly reliable conclusions. Indeed, if the studies share similar problems, such as bias, aggregating them will actually amplify those problems. Similarly, systematic reviews, which are different from meta-analyses in that they don’t involve trying to aggregate the data from the studies being examined and come to conclusions but rather are what the name implies, systematic reviews of existing evidence, can produce low quality and/or biased results if the input consists of low quality or biased studies. That’s why there are all sorts of rules and best practices for meta-analyses, but even those won’t save a systematic review or meta-analysis from the GIGO phenomenon. Unfortunately, these days, when it comes to meta-analysis and systematic reviews, GIGO all too often seems to stand for “garbage in, gospel out.” We sometimes see this phenomenon in evidence-based medicine (EBM), where meta-analyses and systematic reviews are considered the highest forms of evidence on the pyramid.

That means we have to dive into the weeds of how this systematic review was carried out:

We will now review studies on MBIs (mindfulness, yoga, RR [the relaxation response- DG], Tai Chi, and Qigong) that include gene expression analysis as an outcome measure, in order to assess the evidence for their effects on gene expression, and what changes in gene expression underpin the psychological benefits of MBIs. Studies were identified by searching PubMed through September 2016 using the following combination of keywords: (meditation OR mindfulness OR relaxation response OR yoga OR tai chi OR Qigong) and (gene expression OR microarray OR transcriptome). A total of 716 articles were returned and their titles and abstracts were screened (see Figure 1). We excluded studies that did not meet the following eligibility criteria:

  1. The population studied should only contain adults.
  2. Both clinical and non-clinical samples were allowed (for example, students, cancer patients, and caregivers) and studies with all sample sizes were included.
  3. Studies with experienced practitioners or non-experienced practitioners were allowed, making both cross-sectional and longitudinal studies eligible.
  4. Gene expression changes had to be one of the outcome variables (any number of analyzed genes, cell type and any gene expression technology were eligible).
  5. The independent variables had to be any type of MBI.
  6. Articles should be written in English.
  7. Only research papers were included. Review papers, meta-analyses, commentaries, and conference proceedings were excluded.

Most of these are not unreasonable criteria. However, there is one big problem, and that’s the variety of interventions that fall under mind-body interventions (MBIs). For instance, on what basis does one group a psychological intervention (e.g., mindfulness or relaxation response) with various exercises (yoga, tai chi, quigong)? This is particularly problematic given the relatively low number of studies. The authors screened 716 articles, but only ended up with 18 encompassing 846 participants. None of the studies were particularly large, and most were not randomized. The average number of participants per group averaged over the studies was 23.6, which is not large. The studies were also of mixed design: cross-sectional (17%), longitudinal (50%), rapid response (11%), and mixed (22%). A cross-sectional study is an observational study in which data are collected at a specific point in time and compared between groups. A longitudinal study is also an observational study, except that data are collected repeatedly over time from the same subjects, specifically at different time points. Rapid response studies, as the name implies, observe what happens at brief time points after an intervention.

There were a lot of differences in experimental methodology as well, for instance, how gene expression changes were measured. Most of the studies (72%) harvested RNA from peripheral blood monocytes (PBMCs) to measure changes in gene expression, while others (17%) used lymphocytes. A variety of techniques for measuring gene expression were used as well, from just using reverse transcriptase polymerase chain reaction (RT-PCR) to examine two to twenty-three genes to using Affymetrix or Illumina cDNA microarrays, to using next generation sequencing methods like RNASeq. Of course, as we’ve learned from John Ioannidis, reproducibility is a major problem in whole genome gene expression studies. Also, psychology studies suffer greatly from this problem as well, and mindfulness studies are no exception.

Of course, the studies one really wants to look at are randomized controlled clinical trials, and less than a third of the studies examined were randomized, specifically these:

  1. Creswell JD, Irwin MR, Burklund LJ, Lieberman MD, Arevalo JM, Ma J, et al. Mindfulness-based stress reduction training reduces loneliness and proinflammatory gene expression in older adults: a small randomized controlled trial. Brain Behav Immun (2012) 26(7):1095–101. doi:10.1016/j.bbi.2012.07.006
  2. Black DS, Cole SW, Irwin MR, Breen E, Cyr NMS, Nazarian N, et al. Yogic meditation reverses NF-κB and IRF-related transcriptome dynamics in leukocytes of family dementia caregivers in a randomized controlled trial. Psychoneuroendocrinology (2013) 38(3):348–55. doi:10.1016/j.psyneuen.2012.06.011
  3. Irwin MR, Olmstead R, Breen EC, Witarama T, Carrillo C, Sadeghi N, et al. Tai chi, cellular inflammation, and transcriptome dynamics in breast cancer survivors with insomnia: a randomized controlled trial. J Natl Cancer Inst Monogr (2014) 2014(50):295–301. doi:10.1093/jncimonographs/lgu028
  4. Bower JE, Greendale G, Crosswell AD, Garet D, Sternlieb B, Ganz PA, et al. Yoga reduces inflammatory signaling in fatigued breast cancer survivors: a randomized controlled trial. Psychoneuroendocrinology (2014) 43:20–9. doi:10.1016/j.psyneuen.2014.01.019
  5. Bower JE, Crosswell AD, Stanton AL, Crespi CM, Winston D, Arevalo J, et al. Mindfulness meditation for younger breast cancer survivors: a randomized controlled trial. Cancer (2014) 121(8):1231–40. doi:10.1002/cncr.29194
  6. Epel ES, Puterman E, Lin J, Blackburn EH, Lum PY, Beckmann ND, et al. Meditation and vacation effects have an impact on disease-associated molecular phenotypes. Transl Psychiatry (2016) 6(8):e880. doi:10.1038/tp.2016.164

These, and the other studies, had a variety of controls as well, ranging from basically no controls to “active” controls to “passive” controls. Curious what constitutes an “active” control, I examined a couple of the studies above

“Mind-body” interventions: Less there than meets the eye

The first study I looked at was reference #2 (Black et al). This was a study of people who were caring for a frail or demented family member, based on the rationale that caregivers tend to have worse mental and physical health than matched controls probably due to stress-induced upregulation of inflammation-related genes and downregulation of innate antiviral genes. The two groups were Kirtan Kriya Meditation (KKM) or Relaxing Music (RM) listening for 12 minutes daily for 8 weeks, after which subjects had blood drawn for gene expression profiling studies. The authors reported that in the KKM treatment group, 68 genes were found to be differentially expressed (19 up-regulated, 49 down-regulated) compared to the RM control group after adjusting for potentially confounded differences in sex, illness burden, and BMI. Up-regulated genes included immunoglobulin-related transcripts. Down-regulated transcripts included pro-inflammatory cytokines and activation-related immediate-early genes. One thing I noticed right away in this study is that the effect sizes were quite small. The authors set their criteria for a positive result as genes whose transcript levels change by more than 1.2-fold, which is casting a very wide net indeed. None of the genes examined changed by more than 1.65-fold, something that always sends up a red flag to me as far as whether the effect observed is truly biologically relevant. Moreover, one of the sets of genes, the set that responds to Interferon Response Factors (IRF) barely achieved statistical significance (p=0.04, which for a study of this type is pretty lousy). As for NF-κB, having worked with this particular transcription factor in the past rather extensively, I know that it’s hyper-responsive to many things; indeed, I used to joke that if I just looked at my cells funny it would activate NF-κB signaling.

Another interesting publication was the last one, which describes the result of a collaboration with Deepak Chopra, whose foundation funded the study. The study itself took place at the Chopra Center for Wellbeing at the OMNI La Costa Resort and Spa in Carlsbad, CA. It looks very much to be a publication that flowed from a collaboration between Chopra and various academic medical centers prone to quackademic medical studies. Healthy women aged 30–60 were recruited to stay at the resort and then randomized to a vacation arm or novice meditator arm, for a total of 94 participants who completed the program. A comparison group of regulator meditators was recruited from women aged 30–60 who had already enrolled in the retreat. According to the supplemental data, the sample included women who were non-meditators and lived in California who were randomized to a “vacation” arm or a meditation arm. They did not pay for their stay at the resort. The comparison group of non-randomized women (the “experienced” meditators) was recruited from the pool of those already registered for the retreat and paid for their own expenses. I’m sure you can see part of the problem right here. First, it’s a bit dicey ethically to charge anyone for their stays. Second, the experienced meditators were not randomized, and they could easily have confounding differences that could affect their results. In any event, although it wasn’t described in the methods of the paper itself, promotional literature described the retreat goal as “to promote an intensive period of learning and psychological change.” The vacation group was hosted at the same resort, but they did not participate in any retreat activities. Subjects reported on well-being immediately after (on day 5), 1 month later, and 10 months later to assess maintenance of benefits.

Here’s an ad for the study:


And here’s a promotional video Chopra made for the study:


In the video, Chopra states:

The idea is to prove—scientifically—that your biology is a product of the choices that you make, and these choices are made every day, and that we can actually consciously create the experience of a joyful energetic body, a loving compassionate heart, a restful reflective mind, and lightness of being.

Very scientific sounding, isn’t it? But what does this even mean? How, specifically, do you measure whether or not the participants have consciously created “the experience of a joyful energetic body, a loving compassionate heart, a restful reflective mind, and lightness of being”? Besides, it’s trivial to say that our choices affect our biology. Of course they do! If you smoke, you hurt your heart and lungs and vastly increase your risk of heart disease and cancer. If you drink to excess, you vastly increase your risk of liver disease and, to a lesser degree, your risk of variety of cancers. If you’re a sedentary slug who never exercises (as I was and still sometimes am), you vastly increase your risk of developing any number of harmful conditions and diseases, such as hypertension and type II diabetes. These are trivial observations.

Of course, confirming such trivial observations, long confirmed by medical science, was not what Chopra was about. Notice his unscientific language. He didn’t say “test the hypothesis” or “see if this program results in X and Y.” He didn’t know what he’s looking for before the experiment was done. He only knew that it would be good. Basically, there was no hypothesis except that his center’s program is good for your health. So he said things like “prove scientifically” that his program does all sorts of wonderful things that aren’t specified in concrete, measurable ways. Instead, he planned to shotgun measure a whole boatload of markers and endpoints, including next generation whole genome sequencing. Here’s another problem. Because a “whole program” was tested, it is impossible to tell what’s responsible for the changes, as subjects in the vacation group could do whatever they wished. In fairness, “fishing expeditions” like this can be good for generating hypotheses, but the design of this trial makes me doubt the hypotheses generated from the conclusions of the study.

In any case, the study noted improvements in psychological well-being in all groups. (Surprise! Surprise!) Investigators also identified gene expression signatures that changed in all three groups consisting of 390 genes. This is code for saying that there wasn’t a detectable difference between the groups in changes in this gene signature, which the authors attribute to the “vacation effect.” They did, however, find differences in another gene signature whose change differed between the vacation and novice meditator arms versus the regular meditator arm. That’s nice, but it doesn’t really tell us anything, given that the regular meditators could easily have confounding factors at play based on behaviors and diet different from the vacation and novice meditator groups.

I didn’t go through each and every study in the systematic review, but I think you get the idea. The studies were generally preliminary and often not of the best quality. The Chopra study didn’t even really compare the effects of meditation versus not meditating in a rigorous fashion. That doesn’t stop the authors from concluding:

The results of 18 studies that used gene expression analysis in research on meditation and related MBIs have overall found downregulation of NF-κB-targeted genes, which can be understood as the reversal of the molecular signature of the effects of chronic stress. Even though the study designs, the population, and the types of MBI used in the studies included in this review vary, it indicates that some of the psychological and physical benefits of MBIs are underpinned by biological changes in NF-κB genes. These results need to be replicated in larger samples and with stronger research designs that control for non-specific effects of these practices and for as confounding lifestyle factors, such as sleep, diet, and exercise.

Ya think? I can’t help but note that NF-κB is basically at the heart of inflammation, but it isn’t alone. In any case, I also can’t help but note that a recent review of 20 randomized clinical trials of mindfulness meditation and the immune system was a lot more tentative. Only three of the studies found decreased NF-KB, for instance.

None of this means that I don’t find it fairly plausible that various exercise regimens could decrease the level of pro-inflammatory gene expression. I even find it plausible that relaxation and meditation might result in similar effects. It is, however a long stretch to claim that such activities “reprogram your DNA” in that changing gene expression doesn’t require doing anything to the DNA itself, just changing levels of the proteins that regulate expression of the genes whose levels change. In any case, this is preliminary research massively overhyped, and I didn’t even mention until now that it’s in a Frontiers journal, and Frontiers journals are known among my colleagues for their poor peer review.

But that’s not all.

Publication bias a-go-go

Remember how I started out this post with the aphorism “GIGO”? There’s plenty of reason to suspect bias in much of the research into “mindfulness” and “mind-body” interventions. Most recently, Tim Caulfield pointed me to an interesting paper:

 The study looked at publication bias in mindfulness research and noted as a rationale for undertaking an investigation:

MBSR [mindfulness-based stress reduction] and MBCT [mindfulness-based cognitive therapy] have been reported to improve mental health outcomes among patients with psychiatric conditions (e.g., depression [1, 10], anxiety [11, 12], posttraumatic stress disorder [13], eating disorders [14], substance use disorders [15]), and other medical conditions (e.g., diabetes [16], hypertension [17], cancer [18], arthritis [19], obesity [20], heart disease [21], stroke [22]). In the United Kingdom, MBCT has been recommended by the National Institute for Health and Care Excellence to prevent depression relapse [23].

A concern, however, is that the overwhelmingly statistically significant results in favor of MBSR and MBCT interventions that can be seen in the published literature, despite very low power in many studies, may be influenced by reporting biases. Reporting biases are said to occur when statistically significant or “positive” outcomes have been preferentially published compared to non-significant or “negative” outcomes [24–26]. Reporting biases include (1) study publication bias, in which positive studies tend to be published, whereas negative studies are not; (2) selective outcome reporting bias, in which outcomes published are chosen based on statistical significance with non-significant outcomes not published; (3) selective analysis reporting bias, in which data are analyzed with multiple methods but are reported only for those that produce positive results; and (4) other biases, such as relegation of non-significant primary outcomes to secondary status when results are published [24–28].

In other words, there are way more positive studies of mindfulness therapies out there than the statistical power of the studies as reported would predict. The authors, Coronado-Montoya et al, from the Lady Davis Institute for Medical Research, Jewish General Hospital, Montréal, Québec and the Department of Psychiatry, Oregon Health & Science University, note that previous meta-analyses of the topic have not done an adequate job assessing the literature summarized for bias. They note that the methods generally used are not likely to detect reporting biases when there are fewer than 10-20 included trials and might require large numbers of trials in some circumstances. They also note that such methods “are also not appropriate when most studies have limited sample sizes, or when there is relatively little variance in sample sizes [45, 46], all of which are common in MBT [mindfulness-based therapy] trials” and observe drolly that they have “observed anecdotally that there seem to be few examples of published MBT trials without statistically significant results, even though many existing trials appear to have been conducted with very low statistical power.”

So Coronado-Montoya et al did their own systematic review and also assessed recent systematic reviews and meta-analyses to determine whether reporting biases were identified. They began by searching the CINAHL, Cochrane CENTRAL, EMBASE, ISI, MEDLINE, PsycInfo, and SCOPUS databases for randomized controlled trials of MBTs. What they did next was simple in concept, but difficult in execution. They counted the number of positive trials and compared that number to the number of positive trials that that might be expected if mindfulness-based therapy were similarly effective compared to individual therapy for depression. They also searched trial registries like ClinicalTrials.gov for mindfulness-based therapy registrations.

Guess what they found:

108 (87%) of 124 published trials reported ≥1 positive outcome in the abstract, and 109 (88%) concluded that mindfulness-based therapy was effective, 1.6 times greater than the expected number of positive trials based on effect size d = 0.55 (expected number positive trials = 65.7). Of 21 trial registrations, 13 (62%) remained unpublished 30 months post-trial completion. No trial registrations adequately specified a single primary outcome measure with time of assessment. None of 36 systematic reviews and meta-analyses concluded that effect estimates were overestimated due to reporting biases.

Only three were presented unequivocally as negative trials “without alternative interpretations or caveats to mitigate the negative results and suggest that the treatment might still be an effective treatment.”

The authors also point out:

Our review of trial registration records also suggest the possibility that reporting biases may have been an important factor. Of the 124 RCTs reviewed, only 21 (17%) were registered prior to data collection, even though 80 of the eligible RCTs were published recently (since 2010). When we examined trial registries, we identified 21 registrations of MBT trials listed as completed by 2010 and found that 13 (62%) remained unpublished 30 months after completion; of the published trials, all conveyed a positive conclusion. None of the 21 registrations, however, adequately specified a single primary outcome (or multiple primary outcomes with an appropriate plan for statistical adjustment) and specified the outcome measure, the time of assessment, and the metric (e.g., continuous, dichotomous). When we removed the metric requirement, only 2 (10%) registrations were classified as adequate. We evaluated more than 30 published systematic reviews and meta-analyses of MBTs, and none concluded that reporting biases likely exaggerated estimates of effect (see S6 Appendix).

The authors concede that they could not definitively determine whether there was significant reporting bias. One reason is that they could not conduct a statistical test to determine if there was excess significance bias and had to choose a reference point of a known effective therapy to compare to the MBT trials. However, from my perspective, something sure does smell bad here, particularly given that almost none of the MBT trial registrations defined outcome variables with sufficient precision to compare to subsequently published trial results, which would reduce the likelihood of selective outcome reporting. The authors themselves suggest that reporting biases are likely to be the “driving force” causing the disconnect between the expected number of positive trials versus the observed number.

As Tim Caulfield put it:

Indeed. Perhaps a Speedo on an obese old man.

Mind-body: Fantasy versus reality

Mind-body therapies, such as mindfulness-based therapies, are very attractive because they are low cost, don’t involve pharmaceuticals, and, above all, provide a sense of control to the patient. Indeed, as ever more rigorous clinical trials find that most of the “unconventional” therapies (i.e., quackery) that “integrative medicine” integrates with conventional medicine are elaborate placebos, proponents of integrative medicine increasingly fall back to pointing to modalities like yoga, tai chi, and the like and “mindfulness.” Unfortunately, they increasingly oversell both. After all, yoga, tai chi, and qigong are basically forms of exercise. Of course it’s expected that they would be likely to have health benefits, but so do more conventional aerobic exercise and even moderate housework. It’s also not implausible that “mindfulness” or other so-called “mind-body” interventions could also have health benefits.

However, we have to be very careful here. I once said that the “central dogma” of alternative medicine is that wishing makes it so; that is, if you think about or wish for something hard enough, the universe will provide it. That’s why placebo effects are so seductive and attractive to integrative medicine proponents; they can be misrepresented as the “power of the mind over the body.” You can see how mindfulness and other mind-body interactions are just as attractive, because many of them involve actual thinking and meditating. Unfortunately, the science does not (as yet) support many of the overblown claims made for these practices. It’s not clear whether it ever will. Certainly it’s unlikely that they “reprogram” our DNA.

This article originally appeared on Science Based Medicine here and was reposted with permission from the author.

David H. Gorski, MD, PhD, FACS is a surgical oncologist at the Barbara Ann Karmanos Cancer Institute specializing in breast cancer surgery, where he also serves as the American College of Surgeons Committee on Cancer Liaison Physician as well as an Associate Professor of Surgery and member of the faculty of the Graduate Program in Cancer Biology at Wayne State University.

Want to stay up to date on the latest epigenetics news? Follow ELP on Twitter and like the Facebook