The Effects of a Single Bout of High- or Moderate-Intensity Yoga Exercise on Circulating Inflammatory Mediators: A Pilot Feasibility Study

Ethics Statement

The study was approved by the Mass General Brigham/Partners Institutional Review Board (Protocol #: 2019P001716). Written informed consent was obtained from all subjects during the first visit before any intervention. The study was registered at clinicaltrials.gov ID# NCT04444102.

Study Design

This study was a three-arm, pre-post pilot randomized clinical trial of a single bout of yoga exercise intervention. The 3 groups were: high-intensity yoga exercise (HY), moderate-intensity yoga exercise (MY), and a sedentary, no intervention control (CON). In-person outcome assessments were conducted at baseline and post-intervention time points (0-, 30-, 60-, 120-, 180-minutes, and 24-hours post-intervention).

Eligibility Criteria

Healthy participants, relatively sedentary (<30-minutes twice a week of high-intensity exercise, e.g., running), yoga- and MBMT-naïve, ≥40 to ≤60 years old, non-smoking, and with BMI ≥19 and ≤29, were recruited using Partners Rally, an online platform that supports collaboration between the public and research community to help advertise and recruit participants for different clinical trials, and flyers publicly posted in the Boston area. Exclusion criteria included: any history of chronic inflammatory disease or recent acute illness (<1 month) (e.g., rheumatoid arthritis, cancer, acute or chronic musculoskeletal pain), use of any medication in the preceding week (both prescribed or over-the-counter, except vitamins, supplements, herbal medicine, and birth control pills), and practice of structured exercise at least twice a week for more than 30 minutes.

Randomization

An allocation sequence (i.e., random permuted blocks of sizes 3 and 6) was generated and uploaded to the Research Electronic Data Capture (REDCap) online system. After obtaining informed consent, baseline data, and the first blood samples, the study coordinator randomized subjects within the REDCap platform.

Study Visits

Blinding

Participants and the yoga instructor were not blinded to group allocation. However, nurses and laboratory technicians were blinded during sample collection and laboratory analyses. The data analyst was unblinded to the allocation after all laboratory analyses were completed.

Outcomes

The primary goal of this study was to assess a priori-defined metrics of feasibility. Feasibility was evaluated based on: (1) percentage of eligible to enrolled participants (>60%); (2) the number of months required to complete the recruitment process (within 9 months); (3) retention rate (i.e., proportion of participants completing 2 study visits >70%); (4) completion rate of questionnaires (>75% participants with full completion); (5) completeness of physiological measures (i.e., >80% of HR and blood measures collected and processed); and (6) the adherence rate to the yoga exercise protocol, defined as the percentage of the total number of poses each participant was capable of performing (>75%).

Adverse events (AEs) were systematically queried by the study coordinator immediately after the intervention and during the qualitative interview. Participants were also instructed to contact the study coordinator if they experienced physical symptoms that concerned them after the intervention and testing session ended.

Secondary outcomes included serum levels of a battery of cytokines measured at multiple time points to characterize the impact of a single bout of yoga exercise and to inform potential markers to be used in a future, fully powered trial. At baseline and 24 hours after intervention, patients completed online (REDCap) the PROMIS Pain intensity questionnaire, a 3-item measure assessing pain intensity using a 5 point Likert-type scale with anchors, no pain to very severe pain (Q1: worse pain intensity in the past 7 days, Q2: average pain intensity in the past 7 days, and Q3: current pain intensity). ³³ The PROMIS Pain intensity instrument was used to characterize pre-post changes in body soreness after the yoga exercise class. HR was assessed to characterize the metabolic intensity of the yoga exercise interventions. A baseline questionnaire collected information on age, gender, ethnicity, education, alcohol consumption, and BMI.

Laboratory Data Collection

Statistical Methods

Demographic and HR data were analyzed with descriptive statistics. Continuous raw data were tested for normality using the Shapiro-Wilk Test. Means and standard error of the mean (SEM) were reported for normally distributed data, and medians and interquartile ranges were reported for non-normally distributed data. Nominal data were reported as absolute numbers and percentages.

Exploratory analysis of cytokine levels utilized all available data per participant. Undetectable cytokine levels at a given time point were not included in analyses and treated as truncated data. ³⁶ Levels of each cytokine at each timepoint were summarized with means and SEM per group. Exploratory longitudinal inferential analyses were calculated with SAS package (version 9.4, SAS Institute Inc, Cary, NC, USA) for each cytokine to explore time-by-group interactions using unadjusted analysis of response profile commonly used when dealing with an incomplete data set. To further characterize cytokine responses, raw data from the first 6 time points were used to calculate net Area Under the Curve (netAUC) adjusted for each participant’s baseline values. This analysis compares increases and decreases of each cytokine over time using the trapezoid rule to summarize the change in each cytokine level with a single value.^37,38 Comparisons of netAUCs between groups were assessed using an unadjusted general linear regression model. Finally, to characterize correlations between the magnitude of initial changes (baseline to first timepoint after the intervention) between 7 cytokines commonly evaluated in myofascial exercise studies,^10,11,39,40 Pearson correlations were used to generate matrices of association between these 7 cytokines and plotted using heatmaps. Positive or negative relationships between changes among the 7 cytokines were then represented with heptagon strength diagrams. ⁴¹

The Prism 9 software (GraphPad Software, Inc, San Diego, CA, USA) was used to calculate descriptive statistics for HR, PROMIS Pain Intensity score, and exploratory cytokine analyses. Given the pilot nature of this study, these analyses were intended to be exploratory to inform future hypotheses and study design features.

Based on a prior pilot in our laboratory, the sample size for this study was expected to be sufficient for evaluating a priori-defined metrics of feasibility.^42,43 Data from all subjects were reported and analyzed according to the intention-to-treat principle.

Feasibility (Aim 1)

We fulfilled 4 of 6 a-priori defined feasibility metrics (retention rate, completion rates of questionnaires, completion rate of physiological measures, and protocol adherence). The remaining 2 feasibility metrics were close to target; the eligible to enrollment rate was 55% vs. 60%, and the number of months to complete the recruitment process was 11 vs. 9 months. Despite concerns about engaging in any social activity during COVID-19 social distancing measures, especially activities requiring visits to hospital environments, we were still able to recruit 30 yoga-naïve, eligible individuals who were willing to be randomized and provide biological samples at multiple time points. Once enrolled, all but 1 participant returned for visit 2, and the majority were highly compliant with all protocols. Of note, 14 (47%) participants of our sample were other than Caucasian, supporting our ability to recruit a diverse population. This success was due to active recruitment in highly diverse neighborhoods in the Boston area using flyers posted in public spaces. Collectively, these metrics suggest that recruitment and retention in future larger, powered trials with a similar design should be feasible.

The adherence rate to the yoga exercise protocol also indicates feasibility of using the same posture sequence and pacing in future trials. The inclusion of posture modifications (i.e., less physically demanding versions of the original postures) allowed participants and/or the yoga instructor to opt for them when the original version was difficult to achieve. We also noticed that adherence was impacted by differences in BMI at baseline (Table 1). A BMI >25 at baseline in the MY may have caused this group to opt for more posture modifications. Higher BMI may also explain the counterintuitive slightly higher HR in the MY compared to the HY (30% and 24.6% compared to CON, respectively). ⁴⁵ We consider this finding relevant because previous studies among yoga naïve individuals with chronic health conditions and physical limitations have highlighted the importance of the protocol’s design for adherence.^46,47 For example, a study among women with abdominal obesity allowed for the adapting of each posture to the needs, capabilities, and limitations of this population and gradually increased the difficulty of the postures over the course of the intervention. ⁴⁸ In addition to BMI, gender was also skewed in the MY towards females; future larger randomized studies should stratify by BMI and gender to minimize confounding effects of the participants’ characteristics. Future studies might also consider assessing postural range of motion (ROM), heart rate variability (HRV) metrics, and more detailed metabolic activity (e.g., mean VO2, peak VO2, %VO2max, and kcal expenditure) to better characterize the yoga interventions.^5,49 Finally, with respect to delayed onset of muscle soreness (DOMS), our post-intervention results for body soreness were not surprising. Future studies should include and compare subjective and objective tools to evaluate body soreness after short- and long-term yoga exercise interventions. These may include the Likert scale of muscle soreness and quantitative sensory testing (QST), as well as metrics that account for participants’ level of sedentarism and intervention intensity, 2 relevant variables for the evaluation of DOMS.^50,51

Exploratory Characterization of the Temporal Response of Circulating Inflammatory Cytokines to a Short-Term Yoga Exercise Intervention (Aim 2)

We demonstrated that collecting and analyzing blood samples taken at multiple time points (7 in total per participant) for quantitative time-course descriptive analysis is feasible. However, in this pilot study, we observed a great deal of heterogeneity within and between groups, and over time. We speculate that a number of factors may have contributed to this variability. First, the lack of an adequate physical activity washout period between participants’ variably active travel mode (e.g., time amount and intensity of walking required) to the study site and the first baseline blood sample likely contributed to variability across individuals at baseline. This factor also may explain why netAUC was more negative in the control group, as they experienced a longer period of sedentary activity. This bias may have masked other influences of the yoga intervention cytokine dynamics.^9,35 Second, because of our small sample size and chance, randomization with respect to biological factors known to impact systemic inflammation (e.g., BMI and age) was not balanced. ⁵² Of note, differences in BMI and age may have contributed to the lack of observed difference between the HY and MY groups. Third, it is possible that circadian variation also contributed to variability in cytokine levels. It is well established that some cytokines have circadian cycles. Therefore, it is possible that baseline values varied between individuals due to their variability in their circadian cycles. Additionally, these circadian cycles (e.g., increases in the early morning) could have confounded the effect of exposure to yoga exercise on cytokines.^6,53

Few yoga studies have evaluated the impact of short-term (e.g., single bout) interventions on systemic levels of cytokines for comparison with this study. Three studies measured changes in systemic levels of different inflammatory cytokines (e.g., IL-6, IL-10, and TNF-α). The first 1 compared Bikram (hot) yoga exercise vs. room temperature (RT) yoga exercise among yoga experts. Only the hot yoga group experienced a significant increase in IL-6 after the intervention. ⁵ The second study exposed healthy yoga female practitioners to 3 conditions during 3 separate visits (restorative yoga, movement control, and passive-video control) and measured changes in levels of IL-6 and TNF-α before and after each intervention. Findings indicated increases for both cytokines post-intervention. ⁶ Another yoga-related study measured, after a single bout of yogic breathing, levels of inflammatory cytokines in saliva at baseline and 5 timepoint post-intervention (0-, 5-, 10-, 15-, and 20-minutes). Their results show an overall reduction of IL-1β, IL-8, and MCP-1 in the yogic breathing group compared to the control group.7 2 main differences between our study and the first 2 yoga exercise trials are: (1) differences in study design (i.e., cross-over vs. parallel); and (2) previous exposure to yoga exercise (i.e., yoga-naïve vs. yoga experts or beginners). With respect to the third yoga breathing study: (1) they used saliva instead of serum to quantify cytokines levels;^54,55 (2) their intervention targeted the respiratory system, whereas our protocol used yoga’s physical postures without any systematic breathing exercise technique; and (3) the intervention’s duration was different (20- vs. 60-minutes). Future short-term yoga studies should more systematically explore biological interactions between yoga postures and breathing exercises on systemic levels of inflammatory mediators.

While there are few studies of the impact of a single bout of yoga on circulating inflammatory mediators, it is broadly known that a single bout of more conventional exercise (e.g., endurance, resistance, and sprint training) stimulates systemic changes in cytokine levels during physical activity and recovery among healthy individuals, and this immune response reflects the magnitude of physiological stress (e.g., intensity) experienced by the exerciser.^35,56-60 Other research on mechanical forces targeting the myofascial system has also observed changes in the dynamics of circulating cytokines (often referred to as myokines). ⁹

In contrast to short-term studies, there are more robust studies on the effects of long-term yoga practice on inflammation. A systematic review summarized results from 15 studies assessing the long-term effects of yoga on inflammatory markers among healthy and/or disease individuals. Results showed no agreement on the effect of yoga on levels of C-reactive protein (CRP), TNF-α, and IL-6. ⁶¹

Our pilot data reveal a number of trends/patterns worthy of further investigation in fully powered studies. Although serum values of individual cytokines varied considerably within and between groups over a 24-hour period, exploratory analyses revealed trends toward group-by-time effects for IFN-α2, IL-6, and IL-33, suggesting that an acute bout of yoga exercise may impact circulating levels of cytokines. These findings reflect differences in temporal patterns between MY and CON for IFN-α2 and IL-33, while for IL-6, this interaction effect reflects differences between the 2 yoga groups.

To determine whether different cytokines show unique responses to yoga exercise, the first 6 timepoints were used to compute netAUC. Although, in all cases, the average netAUC was negative, a closer look at the 12 graphs reveals a subset of cytokines showing unique trends of responsiveness to the yoga exercise intervention (IFN-γ, TNF-α, IL-8, IL-10, IL-12p70, IL-17A, and IL-33). Specifically, these cytokines exhibited a lower rate of decline in both yoga exercise groups compared to the control group. Of note, 6 of these cytokines have been reported to be important myokines involved in metabolic, inflammatory, and regenerative processes occurring in the muscle microenvironment through the co-dependent cross-talk between resident and migrant immune cells during and after exercise.^11,12,39,62 IL-12p70, a potent driver of Th1 immune responses, has relevant roles in fibrosis and tissue regeneration and warrants further investigation regarding its potential functions on myofascial and yoga physiology. ⁶³ Our findings also suggest a difference between the yoga exercise and control groups for IL-33, a crucial cytokine released by fibro-adipogenic progenitor cells (FAP) to enhance the recruitment of Treg cells to help in the regeneration process while contributing to the extracellular matrix production needed to support myofascial repair. ⁶⁴

Changes in cytokines after a bout of yoga exercise were positively correlated, suggesting that a single bout of yoga exercise may cause coordinated and inter-dependent post-interventional dynamic changes across different cytokines.^10,59,65,66 IL-6 and IL-33, 2 important cytokines with medium to strong positive correlations between these 7 myofascial-associated cytokines, indicate that understanding their dynamic changes, both individually and in relation to other cytokines, will provide granular insights into the temporal, local, and systemic dynamics associated with physical adaptation of the myofascial system through inflammation, tissue remodeling and regeneration processes during and after yoga practice.^11,12,39,62

Neither netAUC nor correlation matrices analyses reveal a difference in the response of systemic levels of cytokines to 2 different intensities of yoga exercise (high vs. moderate). This finding differs from studies in the sports medicine literature, where the intensity of single bouts of different exercise modalities led to differences in circulating levels of cytokines. As noted above, heterogeneity caused by multiple factors may have obscured detection of yoga intensity effects. This parameter of yoga should be systematically explored in future larger trials.^35,56-60

The Relevance of Short-Term Yoga Exercise Studies for Understanding the Physiology of Yoga

How yoga promotes preventive or therapeutic effects during health and disease remains incompletely understood. Likely, the multimodal nature of yoga (i.e., breathing exercises, active body movements, and meditation), ⁶⁷ and the multifactorial etiological nature of these disease syndromes challenge our understanding of the physiological networks triggered during yoga practice.^68-72 Our study design purposely emphasized the physical component of yoga (i.e., body postures) among healthy and yoga-naïve individuals in order to begin understanding the response of the myofascial system to an acute physical challenge expressed through changes in systemic levels of inflammatory mediators (i.e., bottom-up effects).^{27,30,62,73,74} Overall, we postulate that these inflammatory mediators might reflect how yoga practice stresses the myofascial system, allowing the recruitment of circulating immune cells towards myofascial stressed locations, along with activation of resident cells to start mechanisms of tissue remodeling, regeneration, and pro-resolution. ⁷⁵ All these beneficial effects are thought to be modulated by the same inflammatory pathways involved in acute/chronic inflammation and pain. ⁷⁶ In other words, a controlled increase of inflammatory mediators after an acute bout of exercise would be an allostatic mechanism to allow the body to adapt to a subsequent physical challenge with potential beneficial effects in the longer term. The widespread belief that only reductions in inflammatory mediators expression represent healthy change ignores the nuance of the complex role that physiological inflammation plays in maintaining homeostasis and dynamically adapting to an evolving environment. The context in which these increases or decreases occur is crucial. Indeed, recent exercise studies suggest that measurements of systemic biomarkers during and after short-term interventions (i.e., single bout of exercise) are fundamental to understanding training adaptation to exercise among healthy individuals. ⁵⁸ Currently, the best example to elaborate on these complex dynamics is given by the classical pleiotropic inflammatory mediator IL-6. Basal levels of IL-6 have multiple origins (e.g., adipose tissue, liver, brain, muscle, and blood leukocytes). When secreted by the myofascial system after a bout of physical activity, it is considered a myokine/exerkine with potent anti-inflammatory effects. In contrast, it is regarded as a pro-inflammatory cytokine when secreted by other tissues, such as adipose tissue. ²⁸ Moreover dynamic and sometimes opposite changes may occur over time, as with myofascial training adaptation (i.e., IL-6 increases in short peaks after each bout of exercise) vs. co-occurring long-term decrease of basal IL-6 levels produced by non-myofascial tissues. Unfortunately, these processes are not yet fully understood and warrant future research. ²⁹ Thus, a longitudinal future study exploring both effects (short- and long-term) might hypothesize that the time-effect on basal levels of IL-6 is not linear. We acknowledge that while yoga-based exercise and conventional exercise may have important differences, they likely evoke similar physiological responses. Although not observed in this pilot study, the magnitude of the myofascial effects likely depends on the modulation of 3 parameters: intensity, duration, and frequency of the intervention.^77,78

Limitations

Our pilot study has multiple limitations. First, the randomization process did not prevent imbalances due to the small sample size, which likely influenced the yoga exercise protocol’s adherence, the HR data, and potentially systemic levels of cytokines. Other baseline characteristics, such as age, BMI, diet, and even the level of cognitive stress, anxiety, and depression due to the COVID-19 crisis, might have influenced the systemic level of cytokines.^79,80 Additional sources of heterogeneity include the insufficient physical activity washout period prior to the start of the study, the lack of appreciation for potential circadian rhythms in specific cytokines, top-down neuro-muscular influences, and the potentially confounding added element of mental engagement (e.g., higher levels of focused attention) which was not isolated from the physical component. ⁸¹ Second, pilot studies are not powered to formally evaluate differences between groups.^52,82 Hence, inferential findings should not be over-interpreted, and future fully-powered studies should be considered for these purposes. Third, although feasible in this study, multiple timepoint sampling imposes logistic challenges for nurses and lab technicians, and it may be a cumbersome procedure for some participants. Currently, non-invasive and painless electrochemical biosensors for automatic, wireless, and real-time glucose detection levels are available. Future yoga exercise studies might benefit from this technology for cytokine detection at multiple body sites. ⁸³