A systematic review of in vivo stretching regimens on inflammation and its relevance to translational yoga research

Objective To conduct a systematic review evaluating the impact of stretching on inflammation and its resolution using in vivo rodent models. Findings are evaluated for their potential to inform the design of clinical yoga studies to assess the impact of yogic stretching on inflammation and health. Methods Studies were identified using four databases. Eligible publications included English original peer-reviewed articles between 1900–May 2020. Studies included those investigating the effect of different stretching techniques administered to a whole rodent model and evaluating at least one inflammatory outcome. Studies stretching the musculoskeletal and integumentary systems were considered. Two reviewers removed duplicates, screened abstracts, conducted full-text reviews, and assessed methodological quality. Results Of 766 studies identified, 25 were included for synthesis. Seven (28%) studies had a high risk of bias in 3 out of 10 criteria. Experimental stretching protocols resulted in a continuum of inflammatory responses with therapeutic and injurious effects, which varied with a combination of three stretching parameters––duration, frequency, and intensity. Relative to injurious stretching, therapeutic stretching featured longer-term stretching protocols. Evidence of pro- and mixed-inflammatory effects of stretching was found in 16 muscle studies. Evidence of pro-, anti-, and mixed-inflammatory effects was found in nine longer-term stretching studies of the integumentary system. Conclusion Despite the overall high quality of these summarized studies, evaluation of stretching protocols paralleling yogic stretching is limited. Both injurious and therapeutic stretching induce aspects of inflammatory responses that varied among the different stretching protocols. Inflammatory markers, such as cytokines, are potential outcomes to consider in clinical yoga studies. Future translational research evaluating therapeutic benefits should consider in vitro studies, active vs. passive stretching, shorter-term vs. longer-term interventions, systemic vs. local effects of stretching, animal models resembling human anatomy, control and estimation of non-specific stresses, development of in vivo self-stretching paradigms targeting myofascial tissues, and in vivo models accounting for gross musculoskeletal posture.


Introduction
Stretching is an integral component of mind-body exercises such as yoga, tai chi, and qigong [1][2][3]. While clinical studies support the use of mind-body exercises for a range of health conditions, including chronic pain [4], metabolic [5], and affective disorders [6], few experimental studies have tried to isolate the impact of stretching from other potentially therapeutic components (e.g., focused attention, breathing, imagery, psychosocial support) [7], Multifactorial additive designs and/or dismantling studies represent possible experimental clinical approaches for isolating and mechanistically evaluating the therapeutic impact of a single component, such as stretching, from other intervention components. However, these approaches can be costly and premature, especially when preclinical data related to dose and key mediating physiological processes are still poorly understood [8].
One alternative preliminary strategy to inform future mechanistic clinical research design is to leverage in vivo rodent studies. A significant body of rodent research has explored the biological basis of both the therapeutic and injurious effects of stretching on several physiological systems. However, these studies have not been systematically reviewed to inform clinical research [9,10]. This systematic review draws on this body of scientific literature with a specific focus on in vivo rodent studies that have investigated the impact of stretching on pro-and anti-inflammatory processes, as well as processes related to inflammation resolution, in the musculoskeletal (MSK) and integumentary systems. We focus on inflammation because it is an evolutionarily preserved mechanism involved in tissue remodeling (e.g., tissue repair) [11][12][13], it is believed to play a central role across the spectrum of therapeutic and injurious stretching, and it has also been observed to change in response to physical activity and mind-body movement practices like yoga and tai chi [14].
More broadly, through this systematic review of findings from controlled preclinical studies, we aim to inform a basic framework for designing and interpreting clinical studies that evaluate the impact of yogic stretching on inflammation and health. Accordingly, we highlight five features of rodent stretching studies that are particularly relevant to translational research: 1) Use of passive vs. active stretching techniques; 2) Stretching parameters (intensity, duration, frequency, and whole-body posture); 3) Use of shorter-vs. longer-term stretching protocols; 4) Use of injurious vs. therapeutic stretching techniques; and 5) Relevance of inflammatory outcomes, including macroscopic, histopathologic, genetic, immune cells/particles sorting, cytokines, and protective lipids (e.g., resolvins). Our discussion includes an exploration of a coordinated and bidirectional translation model of research, integrating pre-clinical and human studies to elucidate biological processes and guide clinical applications of multimodal interventions that include stretching, with an emphasis on yoga, in health and rehabilitation.

Protocol elaboration
The protocol for this systematic review follow the template of the Systematic Review Centre for Laboratory Animal Experimentation [15].

Information sources and electronic search strategy
The studies included in this systematic review were identified using PubMed, Embase, the Web of Science search engines, and the Harvard University library search platform (Hollis). A broad search strategy was used to identify all relevant studies. PubMed and Embase included citations through April 22, 2020, and Web of Science through May 7, 2020. Hollis identified six additional studies. Keywords used to guide search terms across all three databases are listed in Table 1. A complete electronic search strategy listed in S1 Table.

Study inclusion and exclusion criteria
Studies included investigated the effect of different stretching techniques (i.e., passive and/ or active stretching techniques with injurious and/or therapeutic intent, using either single or multiple bouts of stretching, and having a short-or long-term duration); administered to the MSK and integumentary systems of whole-rodent models (i.e., in vivo); and evaluating at least one outcome related to inflammation using at least one of the following five laboratory technique groups: microscopic, genetic, cell/particle sorting, enzymatic, and macroscopic techniques. Healthy populations of rodents (i.e., rats and mice) of any gender or age were considered. Inclusion also required either an independent non-stretched control group or the use of matched non-stretched contralateral muscles. Exclusion criteria included rodent studies stretching other body systems (e.g., reproductive or respiratory systems), in vitro and ex vivo paradigms, and clinical trials. In addition, eligible publications included all original peer-reviewed articles written in English between 1900 -May 2020. Pre-1900 articles were excluded because the laboratory techniques to evaluate inflammatory outcomes were not yet fully developed. Editorial materials, book and book chapters, biographical items, reviews, notes, letters, or social media news were excluded. Table 1. Summary keywords employed in the search strategy.

Animal Stretching Inflammation
In vivo Stretching exercise Pro-resolving mediators

Duplicates removal, screening, and data extraction process
Two reviewers used the Covidence © software to remove duplicates, screen abstracts, and fulltext reviews from articles identified (DMV and WG). A third reviewer resolved discrepancies (PW). The data extraction template contained nine domains: authors, study aims, type and intention of stretching, animal model information (strain, sex, species, age, N), intervention, stretching parameters, inflammatory outcomes, experimental groups, and results. The study team agreed on these fields, and the data extraction was conducted by two study members (DMV and WG) and reviewed by two independent reviewers (PW and GY).

Quality assessment
The SYRCLE's risk of bias tool for animal studies was used to assess the reported methodological quality of included studies [16]. Each criterion was classified as low, high, or unclear risk of bias.

Search results
Our search strategy identified 766 studies. After removing duplicate citations, 695 publications were screened using titles and abstracts. Fifty-eight studies had a full assessment of eligibility. After excluding ineligible articles, 25 studies underwent a complete systematic synthesis (Fig 1). Eight categories exhibited a low risk of bias in more than 50% of the studies. Outcome assessment blinding and other sources of bias were high and unclear risk of bias, respectively.

Categorization of study features
Across the 25 studies, there was significant heterogeneity in the study designs used. We categorized studies according to key characteristics of reported data: passive versus active stretching, stretching parameters, stretching protocol (shorter-versus longer-term), injurious versus therapeutic intent, and rodent species. Table 2 elaborates on specific terms used in the following subsections.
Passive versus active stretching. Twenty-two studies used passive stretching, satisfying the definition of a movement applied by an external force . All passive stretching studies anesthetized the animals to manual stretching or stretching achieved with a mechanical device . Of the 22 studies, fourteen explored the effect of dynamic muscle activation with a protocol of lengthening and shortening contractions/cycles. One study directly stimulated the nerve tissue associated with muscles [24]. A third study performed passive elongation under anesthesia [20]. The remaining six studies examined the mechanical effect of stretching on the integumentary system of rats and mice. Of note, five studies used devices either on top of the skin or subcutaneously to deliver a static stretching protocol [25][26][27][28]40]. Four studies performed an active stretching intervention by lifting rats and mice by the tail until reaching an approximately 45˚angle and increasing the distance between the shoulders and hips bỹ 25% [8,[41][42][43]. Stretching parameters (intensity, duration, frequency, and posture). Intensity is defined as the magnitude of force generated during stretching. Duration corresponds to the amount of time in which the stretching occurs. Frequency refers to the number of stretching bouts in a given time, and posture is defined as the spatial body position during stretching [17]. Fourteen studies that stretched the musculoskeletal system controlled three out of four stretching parameters (intensity, duration, and frequency) [20,21,24,[29][30][31][32][33][34][35][36][37][38][39]. One only controlled two of the four parameters (duration and frequency) [22], and another controlled for one parameter (frequency) [23]. Three studies that stretched the skin and connective tissue held constant three out of four stretching parameters (intensity, duration, and frequency) [26][27][28]. In contrast, six studies controlled two out of four parameters (duration and frequency) [8,25,[40][41][42][43] The posture parameter was not included by any of the studies included in this systematic review.

Term Definition
Active stretching (AS) [17,18] Refers to a movement applied by an external and/or internal force causing the interaction of the actin and myosin filaments of muscle due to the activation of muscle innervation and targeting the range of motion, i.e., flexibility.
Passive stretching (PS) [17] Refers to a movement applied by an external force causing the elongation of the myofascial and integumentary systems beyond their resting length and targeting the range of motion, i.e., flexibility.
Stretch-shortening contractions/ cycles (SSCs) [19] Refers to the muscle action when active muscle lengthening is immediately followed by active muscle shortening. This combination of eccentric and concentric contractions is one the most common type of muscle action during locomotion.
In vivo model of an acute muscle stretch injury [20][21][22][23] These procedures include: A muscle stretch injury applied by a joint torque system which precisely controls the amount of tendon shortening; Muscle elongation without surgical exposure; Muscle tetanic contractions through electrical stimulation.
In vivo model of an acute nerve stretch injury [24] Refers to the expose of nerves for stretching (e.g., median nerve between flexor digitorum superficialis and flexor carpi radialis).
Skin-stretching device model [25][26][27][28] Refers to the use of devices attached to the skin or implanted under the skin to deliver controlled stretching to the skin and surrounding connective tissue. https://doi.org/10.1371/journal.pone.0269300.t002 Injurious versus therapeutic stretching. Eighteen studies intended to use an injurious passive stretching model. Of these, thirteen applied it directly to muscle and two via muscle innervation [20][21][22][23][24][29][30][31][32][33][34][35][36]38,39] The remaining three studies targeted the skin and the connective tissue to examine the injurious effect of stretching with two different mechanical stretching devices delivering static stretching [25,27,40]. Seven studies investigated the therapeutic effect of stretching. One study employed a muscle model using one bout of lengthening contractions, followed by either passive stretches or isometric contractions [37]. Six studies used a skin model. One study used a subcutaneous stretching device inflated over 7-days delivering a static stretching on the surrounding tissue [28]. A second study utilized an external stretching device glued to the dorsal midline to create 0%, 20%, 33%, and 40% of a static stretch [26]. Four studies corresponded to active stretching models. One of them includes both active and passive stretching paradigms [8,[41][42][43].
Cell/Protein/Lipid outcomes. Three studies used flow cytometry to quantify inflammatory cells and cytokines [8,26,41]. One used liquid chromatography-tandem mass spectrometry to measure lipid mediators [41]. One study explored the effect of stretching on mesenchymal stem cells (MSCs) movement using an in vivo cell migration bioluminescent imaging model [28].
Macroscopic outcomes. Three studies employed ultrasound to measure lesion size. One used a caliper for tumor size. One study injected Evans blue as a marker of vascular permeability, and another performed tensile testing of unwounded skin [8,20,27,[41][42][43].

Effects of stretching on inflammation
Studies targeting the musculoskeletal system. Of the 16 studies exploring the effect of stretching on different muscle groups, 15 used an shorter-term injurious passive stretching protocol to cause a muscle inflammation [20,22,24,[29][30][31][32][33][34][35][36][37][38][39]. The stretching protocols shared several characteristics and were relatively consistent across the studies (Table 3). Studies consistently found that muscle injury by stretching led to plasma extravasation, edema, necrosis, and myofiber degeneration. Six studies characterized cell infiltration [20,21,29,30,32,35,36,38,39]. Of these studies, two found a consistent increase of neutrophils and macrophages after an shorter-term bout of stretching [35,37], Less evidence exists regarding the upregulation and secretion of different cytokines and chemokines. Out of the five studies measuring cytokines and chemokines, three reported that IL-1b, TNF-a, and IL-6 increased after an shorter-term stretching protocol [20,35,38]. Of note, one study found that prior muscle stretching conditioning, using a protocol of lengthening or isometric contractions or passive stretches, was able to reduce inflammatory cell (neutrophils and macrophages) infiltration after subsequent (longer-term) delivery of the same stretching protocol. [37] This study's main finding was that these stretching protocols elevated neutrophil infiltration without causing overt signs of injury, suggesting that the inflammatory process itself may contribute to the induction of a protective mechanism [37]. A second study using a shorter-term stretching intervention observed that an inflammatory process was evident in young rats on day three after the intervention and returned to control values by day 10. However, these responses were not observed in the older rats, indicating that age blunted or muted the inflammatory process [38]. The only study exploring the longer-term effect of stretching in muscle found an increase in the cellular interstitium volume in the exposed limb of old rats suggesting inflammation (Table 3) [34].
Studies targeting the integumentary system. Nine studies explored the effect of stretching employing different longer-term stretching protocols [8,[25][26][27][28]40,42,43,47]. These studies performed the inflammatory assessment at the end of the stretching intervention, or 12-to 24-hours post-intervention. Two studies explored the therapeutic effect of stretching on the lower back skin and connective tissue of rats using a carrageenan-induced inflammation model [41,42]. The first study established an active stretching model applied twice a day for 12 days to mitigate the carrageenan-induced inflammation. This study found stretching decreased lower back inflammation, leading to restored stride length and intrastep distance, decreased mechanical sensitivity of the back, and reduced macrophage expression in that area's nonspecialized connective tissue. The second study employed a similar carrageenan model; however, it included both shorter-term (twice a day for 48 hours) and longer-term (once a day for two weeks) stretching protocols, as well as active and passive stretching interventions [41]. They found stretching reduced inflammatory lesion thickness and neutrophil count and increased resolvin (RvD1) concentration within lesions [41,42]. A third study applied 10-minutes of active stretching once a day for four weeks on a p53/PTEN mouse mammary tumor cells model. The active stretching procedure was similar to the one used in the previous rat studies. Results showed that the stretched group had a 52% reduction in tumor volume, activation of the cytotoxic immune response, and increased SPMs [8,41,42]. A fourth study employed a mouse model of systemic sclerosis to test the same active stretching methodology. The animals were stretched once a day for 10 minutes/5 days a week for 4 weeks. Using ultrasound, they found, compared to control, stretching reduced skin thickness, increased subcutaneous tissue mobility, and downregulated genes related to systemic sclerosis in mice (Ccl2 and Adam8) [43]. Nonetheless, this model of autoimmune inflammation did not show evidence that stretching attenuated inflammation. Across these four studies, 10 minutes of active stretching was enough to trigger localized inflammatory changes [8,[41][42][43].
Five studies used a model of passive stretching delivered by different skin-stretching devices. The first study used an external device, allowing them to control for stretching duration and strain. Results showed that hair stem cells proliferate and hair regenerates when Quantify the acute phase myofiber response in young and old rats exposed to an acute bout of SSCs. • Increase in the volume density and average thickness of degenerating myofibers over time in the muscle exposed to SSCs that was greater than in muscle exposed to isometric contractions at 24, 48, and 72 h postexposure.
Baker et al. 2007 [31] Investigate the effect of repetitive SSC on rat skeletal muscle exposed in vivo using the left limb muscle groups.    applying between 33% and 40% of strain for at least 7 days. This study also reports that macrophages are first recruited by chemokines produced by stretch and then polarized to the M2 phenotype. Subsequently, growth factors released by these macrophages activate stem cell proliferation and hair regeneration [26,28]. Likewise, another skin model surgically implanted a  silicone stretching expander to promote allogeneic luciferase-mesenchymal stem cells (Luc-MSCs) migration. Results suggested that the inflammatory process caused by stretching is responsible for the chemoattraction of MSCs [28]. By comparison, three passive stretching studies reported a pro-inflammatory effect of stretching. Two studies employed an established murine hypertrophic scar model using an external skin-stretching device for ten days, applying tension (0.27 until 0.96 N/mm 2 ), which triggered fibroblast proliferation and inflammatory cell recruitment [27,40]. A third study used a subcutaneous minidilator stretching implanter injected with up to 4 mL of H 2 O (0.5 mL per day) to establish a mouse model of psoriasis. This procedure caused a prominent epidermal hyperproliferation, impairing skin barrier function, and upregulation of psoriasis-associated cytokines in epidermal keratinocytes [25]. Notably, all five studies mentioned the role of cytokines and chemokines in these inflammatory events [25][26][27][28]40]. However, only two studies indicated the relevance of monocytes infiltration and subsequent differentiation of macrophages (Table 4) [26,27].

Discussion
This systematic review summarizes the findings of experimentally controlled rodent studies employing mechanical stretching forces on the musculoskeletal and integumentary (i.e., skin and surrounding connective tissue) systems to evaluate its potential impact on inflammatory processes. To our knowledge, this is the first systematic review synthesizing the scientific body of literature on the impact of injurious and therapeutic stretching on inflammatory responses using experimental in vivo rodent models. The majority of identified studies focused on the impact of passive and shorter-term injurious stretching on muscle. Studies consistently controlled three key stretching parameters--stretching intensity, duration, and frequency--but not overall whole-body posture. A smaller number of studies evaluated the impact of active and passive therapeutic stretching on the integumentary system, with some attention to stretching intensity, duration, and frequency, but not body posture. Studies evaluating either the injurious and therapeutic effects of stretching reported a diversity of inflammatory outcomes, including macroscopic, histopathologic, genetic, immune cells sorting, cytokines, and protective lipids (e.g., resolvins). Few studies evaluated a long course of stretching (i.e., multiple sessions repeated over a period of weeks or months). Below, we discuss the potential translational relevance of these rodent model findings to human studies. We highlight how addressing current gaps in the pre-clinical in vivo literature might further inform clinical research on yogic stretching. We also discuss how a coordinated bi-directional translational research strategy, including both rodent and human studies, may be an effective approach for studying the effect of yogic stretching on inflammation and health.

Current knowledge gaps between the isolated physical component of yoga (i.e., yogic stretching) and its interaction with systemic inflammatory, regenerative and remodeling processes
Yogic stretching primarily happens within the context of specific postures (called asanas) and during transitions between postures [48,49]. The impact of yogic stretching on specific body tissues can differ widely depending on the yoga style, teaching methods, the practitioner's experience and health status, and the duration, frequency, pace, intensity, sequence arrangement, and individual anatomical characteristics during each body posture [50]. These aspects of yogic stretching have received little to no research attention, hence the motivation of this review to mine rodent model stretching research. One set of key physiological events believed to underlie the therapeutic effects of yogic stretching and conventional exercises are local and systemic immune-mediated processes, including inflammation, tissue remodeling, and regeneration occurring mainly in the myofascial and integumentary systems [51]. Recent systematic reviews evaluating the impact of yoga on inflammatory outcome suggest a link between yoga practice and inflammatory processes, however, these findings need to be interpreted carefully as the specific impact of the biomechanical processes associated with yogic stretching is typically embedded within a complex multimodal activity, including focused mental attention, breathing, imagery, and psychosocial interactions (e.g., group practice) which might also impact inflammation via alternative physiological pathways [7].
One systematic review summarized results from 15 studies assessing the long-term effects of yoga on inflammatory markers among healthy and/or disease individuals. Need a bit more here: One study of healthy participants found significant reduction in stimulated levels of IL-6, TNF-α, and IL-1β in ex vivo cultured blood in yoga group compared to controls [52]. However, another study of individuals reporting psychological distress found no significant reduction of IL-6, TNF-α, and CRP post intervention or at follow-up [53]. Overall, results showed no general agreement on the effect of yoga on levels of C-reactive protein (CRP), TNF-α, and IL-6 [54]. This should not be surprising given the heterogeneity of conditions, interventions, and time frames evaluated. Another earlier systematic review of randomized control trials (RCT) on the effects of yoga on stress and mood reported similar findings with respect to inflammatory outcomes [55].
The available clinical research literature on yoga is limited in its ability to inform the specific effects of stretching on inflammatory biomarkers and clinical outcomes related to function and health [56]. This evidence gap highlights the value of summarizing findings related to design features and inflammatory outcomes in rodent stretching studies and their possible relevance to translational yogic stretching research. It is important to highlight that in vivo rodent stretching studies included in this review do not provide interventions that mimic the complexities of yoga practice. However, in both rodent stretching and human yoga studies, stretching events occur at the integumentary and myofascial systems, likely triggering inflammatory, regenerative, and remodeling processes. For these reasons we believe it is helpful to consider these studies together, and as part of a potential bi-directional translational approach to understanding the impact of yogic stretching on inflammation and health.

Key protocol features of stretching paradigms
Passive versus active stretching. While yoga commonly utilizes both passive and active stretching, rodent studies identified for this review primarily employed passive stretching paradigms (89%; n = 22). To date, an agreed upon terminology for the types of stretching that spans humans and animals does not exist. In this systematic review, most studies considered active stretching as a dynamic activity in which the experimental rats or mice are trained and acclimated to be manipulated without any anesthetic protocol [42,57]. In contrast, passive stretching in rodent studies included: static mechanical forces employed either without the use of anesthetics (e.g., via an implanted stretching apparatus) or stretching while animals were anesthetized employing dynamic mechanical forces, for example during stretch-shortening contractions/cycles (i.e., SSC protocols including both concentric and eccentric stretching muscle contractions) [25,40,41] Stretching terminology used in human studies is primarily derived from the field of sports medicine and differs in meaningful ways [58]. For example, using sports medicine terminology, yogic stretching includes passive or static-passive stretching where the posture is held-for elongation-with support from some other part of the body or with the assistance of a partner or some other apparatus (i.e., props) between 10 seconds and less than one minute [59]. However, this passive stretching differs from rodent studies because humans purposely cooperate and are receptive (i.e., they try to relax the targeted area). Yogic stretching also includes 'static' or 'active-static' stretching (i.e., the body assumes a position and then holds it with the support of surrounding tissues, including the agonist muscles' strength) [2]. Furthermore, yogic stretching also utilizes dynamic stretching during flow sequences, which involve moving parts of the body and gradually increasing reach, speed of movement, or both [60]. Inside this dynamic category, specific yogic stretching sequences can be considered ballistic, active, resistance, and loaded stretching modalities, all occurring, for example, during a widely used sequence call the "sun salutation." A growing body of research in physical therapy and sports medicine supports that both passive and active stretching in humans show clinical benefits (e.g., enhancing function in multiple musculoskeletal pain conditions) [61][62][63]. Less research has evaluated the impact of these stretching techniques on inflammatory processes [28].
There are several advantages to the use of passive stretching protocols in rodent studies. They provide more precise control over several aspects of the stretching procedure, such as stretching intensity, duration, and specific localization, and eliminate the role of top-down cognitive/affective contributions to stretching. Several studies identified in this review performed passive stretching under anesthesia to eliminate variability in forces, as well as cognitive /affective contribution [25,40,41]. For example, studies by Baker et al. precisely applied different intensity and frequency of stretch-shortening contractions (SSCs) to deliver an injurious stretching intervention in muscles, which led to muscle inflammation and myofibers degeneration, mainly in older rats [29][30][31][32]. Others, such as Chu et al. used an external device to stretch the skin and surrounding connective tissue and found that stretching promotes hair stem cell proliferation and regeneration, driven by macrophages recruited by inflammatory chemokines and M2 polarization [26]. However, it should be noted that findings from passive rodent stretching models obviously do not reflect the impact of the more complex yogic stretching, which through the use of precise overall body postures, integrates forces from other body segments. To better inform yoga-related research, future passive stretching animal studies should include stretching protocols with postures that simulate whole body yogic postures, such as the those used in active stretching rats, mice and pigs paradigms [8,41,57]. These studies should also give more consideration to control for potential animal's stresses associated with handling and/or anesthesia, and/or provide estimates of the magnitude of these stressors [26,33,40].
A small proportion of studies in this review (14%, n = 4) utilized active stretching techniques and the majority of these evaluated therapeutic effects on inflammatory processes. For example, Berrueta et al., Corey et al., and Xiong et al. actively trained mice and rats to be lifted by the tail to engage in whole body stretching, thus including different muscle groups, skin, and connective tissue [8,[41][42][43]. This experimental stretching protocol partially simulates postures used in yogic stretching, such as backbends and inversions [64]. These rodent studies used a Carrageenan model of subcutaneous inflammation and found that stretching reduces lesion size, neutrophil recruitment, and increased secretion of protective lipids (e.g., resolvins). More recently, Vergara et al. adopted a wheelbarrow active stretching paradigm in a porcine model, using a similar carrageenan-subcutaneous inflammation model. They reported that pigs learned and tolerated the active stretching procedure well and that compared to a control group, the average lesion area was significantly smaller in the stretching group [57]. Collectively, these studies suggest some form of active stretching, partially mimicking yogic stretching, could impact inflammatory, remodeling and regenerative processes commonly present in different inflammatory diseases or after a programmed tissue injury (e.g., surgery) [8].
Compared with passive stretching, rodent studies using active stretching paradigms are more likely to inform stretching that takes place during yoga training. As in passive stretching, future animal research should develop novel active stretching paradigms that better simulate the diversity of postures typically encountered in yogic stretching. Consideration should also be given to the stresses introduced to animals due to human contact during active stretching paradigms [42,43,57]. A possible alternative to manually imposed animal stretching would be to develop models of active self-stretching. For example, the magnitude, intensity, and frequency of active whole-body stretching could be achieved by using physical structures introduced to the animal's environment to promote specific postures and movements (e.g., tubemazes and running wheels of different dimensions, strategic placement of water, or hidden food treats) [65,66]. Wearable devices could also be deployed. For example, Langevin et al. and Bishop et al. utilized harnesses that systematically controlled the magnitudes of trunk and limb mobility in pigs to study wound healing after a fascia injury in the dorsal truck at the L3-4 vertebral level [67,68]. Parallel studies could be designed in humans, with an aim to either minimize or enhance short-term mobility (e.g., kinesio taping-a technique widely used in sports medicine to restrict movements around joints and soft tissues) [69].
Stretching parameters (intensity, duration, frequency, and posture) and stretching protocol duration (shorter-vs. longer-term). In addition to the active vs. passive nature of stretching, key parameters such as intensity, duration, frequency, and overall body posture are likely to influence to the impact of stretching. Clinical yoga research protocols have begun to systematically control the types of asanas (i.e., postures) used, the postures' holding duration, and the practice frequency to inform therapeutic effect [70]. However, clinical studies have not yet systematically controlled the intensity of stretching or the impact of specific postures on local or systematic biological or clinical outcomes (e.g., inflammation and chronic pain) [17].
The rodent studies included in this review report a variety of experimentally controlled stretching protocols accounting for three key parameters: intensity, duration, and frequency-parameters that are appealing because of their quantitative nature [17]. The majority of these studies employed shorter-term stretching interventions with a focus on injury. For example, with respect to an injurious intervention, Baker et al. precisely controlled three stretching parameters during the stretch-shortening contractions (SSCs) of the tibialis anterior muscle in rats [29][30][31][32]. As an example of a therapeutic intervention, Pizza et al. controlled intensity (20% amplitude/strain), duration (each set of stretching included 75 repetitions), and frequency (1 or 2 sets separated by 2-weeks) and reported that acute isometric contractions or passive stretches elevated neutrophils, whereas a longer-term intervention, separated by two weeks, showed a therapeutic effect by reducing neutrophil infiltration [37].
Ten of 25 studies employed carefully controlled longer-term stretching protocols. Nine of these stretched the integumentary system with a focus on both therapeutic and injurious outcomes. For example, Berrueta et al. actively stretched female mice 10 minutes (i.e., duration) daily for four weeks (i.e., frequency) using the previously mentioned tail lifting paradigm. They reported tumor volume reduction (52%), higher cytotoxic immune response, and elevated levels of protective lipids (i.e., resolvins) in the stretch compared to the no-stretch group [8]. Of note, Berrueta et al., Corey et al., and Xiong et al. controlled two stretching parameters (duration and frequency) quantitatively. However, posture was only controlled qualitatively, and they did not control for stretching intensity [8,[41][42][43]. The remaining longer-term stretching studies did not stretch the whole animal body, targeting only a portion of their skin and surrounding connective tissue. For example, Quiao et al. used subcutaneous devices for continuous and static skin stretching (i.e., duration) for eight days to explore its injurious effects. They only controlled for stretching duration and frequency, reporting epidermal hyperproliferation, impaired skin barrier function, and up-regulation of psoriasis-associated cytokines in epidermal keratinocytes [25]. Although studies using skin stretching devices do not directly link to yoga interventions, these studies offer insights into basic mechanisms associated with mechanotransduction, wound healing, and tissue regeneration processes, which might more broadly inform fundamental biological mechanisms associated with yogic stretching [26,71].
Regarding the posture parameter, the ability to adopt precise body posture during yogic stretching practice is crucial [72]. Of note, none of the rodent studies included in this systematic review objectively accounted for overall whole-body posture during stretching of body segments. A few studies provided a qualitative explanation of the animals' position. For example, Brickson et al. placed the rats lying side-wise to manipulate the left hind limb [33].
Future animal studies should emphasize longer-term interventions, their potential therapeutic effects and assume better control of stretching intensity, especially with protocols stretching the whole animal's body since none of the studies included here designed a model with these components [8,[41][42][43]. Laboratory-based studies could utilize widely available kinematics and motion detection technology to objectively and quantitatively characterize posture and magnitudes of tissue displacement [73]. These technologies would also permit the evaluation of how different stretching postures impact the activity of specific muscle groups and tensile properties of connective tissue [74]. Parallel experimental human studies could rely on wearable sensors or kinematic systems capable of measuring tissue displacement, joint angles, and symmetry, as well as employ protocols that mechanically and systematically deliver a more precise magnitude of forces targeting specific body parts (e.g., Cox chiropractic table to control for flexion/distraction in the thoracolumbar region) [75].
Injurious versus therapeutic stretching. While the primary focus of clinical yoga research to date has been to evaluate its therapeutic effects [55], a growing body of studies has begun to evaluate its injurious side effects systematically [76]. Specific postures, such as hand-, shoulder-and headstands, have been associated with adverse events, with the most commonly reported being muscle or joint pain, soreness, and strains. Adverse events appear to be higher among participants with chronic diseases and self-study practitioners without specialized supervision [77]. These injuries are typically minor and affect mainly muscle groups and the surrounding connective tissue, including fascia [76,78]. Findings from injurious stretching interventions in animal studies may inform strategies for minimizing such adverse events by improving our understanding of the systemic and localized effects of yogic stretching-related injuries in humans [79,80]. Currently, muscle injurious animal studies identified in this review were motivated by stretching issues specific to sports medicine and rheumatology. They typically studied the impact of passive elongation and stretch-shortening cycles (SSCs) on muscle injury. For example, Rader et al. used an injurious SSCs protocol to compare muscle fiber changes in young and old rats, and Ramos et al. used passive elongation in an experimental rat model of muscle strain [20,38]. These controlled muscle studies offer several advantages. First, they allow evaluation of tissue adaptation using microscopic techniques, which are less feasible in human studies. Second, they enable ex-vivo or in-vitro studies to focus on the effect of stretching on specific tissues or cell types. Third, they offer the possibility for controlling different stretching parameters (i.e., duration, frequency, and intensity). Of note, no injurious studies evaluated the impact of stretching on inflammation resolution, a key factor in injury recovery. Future animal studies could be designed to focus on the most common types of injuries observed in yoga clinical trials.
Only seven of the 25 studies included in this review specifically evaluated the potential therapeutic effect of stretching [8,26,28,37,[41][42][43]. In all cases, stretching regimens indicated positive effects with respect to inflammation-related outcomes. Of those seven studies, four studies utilized active stretching [8,[41][42][43], three evaluated passive stretching protocols [26,28,37], and six out of seven utilized long-term interventions [8,26,28,[41][42][43]. The tissues targeted included muscle, skin, and connective tissue. The inflammatory outcomes ranged from immune cell infiltration, cytokines, resolvins, gene expression, and lesion size (Tables 3 and 4). The small number of studies and the heterogeneity across them make it premature to draw general conclusions about the therapeutic effect of stretching on inflammation-related outcomes. However, it is noteworthy that studies employing the whole body in active stretching protocols for an extended period of time, showed a positive effect. More generally, these studies support the use of rodent models to simulate some aspects of yoga-like stretching.
Hypothesized mechanisms underlying observed therapeutic effects varied with study design. For example, Pizza et al. suggested that after stretching, inflammatory events afford protection from contraction-induced muscle injury through inflammatory protective mechanisms (e.g., the relationship between stretching and immune cell infiltration) [37]. Chu et al. suggested that precise control of stretching parameters is involved in hair regeneration through the alternative activation of macrophages [26]. Zhou et al. suggested that the therapeutic effect of mechanical stretching on the skin and connective tissue can induce the release of specific cytokines that facilitate stem cell migration involved in tissue repair and regeneration [28]. Compared with human theories that suggest stretching elicits its benefits by promoting proper muscle and connective tissue function [81], these rodent studies targeted specific biological extracellular pathways. Therefore, these in vivo studies further support the value of bidirectional research on unveiling similar mechanisms among humans. Future animal studies might expand upon investigations evaluating stretching protocols with therapeutic intention. Whereas in human studies, there is a need for well controlled studies evaluating the impact of yogic-stretching on inflammatory processes, including those associated with acute and chronic musculoskeletal pain, to inform rehabilitation and prevention. One specific opportunity would be to evaluate the therapeutic benefits of stretching delivered perioperatively for surgical procedures, to test if stretching can reduce the risk of postsurgical chronic pain and long-term use of opioids [82]. Provocative preliminary results indicating that stretching can reduce the rates of tumor growth also suggest that well designed human stretching studies should go beyond studying the impact of yogic stretching on symptoms and quality of life in cancer patient [83], and begin to explore mechanotransduction processes underlying cancer progression [54].
Inflammatory outcomes used in rodent stretching studies. Studies included in this review evaluated a wide range of inflammatory outcomes. The most consistent inflammatory outcomes among the injurious muscle studies were histopathological changes (i.e., plasma extravasation, edema, necrosis, and myofibers degeneration), confirming the acute injurious effect of stretching [20][21][22]33,[35][36][37][38][39]. Some exceptions include Gluck et al. [24], who studied histopathological changes after stretching the nerve innervating specific muscles, and Pizza et al. [37], who utilized immunohistochemistry (IHC) to quantify histopathological changes in neutrophil and macrophage populations during shorter-or longer-term stretching protocols. Of note, this group of studies did not describe histological changes in the connective tissue and fascia surrounding muscle groups. While histopathology is a standard laboratory technique used in basic science, its translation into a clinical trial is less common and feasible. Instead, clinical researchers used other mechanisms to explore muscle changes, such as ultrasound, dynamometers, and electromyographic activity (EMG) [84].
Fewer rodent studies reported inflammatory outcomes using inflammatory markers (e.g., cytokines), immune cell infiltration, and gene expression within the muscle [20,35,38]. As noted above, there was no mention of protective lipids measurements (e.g., resolvins). Noninvasive outcomes, similar to those used in clinical studies, such as dynamic ultrasound imaging, were not used in animal studies to evaluate age-related changes in myofibers, connective tissue, and fascia after stretching [85,86].
To date, no yoga studies have explored the local effect of isolated stretching on inflammatory processes using any of the techniques mentioned above. Clinical studies exploring the effect of yoga on systemic levels of cytokines have found evidence, for example, in decreased levels of IL-6 and TNF-α and increase adiponectin [87]. Of note, these findings are different from a large body of sports studies suggesting acute exercise results in a transient increase of IL-6, IL-10, IL-8, IL-18, and IL-1ra [88].
In the subset of rodent studies, in which stretching targeted the integumentary system, 7 out of 9 studies performed histopathological, immunofluorescence, or IHC analysis to describe microscopic changes in the skin subjected to longer-term protocols of stretching [25][26][27][28]40,42,43]. As noted above in studies targeting muscle, the use of these techniques in clinical trials poses challenges; however, in vitro studies using resident human dermal fibroblast for stretching could be one way to overcome the limitations of histopathology [89]. One critical need to inform the translation between animal and human studies is the development of human experimental models to evaluate the local impact of stretching at the site of injuries or lesions. One example of such an approach would be to intradermally or subcutaneously inject a low dose of endotoxin (i.e., lipopolysaccharide-LPS) to trigger a focal inflammatory process, in which a precise skin stretching modality could be applied. Later, microbiopsies might help unravel changes in the subcutaneous stroma (i.e., collagen, fibroblast, and other stromal cells).
These models could inform surgeons about the future repair and regenerative processes involve after surgery.
Another approach to quantify the effect of stretching on inflammation resolution was introduced by a subgroup of rodent studies using active stretching paradigms to study lesion size, measured by ultrasound, or tumor volume using a caliper, in rats and mice, respectively. These studies, after animal euthanasia, also processed different biological samples for histology, gene expression, cell sorting, and protective lipids detection (lipidomic and ELISA) [8,[41][42][43]. The feasibility of using non-invasive assessment of inflammation, such as high-resolution ultrasound, to evaluate stretching's local effect is appealing. A study lead by Ellis et al. employed ultrasound elastographic measurements to assess the feasibility and reliability of this technology to quantify shear strain at the sciatic nerve-hamstring muscle interface during active and passive knee extension-flexion movements in healthy people [90]. Although they did not access inflammation, this model could be extended towards musculoskeletal chronic inflammatory pain conditions. Langevin et al. have used ultrasound to evaluate the effect of stretching in a porcine model of thoracolumbar fascia movement restriction [67,68].
Animal species. The rat paradigm was the primary model used to examine the effect of stretching on muscle or muscle innervation. Conversely, those studies stretching the skin/connective tissue relied on the mouse model because it allows them to increase the number of animals per group and use continuous static stretching devices underneath or on top of the skin [8,25,26,40,43,91]. Animal studies using models that closely resemble human anatomy (e.g., pigs, dogs, primates) are substantially more expensive and complex to execute [92]. These animal models might allow active self-stretching paradigms by proper training, eliminating significant stressors, such as the anesthetic protocol and human handling. Recent studies evaluating stretching in a pig model provide examples in which active stretching was used to evaluate changes in the thoracolumbar fascia, along with functional changes in fascia mobility and gait speed [57, 67,68].

Strengths and limitations
This study has a number of noteworthy methodological strengths. First, it followed the Systematic Review Centre for Laboratory Animal Experimentation guidelines (i.e., the S1 Protocol format for systematic reviews of animal intervention studies). Second, data extraction and methodological assessment involved multiple reviewers. Third, it assessed the quality of each study using the SYRCLE risk of bias tool [15,16].
This review also has multiple limitations. First, it did not include ex vivo or in vitro studies exploring changes in cell proliferation, elongation, migration, apoptosis, or necrosis due to stretching [89]. These studies have advantages over the in vivo models because they allow even more control and precision during the stretching intervention (i.e., duration, frequency, and intensity). However, ex vivo or in vitro studies' relevance to informing the design of clinical studies evaluating yogic-like stretching is quite limited because the natural microenvironment in which tissues and cells are located in the whole animal body is disrupted. Second, while the search strategy for this review employed a range of terms associated with stretching (e.g., active stretching, passive stretching, stretch injury, mechanical stretch), this strategy may have excluded studies evaluating the impact of mechanical forces induced by only eccentric contractions, mobilization techniques, manual physical therapy, massage, exercise or longer-term repetitive strain injury impact inflammatory outcomes [74,93]. Third, this systematic review did not include search terms targeting delayed onset muscle soreness (DOMS), which may have excluded studies evaluating the impact of physical stretching on muscle soreness and inflammation using rodent models [94]. Fourth, this systematic review only included studies written in English. Finally, while this study largely adhered to the S1 Protocol format for systematic reviews of animal intervention studies and SYRCLE risk of bias tool, the final search strategy was not approved by a professional librarian's.
Supporting information S1