A randomized controlled pilot study of Yoga Skills Training versus an attention control delivered during chemotherapy administration

Abstract

Context:

It is important to address fatigue and co-occurring symptoms during chemotherapy to preserve quality of life in patients with gastrointestinal (GI) cancer.

Objective:

To conduct a randomized controlled pilot study of a Yoga Skills Training (YST) intervention compared to an attention control (AC) among adults diagnosed with GI cancer.

Methods:

YST consisted of four 30-minute sessions delivered individually during chemotherapy plus home practice. AC provided empathic attention plus home diaries. Patient-reported (PROMIS T-score) assessments of fatigue, depressive symptoms, sleep disturbances, and psychological stress (Perceived Stress Scale) were collected at chemotherapy visits: baseline, Week 8, Week 10 and Week 14, and analyzed using a mixed effects model. Inflammatory cytokines were assessed at baseline and Week 10.

Results:

Forty-four of 77 adults approached agreed to participate (57%; YST n=23; AC n=21). Participants’ mean age was 58 years and 48% were men. Participants randomized to YST reported a larger decline in fatigue (−2.4 difference, d=0.30) and depressive symptoms (−2.5 difference, d=0.30) than AC participants from baseline to Week 10 and sleep disturbances at Week 8 (−3.9 difference, d=0.50). Differences in magnitude of change in symptoms were consistent with or exceeded a minimally important difference. Psychological stress decreased more in the AC at Week 10 (d=0.30). Reductions in inflammatory cytokines (IL-6, sTNF R1) were larger in the YST group than AC.

Conclusions:

YST showed promise for improving fatigue, depressive symptoms, sleep disturbances, and inflammation. YST is also feasible and reaches patients underrepresented in yoga research (i.e., GI cancer, men), thus warranting further examination.

Introduction

Gastrointestinal (GI) cancers are among the most prevalent cancers for both men and women (over 330,000 diagnosed in the United States each year), and the prevalence of GI cancer is increasing (1). Fatigue is reported by a majority (>50%; 2,3) of patients with GI cancer, co-occurs with other common symptoms such as depression (>30%; 4,5), and is associated with a decline in post-treatment quality of life (6). Further, although chemotherapy is efficacious for improving survival and quality of life in some contexts, it leads to increases in fatigue and depressive symptoms (7) that can persist even five years after the completion of treatment (8,9).

Fatigue and depressive symptoms experienced during treatment are associated with a continued experience of these symptoms (10). Thus, it is important to address elevations of fatigue and commonly co-occurring depressive symptoms during chemotherapy to improve overall quality of life in patients with GI cancer. A biobehavioral framework suggests that fatigue and depressive symptoms tend to co-occur because they share underlying mechanisms including psychological stress, sleep disturbance and inflammation (11,12). Increases in inflammation (i.e., sTNF-R1, IL-6) during treatment has been related to changes in fatigue and other symptoms (3). Despite high symptom burden in GI cancer patients, there are few behavioral interventions designed for these patients that proactively address emerging fatigue and depressive symptoms during chemotherapy (13,14).

Yoga is already recommended in clinical guidelines for managing cancer-related fatigue and depressive symptoms during treatment (15,16). Evidence also supports the biobehavioral framework such that yoga has been shown to reduce psychological stress, sleep disturbances, and inflammation (e.g., IL-6, TNF-α) (17–19). Yet, data showing efficacy of yoga for cancer-related symptoms is primarily from studies in female breast cancer patients that required group-class attendance, which creates barriers to participation (scheduling, transportation, too sick; 18). One study of yoga classes specifically for patients with GI cancer found low enrollment, low intervention adherence, and high attrition rates due partially to scheduling challenges (20). Another methodological limitation of current yoga research is that most studies lack an active control group, raising the possibility that intervention effects may be due to nonspecific factors (18).

To address these limitations, we piloted a yoga intervention adapted to be delivered individually in the chair during the time GI patients were in the clinic for chemotherapy administration. This intervention was designed to reduce barriers to participation and broaden the reach of yoga. The primary objective of this randomized controlled pilot study was to assess the preliminary efficacy of a Yoga Skills Training (YST) through estimating if there was at least a minimally important difference (MID) between YST compared to an attention control (AC). The primary outcome was fatigue; the secondary outcome was depressive symptoms; and proposed mediators were psychological stress, sleep disturbances, and inflammation. We employed a contemporary approach to quantify outcomes (e.g., estimates of variance) to inform a future efficacy a trial as opposed to a traditional approach to reporting underpowered hypothesis tests (21). We also assessed study feasibility and intervention acceptability to inform future research.

Methods

Participants and Trial Design

Participants were recruited from the Vanderbilt Ingram Cancer Center (VICC) and Wake Forest Baptist Comprehensive Cancer Center (WFBCCC). The respective Institutional Review Boards (IRBs) approved study procedures (IRB00038321). Adults (≥21 years of age) who were scheduled to begin first-line Fluorouracil (5FU)-based intravenous chemotherapy treatment for a GI cancer (stages II-IV), had an Eastern Cooperative Oncology Group performance status of ≤1, and were able to understand and willing to sign an informed consent document in English were eligible. Patients were excluded if they had a self-reported history of diagnosed sleep disorders (e.g., obstructive sleep apnea, insomnia), comorbidities associated with poor sleep or fatigue (e.g., chronic fatigue syndrome), or a job with night shifts. This trial was registered (NCT02489422).

To identify potential patients, a study team member communicated with healthcare providers. Participants were told that the study was called “Programs to Support You During Chemotherapy (Pro-You)” and that both treatment arms offered programs to support their experiences coping with chemotherapy. Participants were blinded to their randomization assignment and information about the complete study design was sent to participants after completion of data collection. Eligible, consented, and enrolled participants were randomized to YST or AC after patient-reported baseline assessments. Randomization occurred stratified by both cancer type (colon, rectal/other better prognosis GI cancers, poorer prognosis cancers) and stage (localized stages II/III or metastatic stage IV). Randomization (1:1) lists within each strata were generated via a permuted block program developed and executed by the study biostatistician with blocks of random sizes between 2 and 6. The randomization lists were held by a study team member (not involved with data collection) who allocated each participant to study group. Once assigned, details about that arm were revealed to a different team member who assigned participants to interventions. The team member who collected primary outcome data was blinded to group assignment.

To maximize generalizability, eligibility criteria were expanded after enrollment began from only patients with colorectal cancer to all patients who had gastrointestinal cancer and were receiving Fluorouracil (5FU)-based intravenous chemotherapy. Accordingly, prognosis (good or poor) was added as a strata to the randomization table.

Interventions

Active Intervention.

The Yoga Skills Training (YST) intervention was administered individually in the outpatient clinic. Patients were encouraged to also practice YST daily at home. First-line chemotherapy for GI cancers was generally given during biweekly visits and participants completed baseline at the time of their first chemotherapy visit (week 0). YST was adapted for the current study as detailed in another publication (22). Briefly, YST consisted of four 30 minute in-person sessions (weeks 2, 4, 6, 8) that taught the following skills: awareness, movement, breathing practices, and meditation. Participants were also given a 16-minute audio recording of the YST and devices to play the recording. All yoga instructors (4 total, one primary at each site with backups including S.J.S.) had completed accredited yoga teacher training programs and were experienced teaching patients with chronic illness.

Attention Control (AC).

An empathic attention control group was employed to account for time, added attention, and efficacy expectations of the YST, as used in previous studies (23). The location, number of sessions, recommended frequency of home practice, and home assessments matched YST. The AC was delivered by research staff who were trained to create and maintain a relationship by using techniques of active listening, reflection of statements, and avoiding negative judgments. They utilized standardized instructions to prompt the patient: “What were some of the events or circumstances that affected you in the past two weeks?” The instructions for daily diary entries were the same with reference to “the past day?” Participants were asked to “Think back over the past day and write down up to five events that had an impact on you”(24).

Treatment Fidelity.

A number of steps were taken to ensure treatment fidelity (25,26). The study team trained interventionists during an initial session and ongoing monthly meetings. Interventionists completed a written test to confirm understanding, documented each in-person session with a checklist, and were video recorded during all sessions with 10% of the recordings randomly selected for review by S.J.S. (22). Each interventionist noted any adverse events experienced during sessions. In addition, interventionists asked participants about any problems during home practices.

Measures

Patient-reported assessments of fatigue, depressive symptoms, and potential mediators were collected in person at four time points that coincided with biweekly chemotherapy or other clinic appointments (baseline, before the final intervention on Week 8, Week 10, and Week 14). We determined a priori that fatigue at Week 10 was the primary outcome. Inflammatory cytokines were measured in EDTA plasma samples collected along with patients’ regularly scheduled blood draws at weeks 2 and 10. Additional variables were documented to assess study feasibility (recruitment, retention), intervention acceptability, and sample characteristics.

Patient Reported Assessments.

Fatigue was assessed with the 7-item measure of Fatigue from the Patient-Reported Outcomes Measurement Information System (PROMIS)-Cancer (item responses: 1=never, 5=always over the past 7 days; 27). Items are summed converted to a standardized T-score (28). Higher T-scores on this and all PROMIS measures indicate more of the construct measured. Depressive Symptoms was assessed with the 10-item PROMIS-Cancer Depressive symptoms short-form (28). We measured Psychological Stress with the 10-item Perceived Stress Scale (PSS) and higher scores indicated more stress (29). Sleep Disturbance was assessed with the 8-item PROMIS sleep disturbance short-form (30).

Blood draws (Inflammation).

Inflammatory cytokines (IL-6, sTNF-R1, TNF-α) that have been associated with fatigue during treatment for gastrointestinal cancer (3) or influenced by yoga practice (19) were measured in plasma extracted from whole blood drawn during regularly scheduled clinical blood draws. IL-6 and TNF-α were measured using high-sensitivity Quantikine® immunoassay kits. The inter-assay and intra-assay CVs for IL-6 are 6.5% and 4.1% respectively, and for TNF-α are 6.5% and 2.0%, respectively. sTNF-R1 was measured using standard Quantikine® kits. All samples were assayed in duplicate and the average of the two measures were used for data analyses. Inter-assay variability in this internal control was low (IL-6 CV=3.1%, sTNF-R1 CV=4.2%, TNF-α CV=5.4%), indicating that cytokine levels can be measured reliably over time.

Intervention Acceptability. Intervention Adherence.

High protocol adherence was defined by completing 75% of the planned in-person sessions and ≥3 days of home practice per week. Throughout the intervention, all participants were asked to keep daily paper logs of home practice (weeks 3-8; 31). In addition, one item on home practice that day was sent daily via an automated survey for two weeks after the intervention (weeks 9-10; 32). For daily reports of home practice missing values were treated as zero and reported for those who completed at least one report and for those missing all reports. Intervention Satisfaction was assessed with 3 items (0 not at all to 4 very much).

Demographics and Clinical Data.

Age, gender, race, ethnicity, relationship status, education level, income, comorbidities, use of medication to control symptoms, ability to pay monthly bills, willingness to participate in a group setting, use of psychological services, and use of yoga in the last 2 weeks was self-reported at baseline. Cancer type, stage of disease, recurrence status, receipt of radiation, height, and weight were abstracted from medical charts.

Statistical Analyses

Sample size justification was based on informing standard deviation (SD) estimates for a subsequent larger study (21). With a sample size of 44, the probability is 0.80 that the relative error in the estimate of the standard deviation (SD) will be less than 10%. Data were cleaned and databases locked prior to data analysis by arm. Feasibility statistics were estimated with proportions and 95% confidence intervals. Participant characteristics, intervention satisfaction, and adherence were summarized. Enrolled participants were compared to those who declined using chi-square tests and t-tests. Key statistical tests involved between group differences of the changes from respective baseline values of the PROMIS fatigue, depressive symptoms, and sleep disturbance t-scores and psychological stress. These tests were evaluated using mixed-effects generalized linear modeling procedures with week (categorical), arm, week by arm interaction, and were adjusted for randomization strata. Differences by arm in change from baseline were estimated using linear contrasts. Estimates of MIDs for PROMIS measures of fatigue, depressive symptoms and sleep range from 2.0-5.0 (28,33). The median MID estimate specific to longitudinal analyses of the primary outcome of Fatigue (7-items) in advanced cancer patients was a t-score of 2.4 and effect size of 0.29 (28). Guidelines for interpreting Cohen’s d indicate that a value larger than 0.2 represents a MID (34,35).

Inflammatory cytokine data were skewed and thus natural log transformed prior to analysis using the same mixed model framework. Due to the natural log transformation and back-transformation, the difference of change scores represent the ratio of the arms on the original scale at 10 weeks with adjustment for baseline (YST: AC). Exploratory analyses examined intervention dose (i.e., days of in-person attendance plus total days of home practice reported with values of 0 imputed for incomplete assessments) with outcomes using Pearson correlation. SAS (v9.4, Cary, NC) was used for statistical analysis.

Results

Sample Characteristics.

Enrollment occurred 9/2015 - 10/2019 (S.J.S. relocated mid study). We approached 77 eligible patients and enrolled 44 participants (YST n=23, AC n=21; Figure 1 study flow diagram), for an overall recruitment proportion of 57% (95% Confidence Interval [CI]: 46%, 68%). Ten participants were enrolled after expanding eligibility criteria (3 had poor prognosis cancers). Patients who declined participation and provided demographic information (n=24) did not differ from those enrolled by age, gender, ethnicity, or race (Supplemental Table 1). Frequent reasons for declining included not interested (n=16) or overwhelmed (n=11). Participants had a mean age of 58 years, a majority were White (86%), and had colon cancer (75%). Approximately half of participants were men (48%; Table 1). The study was stopped when the primary objective to determine preliminary efficacy was completed.

Figure 1.

Study flow diagram

Table 1.

Participant Characteristics

	Total Sample (N=44)		Yoga Skills Training (N=23)		Attention Control (N=21)
Characteristics	N	N%	N	N%	N	N%
Gender
Women	23	52%	13	57%	10	48%
Men	21	48%	10	43%	11	52%
Ethnicity
Hispanic or Latino	1	2%	0	0%	1	5%
Not Hispanic or Latino	43	98%	23	100%	20	95%
Race
Black or African American	6	14%	3	13%	3	14%
White	38	86%	20	87%	18	86%
Relationship
Currently married/living with partner	30	68%	18	78%	12	57%
Separated/divorced	5	11%	3	13%	2	10%
Widowed	5	11%	1	4%	4	19%
Never married	4	9%	1	4%	3	14%
Education
High school or equivalent	11	26%	6	26%	5	25%
Technical/vocational school/some college	14	33%	7	30%	7	35%
College graduate/post Graduate	18	42%	10	43%	8	40%
Income
Less than $10,000	5	12%	2	9%	3	15%
$10,000 to $39,000	10	24%	5	23%	5	25%
$40,000 to $59,000	9	21%	5	23%	4	20%
$60,000 to $100,000	13	31%	7	32%	6	30%
More than $100,000	5	12%	3	14%	2	10%
How difficult it is to pay monthly bills
Very difficult	3	7%	2	9%	1	5%
Somewhat difficult	13	31%	7	32%	6	30%
Not very difficult	17	40%	9	41%	8	40%
Not at all difficult	9	21%	4	18%	5	25%
Cancer/Disease Type
Colon	33	75%	17	74%	16	76%
Localized	19	43%	10	43%	9	43%
Metastatic	14	32%	7	30%	7	33%
Rectala	8	18%	4	17%	4	19%
Localized	3	7%	1	4%	2	10%
Metastatic	5	11%	3	13%	2	10%
Poorer prognosis cancers: Localized (i.e., 2 pancreatic, esophageal)	3	7%	2	9%	1	5%
Stage
II	6	14%	2	9%	4	20%
III	18	42%	11	48%	7	35%
IV	19	44%	10	43%	9	45%
Recurrent Disease b	5	11%	3	13%	2	10%
Current Radiation c	1	2%	0	0%	1	5%
Use of medication/supplement for symptoms d	19	48%	9	41%	10	56%
Use of psychological services in last 2 weeks b	3	7%	1	4%	2	10%
Use of spiritual/mind-body therapies in last 2 weeks b, e	27	63%	15	65%	12	60%
Willingness to participate in group setting b	11	26	7	30%	4	20
	Total Sample (N=44)		Yoga Skills Training (N=23)		Attention Control (N=21)
	Mean	SD	Mean	SD	Mean	SD
Age	58.5	11.7	60.8	11.2	56.0	12.0
Body Mass Index (BMI)	27.7 (n=43)	7.0	28.2	6.5	27.21 (n=20)	7.7
Number of Co-morbidities	1.7 (n=43)	2.0	1.3	1.1	2.2 (n=20)	2.7

^aThere were no participants diagnosed with “other better prognosis cancers” enrolled.

^bYST n=23; AC n=20

^cYST n=23; AC n=19

^dYST n=22; AC n=18

^ePrayer n=25; Meditation n=7; Yoga n=1; Other n=3

Outcomes.

Seventy-seven percent of participants completed the primary outcome assessment at Week 10, two weeks after the final YST session (95% CI: 65%, 90%; YST=87%; AC=67%). Common reasons for not completing follow-up assessments were passing away on study (n=4) or change in cancer treatment plan (n=4).

There was a larger change in the primary outcome of fatigue in the YST group than the AC group from baseline Week 10 (−2.4 t-score point difference, d=0.30, Table 2). This difference was consistent with a MID (28). Greater changes in fatigue in YST from baseline were lower than a MID, although in a consistent direction at Weeks 8 and Week 14. There was also a larger decline in the secondary outcome of depressive symptoms in the YST group compared to the AC group at Week 10 (−2.5 t-score point difference, d=0.30). The larger decline in depressive symptoms in the YST group was also evident at Week 8, although no longer present at Week 14. Changes in depressive symptoms were consistent with an estimated MID (28,33,35).

Table 2.

Description of study outcomes and mediators by group.

Measures	Baseline		Week 8		Week 8	Week 8	Week 10		Week 10	Week 10	Week 14		Week 14	Week 14
	Yoga Skills Training (YST)	Attention Control (AC)	YST	AC	Difference in YST and AC	Cohen’s d	YST	AC	Difference in YST and AC	Cohen’s d	YST	AC	Difference in YST and AC	Cohen’s d
	Mean (SE)a	Mean (SE)					Mean (SE)	Mean (SE)			Mean (SE)	Mean (SE)
	n = 23	n = 20			Change Scores (SE)				Change Scores (SE)				Change Scores (SE)
PROMIS Fatigue (Primary; Range 0-100) b	53.1 (18)	56.8 (19)	53.0 (2.1)	57.2 (2.2)	−0.6 (2.5)	−0.1	51.1 (2.2)	57.2 (2.4	−2.4 (2.6)	−0.3	55.4 (2.2)	60.6 (2.3)	−1.6 (3.1)	−0.2
PROMIS Depression (Secondary; Range 0-100)	49.9 (19)	51.4 (2.0)	47.4 (2.0)	51.2 (2.1)	−2.3 (2.5)	−0.3	46.6 (2.3)	50.5 (2.5)	−2.5 (2.6)	−0.3	49.1 (2.0)	50.2 (2.1)	0.3 (2.5)	0.0
PROMIS Sleep Disturbance (Range 0-100)	51.5 (1.9)	53.4 (1.9)	50.4 (2.3)	56.1 (2.4)	−3.9 (2.6)	−0.5	51.4 (2.1)	53.3 (2.4)	−0.0 (2.3)	0.0	52.3 (2.1)	51.8 (2.2)	2.3 (2.2)	0.3
Psychological Stress (Range 0-40)	14.7 (1.5), n=22	16.5 (1.6)	11.4 (1.8)	13.7 (1.9)	−0.5 (1.5)	−0.1	12.1 (2.0)	12.3 (2.1)	1.6 (1.5)	0.3	12.5 (1.8)	13.6 (1.9)	0.7 (1.6)	0.1
	Mean (CI) n=16	Mean (CI) n=13					Mean (CI)	Mean (CI)	Ratioc (CI)
IL-6	5.6 (3.4, 9.2)	4.6 (2.8, 7.8)			N/A		4.8 (3.0, 7.8)	6.9 (4.2, 11.3)	0.58 (0.38, 0.90)				N/A
TNF-α	1.6 (1.1, 2.1)	1.7 (1.2, 2.3)			N/A		1.7 (1.3, 2.3)	1.8 (1.3, 2.5)	1.01 (0.91, 1.13)				N/A
sTNF-R1	1633.8 (1222.1, 2184.1)	1727.5 (1281.6, 2328.6)			N/A		1742.6 (1324.4, 2292.8)	2121.1 (1605.2, 2802.6)	0.87 (0.78, 0.97)				N/A

^aAll mean values reported are least squares means from a mixed model that included arm, week, arm x week interaction, and strata.

^bPatient-Reported Outcomes Measurement Information System (PROMIS) measures are standardized (Mean=50). CI = 95% Confidence Interval, which is reported instead of standard error (SE) due to the skewed nature of cytokine data.

^cWeek 10 ratio is YST: AC with adjustment for baseline and strata on the original scale. Data were modeled on the natural log scale and back-transformed.

Mediators.

There were larger declines in sleep disturbances upon completion of the intervention at week 8 in YST as compared to AC (−3.9 t-score point difference, d=0.54), which exceeded estimates for a MID (28,33,35). The larger improvement in sleep disturbances on the day of completing the YST (Week 8) did not last and reversed such that sleep disturbances improved more in the AC group at Week 14. Changes in psychological stress decreased more in the AC group at Week 10 and did not differ by group at any other time point. Although the change score differed, the magnitude of psychological stress was similar in both groups at Week 10. The ratio of values at week 10 with adjustment for baseline showed that inflammatory cytokines IL-6 and sTNF-R1 were lower in the YST than AC group with confidence intervals that did not include 1 (with a ratio of 1 meaning that the groups were the same). TNF-α did not differ between groups.

Process Assessments.

Adherence to at least three of the four in-person facilitated sessions was 77% (95% CI: 65%, 90%; YST=78%, AC=76%). High adherence to home practice as defined a priori (≥3 days of home practice per week) was similar between groups and participants were more likely to complete electronic surveys (68%; weeks 9-10) than paper home practice logs (41%; weeks 3-8). In the YST group, 33.3% of the participants who completed surveys during weeks 3-8 and 31.2% during weeks 9-10 reported high adherence. In the AC group, 33.3% of participants during weeks 3-8 and 57.1% during weeks 9-10 reported practicing 3 or more days per week. Treatment adherence is further described in Table 3. There were moderate relationships between higher YST total home practice days and reductions in fatigue (r=−0.48, n=17) and depressive symptoms at Week 10 (r=−0.26, n=17), whereas total home practice days was not as highly associated with fatigue or depressive symptoms in the AC group (fatigue r=−0.03, n=11; depressive symptoms r=−0.16, n=12).

Table 3.

Intervention adherence to in person sessions and home practice

			YST		AC
Assessment Method	Range		Days		Days
		n	Mean (SD)	n	Mean (SD)
In Person Attendance	1-4 Possible	23	3.1 (1.3)	21	3.2 (1.3)
Paper Diaries (Weeks 3-8)	0-42 days	9	25.9 (14.1)	9	20.2 (17.7)
Electronic Diaries (weeks 9-10)	0-14 days	16	6.1 (3.5)	14	7.2 (5.7)
Total Home Practice (completed)	Average Days/Week	17	3.6 (1.7)	15	3.6 (2.2)
Total Home Practice (completed)	0-56 days	17	19.5 (16.7)	15	18.9 (17.5)
Total Home Practice (all participants)a	All 8 Weeks	23	14.4 (16.7)	21	13.5 (17.1)

^aValues of 0 were imputed for participants who did not complete any diaries

Treatment fidelity was high in both groups (YST, n=13, Mean Percent=95.9, SD=2.8; AC, n=8, Mean Percent=91.0, SD=10.6). S.J.S. served as the back-up interventionist for 12% of the sessions. One mild adverse event probably related to the interventions was reported; one participant in the AC group experienced tearfulness that resolved with emotional support. Unexpected serious adverse events (n=7) were not related to study interventions. Overall, participants in both groups were highly satisfied with the interventions. A majority of participants (n=29) liked their assigned intervention (YST: 82% quite a bit/very much; AC: 93% quite a bit/very much), found it helpful (YST: 77% quite a bit/very much; AC: 76% quite a bit/very much), and would continue to use what they learned (YST: 80% quite a bit/very much; AC: 92% quite a bit/very much).

Discussion

This study of YST for patients with a GI cancer found that the YST group had a larger decline from baseline to Week 10 in the primary outcome of fatigue as compared to an attention control group. There was also a larger decline in the secondary outcome of depressive symptoms in the YST group at Week 10. These changes were consistent with a MID (28,33,35). It is notable that MIDs in outcomes were evident in this study comparing YST to an active control group at the primary time point since behavioral interventions compared to active control groups have smaller effect sizes than studies compared to usual care (36). Yet, changes did not yet reach MID at Week 8 for fatigue and were attenuated for both outcomes at Week 14.

As suggested by the biobehavioral framework, higher reductions in the proposed mediator of sleep disturbances in the YST group as compared to AC group at Week 8 may explain subsequent larger reductions in fatigue and depressive symptoms in the YST group at Week 10 (11,12). Yet, results for sleep disturbances and psychological distress were inconsistent with expected results at some time points, and also showed that the AC group improved more than YST. Therefore, it is possible that AC could also have been beneficial for these constructs or inconsistencies could be due to the small sample size. Additionally, greater reductions were evident for inflammatory biomarkers IL-6 and sTNF-R1 in the YST as compared to AC group, although no differences were seen for TNF-α. This may be because sTNF-R1 is a better indicator of chronic inflammation, whereas TNF-α is more variable and sensitive to acute changes. Therefore, these data provide partial support for the biobehavioral framework and potential mediators (i.e., sleep disturbances, inflammation) that could explain why YST led to greater improvements in fatigue and depressive symptoms than AC (11,12). Mediation analyses in a future larger study should more definitively examine relationships among these constructs.

This study design enabled us to reach and gather data from participants who have not typically participated in yoga research (i.e., GI cancer patients, men). The overall participation rate (57%) in the present study was somewhat higher than participation rates in other multiple session intervention studies delivered during chemotherapy (e.g., 50%, 44%; 31,37) supporting the feasibility of conducting a YST intervention. Our increased reach to men may have been due to offering supportive programs (rather than “yoga”) and delivering interventions individually, which could have overcome barriers such as beliefs that yoga class participants are mostly women (38). In person intervention adherence rates were high and the interventions were delivered with a high level of fidelity, although home practice rates were lower than desired. This pilot study also demonstrated that it was feasible to obtain follow-up data (77%) and participants were satisfied with the interventions. The relationship between days of home practice and improvements in fatigue and depressive symptoms in the YST group is consistent with other research supporting that self-directed yoga practice is related to better outcomes in people with cancer (32,37). Thus, further supporting participants’ home practice may strengthen outcomes.

Study Limitations

Estimates of effects should be interpreted with the consideration that they may be unstable due to the small sample and patient-reported measures may be biased by social desirability. Although the participation rate demonstrated feasibility, many potential participants were deemed ineligible, which slowed accrual. We chose to maintain homogeneity of the sample in this study to minimize confounds introduced by differences in cancer type. However, expanding eligibility to include other cancer types would enhance feasibility and generalizability of results. Also, while our sample represented patients seen in the recruitment clinics, these clinics had limited racial and ethnic diversity. Since those who agreed to participate did not differ on these characteristics compared to those who declined, recruiting from more diverse clinics will likely improve representation of other groups. Availability of yoga instructors was somewhat limited since interventions were planned around patients’ treatment. Broadening the pool of instructors, hiring a full-time instructor, or flexibility for intervention delivery (e.g., video conferencing) could improve this issue.

Conclusion

In summary, we provide evidence that YST reduces fatigue, depressive symptoms, and inflammation as compared to empathic attention. Yet, results for sleep disturbances and psychological distress were inconsistent. Our study supports that delivering YST in the clinical setting is feasible and can reach those who have higher symptom burden and have not typically participated in yoga research (i.e., GI cancer patients, men). Results warrant a larger and more generalizable efficacy study.

Key Message.

We conducted a randomized controlled pilot study of a Yoga Skills Training (YST) intervention compared to an attention control (AC) among adults diagnosed with gastrointestinal cancer. YST showed promise for improving fatigue, depressive symptoms, and inflammation as compared to AC and was feasible, thus warranting further examination.