Design and ethical considerations
This ongoing trial uses a prospective two-group, double-blind, randomized, repeated measures, placebo-controlled design. Participants are studied at 0, 4, 8 and 12 weeks after eligibility screening, written informed consent, and enrollment to the study. A subset of participants in the EPA+DHA group with healed CVLUs by 12 weeks will be studied for an additional three months so that we can determine the frequency of CVLU recurrence. The trial was approved by the local institutional review board of the participating medical center (The Ohio State University [OSU] Wexner Medical Center – July 20, 2018). Diagram 1 and Table 1 show the study flow chart and the schedule, respectively.
Diagram 1. Study flow chart.
Insert Diagram 1 here.
Table 1. Study schedule.
Time point
|
Pre-Visit
|
Visit 1 (Week 0)
|
Visit 2
(Week 4)
|
Visit 3
(Week 8)
|
Visit 4
(Week 12)
|
Visit 5*
(3 mos. post Week 12)
|
Enrollment
|
|
|
|
|
|
|
Screening
|
X
|
|
|
|
|
|
Informed Consent
|
|
X
|
|
|
|
|
Allocation
|
|
X
|
|
|
|
|
Interventions
|
|
|
|
|
|
|
Treatment (EPA+DHA)
|
|
X
|
X
|
X
|
X
|
X
|
Placebo (Mineral Oil)
|
|
X
|
X
|
X
|
X
|
|
Assessments
|
|
|
|
|
|
|
Wound Treatments
|
X
|
X
|
X
|
X
|
X
|
X
|
Adherence
|
|
|
X
|
X
|
X
|
X
|
Biological Measures
|
|
|
|
|
|
|
Blood and Wound Fluid
|
|
X
|
X
|
X
|
X
|
X
|
Fatty Acids
|
|
X
|
X
|
X
|
X
|
X
|
n-3, n-6 lipid mediators
|
|
X
|
X
|
X
|
X
|
X
|
Cytokines
|
|
X
|
X
|
X
|
X
|
X
|
PMN activation
|
|
X
|
X
|
X
|
X
|
X
|
Proteases
(wound fluid only)
|
|
X
|
X
|
X
|
X
|
X
|
Wound area
(if wound is present)
|
|
X
|
X
|
X
|
X
|
X
|
Body mass index
|
|
X
|
X
|
X
|
X
|
X
|
CEAP
|
|
X
|
X
|
X
|
X
|
X
|
Revised VCSS
(includes pain score)
|
|
X
|
X
|
X
|
X
|
X
|
Interview/Self-Report
|
|
|
|
|
|
|
Sociodemographic
Questionnaire
|
|
X
|
|
|
|
|
Perceived Stress Scale
|
|
X
|
X
|
X
|
X
|
X
|
Health history
|
|
X
|
|
|
|
|
Food Frequency
Questionnaire
|
|
X
|
|
|
X
|
X
|
VEINES-QoL/Sym
Questionnaire
|
|
X
|
|
|
X
|
X
|
Legend: X = Collected at the given time point. n-3 = Omega 3; n-6 = Omega 6; PMN = polymorphonuclear leukocytes; CEAP = Clinical, Etiologic, Anatomic, and Pathophysiologic classification; VCSS = Venous Clinical Severity Score; VEINES-QoL/Sym = VEnous Insufficiency Epidemiological and Economic Study – Quality of Life/Symptoms. *All assessments at Visit 5 are for participants initially randomized to the treatment group, have healed wounds by Visit 4, and are consented for the additional study. At Visit 4 these participants will be re-randomized to the treatment group or placebo group and have one more study visit, three months following Visit 4.
Setting
We are recruiting participants over a 53-month period from a pool of patients diagnosed with CVLUs who are receiving treatment with standard single- or multi-layer compression bandaging at the OSU Comprehensive Wound Center (CWC). The study visits occur at the OSU Clinical Research Center, located on the medical center’s main campus.
Eligibility criteria
Patients are eligible for the study if they are women and men ≥ 55 years of age 1) with a CVLU between the ankle and knee that has been present for at least 4 weeks, but not longer than 12 months, are prescribed compression therapy with 1-4 layer bandaging; 2) with an ankle brachial pressure index (ABPI) between 0.7 and 1.2; 3) with a target wound area of 2–60 cm2, 4) who can read and understand English or Spanish, and 5) who can provide consent. Patients are included after full, understandable and neutral explanation by the project manager or principal investigator (PI) and after giving written informed consent. Patients are excluded when they report a fish allergy or when they receive a treatment or have a condition known to seriously impair normal wound healing (e.g., corticosteroids, selective COX-2 inhibitors, non-steroidal anti-inflammatory drugs > 2x/week (exception: aspirin 81 mg/day), chemotherapy, autoimmune diseases, diabetes if HbA1c > 12%, or a venous leg ulcer complicated by cellulitis, exposed tendon or bone).
Sample size
Sample size was determined to have sufficient power to detect intervention effect on wound healing – the optimal outcome of the study. Using mixed-effects linear modeling for repeated measures, a total of 248 participants (124 per group) are needed to have 80% power to detect an effect size of 0.4 for between-group comparisons in percent wound area reduction. We assumed 1) a first-order autoregressive covariance structure among the three post-intervention repeated measures, 2) a within-subject correlation of 0.7, and 3) an attrition rate of 20% in the sample size calculation. These assumptions are reasonable based on data from our pilot study [25]. The effect size of 0.4 is corresponding to an average of 16% between-group difference in percent wound area reduction with a common standard deviation of 0.4. Our power analysis also applied Bonferroni method to adjust for three comparisons of primary interest (4-week vs. baseline, 8-week vs. baseline, and 12-week vs. baseline). We used the same approach to conduct power analysis for other outcomes. Our sample size will have over 90% power to detect intervention effects on levels of lipid mediators and inflammatory cytokines and PMN activation, which had medium to large effect sizes based on our pilot study [25, 29]. A two-sided significance level of 0.05 was used for all power analyses. Based on OSU CWC data, we will be able to recruit the required number of people (n = 248) in the 53-month recruitment period, averaging 4-5 per month.
Recruitment and Retention Plans: We are recruiting participants from a pool of patients diagnosed with CVLUs scheduled to begin standard care (1-4 layer compression bandage) at the OSU CWC. Because of the known effect of aging on wound healing, the age distribution of CVLU patients at the CWC and participants in our R21 study [25], our target recruitment is 168 patients aged 55-69 years and 80 patients aged ≥ 70 years. The project manager identifies eligible patients by reviewing CWC clinical records. After CWC clinicians present a letter that introduces our team to the potential participants, the project manager provides study details and confirms eligibility of patients interested in participating in the study. After enrollment, the project manager contacts participants every two weeks to answer questions and remind them of study visits. Participants receive $250 after completing the 12-week study; subgroup participants receive an additional $150 after completing Visit 5.
Randomization and intervention
After written informed consent is obtained, patients eligible for the study are randomly assigned to either the treatment group or the placebo group by the project manager using the stratified permuted block randomization scheme generated by the study statistician. Precisely, we divide participants in each age stratum (55-69 vs. 70+ years) into permuted blocks of varying block sizes (4 or 8 people/block). Those in each block will be randomly allocated with a ratio of 1:1 to either of the two study arms. Those in each arm will consume softgels of allocated therapy (EPA+DHA or placebo) daily for 12 weeks. Stratified permuted block randomization ensures balanced randomization across age groups and over time. The varied block sizes were chosen to prevent predictability in treatment allocation. The study participants, clinicians and researchers are blinded as to treatment.
EPA+DHA softgels: Three opaque EPA+DHA softgels (J.R. Carlson Laboratories, Inc.; Arlington Heights, IL) will provide a total daily intake of 1.87 g EPA and 1.0 g DHA. This dose/ratio is proposed because EPA has relatively stronger anti-inflammatory properties than DHA [31], and a similar dose/ratio significantly raised EPA+DHA plasma levels and reduced n-6:n-3 ratios in humans after 4 weeks in our pilot work [25]. The Federal Drug Administration reports that ≤ 3.0 g/d of EPA+DHA is safe for public use [32]. At Weeks 0, 4 and 8 participants are given 1 bottle with the ensuing 1-month supply of EPA+DHA or placebo softgels. A three month supply of supplements will be supplied at Week 12 (visit 4) for those participants that qualify and enroll in the sub-study.
Placebo softgels: Three opaque placebo softgels will provide a total daily dose of 2.5 mL of mineral oil (well below therapeutic dose of 10 mL for constipation). Mineral oil is chemically inert, and on ingestion 98% remains unabsorbed in feces. We have used the same placebo dose in prior studies [25, 32]. The EPA+DHA and placebo softgels will be the same in appearance, lemon-flavored and packaged in like bottles (J.R. Carlson Laboratories, Inc.; Arlington Heights, IL).
Clinical management
Monitoring clinic wound treatments: Standard of care for uncomplicated CVLUs in the CWC involves a silver impregnated dressing under a 1-4 layer compression bandage. However, circumstances (e.g., infection) may arise that require a clinician to alter treatment. The project manager conducts weekly wound treatment monitoring checks using a standardized checklist to review participants’ clinic records for changes in wound treatments. Any treatment change (e.g.,oral antibiotic) is recorded. We will control for treatment related variables by considering them covariates as appropriate.
Monitoring adherence: At each study visit, the subsequent month’s supply of softgels (1 bottle) and written and verbal directions for consuming/storing softgels are given to each participant. At subsequent study visits, the bottles are collected. The number of softgels remaining in bottles are counted and logged for each participant. The project manager contacts participants every two weeks to review instructions and promote adherence. Adherence is verified by measuring EPA+DHA levels in blood plasma at each study visit to assess changes over time.
Data and laboratory measurements
All data are anonymized and collected using electronic report forms by investigators or trained research personnel at each study visit who are blinded to the study group assignments. Adverse events are recorded (e.g., gastrointestinal upset). Blood and wound fluid samples are collected at each study visit for use in laboratory tests: quantification of fatty acids, lipid mediators, PMN activation, cytokine levels, and cytokine gene expression.
Wound fluid collection
Wound fluid is collected from unhealed ulcers at each study time point using a standard wound fluid collection protocol [25, 29]. The fluid is collected by the PI experienced in the protocol or CRC nurses trained by the PI. Briefly, after CVLUs are washed with sterile water, a transparent occlusive film (Opsite, Smith & Nephew, UK) is applied over the wound and the leg is placed in a dependent position for approximately 1-1½ hours. While slowly removing the occlusive film and rinsing the wound with 1 mL of sterile saline, the fluid is collected using a 26G x 0.5” angiocatheter attached to a 3 mL syringe (Terumo Medical, Somerset, N.J.). The fluid is transferred into plain collection tubes and analyzed immediately to determine PMN activation or frozen and stored at -800 C until further analysis.
Primary outcome measures
The primary outcome measure with respect to the effectiveness of EPA+DHA oral therapy in the treatment of CVLUs is time to complete wound healing. We define complete wound healing as reepithelialization of the total wound surface. The PI who is blinded to treatment assesses this outcome. The area of unhealed ulcers is quantified at each time point in cm2 using a single digital camera photogrammetry system [34]. We calculate percent reduction in ulcer area at Weeks 4, 8 and 12 compared to Week 0 for each participant and averages for each group. Larger percentages indicate greater healing. The second primary outcome measure is PMN activation in blood and CVLU microenvironments. The determination of PMN activation is accomplished using flow cytometry. Additional measures to assess PMN activation in CVLU fluid involve quantifying reactive oxygen species production during ‘‘respiratory burst’’ by PMNs and two PMN-specific protease biomarkers (matrix metalloproteinase-8, human neutrophil elastase). We also measure free fatty acids in plasma and erythrocyte membranes, lipid mediators in plasma and CVLU fluid, and a panel of pro- and anti-inflammatory cytokines in plasma and CVLU fluid. Finally, inflammatory cytokine gene expression in PMNs (neutrophils and monocytes) is measured.
Secondary outcome measures
Pain is assessed at all time points using the revised Venous Clinical Severity Score (VCSS). This revised VCSS is a valid and reliable instrument to measure severity of venous disease, pain specific to chronic venous disease of the legs, and response to treatment over time that can be compared across studies [35]. Quality of life is measured at Weeks 0 and 12 using the Venous Insufficiency Epidemiological and Economic Study-Quality of Life/Symptoms (VEINES-QOL/Sym) questionnaire, based on the Short Form – 36 (SF-36) questionnaire [36]. VEINES-QOL/Sym is a standardized, 26-item, patient-reported questionnaire to assess severity and frequency of venous insufficiency symptoms.
CVLU recurrence will be assessed in two subgroups derived from the EPA+DHA group. Unblinding will occur at Week 12. If participants in the EPA+DHA Group have healed CVLUs by Week 12, they will be asked to consider participating in a continuing study to determine the preliminary efficacy of continuing EPA+DHA therapy in preventing ulcer recurrence. If participants agree to participate, they will be randomly assigned 1:1 to either of two subgroups: subgroup 1 will continue EPA+DHA therapy for three more months, while subgroup 2 will receive placebo therapy. The PI and participants will be blinded to treatment. Three months later participants in the subgroups will return for a final study visit (Visit 5).
Covariates
To address potential confound of variables that may impact inflammation or wound healing [37], we are collecting data on age, sex, tobacco use, comorbidities, perceived stress, and prescribed/over-the-counter medications. We are also collecting data on dietary intake of nutrients that may affect healing (e.g., vitamin C and protein) using a food frequency questionnaire [38]. All instruments being used to measure tobacco use, perceived stress and dietary intake have been well vetted so data can be compared across studies. We are also determining if wound debridement or adjunct therapies were used between study visits and noting the type of dressing under compression bandaging.
Statistical analysis
Congruent with the RCT design of the study, we will conduct an intent-to-treat analysis. We will first use descriptive statistics to check data distribution, identify outliers, guide appropriate data transformation if needed and summarize sample characteristics. Two-sample T-tests and Chi-square statistics will be used to check the balance of baseline measures between two study arms. Mixed-effects linear modeling for repeated measures will be used to test the intervention effect on each outcome measure (PMN activation, lipid mediators, inflammatory cytokines, inflammatory cytokine gene expression, percent reduction in wound area, and pain and QoL). In each model, we will include fixed-effects of treatment (EPA+DHA vs. control), time, and treatment by time interaction, adjusting for possible covariates (e.g., lipid lowing medications) and data dependancy from repeated measures. From the mixed-effects modeling, we will derive the between-group contrast (EPA+DHA vs. control) estimates of change from baseline in outcome variable at each follow-up time point. Multiple comparison adjustment (e.g., Bonferroni adjustment) will be applied to these between-group contrasts to avoid inflated Type-I error. Logistic regression modeling will be used to analyze wound recurrence. Again, we will estimate between-group difference in the risks of wound recurrence, adjusting for potential covariates and multiple comparison. Lastly, we will use longitudinal mediation modeling [39] to test hypothesized mediating pathways. A significant mediation pathway is indicated if both paths from X (e.g., treatment) to M (mediator, e.g., pro-inflammatory cytokine) and from M to Y (e.g., PMN activation) are statistically significant. Adequate model fit of a mediation model will be indicated with a non-significant Chi-square statistic, comparative fit index and Tucker & Lewis Index ≥ 0.9, and a root mean square error of appromixation of ≤ 0.08 [40]. The effect of biological sex and its interaction with treatment will also be examined in the regression models. Missing Data. We expect missing data due to non-adherence, loss of follow-up, or lack of data due to healed wounds. We will carefully examine the extent and pattern of missing data. We will use an indicator variable to flag missing data due to healed wounds and incoporate the variable in the missing pattern analysis. Mixed-effects modeling allows for missing at random [41]. If missing completely at random (MCAR) exists, multiple imputation will be used in order to use the optimal amount of information in our analysis. For missing not at random, pattern-mixture modeling will be used [42]. We will also conduct sensitivity analysis to examine the robustness of study findings from different methods. In terms of data and safety monitoring, a Safety Officer (SO) has been selected and approved by the funding agency in consultations with the PI. The SO has no financial, scientific, or other conflict of interest with the trial. The SO meets with the PI twice annually to review study progress, data quality, and participants’ safety.