The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Protocol S1
and Checklist S1
A total of 581 type 2 diabetic patients were screened for inclusion at the local community medical service centers in urban districts of Shanghai. Participants were considered eligible if they met the following criteria: 1) 50–79 years of age (women were required to be postmenopausal for at least 1-year); 2) LDL-C level ≥2.9 mmol/L; and 3) diagnosis of type 2 diabetes, but not using exogenous insulin for glycemic control. Exclusion criteria were: 1) current or previous (preceding 6 months) estrogen-use; 2) regularly taking phytoestrogen-containing supplements; 3) antibiotic-use in the preceding 3 months; 4) severe renal, liver, heart, pituitary, thyroid or mental diseases, alimentary tract ulceration or diseases affecting absorption; or 5) history of cancer, history of drug or alcohol abuse.
The study protocol was approved by the Ethics Committee of Institute for Nutritional Sciences, Chinese Academy of Sciences. All participants provided signed informed consents.
Thirty-seven subjects began lignan supplements and 36 subjects started on the placebo for 12 weeks. After an 8-week wash-out period, the participants received the alternative treatment for another 12 weeks. While on study, participants were instructed to maintain their habitual diets, levels of physical activity, and use of prescribed medications. During the 12 weeks, participants assigned to the lignan supplements were instructed to take three lignan capsules (0.6 g/capsule) each day that provided a daily amount of 360 mg isolated flaxseed lignan. The raw materials of lignan were donated by Frutarom Netherlands BV (LinumLife™ Extra, Veenendaal, The Netherlands) and were produced by Jarrow Formulas Inc (Flax Essence™ , Los Angeles). The three capsules provided 3.7 kilocalories and were comprised of 20% SDG, 15.6% fat, 3.2% protein, 2.6% fiber and 30% carbohydrate. Participants randomly assigned to the placebo group were instructed to take 3 placebo capsules per day, which were comprised of rice flour (98%) and provided a total of 5.8 kilocalories. Subjects were asked to return any unused capsules and adherence was assessed by pill counts, as well as by urinary concentrations of lignan metabolites (for SDG-specific adherence).
The objective of the study was to investigate the effect of a flaxseed-derived lignan supplement on indexes of glycemic control, insulin resistance and lipid profiles in type 2 diabetic patients.
All participants were scheduled to visit Huadong Hospital every 3 weeks to obtain new capsules, and their adherence and physical status were evaluated. Dietary intakes were assessed using 3-day food records which ascertained intakes during 2-weekdays and 1-weekend day at four timepoints throughout the study (1 week prior to and during the last week of each intervention period). All food records were reviewed for completeness and coded by the trained dietitians who were blinded to the study arms. Energy and nutrient intakes were calculated using the SY Nutrition Software (Fudan University, Shanghai, China) based on the local food composition database. Physical activity level was evaluated by asking the average times per week spent on several common activities (e.g. running, jogging, dancing, bicycling) in the last month and each activity was assigned a metabolic equivalent value (MET) according to accepted standards 
. After fasting overnight, blood pressure and anthropometric parameters (height, weight, waist and hip circumference) were measured and blood samples were collected at the beginning and the end of each two intervention periods. Fasting venous blood samples were collected in 7-mL EDTA-treated vacutainers for insulin analysis and 7-mL serum separator vacutainers for measurements of fasting glucose and lipid profiles. After centrifugation at 2600 × g for 10 min at 4°C, samples were aliquotted and stored at −80°C until batch analysis. Fasting morning urine samples (50 mL) were collected at identical timepoints using plastic jugs containing 50 mg ascorbic acid. Urine samples were aliquotted and stored at −20°C until processed.
Serum total cholesterol, HDL cholesterol (HDL-C), LDL-C, triacylglycerol, and glucose were measured using reagents purchased from Wako Pure Chemical Industries (Osaka, Japan), serum lipoprotein(a) [Lp(a)], apolipoprotein A-1 and B (apoA1 and apoB) were measured using kits from Roche Diagnostics (Mannheim, Germany). All the above assays were performed on an automatic analyzer (Hitachi 7080, Japan) within one day. HbA1c
was determined by turbidometric immunoinhibition on packed red blood cells on the automatic analyzer using kits from Roche Diagnostics 
. This assay is approved by the US National Glycohemoglobin Standardization Program and by the Food and Drug Administration for clinical use. Plasma insulin was determined by radioimmunoassay (Linco Research, MO). Insulin resistance was calculated using the Homeostasis Model Assessment of Insulin Resistance method (HOMA-IR)
(Insulin (µU/mL)×glucose (mmol/L))/22.5 
. Urinary excretion of lignan metabolites (enterodiol and enterolactone) and isoflavones (genistein and daidzein) was measured using a modified HPLC method 
. The intra- and inter-assay CVs ranged from 2.3% to 15.8% for enterodiol, enterolactone, genistein and daidzein at concentrations of 1.0 and 10 µg/mL.
Lp(a) concentrations were measured in 245 serum samples due to reagent shortage. The values of urinary excretion of lignan or isoflavone metabolites were replaced by 0.05 µg/mL (half of the lowest detectable limit) when not detectable.
As there were no extant studies that had investigated the effect of flaxseed-derived lignans on HbA1c, the sample size was calculated based on the previous studies of flaxseed on LDL-C concentrations. Using a one-sided 5% significance level, a sample of 67 patients was needed for this cross-over trial, assuming a 10% drop-out rate. This gave the study 90% power to detect a 5% difference (0.2 mmol/L) in LDL-C between treatments (assuming a common SD of 0.5 mmol/L).
Randomization and Blinding
The random allocation sequence was developed using computer program by a statistician who was not involved in the study. After enrollment, each participant was randomly given a unique study number. Randomization and allocation to the treatment or placebo group was based on the study number. Participants were randomized to the intervention or the placebo arms using stratification factors of gender and tertiled LDL-C concentrations from screening. In detail, women and men were sorted by their levels of LDL-C, respectively. Then randomization was performed using a block size of 3 and length of 10 for men and length of 16 for women, respectively. Placebo capsules were almost identical to the lignan capsules in size, shape, color, and taste. Each bottle of capsules was marked with the participant's study number, but no product identifier. Participants, health professionals, statisticians, and other research staff involved in the trial were blinded to the assignments.
Differences between values after 12-week intervention were analyzed in Stata 9.2 (Stata ™, Texas) using a mixed model analysis of covariance with treatment (lignan or placebo) and period (first or second period) as fixed factors, subjects as random factors, and baseline measurements as covariates. Further fixed terms corresponding to treatment/period interactions were included to test for any carry-over effect between periods, and treatment/covariate interaction was included to test whether the treatment effect of flaxseed lignan varied according to the baseline values of the covariate. Given the weight-dependent nature of the study endpoints, weight and weight changes were incorporated into the model as covariates. However the results were not affected and the data are therefore not shown. Data that were not normally distributed, as assessed by the Shapiro-Wilks test, were natural-logarithmically transformed prior to analysis (see table footnotes for details). Differences before and after treatments were analyzed using paired Student's t-test. Pre-post differences in physical activity level, and habitual energy and nutrient intake were analyzed using ANOVA. Differences were considered significant at P<0.05.