Clinical Care/Education/Nutrition O R I G I N A L                        A R T I C L EA Low-Fat Vegan Diet Improves Glycem...
Low-fat vegan diet and type 2 diabetesfavor low– glycemic index foods, such as 24-h recalls or incidentally at any point, ...
Barnard and AssociatesTable 1—Baseline demographic and clinical variables                                                 ...
Table 2—Diet effects on clinical measures1780                                                                             ...
Barnard and Associates                                                                                      Data are means...
Low-fat vegan diet and type 2 diabetes                                                                                    ...
Barnard and Associates      domized, controlled, pilot intervention             density lipoprotein cholesterol in plasma ...
Upcoming SlideShare
Loading in …5

Scholarly Article


Published on

1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Scholarly Article

  1. 1. Clinical Care/Education/Nutrition O R I G I N A L A R T I C L EA Low-Fat Vegan Diet Improves GlycemicControl and Cardiovascular Risk Factors ina Randomized Clinical Trial in IndividualsWith Type 2 DiabetesNEAL D. BARNARD, MD1,2 BRENT JASTER, MD2 used in the absence of exercise, was asso-JOSHUA COHEN, MD1 KIM SEIDL, MS, RD2 ciated with increased insulin sensitivityDAVID J.A. JENKINS, MD, PHD3 AMBER A. GREEN, RD2 and reduced body weight in nondiabeticGABRIELLE TURNER-MCGRIEVY, MS, RD4 STANLEY TALPERS, MD1 overweight women (4).LISE GLOEDE, RD, CDE5 We therefore conducted a random- ized controlled trial of a vegan diet with exercise held constant to test the hypoth-OBJECTIVE — We sought to investigate whether a low-fat vegan diet improves glycemic esis that, in individuals with type 2 diabe- Authors credentials (medical doctorscontrol and cardiovascular risk factors in individuals with type 2 diabetes. tes, a low-fat plant-based diet improves and registered dieticians) plasma lipid, and weight con- glycemic, are given.RESEARCH DESIGN AND METHODS — Individuals with type 2 diabetes (n ϭ 99) trol compared with a diet based on cur-were randomly assigned to a low-fat vegan diet (n ϭ 49) or a diet following the AmericanDiabetes Association (ADA) guidelines (n ϭ 50). Participants were evaluated at baseline and 22 rent ADA guidelines.weeks. RESEARCH DESIGN ANDRESULTS — Forty-three percent (21 of 49) of the vegan group and 26% (13 of 50) of the ADA METHODS — Individuals with type 2group participants reduced diabetes medications. Including all participants, HbA1c (A1C) de- diabetes, defined by a fasting plasma glu-creased 0.96 percentage points in the vegan group and 0.56 points in the ADA group (P ϭ cose concentration Ͼ6.9 mmol/l on two0.089). Excluding those who changed medications, A1C fell 1.23 points in the vegan group occasions or a prior diagnosis of type 2compared with 0.38 points in the ADA group (P ϭ 0.01). Body weight decreased 6.5 kg in thevegan group and 3.1 kg in the ADA group (P Ͻ 0.001). Body weight change correlated with A1C diabetes with the use of hypoglycemicchange (r ϭ 0.51, n ϭ 57, P Ͻ 0.0001). Among those who did not change lipid-lowering medications for Ն6 months, were re-medications, LDL cholesterol fell 21.2% in the vegan group and 10.7% in the ADA group (P ϭ Technical cruited through newspaper advertise-0.02). After adjustment for baseline values, urinary albumin reductions were greater in the vegan ments in the Washington, DC, area on language, commongroup (15.9 mg/24h) than in the ADA group (10.9 mg/24 h) (P ϭ 0.013). two occasions (October 2003 through in scholarly articles December 2003 and October 2004CONCLUSIONS — Both a low-fat vegan diet and a diet based on ADA guidelines improved through December 2004) to complete theglycemic and lipid control in type 2 diabetic patients. These improvements were greater with a study from January 2004 through Junelow-fat vegan diet. 2004 and January 2005 through June 2005, respectively. Exclusion criteria Diabetes Care 29:1777–1783, 2006 were an HbA1c (A1C) Ͻ6.5 or Ͼ10.5%, use of insulin for Ͼ5 years, current smok-D iabetes prevalence is relatively low fects. In a 12-week pilot trial of a low-fat ing, alcohol or drug abuse, pregnancy, among individuals following plant- vegan diet in individuals with type 2 dia- unstable medical status, and current use based and vegetarian diets, and betes, conducted without increased exer- of a low-fat vegetarian diet. The protocolclinical trials using such diets have shown cise, fasting serum glucose concentration was approved by the George Washingtonimprovements in glycemic control and dropped 28% compared with 12% in the University Institutional Review Board. Allcardiovascular health (1,2). Most of these control group following a diet based on participants gave written informedtrials have also included exercise, thus American Diabetes Association (ADA) consent.making it impossible to isolate diet ef- guidelines (P Ͻ 0.05) (3). A similar diet, Authors affiliations are given. A1C was assayed using affinity chro- From this and an Abbott IMx analyzer matography on the credentials● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● above we know they arein order of (5). Volunteers were rankedFrom the 1Department of Medicine, George Washington University School of Medicine, Washington, DC;the 2Physicians Committee for Responsible Medicine, Washington, DC; the 3Department of Nutritional experts in their field. randomly as- A1C concentrations andSciences, Faculty of Medicine, University of Toronto, and the Clinical Nutrition and Risk Factor Modification signed in sequential pairs, using a ran-Center, St. Michael’s Hospital, Toronto, Canada; the 4Department of Nutrition, School of Public Health, dom-number table, to a low-fat vegan dietUniversity of North Carolina, Chapel Hill, North Carolina; and 5Private practice, Arlington, Virginia. or a diet following the 2003 ADA guide- Address correspondence and reprint requests to Neal D. Barnard, MD, 5100 Wisconsin Ave., Suite 400, lines (6) for 22 weeks. Because assign-Washington, DC 20016. E-mail: Received for publication 20 March 2006 and accepted in revised form 15 May 2006. ment was done simultaneously, allocation Abbreviations: ADA, American Diabetes Association. concealment was unnecessary. A table elsewhere in this issue shows conventional and Systeme International (SI) units and conversion ` The vegan diet (ϳ10% of energy fromfactors for many substances. fat, 15% protein, and 75% carbohydrate) DOI: 10.2337/dc06-0606. Clinical trial reg. no. NCT00276939, consisted of vegetables, fruits, grains, and © 2006 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby legumes. Participants were asked to avoidmarked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. animal products and added fats and toDIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006 1777
  2. 2. Low-fat vegan diet and type 2 diabetesfavor low– glycemic index foods, such as 24-h recalls or incidentally at any point, ence was measured at the maximal pro-beans and green vegetables. Portion sizes, as saturated fat Յ5% and total fat Յ25% trusion of the intake, and carbohydrate intake of energy, and as average daily cholesterol Blood pressure was measured afterwere unrestricted. intake Յ50 mg on 3-day dietary records participants had rested in a seated posi- The ADA diet (15–20% protein, Ͻ7% at 22 weeks. Adherence for the ADA tion for 5 min using a digital monitorsaturated fat, 60 –70% carbohydrate and group was defined as average daily energy (Omron HEM-711) and a standard cuffmonounsaturated fats, and cholesterol intake on the 22-week 3-day dietary maintained at the level of the heart. ThreeՅ200 mg/day) was individualized, based records being no more than 200 kcal in measurements were taken at 2-min inter-on body weight and plasma lipid concen- excess of the intake prescribed by the reg- vals; the first measurement was disre-trations (6). ADA group participants with istered dietitian and saturated fat Յ10% garded, and a mean was calculated for thea BMI Ͼ25 kg/m2 (all but three ADA of energy. Individuals who attended remaining two participants) were prescribed en- fewer than 10 of the 22 weekly sessionsergy intake deficits of 500 –1,000 kcal. were also considered nonadherent on ei- Statistical analyses No meals were provided. To meet the ther diet. To have an 80% chance of detecting avitamin B12 needs of the vegan group Participants were asked to continue 1.5–percentage point between-groupwhile maintaining the same intervention their preexisting medication regimens, A1C difference as significant (at the two-in the ADA group, all participants were except when fasting plasma glucose deter- sided 5% level), with an assumed SD ofprovided a vitamin B12 supplement (100 minations fell below 4.4 mmol/l or hypo- 1.9% and a loss to follow-up of 26%, 34␮g) to be taken every other day. For both glycemic symptoms were accompanied participants were required per group. Angroups, alcoholic beverages were limited by a capillary glucose reading Ͻ3.6 interim analysis indicated group differ-to one per day for women and two per day mmol/l. In such cases, medications were ences of 0.8% with an SD of 1.3%; there-for men. Participants were asked not to reduced for participant safety by a study fore, a revised power analysis wasalter their exercise habits during the inter- endocrinologist, who remained blind to conducted. To have an 80% chance of de-vention period. group assignment, following an estab- tecting a 0.8% difference as significant Each participant met for 1 h with a lished protocol. with an SD of 1.3% and loss to follow-upregistered dietitian experienced in the use Laboratory measurements were made of 33%, an additional 15 participantsof the assigned diet to establish an appro- after a 12-h fast by technicians blind to were required per group.priate diet plan. Thereafter, participants group assignment. A1C (the primary end Between-subject t tests were calcu-attended weekly 1-h meetings of their as- point) was assayed at 0, 11, and 22 weeks, lated for each measure to determinesigned groups for nutrition and cooking as described above. All other measures whether the changes associated with theinstruction conducted by a physician and were assessed at baseline and 22 weeks, intervention diet were greater than thosea registered dietitian and/or a cooking in- except as noted. Plasma glucose was mea- associated with the control diet. Withinstructor. Sessions for the two groups were sured by the glucose oxidase method us- each diet group, paired comparison t testssimilar in duration and content, except ing an Abbott Spectrum analyzer (Abbott were calculated to test whether thewith regard to dietary details. Group lead- Park, IL) (8). Plasma cholesterol and tri- change from baseline to 22 weeks was sig-ers were instructed to make no comment glyceride concentrations were measured nificantly different from zero. The pri-favoring either diet over the other. by enzymatic methods using an Abbott mary analysis of the main end point was At weeks 4, 8, 13, and 20, a registered Spectrum analyzer (9,10). HDL choles- intention to treat and included all partic-dietitian made unannounced telephone terol was measured after double precipi- ipants. Because medication changes influ-calls to each participant to administer a tation with dextran and MgCl2 (11). LDL ence the dependent measures,24-h diet recall. These recalls were not cholesterol was estimated using the exploratory analyses were performed forstatistically analyzed, but allowed the in- Friedewald equation (12). In individuals the subgroup whose medications re-vestigators to check for poor adherence whose plasma triglyceride concentrations mained constant. An ␣ of 0.05 was usedand provide additional dietary counseling exceeded 400 mg/dl, LDL cholesterol was for all statistical tests, with no adjustmentas needed. measured directly by precipitation and for multiple comparisons. In addition, a 3-day dietary record magnetic separation (LipiDirect; Poly- Regression analyses assessed whetherwas completed by each participant at medco, Cortlandt Manor, NY). Non-HDL the diet group effects on A1C and bodyweeks 0, 11, and 22, on 2 weekdays and 1 cholesterol concentration was calculated weight were significant, while controllingweekend day, using a food scale, after par- as the difference between total and HDL for baseline values, and whether the dietticipants had completed a 3-day practice cholesterol. Urinary albumin was mea- group effect on A1C was significant, whilerecord. Using the Nutrition Data System sured on 24-h samples using an anionic controlling for baseline A1C and changesfor Research software version 5.0 (Food dye– binding assay (13). in body weight. Pearson correlations wereand Nutrient Database 35 [released May Physical activity was assessed over a calculated for the relationship between2004]; Nutrition Coordinating Center, 3-day period by pedometer (Omron HJ- A1C change and weight change.University of Minnesota, Minneapolis, 112) and with the Bouchard 3-Day Phys- An interim analysis was performed af-MN) (7), a registered dietitian certified by ical Activity Record (14). Body weight ter week 11 to assess whether benefits Note the use ofthe Nutrition Coordinating Center ana- was determined at 0, 11, and 22 weeks, or adverse outcomes were statisticallylyzed all 3-day dietary records and diet structured headings before breakfast while participants wore unusual.recalls. For purposes of statistical analy- (research design a digital scale accu- hospital gowns, usingsis, dietary adherence for the vegan group and methods,Waist circumference was rate to 0.1 kg. RESULTS — Of 1,049 individualswas defined as the absence of meat, poul- measured with a tape measure placed 2.5 screened by telephone, 99 met participa-try, fish, dairy, or egg intake reported on results, conclusions, Hip circumfer- cm above the umbilicus. tion criteria and were randomly assigned etc.)1778 DIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006
  3. 3. Barnard and AssociatesTable 1—Baseline demographic and clinical variables g/day, P Ͻ 0.0001; ADA 23 Ϯ 12 to 14 Ϯ 6 g/day, P Ͻ 0.0001 [between-group P Ͻ Vegan group ADA group P value 0.001]) and cholesterol (vegan 291 Ϯ 223 to 24 Ϯ 57 mg/day, P Ͻ 0.0001; ADAn 49 50 317 Ϯ 200 to 189 Ϯ 89 mg/day, P ϽAge (years) 56.7 (35–82) 54.6 (27–80) 0.29 0.0001 [between-group P ϭ 0.002]). Fi-Sex 0.26 ber increased only among vegans (18.8 Ϯ Male 22 (45) 17 (34) 6.4 to 36.3 Tables g/day, P Ͻ 0.0001; Ϯ 13.3 of results Female 27 (55) 33 (66) ADA 19.5 Ϯ 6.9 to 19.0 Ϯ 7.9 g/day, P ϭRace, ethnicity 0.71* 0.73 [between-group P Ͻ in are typical 0.001]). Black, non-Hispanic 22 (45) 22 (44) scholarly articles. Pedometer readings and self-reported White, non-Hispanic 21 (43) 22 (44) energy expenditure revealed no signifi- White, Hispanic 4 (8) 2 (4) cant between-group differences. Group- Asian, non-Hispanic 2 (4) 4 (8) specific dietary adherence criteria wereMarital status 0.08 met by 67% (33 of 49) of vegan group Not married 20 (41) 26 (52) participants and 44% (22 of 50) of ADA Married 29 (59) 24 (48) group participants. During the interven-Education 0.69 tion period, 43% (21 of 49) of vegan High school, partial or graduate 6 (12) 3 (6) group participants and 26% (13 of 50) of College, partial or graduate 26 (53) 25 (50) ADA group participants reduced their di- Graduate degree 17 (35) 22 (44) abetes medications, mainly as necessi-Occupation 0.04 tated by hypoglycemia, while 8% (4 of Service occupation 3 (6) 7 (14) 49) of vegans and 8% (4 of 50) of ADA Technical, sales, administrative 16 (33) 18 (36) participants increased medications with- Professional or managerial 15 (31) 21 (42) out investigators’ authorization. Retired 15 (31) 4 (8) A1C fell 0.96 percentage points (P ϽOn insulin 11 (22) 5 (10) 0.09 0.0001) in the vegan group and 0.56 per-On metformin 34 (69) 39 (78) 0.33 centage points (P ϭ 0.0009) in the ADAOn sulfonylurea 25 (51) 29 (58) 0.49 group (between-group P ϭ 0.089; base-On thiazolidinedione 16 (33) 15 (30) 0.78 line-adjusted P ϭ 0.091; Table 2 and Fig.On other diabetes medications 1 (2) 2 (4) 0.57 1). Among participants whose diabetesOn blood pressure medications 31 (63) 38 (76) 0.17 medications remained unchangedOn lipid-lowering medications 27 (55) 27 (54) 0.88 throughout (n ϭ 24 vegan and n ϭ 33History of eye involvement 9 (18) 10 (20) 0.82 ADA), A1C fell 1.23 points in the veganHistory of renal involvement 6 (12) 4 (8) 0.48 group and 0.38 points in the ADA groupHistory of neuropathy 18 (37) 24 (48) 0.25 (P ϭ 0.01; baseline-adjusted P ϭ 0.007).Mean BMI (kg/m2) 33.9 35.9 0.18 Subanalyses were conducted to assess the Ͻ25 kg/m2 5 (10) 2 (4) effects of dietary adherence. For those 25–29.9 kg/m2 14 (29) 5 (10) who met adherence criteria (n ϭ 33 vegan Ն30 kg/m2 30 (61) 43 (86) and n ϭ 22 ADA), the A1C changes wereData are mean (range) or n (%) unless otherwise indicated. P values refer to t test for continuous variables and Ϫ1.20% for the vegan group and␹2 for categorical variables. *P value calculated for race distribution; for ethnicity (Hispanic vs. non-His- Ϫ0.88% for the ADA group (P ϭ 0.31).panic), P ϭ 0.39. For those who were both adherent and medication stable (n ϭ 17 vegan and n ϭto the vegan (n ϭ 49) or ADA (n ϭ 50) duced energy intake (vegan 1,759 Ϯ 468 12 ADA), A1C changes were Ϫ1.48% forgroups. The reasons for exclusion were to 1,425 Ϯ 427 kcal/day, P Ͻ 0.0001; the vegan group and Ϫ0.81% for the ADAA1C values outside the required range ADA 1,846 Ϯ 597 to 1,391 Ϯ 382 kcal/ group (P ϭ 0.15).(n ϭ 201), failure to meet other participa- day, P Ͻ 0.0001 [between-group P ϭ To test whether the effect of diet ontion criteria (n ϭ 279), inability to attend 0.22]) and protein intake (vegan 77 Ϯ 27 A1C was mediated by body weightscheduled meetings (n ϭ 187), failure to to 51 Ϯ 16 g/day, P Ͻ 0.0001; ADA 85 Ϯ changes, a regression model was con-keep interview appointment (n ϭ 153), 27 to 73 Ϯ 23 g/day, P ϭ 0.002 [between structed, including baseline A1C, weightreluctance to change diet (n ϭ 72), and group P ϭ 0.01]). Carbohydrate intake change, and diet group as predictors ofother or unspecified (n ϭ 58). The partic- increased in the vegan group from 205 Ϯ A1C change, among those whose hypo-ipants’ demographic and clinical charac- 69 to 251 Ϯ 70 g/day (P Ͻ 0.0001) but glycemic medications remained constant.teristics (Table 1) were similar to those of fell in the ADA group from 213 Ϯ 70 to In this model, the effect of diet group wasindividuals with type 2 diabetes in the 165 Ϯ 51 g/day (P Ͻ 0.0001 [between- no longer significant (P ϭ 0.23). Control-Washington, DC, area. All participants group P Ͻ 0.001]). ling for diet group and for baseline A1Ccompleted laboratory assessments at 22 Fat intake fell in both groups (vegan scores, weight change was significantlyweeks. 72 Ϯ 28 to 30 Ϯ 19 g/day, P Ͻ 0.0001; associated with A1C change; each kilo- Three vegan participants and eight ADA 73 Ϯ 35 to 52 Ϯ 21 g/day, P Ͻ gram of weight loss was associated with aADA participants failed to complete 22- 0.0001 [between-group P ϭ 0.002]), as 0.12% drop in A1C. For the subgroupweek dietary records. Both groups re- did saturated fat (vegan 23 Ϯ 10 to 6 Ϯ 4 that did not change diabetes medications,DIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006 1779
  4. 4. Table 2—Diet effects on clinical measures1780 Vegan group (n ϭ 49, except as noted) ADA group (n ϭ 50, except as noted) Baseline 22 weeks Change Baseline 22 weeks Change Effect size P value Anthropometric and glycemic variables All participants A1C (%) 8.0 Ϯ 1.1 7.1 Ϯ 1.0 Ϫ1.0 Ϯ 1.2* 7.9 Ϯ 1.0 7.4 Ϯ 1.0 Ϫ0.6 Ϯ 1.1† Ϫ0.4 (Ϫ0.9 to 0.1) 0.09 Fasting plasma glucose (mmol/l) 9.08 Ϯ 2.95 7.11 Ϯ 1.97 Ϫ1.97 Ϯ 2.68* 8.90 Ϯ 2.26 6.98 Ϯ 1.91 1.92 Ϯ 2.48* Ϫ0.05 (Ϫ1.08 to 0.98) 0.92 Fasting plasma glucose (mg/dl) 163.5 Ϯ 53.2 128.0 Ϯ 35.5 Ϫ35.5 Ϯ 48.3* 160.4 Ϯ 40.7 125.8 Ϯ 34.4 Ϫ34.6 Ϯ 44.7* Ϫ0.9 (Ϫ19.5 to 17.6) 0.92 Weight (kg) 97.0 Ϯ 22.9 91.1 Ϯ 22.4 Ϫ5.8 Ϯ 4.4* 99.3 Ϯ 21.0 95.0 Ϯ 20.9 Ϫ4.3 Ϯ 4.4* Ϫ1.6 (Ϫ3.3 to 0.2) 0.08 BMI (kg/m²) 33.9 Ϯ 7.8 31.8 Ϯ 7.5 Ϫ2.1 Ϯ 1.5* 35.9 Ϯ 7.0 34.3 Ϯ 7.3 Ϫ1.5 Ϯ 1.5* Ϫ0.6 (Ϫ1.2 to 0.1) 0.08 Waist (cm) 110.8 Ϯ 18.4 105.5 Ϯ 18.1 Ϫ5.3 Ϯ 4.4* 112.3 Ϯ 14.9 109.5 Ϯ 14.7 Ϫ2.8 Ϯ 4.7* Ϫ2.5 (Ϫ4.3 to Ϫ0.7) 0.008 Low-fat vegan diet and type 2 diabetes Hip (cm) 118.4 Ϯ 17.8 114.5 Ϯ 17.8 Ϫ3.9 Ϯ 3.4* 121.3 Ϯ 12.7 117.5 Ϯ 12.2 Ϫ3.8 Ϯ 3.9* Ϫ0.1 (Ϫ1.5 to 1.4) 0.94 Waist-to-hip ratio (cm) 0.94 Ϯ 0.08 0.92 Ϯ 0.07 Ϫ0.02 Ϯ 0.03‡ 0.93 Ϯ 0.07 0.93 Ϯ 0.07 0.01 Ϯ 0.04 Ϫ0.02 (Ϫ0.03 to 0.01) 0.003 Participants whose diabetes medications remained unchanged (n ϭ 24 vegan, n ϭ 33 ADA) A1C (%) 8.07 Ϯ 1.24 6.84 Ϯ 0.84 Ϫ1.23 Ϯ 1.38† 7.88 Ϯ 0.93 7.50 Ϯ 1.03 Ϫ0.38 Ϯ 1.11 Ϫ0.85 (Ϫ1.51 to Ϫ0.19) 0.01 Fasting plasma glucose (mmol/l) 9.85 Ϯ 2.95 7.12 Ϯ 1.80 2.73 Ϯ 3.05† 8.90 Ϯ 2.05 7.34 Ϯ 2.05 1.57 Ϯ 2.50‡ Ϫ1.17 (Ϫ2.64 to 0.31) 0.12 Fasting plasma glucose (mg/dl) 177.4 Ϯ 53.2 128.2 Ϯ 32.4 Ϫ49.2 Ϯ 55.0† 160.3 Ϯ 37.0 132.2 Ϯ 36.9 Ϫ28.2 Ϯ 45.0‡ Ϫ21.1 (Ϫ47.6 to 5.5) 0.12 Weight (kg) 102.4 Ϯ 23.6 95.9 Ϯ 22.4 Ϫ6.5 Ϯ 4.3* 100.0 Ϯ 19.4 96.9 Ϯ 19.1 Ϫ3.1 Ϯ 3.4* Ϫ3.4 (Ϫ5.5 to Ϫ1.4) 0.001 BMI (kg/m²) 36.1 Ϯ 7.5 33.8 Ϯ 7.2 Ϫ2.3 Ϯ 1.5* 36.0 Ϯ 5.8 34.9 Ϯ 5.9 Ϫ1.1 Ϯ 1.2* Ϫ1.2 (Ϫ1.9 to Ϫ0.5) 0.001 Waist (cm) 115.3 Ϯ 17.9 110.3 Ϯ 17.8 Ϫ5.0 Ϯ 3.7* 113.3 Ϯ 13.1 111.0 Ϯ 13.5 Ϫ2.3 Ϯ 4.2‡ Ϫ2.7 (Ϫ4.8 to Ϫ0.5) 0.017 Hip (cm) 123.3 Ϯ 17.4 119.1 Ϯ 17.3 Ϫ4.1 Ϯ 2.8* 121.7 Ϯ 12.1 118.6 Ϯ 11.6 Ϫ3.1 Ϯ 3.3* Ϫ1.0 (Ϫ2.7 to 0.7) 0.23 Waist-to-hip ratio (cm) 0.94 Ϯ 0.08 0.93 Ϯ 0.07 Ϫ0.01 Ϯ 0.03 0.93 Ϯ 0.07 0.94 Ϯ 0.08 0.00 Ϯ 0.04 Ϫ0.02 (Ϫ0.03 to 0.00) 0.10 Renal variable Urinary albumin/24 h 33.0 Ϯ 51.8 14.6 Ϯ 17.8 Ϫ18.4 Ϯ 39.0‡ 55.0 Ϯ 263.1 43.7 Ϯ 212.0 Ϫ11.3 Ϯ 53.9 Ϫ7.1 (Ϫ25.9 to 11.7) 0.45 With albumin Ͼ30 mg/24 h (m) 12 5 Ϫ7 8 6 Ϫ2 NA NA Blood pressure (n ϭ 48 vegan, n ϭ 50 ADA) (mmHg)§ Systolic 123.8 Ϯ 17.1 120.0 Ϯ 18.3 Ϫ3.8 Ϯ 12.6ʈ 122.9 Ϯ 15.1 119.4 Ϯ 16.5 Ϫ3.6 Ϯ 13.7 Ϫ0.2 (Ϫ5.5 to 5.1) 0.93 Diastolic 77.9 Ϯ 11.1 72.8 Ϯ 10.2 Ϫ5.1 Ϯ 8.3* 80.0 Ϯ 10.5 76.7 Ϯ 11.1 Ϫ3.3 Ϯ 8.8ʈ Ϫ1.8 (Ϫ5.2 to 1.6) 0.30 All participants Total cholesterol (mg/dl) 187.0 Ϯ 37.4 159.3 Ϯ 31.9 Ϫ27.7 Ϯ 28.5* 198.9 Ϯ 44.0 174.6 Ϯ 36.2 Ϫ24.2 Ϯ 30.5* Ϫ3.5 (Ϫ15.3 to 8.3) 0.56 HDL cholesterol (mg/dl) 52.3 Ϯ 19.7 47.3 Ϯ 16.9 Ϫ5.0 Ϯ 7.1* 49.8 Ϯ 14.5 46.6 Ϯ 11.8 Ϫ3.2 Ϯ 11.0ʈ Ϫ1.8 (Ϫ5.5 to 1.9) 0.34 Non-HDL cholesterol (mg/dl) 134.7 Ϯ 39.2 112.0 Ϯ 31.9 Ϫ22.7 Ϯ 28.2* 149.0 Ϯ 44.1 128.0 Ϯ 35.0 Ϫ21.0 Ϯ 31.5* Ϫ1.7 (Ϫ13.6 to 10.2) 0.78 Total cholesterol/HDL (mg/dl) 4.0 Ϯ 1.6 3.7 Ϯ 1.2 Ϫ0.3 Ϯ 0.9ʈ 4.3 Ϯ 1.7 3.9 Ϯ 1.2 Ϫ0.4 Ϯ 1.2ʈ Ϫ0.0 (Ϫ0.4 to 0.4) 0.92 LDL cholesterol (n ϭ 49 vegan, n ϭ 48 104.4 Ϯ 32.9 88.0 Ϯ 27.8 Ϫ16.4 Ϯ 30.6† 118.5 Ϯ 41.5 103.1 Ϯ 33.3 Ϫ15.4 Ϯ 25.1* Ϫ1.0 (Ϫ12.3 to 10.3) 0.86 ADA) (mg/dl)¶ VLDL cholesterol (n ϭ 47 vegan, n ϭ 47 26.2 Ϯ 14.4 23.0 Ϯ 10.2 Ϫ3.2 Ϯ 10.0ʈ 26.8 Ϯ 13.8 22.3 Ϯ 10.0 Ϫ4.4 Ϯ 12.4ʈ 1.2 (Ϫ3.4 to 5.8) 0.60 ADA) (mg/dl) Triglycerides (mg/dl) 148.1 Ϯ 112.5 119.7 Ϯ 56.0 Ϫ28.5 Ϯ 80.0ʈ 158.1 Ϯ 133.1 132.9 Ϯ 114.4 Ϫ25.1 Ϯ 124.7 Ϫ3.3 (Ϫ45.2 to 38.6) 0.87 Log triglycerides 2.08 Ϯ 0.28 2.03 Ϯ 0.21 Ϫ0.05 Ϯ 0.17ʈ 2.11 Ϯ 0.25 2.05 Ϯ 0.23 Ϫ0.07 Ϯ 0.20ʈ 0.02 (Ϫ0.05 to 0.09) 0.61DIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006
  5. 5. Barnard and Associates Data are means Ϯ SD unless otherwise indicated. Listed P values are for comparisons of between-group (vegan vs. ADA) changes (baseline to 22 weeks).*P Ͻ 0.0001, †P Ͻ 0.001, ‡P Ͻ 0.01, and ʈP Ͻ 0.05 for within-group changes. §Blood pressure was not determined on one vegan group participant due to equipment failure. ¶When triglycerides exceeded 400 mg/dl, LDL was calculated via direct-LDL; two ADA-group the Pearson’s correlation of weight change 0.01 0.14 0.05 0.98 0.02 0.89 0.98 0.68 with A1C change was r ϭ 0.51, P Ͻ 0.0001 (within the vegan group [n ϭ 24], r ϭ 0.39, P ϭ 0.05; within the ADA group Ϫ14.5 (Ϫ25.8 to Ϫ3.2) Ϫ11.9 (Ϫ22.2 to Ϫ1.7) 0.6 (Ϫ45.9 to 47.2) Ϫ11.3 (Ϫ22.9 to 0.3) [n ϭ 33], r ϭ 0.49, P ϭ 0.004). individuals were excluded due to lack of sufficient plasma samples. SI conversion: to convert HDL, LDL, and total cholesterol to mmol/l, multiply by 0.0259; for tryiglycerides, multiply by 0.0113. Ϫ3.2 (Ϫ7.5 to 1.1) Ϫ0.0 (Ϫ0.4 to 0.4) 0.4 (Ϫ4.8 to 5.6) 0.02 (Ϫ0.1 to 0.1) Body weight fell 5.8 kg in the vegan group (P Ͻ 0.0001) and 4.3 kg in the ADA group (P Ͻ 0.0001) (between-group P ϭ 0.082; baseline-adjusted P ϭ 0.066). Among medication-stable participants, vegan participants lost 6.5 kg compared with 3.1 kg for ADA participants (P Ͻ 0.001; baseline-adjusted P ϭ 0.001). Ϫ19.0 Ϯ 28.5* Ϫ2.8 Ϯ 11.6 Ϫ16.3 Ϯ 30.1ʈ Ϫ0.3 Ϯ 1.2 Ϫ10.7 Ϯ 23.3ʈ Ϫ3.8 Ϯ 12.1 134.6 Ϯ 122.9 Ϫ22.8 Ϯ 134.3 Ϫ0.06 Ϯ 0.21 The reduction in urinary albumin was significant in the vegan group (P ϭ 0.002) but not in the ADA group (P ϭ 0.14). The unadjusted between-group difference was not significant. However, after adjust- ment for baseline values, the effect of diet 175.9 Ϯ 36.2 46.4 Ϯ 12.2 129.4 Ϯ 35.9 4.0 Ϯ 1.3 104.6 Ϯ 33.7 21.7 Ϯ 9.0 2.05 Ϯ 0.24 was significant (P ϭ 0.013). For the entire sample, there were no between-group differences in lipid val- ues. Among those whose lipid-controlling medications remained constant (80% [39 of 49] of vegan group, 82% [41 of 50] of 25.5 Ϯ 13.2 157.4 Ϯ 143.0 2.11 Ϯ 0.25 194.9 Ϯ 40.9 49.2 Ϯ 15.5 145.7 Ϯ 42.9 4.4 Ϯ 1.8 115.3 Ϯ 40.4 ADA group), reductions in total choles- terol were Ϫ0.866 mmol/l (Ϫ33.5 mg/dl, Ϫ17.6%) for the vegan group and Ϫ0.491 mmol/l (Ϫ19.0 mg/dl, Ϫ9.7%) for the ADA group (P ϭ 0.0125). Changes in LDL cholesterol were Ϫ0.58 mmol/l Ϫ3.5 Ϯ 10.5ʈ Ϫ22.2 Ϯ 58.5ʈ Ϫ0.04 Ϯ 0.16 Ϫ33.5 Ϯ 21.5* Ϫ6.0 Ϯ 6.8* Ϫ27.6 Ϯ 21.1* Ϫ0.3 Ϯ 0.6‡ Ϫ22.6 Ϯ 22.0* (Ϫ22.6 mg/dl, Ϫ21.2%) for the vegan group and Ϫ0.277 mmol/l (Ϫ10.7 mg/dl, Ϫ9.3%) for the ADA group (P ϭ 0.023). Changes in HDL cholesterol were Ϫ0.16 mmol/l (Ϫ6.0 mg/dl, Ϫ11.0%) for the vegan group and Ϫ0.07 mmol/l (Ϫ2.8 23.1 Ϯ 10.9 118.2 Ϯ 57.3 2.02 Ϯ 0.22 156.9 Ϯ 25.1 48.6 Ϯ 18.4 108.3 Ϯ 25.6 3.6 Ϯ 1.2 84.6 Ϯ 22.5 mg/dl, Ϫ5.7%) for the ADA group (P ϭ 0.14). The total-to-HDL cholesterol ratio fell for both groups, as did triglyceride concentrations. There were no treatment-related seri- 26.6 Ϯ 15.4 140.3 Ϯ 89.1 2.06 Ϯ 0.28 190.5 Ϯ 36.8 54.6 Ϯ 21.0 135.9 Ϯ 38.4 3.9 Ϯ 1.5 107.3 Ϯ 34.3 ous adverse events. CONCLUSIONS — Both diets were associated with significant clinical im- provements, as indicated by reductions in A1C, body weight, plasma lipid concen- controlling medications (n ϭ 39 vegan, VLDL cholesterol (n ϭ 38 n ϭ vegan, 38 LDL cholesterol (n ϭ 39 vegan, n ϭ 39 trations, and urinary albumin excretion. Among medication-stable participants, Participants with no changes to lipid- changes in A1C, weight, BMI, waist cir- n ϭ 41 ADA, except as noted) Total cholesterol/HDL (mg/dl) cumference, total cholesterol, and LDL Non-HDL cholesterol (mg/dl) cholesterol were significantly greater in Total cholesterol (mg/dl) HDL cholesterol (mg/dl) the vegan group. The magnitude of A1C Triglycerides (mg/dl) reduction in medication-stable vegan group participants, 1.23 percentage Log triglycerides ADA) (mg/dl)¶ points, compares favorably with that ob- ADA) (mg/dl) served in single-agent therapy with oral diabetes drugs (15). A low-fat plant-based diet influences nutrient intake and body composition in several ways that may, in turn, affect in-DIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006 1781
  6. 6. Low-fat vegan diet and type 2 diabetes portions can, as a result, easily exceed rec- ommended limits on saturated fat. In con- trast, the vegan diet includes no animal fat, so variations in food quantity are less likely to result in substantial increases in saturated fat intake. Because the vegan diet is based on the elimination of certain foods, it may be easier to understand than regimens that limit quantities of certain foods without proscribing any. The ac- ceptability of low-fat vegan diets in clini- cal studies is similar to that of seemingly more moderate diets (27). This study’s strengths include its analysis of dependent measures without regard to variations in dietary adherence and applicability outside the research set- ting. A study limitation was that both di- ets made participants vulnerable to theFigure 1—A1C at baseline and at 11 and 22 weeks. Open circles: all vegan group participants hypoglycemic effect of their diabetes(n ϭ 49); closed circles: medication-stable vegan group participants (n ϭ 24); open squares: allADA group participants (n ϭ 50); closed squares: medication-stable ADA group participants (n ϭ medications, resulting in medication re-33). Error bars represent SE of the mean. P ϭ 0.09 for between-group comparison from baseline ductions that confound the interpretationto 22 weeks for full sample; P ϭ 0.01 for medication-stable participants (vegan vs. ADA). of A1C changes and necessitating a sub- group analysis of medication-stable par- ticipants. Because these episodessulin sensitivity. First, because such diets fore diabetes manifests (20). This lipid occurred early in the trial, there was noare low in fat and high in fiber, they typ- accumulation may be responsive to diet. opportunity to bring interim laboratoryically cause associated reductions in di- High-fat diets appear to downregulate the values forward. Also, most study partici-etary energy density and energy intake, genes required for mitochondrial oxida- pants were taking antihypertensive med-which are not fully compensated for by tive phosphorylation in skeletal muscle ications, which may have blunted theincreased food intake (16,17). Our data (21). In contrast, a case-control study effect of diet on blood pressure.suggest that the weight-reducing effect of found that soleus muscle intramyocellu- In conclusion, in individuals withthe vegan diet (4) is responsible for a sub- lar lipid concentrations were significantly type 2 diabetes participating in a 22-weekstantial portion of its effect on A1C. lower in a group of 21 vegans compared clinical trial, both a low-fat vegan diet and Independent of their effect on body with 25 omnivores (Ϫ9.7 [95% CI Ϫ16.2 a diet following ADA guidelines improvedweight, reductions in total fat intake and to Ϫ3.3], P ϭ 0.01) (22). glycemic control; however, the changesin the proportion of dietary saturated to The lipid-lowering effect of vegan di- were greater in the vegan group. Furtherunsaturated fat increase insulin sensitivity ets, attributable to their absence of dietary research is necessary to establish longer-(18), as do increased intake of low– cholesterol, low saturated fat content, and term diet effects and sustainability.glycemic index and high-fiber foods (1). a specific cholesterol-reducing effect of Finally, limited evidence suggests soluble fiber and other plant constituentsthat reductions in iron stores increase in- (23), is particularly important given that Acknowledgments — The study was sup-sulin sensitivity (19). A vegan diet pro- cardiovascular complications are the pri- ported by grant R01 DK059362-01A2 fromvides iron in its nonheme form, which is mary cause of morbidity and mortality in the National Institute of Diabetes and Diges- tive and Kidney Diseases and by the Diabetessomewhat less absorbable than heme diabetes. While diets high in refined car- Action Research and Education Foundation.iron. A study comparing 30 ovolactoveg- bohydrate may increase triglyceride con- The authors express appreciation to Pauletarians and 30 meat eaters, all of whom centrations, high-fiber and low– glycemic Poppen, PhD, for statistical analyses.were healthy and had BMIs Ͻ23 kg/m2, index foods appear to have the oppositeshowed that vegetarians had adequate, result (24).but lower, body iron stores, compared The limited compliance of the ADA Referenceswith meat eaters (serum ferritin concen- group merits comment. Researchers have 1. Jenkins DJA, Kendall CWC, Marchie A,tration 35 ␮g/l [95% CI 21– 49] vs. 72 long lamented the difficulties in adhering Jenkins AL, Augustin LSA, Ludwig DS,␮g/l [45–100]). The vegetarians also to diets for diabetes (25). The A1C reduc- Barnard ND, Anderson JW: Type 2 diabe-demonstrated less insulin resistance tion observed in the ADA group was sim- tes and the vegetarian diet. Am J Clin Nutr(steady-state plasma glucose concentra- ilar to that found in previous studies (26). 78:610S– 616S, 2003tion 4.1 mmol/l [3.5–5.0] vs. 6.9 mmol/l A potential weakness of the ADA guide- 2. Fraser GE: Vegetarianism and obesity, hy- pertension, diabetes, A comprehensive and arthritis. In Diet,[5.2–7.5], respectively) (19). lines is that they require portion size lim- Life Expectancy, and Chronic Disease. Ox- Insulin resistance is related to lipid its for overweight individuals, and bibliography is ford, U.K., Oxford University Press,accumulation within muscle cells (in- limitations on saturated-fat intake are 2003, p. 129 –148 typical in scholarlytramyocellular lipid), apparently due to a based on these limited energy intakes. In- 3. Nicholson AS, Sklar M, Barnard ND, Gore publicationsgenetically based reduction in mitochon- dividuals who exceed their prescribed en- S, Sullivan R, Browning S: Toward im-drial activity identifiable many years be- ergy intake limits with overly large proved management of NIDDM: a ran-1782 DIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006
  7. 7. Barnard and Associates domized, controlled, pilot intervention density lipoprotein cholesterol in plasma phosphorylation in skeletal muscle. Dia- using a low-fat, vegetarian diet. Prev Med without use of the preparative ultracentri- betes 54:1926 –1933, 2005 29:87–91, 1999 fuge. Clin Chem 18:499 –502, 1972 22. Goff LM, Bell JD, So PW, Dornhorst A, 4. Barnard ND, Scialli AR, Turner-McGrievy 13. Corcoran RM, Durnan SM: Albumin de- Frost GS: Veganism and its relationship G, Lanou AJ, Glass J: The effects of a low- termination by a modified bromocresol with insulin resistance and intramyocel- fat, plant-based dietary intervention on green method (Letter). Clin Chem 23: lular lipid. Eur J Clin Nutr 59:291–298, body weight, metabolism, and insulin 765–766, 1977 2005 sensitivity. Am J Med 118:991–997, 2005 14. Bouchard C, Tremblay A, LeBlanc C, 23. Jenkins DJ, Kendall CW, Marchie A, 5. Wilson DH, Bogacz JP, Forsythe CM, Lortie G, Savard R, Theriault G: A method Faulkner DA, Wong JM, de Souza R, Turk PJ, Lane TL, Gates RC, Brandt DR: to assess energy expenditure in children Emam A, Parker TL, Vidgen E, Lapsley Fully automated assay of glycohemoglo- and adults. Am J Clin Nutr 37:461– 467, KG, Trautwein EA, Josse RG, Leiter LA, bin with the Abbott IMx analyzer: novel 1983 Connelly PW: Effects of a dietary portfolio approaches for separation and detection. 15. Krentz AJ, Bailey CJ: Oral antidiabetic Clin Chem 39:2090 –2097, 1993 agents: current role in type 2 diabetes on cholesterol-lowering foods vs lova- 6. American Diabetes Association: Evi- mellitus (Review). Drugs 65:385– 411, statin on serum lipids and C-reactive pro- dence-based nutrition principles and rec- 2005 tein. JAMA 290:502–510, 2003 ommendations for the treatment and 16. Kendall A, Levitsky DA, Strupp BJ, Liss- 24. Jenkins DJ, Wolever TM, Kalmusky J, prevention of diabetes and related com- ner L: Weight loss on a low-fat diet: con- Guidici S, Giordano C, Patten R, Wong plications (Position Statement). Diabetes sequence of the imprecision of the control GS, Bird JN, Hall M, Buckley G, et al.: Care 26 (Suppl. 1):S51–S61, 2003 of food intake in humans. Am J Clin Nutr Low-glycemic index diet in hyperlipid- 7. Schakel SF, Sievert YA, Buzzard IM: 53:1124 –1129, 1991 emia: use of traditional starchy foods. Sources of data for developing and main- 17. Howarth NC, Saltzman E, Roberts SB: Di- Am J Clin Nutr 46:66 –71, 1987 taining a nutrient database. J Am Diet As- etary fiber and weight regulation (Re- 25. Laitinen JH, Ahola IE, Sarkkinen ES, Win- soc 88:1268 –1271, 1988 view). Nutr Rev 59:129 –139, 2001 berg RL, Harmaakorpi-Iivonen PA, Uus- 8. Barthelmai W, Czok R: Enzymatic deter- 18. Lovejoy JC, Windhauser MM, Rood JC, itupa MI: Impact of intensified dietary minations of glucose in the blood, cere- de la Bretonne JA: Effect of a controlled therapy on energy and nutrient intakes brospinal fluid and urine. Klin Wochenschr high-fat versus low-fat diet on insulin sen- and fatty acid composition of serum lipids 40:585–589, 1962 [in German] sitivity and leptin levels in African-Amer- in patients with recently diagnosed non- 9. Allain CC, Poon LS, Chan CSG, Rich- ican and Caucasian women. Metabolism insulin-dependent diabetes mellitus. J Am mond W, Fu PC: Enzymatic determina- 47:1520 –1524, 1998 Diet Assoc 93:276 –283, 1993 tion of total serum cholesterol. Clin Chem 19. Hua NW, Stoohs RA, Facchini FS: Low 26. Franz MJ, Splett PL, Monk A, Barry B, 20:470 – 475, 1974 iron status and enhanced insulin sensitiv- McClain K, Weaver T, Upham P, Bergen-10. Wieland H, Seidel D: A simple specific ity in lacto-ovo vegetarians. Br J Nutr 86: method for precipitation of low density 515–519, 2001 stal R, Mazze RS: Cost-effectiveness of lipoproteins. J Lipid Res 24:904 –909, 20. Petersen KF, Dufour S, Befroy D, Garcia medical nutrition therapy provided by di- 1983 R, Shulman GI: Impaired mitochondrial etitians for persons with non-insulin-de-11. Finley PR, Schifman RB, Williams RJ, activity in the insulin-resistant offspring pendent diabetes mellitus. J Am Diet Assoc Licht DA: Cholesterol in high-density li- of patients with type 2 diabetes. N Engl 95:1018 –1024, 1995 poprotein: use of Mg2ϩ/dextran sulfate J Med 350:664 – 671, 2004 27. Barnard ND, Scialli AR, Turner-McGrievy in its enzymatic measurement. Clin Chem 21. Sparks LM, Xie H, Koza RA, Mynatt R, GM, Lanou AJ: Acceptability of a low-fat 24:931–933, 1978 Hulver MW, Bray GA, Smith SR: A high- vegan diet compares favorably to a step II12. Friedewald WT, Levy RI, Fredrickson DS: fat diet coordinately downregulates genes diet in a randomized, controlled trial. Estimation of the concentration of low- required for mitochondrial oxidative J Cardiopulm Rehabil 24:229 –235, 2004DIABETES CARE, VOLUME 29, NUMBER 8, AUGUST 2006 1783