Blood-type diet and ABO genotype: a 2014 study

In a cross-sectional analysis of 1,455 young adults, some relative blood-type diet-score patterns were associated with cardiometabolic biomarkers, but matching a score to ABO genotype did not add a biomarker benefit. The study measured no Candida or infection outcome and cannot inform Candida treatment.

The study tests part of the blood-type diet hypothesis by comparing relative pattern scores with ABO genotype across cardiometabolic markers. Its results are relevant to that cardiometabolic question only. The broad Candida claims evidence map keeps it separate from evidence about Candida treatment.

The Toronto Nutrigenomics and Health study enrolled young adults ages 20–29. Of 1,639 participants, 1,455 were analyzed (993 women and 462 men; mean age about 22.7 years). The analyzed group included 703 White, 491 East Asian, 155 South Asian, and 106 other participants. The paper's wording does not make clear whether all exclusions for energy under- or over-reporting are included in the 184 exclusions it attributes to missing biomarkers or ABO genotype.

Participants completed a one-month, 196-item food-frequency questionnaire. Researchers calculated four separate scores for every participant from foods recommended or discouraged by each promoted blood-type diet, then compared tertiles; T3 meant closer relative resemblance, not confirmed adherence to a prescribed diet. ABO group was assigned by genotyping. Adjusted outcomes included body measurements, blood pressure, fasting blood biomarkers, and calculated HOMA-IR and HOMA-Beta. No diet was assigned, no follow-up was conducted, and no Candida or infection outcome was measured.

The values below are adjusted means ± SE as printed. The overall analyzed denominator was N=1,455, but Tables 2–5 do not report the number of participants in each diet-score tertile. All p values from Tables 2–5 cited here are overall ANCOVA tests across T1/T2/T3, not pairwise p values for displayed T1/T3 means; Tukey–Kramer footnotes identify the specific pairwise differences. For the Type-A score, T1 versus T3 means were BMI 23.7±0.2 versus 23.1±0.2 kg/m² (overall p=.03), systolic pressure 117.6±0.5 versus 115.4±0.5 mm Hg (p<.001), diastolic pressure 70.4±0.4 versus 69.4±0.4 mm Hg (p=.009), waist 77.0±0.4 versus 75.6±0.4 cm (p=.02), insulin 52.7±1.8 versus 46.3±1.8 pmol/L (p=.002), HOMA-IR 1.61±0.06 versus 1.41±0.06 (p=.002), HOMA-Beta 111.7±3.7 versus 101.2±3.8 (p=.007), total cholesterol 4.26±0.04 versus 4.14±0.04 mmol/L (p=.02), and triglycerides 1.00±0.02 versus 0.92±0.02 mmol/L (p=.005). The table footnotes show T1/T2 greater than T3 for systolic pressure, insulin, and HOMA measures; for BMI, waist, total cholesterol, and triglycerides, the reported pairwise difference is T1 greater than T3. These means are not individual changes.

For Type AB, T1 versus T3 means were systolic pressure 117.3±0.5 versus 115.5±0.5 mm Hg (overall p=.006), diastolic pressure 70.8±0.4 versus 69.4±0.4 mm Hg (p=.02), insulin 56.1±1.7 versus 45.6±1.9 pmol/L (p<.001), HOMA-IR 1.72±0.06 versus 1.39±0.06 (p<.001), HOMA-Beta 119.0±3.7 versus 98.0±3.9 (p<.001), total cholesterol 4.27±0.04 versus 4.16±0.04 mmol/L (p=.03), and triglycerides 1.01±0.02 versus 0.93±0.02 mmol/L (p=.004). These p values are overall tests across three tertiles. Tukey–Kramer notes show T1>T3 for blood pressure, total cholesterol, and triglycerides; all insulin and HOMA-IR tertiles differed, while T1/T2 exceeded T3 for HOMA-Beta. Insulin means for T1/T2/T3 were 56.1±1.7, 49.8±1.8, and 45.6±1.9 pmol/L. BMI and waist were not significant (p=.1 and p=.08). For Type O, the overall triglyceride p value was <.001; means were 1.04±0.03, 0.96±0.02, and 0.91±0.02 mmol/L across T1/T2/T3, with Tukey–Kramer notes showing T1>T2/T3. Other Type-O outcomes were not significant. For Type B, HDL had an overall p=.04 but no pairwise tertile difference; means were 1.47±.02, 1.47±.02, and 1.43±.02 mmol/L, and no other pairwise differences were reported.

In matching analyses, subgroup counts across T1/T2/T3 were 177/182/185 matched to Type A and 307/295/309 with other genotypes; for Type AB, 26/35/30 were matched and 464/441/459 had other genotypes. These counts total 1,455 per score, but the paper does not give separate denominators for every outcome. Type-A score interactions were reported for glucose (p=.02), insulin (p=.02), and HOMA-IR (p=.01). Only in T2 were insulin (60.5±2.7 versus 48.6±2.2 pmol/L) and HOMA-IR (1.87±0.09 versus 1.48±0.07) higher in matched group A than the other genotypes; glucose showed no within-tertile difference. The Type-AB glucose interaction had p=.02, but no within-tertile matched-versus-other difference was found; matched-AB means increased from 4.74±0.07 to 4.95±0.06 mmol/L across tertiles. Type-B and Type-O interactions were not significant. These nominal interaction results arose among many tests, and the interaction findings were not Bonferroni-corrected. Overall, matching did not add benefit to the cardiometabolic associations. No clinical infection outcome was measured.

The findings apply to relative diet-score patterns and cardiometabolic biomarkers in this selected young-adult Canadian cohort. They do not establish effects of a prescribed or strictly followed diet, do not describe Candida or any infection, and do not show whether anyone would benefit from choosing foods by blood type.

The cross-sectional design cannot determine which came first or whether diet caused the biomarker associations. Food intake was self-reported, strict adherence was not counted, and each person received all four relative diet scores. The study examined many outcomes and interactions; the authors did not apply a more conservative Bonferroni correction to interaction findings. The matched-AB subgroups were small (n=26, 35, and 30). Confidence intervals and main-tertile participant counts were not reported. Participant-flow wording leaves unclear whether energy-report exclusions are included in the stated 184 missing-data exclusions. The cohort was limited to ages 20–29, and adverse events, Candida, infections, symptoms, and treatment outcomes were not assessed.

Contextual. This is a cross-sectional cardiometabolic study of relative blood-type diet scores and ABO-genotype matching, not a Candida study. It found no added biomarker benefit from matching overall, but it provides no evidence about Candida treatment efficacy.

No practical change for Candida-related care. These cardiometabolic associations do not support or refute a Candida-treatment effect and should not guide an individual's Candida care or diet choice.

Funding came from Advanced Foods and Materials Network grant 305352 (to Ahmed El-Sohemy); Jingzhou Wang held an Ontario Graduate Scholarship. The funders had no role in the study or publication. Ahmed El-Sohemy held shares in Nutrigenomix Inc., a personalized-nutrition genetic-testing company.

Read the original PLOS ONE article (published January 15, 2014; DOI: 10.1371/journal.pone.0084749). The full text is also available from PubMed Central.