Associations of obesity and body shape with erythrocyte and reticulocyte parameters in the UK Biobank cohort

Background Obesity is associated with type 2 diabetes mellitus and chronic low-grade inflammation. Although chronic inflammatory conditions and diabetes are associated with anaemia, less is known about associations of obesity and body shape, independent of each other, with erythrocyte and reticulocyte parameters. Methods We investigated the associations of body mass index (BMI) and the allometric body shape index (ABSI) and hip index (HI), which are uncorrelated with BMI, with erythrocyte and reticulocyte parameters (all continuous, on a standard deviation (SD) scale) in UK Biobank participants without known metabolic, endocrine, or major inflammatory conditions (glycated haemoglobin HbA1c < 48 mmol/mol, C-reactive protein CRP < 10 mg/L). We examined erythrocyte count, total reticulocyte count and percent, immature reticulocyte count and fraction (IRF), haemoglobin, haematocrit, mean corpuscular haemoglobin mass (MCH) and concentration (MCHC), mean corpuscular and reticulocyte volumes (MCV, MRV), and red cell distribution width (RDW) in multivariable linear regression models. We additionally defined body shape phenotypes with dichotomised ABSI (≥ 73 women; ≥ 80 men) and HI (≥ 64 women; ≥ 49 men), including “pear” (small-ABSI-large-HI) and “apple” (large-ABSI-small-HI), and examined these in groups according to BMI (18.5–25 normal weight; 25–30 overweight; 30–45 kg/m2 obese). Results In 105,853 women and 100,854 men, BMI and ABSI were associated positively with haemoglobin, haematocrit, and erythrocyte count, and more strongly with total reticulocyte count and percent, immature reticulocyte count and IRF. HI was associated inversely with all, but least with IRF. Associations were comparable in women and men. In groups according to obesity and body shape, erythrocyte count was ~ 0.6 SD higher for obese-“apple” compared to normal-weight-“pear” phenotype (SD = 0.31*1012/L women, SD = 0.34*1012/L men), total reticulocyte count was ~ 1.1 SD higher (SD = 21.1*109/L women, SD = 23.6*109/L men), immature reticulocyte count was ~ 1.2 SD higher (SD = 7.9*109/L women, SD = 8.8*109/L men), total reticulocyte percent was ~ 1.0 SD higher (SD = 0.48% women and men), and IFR was over 0.7 SD higher (SD = 5.7% women and men). BMI but not ABSI or HI was associated more weakly inversely with MCV, MRV, and MCH, but positively with MCHC in men and RDW in women. Conclusions In obesity uncomplicated with diabetes, larger BMI and ABSI are associated with increased erythropoiesis and reticulocyte immaturity. Supplementary Information The online version contains supplementary material available at 10.1186/s12902-023-01423-1.

Vitamin B12 or folic acid (for exclusions) was based on Fields [6155-0.0…5] "Vitamin and mineral supplements", coding values as Yes for Answer: 6 "Folic acid or Folate (Vit B9)", as No for answers 1,2,3,4,5,7, or -7, and   For all meat variables above, the categorical answers were converted to a continuous scale as follows: Answer 0 "Never" remained 0; Answer 1 "Less than once a week" was coded as 0.5; Answer 2 "Once a week" was coded as 1; Answer 3 "2-4 times a week" was coded as 3; Answer 4 "5-6 times a week" was coded as 5.5; Answer 5 "Once or more daily" was coded as 7.
Cheese intake (for adjustment) was based on Field [1408-0.0] "Cheese intake". The categorical answers were converted to a continuous scale as for meat variables. Additional information was obtained from Field [6144-0.0] "Never eat eggs, dairy, wheat, sugar". Participants providing Answer 2: "Dairy products" were recoded as 0.
For all dietary variables, Answer: -10 "Less than one" was re-coded to 0.5 and Answers: -1 "Do not know" and -3 "Prefer not to answer" were considered missing.
Diet change (for exclusions and adjustment) was based on Field [1538-0.0] "Major dietary changes in the last 5 years". Participants providing Answers: 1 "Yes, because of illness"; -3 "prefer not to answer", and those with missing values were excluded from the study. Answer 2: "Yes, because of other reason" was coded as Yes. Answer: 0 "No" was coded as No.

S8
Statistical tests for heterogeneity ABSI -a body shape index; BMI -body mass index; HI -hip index.
To evaluate associations with body shape phenotypes overall, we used a likelihood ratio test comparing a model including BMI categories and covariates with a model additionally including ABSI-by-HI cross-classification, separately in women and men.
To evaluate heterogeneity of the associations with body shape phenotypes according to BMI, we used a likelihood ratio test comparing the additive model, including body shape phenotypes (ABSI-by-HI cross-classification), BMI categories, and covariates, with the interaction model, including BMI-by-ABSI-by-HI cross-classification and covariates, separately in women and men.
To evaluate heterogeneity in the associations with BMI according to sex, we used the interaction term between BMI (sex-specific z-scores) and sex (women reference) from a joint model, including women and men, adjusted for ABSI and HI (sex-specific z-scores) and all covariates (except female-specific). To evaluate heterogeneity in the associations with body shape indices according to sex, we used the interaction terms between ABSI and sex and HI and sex from a joint model, including women and men, adjusted for BMI and all covariates (except female-specific). To account for potential differences by menopausal status in women, we additionally included an interaction term between age and sex in all models examining differences according to sex.
To evaluate heterogeneity in the associations with BMI according to menopausal status, we used the interaction term between BMI (sex-specific z-scores) and menopausal status (premenopausal reference) from a joint model, including pre-menopausal and post-menopausal women, adjusted for ABSI and HI (sex-specific z-scores) and all covariates (including femalespecific). To evaluate heterogeneity in the associations with body shape indices according to menopausal status, we used the interaction terms between ABSI and menopausal status and HI and menopausal status from a joint model, including pre-menopausal and post-menopausal women, adjusted for BMI and all covariates (including female-specific).

S10
The exclusion criteria were applied sequentially in the displayed order, counting each excluded individual only once. i -see lists of medications in Supplementary Methods (above).

Supplementary Material
Obesity, body shape and red blood cells in UK Biobank Christakoudi et al. 2023 S12 Supplementary  SDchange (95% CI) -regression coefficient (95% confidence interval) interpreted as the change in haematological parameters on SD scale per one SD increase of each anthropometric index. Estimates were obtained from multivariable linear regression models including each erythrocyte or reticulocyte parameter as an outcome variable (sex-specific z-scores, value minus mean divided by SD). Exposure variables included BMI, ABSI, and HI (sexspecific z-scores). Adjustment variables included height, age, recent weight change, smoking status, alcohol consumption, physical activity, Townsend deprivation index, dietary intake (fruit, vegetables, red meat, processed meat, poultry, fish, cheese, bread, cereals, tea, coffee), recent major dietary change, region of the assessment centre, time of blood collection, fasting time, use of nonsteroidal anti-inflammatory drugs, use of vitamins ACE, vitamin D and minerals, or multivitamins, and for women, menopausal status, hormonal replacement therapy use, oral contraceptives use, and age at the last live birth.
Plots are shown in Figure 1.
p sex BMI -p-value for the interaction term between BMI and sex, from a joint model including women (reference) and men, with adjustment for ABSI, HI, and all covariates, except female-specific, but including an interaction term between age and sex, to account for potential differences by menopausal status in women (p<1*10 -6 bold).
p sex ABSI / HI -p-value for the interaction term between ABSI and sex or HI and sex, from a joint model including women (reference) and men, with adjustment for BMI and all covariates, except female-specific, but including an interaction term between age and sex. * p<0.001; ** p<1*10 -6 from Wald test for the individual term.

Supplementary Material
Obesity, body shape and red blood cells in UK Biobank Christakoudi et al. 2023