Different Associations of Anthropometric Indices with Diabetic Retinopathy and Diabetic Kidney Disease in Chinese Patients with Type 2 Diabetes Mellitus

Aims: To investigate the associations of anthropometric indices, including body mass index (BMI), waist-to-hip ratio (WHR), waist-to-height ratio (WHtR), waist circumference (WC) and hip circumference (HC), with diabetic retinopathy (DR) and diabetic kidney disease (DKD) in Chinese patients with type 2 diabetes mellitus (T2DM).

This cross-sectional study evaluated 5226 participants with T2DM at Shanghai General Hospital between 2005 and 2016. Logistic regression models and restricted cubic spline analysis were used to assess the associations of anthropometric indices with DR and DKD.

A BMI of around 25 kg/m2 was related to a low risk of DR (OR based on the third fifth: 0.752, 95%CI: 0.615-0.920). Besides, HC had an inverse association with DR in men independently of BMI (OR based on the highest fifth: 0.495, 95%CI: 0.350-0.697). In the restricted cubic spline models, BMI, WHtR, WC, and HC showed J-shaped associations with DKD, while WHR showed an S-shaped association with DKD. Compared to the lowest fifth, the . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 4 odds ratios (OR) based on the highest fifth of BMI, WHR, WHtR, WC and HC for DKD were 1.927 (1.572-2.366), 1.566 (1.277-1.923), 1.91 (1.554-2.351), 1.91 (1.554-2.351) and 1.585 (1.300-1.937) respectively in multivariable models.

A median BMI and a large hip might be related to a low risk of DR, while lower levels of all the anthropometric indices were associated with a lower risk of DKD. Our findings suggested maintain a median BMI, a low WHR, a low WHtR and a large hip for prevention of DR and DKD.

anthropometric indices, associations, diabetic retinopathy, diabetic kidney disease. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 5 1. Introduction The global prevalence of diabetes has increased over the past decades and is expected to rise from an estimated 422 million people in 2014 to 783 million by 2045 12. Type 2 diabetes mellitus (T2DM) accounts for around 90% of all diabetes cases. Diabetic retinopathy (DR) and diabetic kidney disease (DKD) are the major microvascular complications of diabetes, which can greatly reduce the quality of life and place a significant burden on health care costs. DR is the leading cause of vision loss and blindness worldwide. More than 60% of patients will eventually develop DR after 20 years of T2DM 3. DKD can develop in about 40% of patients with diabetes, leading to an increased risk of frailty, end-stage renal disease (ESRD), and premature mortality 4. Therefore, early detection and treatment of microvascular complications of diabetes is an important public health problem. DR and DKD are usually comorbid, and they share some common pathogenic mechanisms, such as angiogenesis, oxidative stress and inflammation 5. However, there are apparent differences between the pathophysiologic . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 6 features of DR and DKD: DR is characterized by vascular permeability and retinal neovascularization, while DKD is featured with glomerular sclerosis. It has been revealed that the associations of several clinical indicators with DR and DKD are different, for example, C peptide 6, glycated albumin to glycated hemoglobin (HbA1c) ratio 7, and age 8. Obesity has been established as a risk factor for T2DM 910. However, the associations of obesity-related anthropometric indices, including body mass index (BMI), waist-to-hip ratio (WHR), waist-to-height ratio (WHtR), waist circumference (WC) and hip circumference (HC), with DR and DKD have not been systemically reported. According to our knowledge, evidence for these anthropometric indices with DR and DKD has been inconclusive. For instance, conflicting results were reported for the associations of BMI, WC, WHR and WHtR with DR, as positive 111213, inverse 141516 and no associations 17. Although numerous studies suggested that BMI, WC, WHR and WHtR were associated with a greater likelihood of having DKD 18, one study found that WC, WHR and WHtR had no significant associations with DKD 19. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 7 Heterogeneous results were observed in previous studies where obesity- related parameters were analyzed as continuous variables or based on different categories 19, which probably indicated the existence of a non-linear trend. However, only a few studies were conducted to specifically analyze the non-linear associations between anthropometric indices and DR and DKD. In this study, we aim to assess the associations of obesity-related anthropometric indices, including BMI, WHR, WHtR, WC and HC, with the prevalence of DR and DKD in 5226 Chinese individuals with T2DM. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 8 2. Materials and methods 2.1 Study design and participants This cross-sectional study was designed to investigate the associations of BMI, WHR, WHtR, WC and HC with DR and DKD among Chinese adults. Participants in this study were patients diagnosed with T2DM at Shanghai General Hospital between 2005 and 2016. Patients who were missing weight, height, WC or HC measurement or missing triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), HbA1c or systolic blood pressure (SBP) data or missing DR assessment information, or missing urinary albumin: creatinine (UACR) or estimated glomerular filtration rate (eGFR) information were excluded. Overall, 5226 patients were involved in this analysis. The study was performed following the guidelines of the World Medical Association Declaration of Helsinki and was approved by the Institutional Review Board of the Shanghai General Hospital, Shanghai Jiao Tong . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 9 University School of Medicine. Written informed consent was obtained from all patients or their legal guardians. 2.2 Data collection The information on sociodemographic characteristics and medical history was collected by trained doctors through a standardized interview. Height and weight were measured in cm and kg respectively with participants standing without shoes and in lightweight clothes. WC (in cm) was measured on the mid-axillary line between the lowest border of the rib cage and the top of the iliac crest. HC (in cm) was measured at the widest part of the hip at the level of the greater trochanter. BMI was calculated as weight in kilograms divided by squared height in meters. WHR was calculated as waist circumference divided by hip circumference. WHtR was calculated as waist circumference divided by height. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 10 SBP and diastolic blood pressure (DBP) values were measured with participants in a sitting position after 10 minutes of rest. Venous blood samples were drawn between 6:00 am and 9:00 am after an overnight fast. HbA1c, TC, TG, LDL, HDL, fasting blood glucose (FBG) and serum creatinine were assessed using a conventional automated blood analyzer. The 24-hour urine samples were collected during the period of hospitalization. The concentration of urine albumin and creatinine were measured with a turbidimetric immunoassay and an enzymatic method in a single 24-hour urine sample, respectively. Then the UACR was calculated. 2.3 Definition of Variables Hypertension was defined as SBP≥140 mmHg, DBP≥90 mmHg, or a self- reported previous diagnosis of hypertension. The eGFR was calculated according to the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation for “Asian origin”20. The . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 11 definition of DKD was UACR ≥30 mg/g or eGFR<60 mL/min*1.73 m^2, as suggested by the statement from the American Diabetes Association21. DR screening was conducted by experienced ophthalmologists, including a review of ophthalmologic history, measurement of visual acuity and intra- ocular pressure, slit lamp examination, and dilated fundus examination. The digital retinal photography was undertaken after pupil dilation by a digital retinal camera (Carl Zeiss Meditec AG, Jena, Germany), and the Optical coherence tomography was obtained by Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). Retinopathy was graded based on the worst eye by well-trained assessors according to the Early Treatment Diabetic Retinopathy Study grading system 22. 2.4 Statistical analysis Data analyses were performed with R 4.1.2 version. Continuous variables were expressed as the mean ± standard deviation (SD) or the median with an . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 12 interquartile range (25%, 75%), and categorical variables were presented as percentages (%). Baseline characteristics were compared using the Chi- square test for dichotomous variables, the Mann-Whitney U test or the Student’s t test for continuous variables as appropriate. P value (two-sided) <0.05 indicated significance. We calculated a Spearman correlation between every two variables of BMI, WHR, WHtR, WC and HC (Table S1). We used binomial logistic regression models to analyze the associations of these anthropometric indices with DR and DKD. BMI, WHR, WHtR, WC and HC were analyzed continuously and categorically in fifths. Data were summarized as odds ratios (OR) and 95% confidence intervals (95%CI). For the primary analysis, we conducted three adjusted regression models: model 1 adjusted for age and sex; model 2 adjusted for age, sex, duration of diabetes, SBP, TG, LDL, HDL and HbA1c; model 3 further including BMI for analyses of WHR, WHtR, WC and HC or including WHR for analyses of BMI. Ordinal logistic regression models were . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 13 used to assess the overall trend of categorical exposure with the presence of DR and DKD. We also used restricted cubic splines to examine the shape of the associations of BMI, WHR, WHtR, WC and HC with DR and DKD. The numbers of knots were set between 3 and 5 based on Akaike’s information criteria 23. The locations of knots were prespecified based on the quantiles of the continuous variable (Table S2). The median values of these variables were used as references. Age, sex, duration of diabetes, SBP, TG, LDL, HDL and HbA1c were adjusted in the spline models. P for non-linearity was tested with the Wald test, and a P value<0.05 was considered to show a significant non-linear trend. We also used a linear model to calculate OR per SD increase in a specific range of these variables. Because evidence indicated a significant interaction between sex and BMI, WHtR, WC or HC with DR and DKD (table S3), we conducted sex-stratified analyses of all these anthropometric indices with the presence of DR and DKD. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 14 We tested the consistency of our findings by using different categories for BMI. And we also did several sensitivity analyses of WC and HC with DR in men with additional adjustments for WHR, WC or HC. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 15 3. Results 3.1 General characteristics of study participants by DR and DKD We included 5226 patients with T2DM in this analysis. Overall, the mean age was 59.2 ± 12.0 years and of these patients, 1662 (31.8%) had DR and 1742 (33.3%) had DKD. Table1 summarizes the clinical and demographic characteristics of the study population in those with or without DR, and with or without DKD. No difference in BMI, WHtR and WC were found between patients with and without DR (all P >0.05). And patients with DR were likely to have a larger WHR and a smaller HC than those without DR (P <0.05). However, compared with patients without DKD, BMI, WHR, WHtR, WC and HC were significantly higher in those with DKD (all P <0.01). 3.2 Associations of anthropometric indices with DR and DKD . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 16 Table 2 and 3 show the associations of anthropometric indices with DR and DKD, which involved BMI, WHR, WHtR, WC and HC. HC showed an independent inverse association with DR. The OR based on the highest fifth of HC were 0.730 (0.600-0.889) in the multivariable model 2 adjusted for age, sex, duration of diabetes, SBP, TG, LDL, HDL and HbA1c, and 0.600 (0.463-0.776) in the model 3 further adjusted for BMI. Moreover, participants in the third fifth of BMI had the lowest OR for DR in the multivariable model 2 (OR 0.779, 95%CI 0.640-0.949) and model 3 further adjusted for WHR (OR 0.722, 95%CI 0.589-0.886). In contrast, a positive association between all these anthropometric indices and DKD was shown in the multivariable model 2. In model 2, the OR based on the highest fifth of BMI, WHR, WHtR, WC and HC were 1.927 (1.572- 2.366), 1.566 (1.277-1.923), 1.91 (1.554-2.351), 1.624 (1.312-2.012) and 1.585 (1.300-1.937) for DKD respectively. After further adjustment for BMI, the positive associations of WHR and WHtR with DKD were attenuated but still existed, while the positive associations of WC and HC disappeared. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 17 Besides, the positive association between BMI and DKD remained in the model 3 further adjusted for WHR. In Figure1, we further used restricted cubic splines to flexibly model and visualize the relation of anthropometric indices with DR and DKD, after adjustment for age, sex, duration of diabetes, HbA1c, SBP, TG, LDL and HDL. The analyses provided significant evidence of the non-linear associations between BMI and DR (p 0.05). The OR per SD higher of WHR, WHtR, WC and HC was 1.056 (0.992-1.125), 1.001(0.938-1.068), 0.950 (0.891-1.013) and 0.891 (0.836-0.949) respectively. Moreover, the OR of DR was relatively low at the median value of BMI. We also found evidence of non-linear associations of all these anthropometric indices with DKD (all p <0.05). Beyond the threshold at about the median value of parameters, higher BMI, WHtR, WC and HC contribute significantly to DKD development. But there was no increased risk in the lower range of BMI, . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 18 WHtR, WC and HC. Above the median value, the OR per SD higher of BMI, WHtR, WC and HC were 1.316 (1.207-1.438), 1.245 (1.145-1.353), 1.271 (1.174-1.377) and 1.248 (1.150-1.355) respectively. Besides, an increased risk was shown in the middle range of WHR. The OR per SD higher of WHR was 1.174 (1.093-1.262) between 0.85 and 1.0. 3.3 Associations of anthropometric indices with DR and DKD stratified by sex. We further examined how the associations of anthropometric indices changed by sex in Table 4, Table S3 and Table S4. And the baseline characteristics of the male and female participants were shown in table S5 and table S6 respectively. The OR for DR based on the third fifth of BMI was the lowest across gender groups, although the OR in women was higher than that in men. We also found significant differences in WC (p=0.001 for interaction) and HC (p<0.001 for interaction) for DR across strata of sex (Table S3). The inverse . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 19 associations of WC and HC with DR were observed in men, and still existed in model 3 further adjusted for BMI, whereas it was not found in women. The positive associations of anthropometric indices with DKD were generally consistent across gender groups. Further adjustment for WHR did not alter the significant positive association between BMI and DKD in men and women. Moreover, when further adjusted for BMI, the associations of WHR, WHtR, WC and HC with DKD were attenuated, among which only the positive association between WHtR and DKD in women remained significant. We did several sensitivity analyses to verify our findings. The associations of BMI with DR and DKD remained robust with different cut points of BMI (Table S7). After mutually adjusting for WC and HC, HC was still inversely associated with DR, while the inverse association between WC and DR in men was attenuated (Table S8). . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 20 4. Discussion In this cross-sectional study of Chinese participants with T2DM, we used logistic regression models and restricted cubic spine analysis to examine the different associations of anthropometric indices with DR and DKD. We found non-linear positive associations of all these anthropometric indices with DKD, while the associations of different anthropometric indices with DR varied dramatically. After taking WHR into account, BMI was still positively associated with DKD. In contrast, a BMI of around 25 kg/m2 was associated with the low risk of DR, and HC showed an independent and inverse association with DR in men. To the best of our knowledge, our study is the first to analyze the non-linear associations of BMI, WHtR, WC and HC with DKD and DR using restricted cubic spline analysis in the Chinese population. As is known, the associations between BMI and DR have been inconclusive. For example, in the Australian Diabetes Management Project 11, BMI showed a significantly positive association with DR, which was also reported in other . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 21 western studies like the Netherlands Hoorn Study 24, whereas the Singapore Malay Eye Study (SiMES) reported an inverse association between BMI and DR using the quartiles of BMI 25. Besides, a meta-analysis of 27 clinical studies reported that compared with normal weight (BMI 18.5-24.9 kg/m2), neither overweight (BMI 25-29.9 kg/m2) nor obesity (BMI >=30 kg/m2) increased the risk of DR 17. Distinct with these findings, our research indicated that participants with a BMI of around the median value (25 kg/m2) have the lowest risk of DR, which added a new sight into the association between BMI and DR. The mechanism behind the association between BMI and DR is unclear. Too low BMI may reflect unintentional weight loss resulting from poorly controlled diabetic mellitus, which increases the risk of DR. And too high BMI may be associated with increased production of proinflammatory cytokines in adipocytes, leading to increased vascular permeability and a high risk of DR 26. However, due to the lack of cause and effect evidence in our cross-sectional study, more prospective studies focused on the relationship between BMI and DR are needed to verify our findings. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 22 It has been established that a larger hip is associated with a lower risk of T2DM independently of WC and BMI according to previous studies 272829. In our study, we found that HC also had a significant inverse association with DR independently of BMI and WC in men. To our knowledge, there has been only one other study that investigated the independent relationship of HC with DR: a cross-sectional study involving 1773 patients with T2DM in India found a significant negative correlation between HC and DR 30. However, this previous study did not take the interdependence of BMI, WC or WHR with HC into consideration, and it also did not perform any sex-stratified analysis. The mechanism underlying the negative association of HC on DR in men is yet to be explored, but there is evidence that may partly explain this association. A larger hip could reflect a greater muscle mass in the gluteal region 29. Skeletal muscle mass is the main target of insulin and one major site of insulin resistance. Besides, a larger hip also indicates increased femoral and gluteal fat mass, which is less sensitive to lipolytic stimuli 31 and has an inverse relationship with arterial stiffness 32. One reason for the gender difference in . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 23 the associations of HC with DR may be the distinction in fat distribution between men and women: men tend to store more visceral fat leading to an apple-shaped body while premenopausal women tend to accumulate more fat in the subcutaneous regions leading to a pear-shaped body 33. Furthermore, the positive association between adiposity and low-grade systemic inflammation was stronger in women than in men 34, which might partly explain why the inverse association between HC and DR disappeared in women. Our findings on the positive associations of anthropometric indices with DKD were broadly in line with the majority of previous studies. For instance, BMI has been identified as a risk factor for DKD development based on a systematic review and meta-analysis of 20 cohorts 35. Another meta-analysis involving 15 cross-sectional studies reported that WC, WHR and WHtR were associated with a greater risk of DKD 18. Our study also elaborated the approximately J-shaped associations of BMI, WHR, WHtR, WC and HC with DKD and the S-shaped association between WHR and DKD. J-shaped . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 24 associations have been reported in numerous previous studies related to obesity, while S-shaped associations are rarely observed. For instance, one cohort study of 3.6 million adults in the UK found J-shaped associations of BMI with overall mortality and most specific causes of death 36. WC also showed a J-shaped association with total mortality in a 12-year prospective cohort study of US adults 37. Moreover, there are several studies suggesting J-shape associations between BMI and kidney diseases. A cross-sectional study in Norway suggested a J-shaped association between BMI and risk of CKD 38. Another cross-sectional study of the Southeast Asian population reported a J-shaped association between BMI and proteinuria 39. Because there are few researches on the non-linear associations of anthropometric indices with DKD, more studies are necessary to verify our findings. Besides, our research indicated that BMI was positively associated with DKD independent of WHR. When further adjusted for BMI, the positive association of WHR and WHtR with DKD still existed but attenuated, while positive associations of WC and HC were not significant. In contrast to our findings, . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 25 one cross-sectional study and another 5-year prospective study in China reported that WC was significantly associated with risk of DKD after adjustment for BMI, and BMI was not related to the risk of DKD in WHtR- adjusted model 40. The inconsistency between our research and previous studies may partly be due to the different covariates involved in the adjusted models, for the BMI-adjusted model and WHtR-adjusted model used in the previous studies did not include any other covariate. Further studies considering the mutually confounding effect between generalized and abdominal obesity are necessary to examine the associations of anthropometric indices with DKD. Our study revealed remarkable differences between the association of anthropometric indices with DKD and that with DR, which might indicate the tissue-specific mechanisms underlying the pathogenesis of DR and DKD. Lipid metabolism has been proven to play different roles in the development of DR and DKD 41. Tissue-specific lipid alterations have been observed in mice with diabetes 42. Besides, analysis of metabolic flux revealed the . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 26 difference of glucose and fatty acid metabolism in the progression of DR and DKD 43. Further researches investigating the difference between the pathogenesis of DR and DKD are warranted. Our study has several strengths. Firstly, it is the first study to assess the non- linear associations of anthropometric indices with DR and DKD using restricted cubic spline concurrently. Secondly, the interdependence between general obesity and abdominal obesity was taken into account in our study. Thirdly, we did several sex-stratified and sensitivity analyses to verify the robustness of findings. In addition, we recruited a relatively large clinical sample of patients with T2DM and performed this study with a comprehensive and standardized clinical assessment protocol. There are also some limitations in our study. Firstly, because of the cross- sectional nature of this study, causal inference of these anthropometric indices with DR and DKD cannot be determined. Secondly, our study was conducted in a hospital, which may affect the generalizability of findings. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 27 Lastly, because of the low numbers of participants with ESRD or proliferative diabetic retinopathy (PDR), we did not investigate the associations of anthropometric indices with ESRD and PDR. In conclusion, this study revealed different associations of anthropometric indices with DR and DKD in Chinese patients with T2DM. A BMI of around 25 kg/m2 and a large hip might be related to the relatively low risk of DR, while a higher BMI, WHR, WHtR, WC or HC was associated with a higher risk of DKD. Our findings support recommendations to maintain a BMI of around 25 kg/m2, a low WHR, a low WHtR and a large hip for prevention of DR and DKD. More longitudinal researches are warranted to verify our findings. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 28

The authors thank the clinical staffs of diabetes and ophthalmology clinics of the Shanghai General Hospital and the participants in this study for their valuable contributions. Funding: This work was supported by the National Natural Science Foundation of China (grant number 82271111, 81770947). Declarations of interest: The authors declare no conflict of interest. Author Contributions: YJW, HBC and ZZ designed the research. XP, CFG, CXL and HBC collected the research data. YJW, XP and BL analyzed the data. YJW drafted the manuscript. ZZ, HBC and CDZ assumed primary responsibility for the final content. YJW and XP contributed equally to this work. All authors read and approved the final manuscript. Colors in Figures: #C15858 was used in Figure 1 (Figure1 can be printed without color). . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 29

1. Sun H, Saeedi P, Karuranga S, et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183:109119. doi:10.1016/j.diabres.2021.109119 2. Roglic. WHO Global report on diabetes: A summary. Accessed Apirl 21, 2016. https://www.who.int/publications/i/item/9789241565257 3. Yau JWY, Rogers SL, Kawasaki R, et al. Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care. 2012;35(3):556-564. doi:10.2337/dc11-1909 4. Thomas MC, Brownlee M, Susztak K, et al. Diabetic kidney disease. Nat Rev Dis Primers. 2015;1:15018. doi:10.1038/nrdp.2015.18 5. Lin W, Luo Y, Liu F, et al. Status and Trends of the Association Between Diabetic Nephropathy and Diabetic Retinopathy From 2000 to 2021: Bibliometric and Visual Analysis. Front Pharmacol. 2022;13:937759. doi:10.3389/fphar.2022.937759 6. Huang Y, Wang Y, Liu C, et al. C-peptide, glycaemic control, and diabetic complications in type 2 diabetes mellitus: A real-world study. Diabetes Metab Res Rev. 2022;38(4):e3514. doi:10.1002/dmrr.3514 7. Umayahara Y, Fujita Y, Watanabe H, et al. Association of glycated albumin to HbA1c ratio with diabetic retinopathy but not diabetic nephropathy in patients with type 2 diabetes. Clinical Biochemistry. 2017;50(6):270-273. doi:10.1016/j.clinbiochem.2016.11.032 8. Krakoff J, Lindsay RS, Looker HC, Nelson RG, Hanson RL, Knowler WC. Incidence of retinopathy and nephropathy in youth-onset compared with adult-onset type 2 diabetes. Diabetes Care. 2003;26(1):76-81. doi:10.2337/diacare.26.1.76 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 30 9. Lee CMY, Woodward M, Pandeya N, et al. Comparison of relationships between four common anthropometric measures and incident diabetes. Diabetes Res Clin Pract. 2017;132:36-44. doi:10.1016/j.diabres.2017.07.022 10. Nguyen NT, Nguyen XMT, Lane J, Wang P. Relationship between obesity and diabetes in a US adult population: findings from the National Health and Nutrition Examination Survey, 1999-2006. Obes Surg. 2011;21(3):351-355. doi:10.1007/s11695-010-0335-4 11. Dirani M, Xie J, Fenwick E, et al. Are Obesity and Anthropometry Risk Factors for Diabetic Retinopathy?: The Diabetes Management Project. Investigative Ophthalmology & Visual Science. 2011;52(7):4416-4421. doi:10.1167/iovs.11-7208 12. Yi QX, Zhu LN, Ma J, Yu XJ, Liu L, Shen J. Use of Anthropometric Measures of Obesity to Predict Diabetic Retinopathy in Patients with Type 2 Diabetes in China. Diabetes Metab Syndr Obes. 2021;14:4089-4095. doi:10.2147/DMSO.S321030 13. Zhu W, Wu Y, Meng YF, Xing Q, Tao JJ, Lu J. Association of obesity and risk of diabetic retinopathy in diabetes patients. Medicine (Baltimore). 2018;97(32):e11807. doi:10.1097/MD.0000000000011807 14. Man REK, Sabanayagam C, Chiang PPC, et al. Differential Association of Generalized and Abdominal Obesity With Diabetic Retinopathy in Asian Patients With Type 2 Diabetes. JAMA Ophthalmology. 2016;134(3):251- 257. doi:10.1001/jamaophthalmol.2015.5103 15. Raman R, Rani PK, Gnanamoorthy P, Sudhir RR, Kumaramanikavel G, Sharma T. Association of obesity with diabetic retinopathy: Sankara Nethralaya Diabetic Retinopathy Epidemiology and Molecular Genetics Study (SN-DREAMS Report no. 8). Acta Diabetol. 2010;47(3):209-215. doi:10.1007/s00592-009-0113-8 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 31 16. Chan JCY, Chee ML, Tan NYQ, Cheng CY, Wong TY, Sabanayagam C. Differential effect of body mass index on the incidence of diabetes and diabetic retinopathy in two Asian populations. Nutr & Diabetes. 2018;8(1):1-11. doi:10.1038/s41387-018-0018-0 17. Zhou Y, Zhang Y, Shi K, Wang C. Body mass index and risk of diabetic retinopathy. Medicine (Baltimore). 2017;96(22):e6754. doi:10.1097/MD.0000000000006754 18. Zhao Q, Yi X, Wang Z. Meta-Analysis of the Relationship between Abdominal Obesity and Diabetic Kidney Disease in Type 2 Diabetic Patients. OFA. 2021;14(4):338-345. doi:10.1159/000516391 19. Man REK, Gan ATL, Fenwick EK, et al. The Relationship between Generalized and Abdominal Obesity with Diabetic Kidney Disease in Type 2 Diabetes: A Multiethnic Asian Study and Meta-Analysis. Nutrients. 2018;10(11):E1685. doi:10.3390/nu10111685 20. Stevens LA, Claybon MA, Schmid CH, et al. Evaluation of the Chronic Kidney Disease Epidemiology Collaboration equation for estimating the glomerular filtration rate in multiple ethnicities. Kidney Int. 2011;79(5):555- 562. doi:10.1038/ki.2010.462 21. American Diabetes Association. 11. Microvascular Complications and Foot Care: Standards of Medical Care in Diabetes-2019. Diabetes Care. 2019;42(Suppl 1):S124-S138. doi:10.2337/dc19-S011 22. Grading diabetic retinopathy from stereoscopic color fundus photographs-- an extension of the modified Airlie House classification. ETDRS report number 10. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology. 1991;98(5 Suppl):786-806. 23. Regression Modeling Strategies. Accessed July 3, 2022. https://link.springer.com/book/10.1007/978-1-4757-3462-1 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 32 24. van Leiden HA, Dekker JM, Moll AC, et al. Blood pressure, lipids, and obesity are associated with retinopathy: the hoorn study. Diabetes Care. 2002;25(8):1320-1325. doi:10.2337/diacare.25.8.1320 25. Lim LS, Tai ES, Mitchell P, et al. C-reactive Protein, Body Mass Index, and Diabetic Retinopathy. Investigative Ophthalmology & Visual Science. 2010;51(9):4458-4463. doi:10.1167/iovs.09-4939 26. Shoelson SE, Herrero L, Naaz A. Obesity, Inflammation, and Insulin Resistance. Gastroenterology. 2007;132(6):2169-2180. doi:10.1053/j.gastro.2007.03.059 27. Snijder MB, Dekker JM, Visser M, et al. Associations of hip and thigh circumferences independent of waist circumference with the incidence of type 2 diabetes: the Hoorn Study. Am J Clin Nutr. 2003;77(5):1192-1197. doi:10.1093/ajcn/77.5.1192 28. Snijder MB, Zimmet PZ, Visser M, Dekker JM, Seidell JC, Shaw JE. Independent and opposite associations of waist and hip circumferences with diabetes, hypertension and dyslipidemia: the AusDiab Study. Int J Obes Relat Metab Disord. 2004;28(3):402-409. doi:10.1038/sj.ijo.0802567 29. Seidell JC, Han TS, Feskens EJ, Lean ME. Narrow hips and broad waist circumferences independently contribute to increased risk of non-insulin- dependent diabetes mellitus. J Intern Med. 1997;242(5):401-406. doi:10.1046/j.1365-2796.1997.00235.x 30. Verma A, Jha A, Alagorie AR, Sharma R. Association of anthropometric parameters as a risk factor for development of diabetic retinopathy in patients with diabetes mellitus. Eye. Published online January 20, 2022:1- 6. doi:10.1038/s41433-022-01934-6 31. Ibrahim MM. Subcutaneous and visceral adipose tissue: structural and functional differences. Obes Rev. 2010;11(1):11-18. doi:10.1111/j.1467- 789X.2009.00623.x . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 33 32. Lee M, Choh AC, Demerath EW, Towne B, Siervogel RM, Czerwinski SA. Associations between trunk, leg and total body adiposity with arterial stiffness. Am J Hypertens. 2012;25(10):1131-1137. doi:10.1038/ajh.2012.92 33. Palmer BF, Clegg DJ. The sexual dimorphism of obesity. Mol Cell Endocrinol. 2015;402:113-119. doi:10.1016/j.mce.2014.11.029 34. Thorand B, Baumert J, Döring A, et al. Sex differences in the relation of body composition to markers of inflammation. Atherosclerosis. 2006;184(1):216-224. doi:10.1016/j.atherosclerosis.2005.04.011 35. Jiang W, Wang J, Shen X, et al. Establishment and Validation of a Risk Prediction Model for Early Diabetic Kidney Disease Based on a Systematic Review and Meta-Analysis of 20 Cohorts. Diabetes Care. 2020;43(4):925-933. doi:10.2337/dc19-1897 36. Bhaskaran K, Dos-Santos-Silva I, Leon DA, Douglas IJ, Smeeth L. Association of BMI with overall and cause-specific mortality: a population- based cohort study of 3·6 million adults in the UK. Lancet Diabetes Endocrinol. 2018;6(12):944-953. doi:10.1016/S2213-8587(18)30288-2 37. Reis JP, Macera CA, Araneta MR, Lindsay SP, Marshall SJ, Wingard DL. Comparison of overall obesity and body fat distribution in predicting risk of mortality. Obesity (Silver Spring). 2009;17(6):1232-1239. doi:10.1038/oby.2008.664 38. Hallan S, de Mutsert R, Carlsen S, Dekker FW, Aasarød K, Holmen J. Obesity, smoking, and physical inactivity as risk factors for CKD: are men more vulnerable? Am J Kidney Dis. 2006;47(3):396-405. doi:10.1053/j.ajkd.2005.11.027 39. Ramirez SPB, McClellan W, Port FK, Hsu SIH. Risk factors for proteinuria in a large, multiracial, southeast Asian population. J Am Soc Nephrol. 2002;13(7):1907-1917. doi:10.1097/01.asn.0000018406.20282.c8 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 34 40. Hu J, Yang S, Zhang A, et al. Abdominal Obesity Is More Closely Associated With Diabetic Kidney Disease Than General Obesity. Diabetes Care. 2016;39(10):e179-e180. doi:10.2337/dc16-1025 41. Eid S, Sas KM, Abcouwer SF, et al. New insights into the mechanisms of diabetic complications: role of lipids and lipid metabolism. Diabetologia. 2019;62(9):1539-1549. doi:10.1007/s00125-019-4959-1 42. Sas KM, Lin J, Rajendiran TM, et al. Shared and distinct lipid-lipid interactions in plasma and affected tissues in a diabetic mouse model. J Lipid Res. 2018;59(2):173-183. doi:10.1194/jlr.M077222 43. Sas KM, Kayampilly P, Byun J, et al. Tissue-specific metabolic reprogramming drives nutrient flux in diabetic complications. JCI Insight. 2016;1(15):e86976. doi:10.1172/jci.insight.86976 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 35 Figures: Figure 1. Associations between anthropometric indices and risk of DR and DKD. a Associations between anthropometric indices and risk of DR. In the analysis of DR, knots placed at 5th, 27.5th, 50th, 72.5th and 95th for BMI, 10th, 50th and 90th percentile for WHR, WHtR, WC and HC. Test for non-linearity: BMI, p=0.018; WHR, p=0.624; WHtR, p=0.928; WC, p=0.592; HC, p=0.853. b Associations between anthropometric indices and risk of DKD. In the analysis of DKD, knots placed at 5th, 35th, 65th and 95th for BMI, WHR, WHtR and HC, 10th, 50th and 90th percentile for WC. Test for non-linearity: BMI, p<0.001; WHR, p=0.045; WHtR, p=0.022; WC, p=0.002; HC, p<0.001. All the anthropometric indices were assessed as a continuous variable using restricted cubic spine regression, adjusted for age, sex, duration of diabetes, HbA1c, SBP, TG, LDL and HDL.

point was median value for each of anthropometric indices (25 for BMI, 0.93 for WHR, 0.54 for WHtR, 90 for WC, 97 for HC). . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 36 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 37 Tables: Table 1. General characteristics of all participants with T2DM by DR and DKD. Characteristics All DR- DR+ P value DKD- DKD+ P value (n=5226) (n=3564) (n=1662) (n=3484) (n=1742) Age 59.2±12.0 59.1±12.5 59.4±10.8 0.315 57.8±11.7 62.0±12.2 <0.001 Female, n (%) 2309 (44.2%) 1538 (43.2%) 771 (46.4%) 0.030 1671 (48.0%) 638 (36.6%) <0.001 Diabetes duration (years) 10.0 [4.00, 14.0] 8.00 [3.00, 13.0] 11.0 [7.00, 17.0] <0.001 8.00 [3.00, 13.0] 10.0 [5.00, 16.0] <0.001 BMI (kg/m2) 25.2±3.57 25.2±3.59 25.2±3.52 0.978 24.9±3.44 25.9±3.71 <0.001 WHtR 0.54 [0.51, 0.59] 0.54 [0.51, 0.59] 0.55 [0.51, 0.59] 0.140 0.54 [0.50, 0.58] 0.56 [0.52, 0.60] <0.001 WHR 0.93 [0.89, 0.97] 0.92 [0.89, 0.96] 0.93 [0.89, 0.97] 0.010 0.92 [0.88, 0.96] 0.94 [0.90, 0.98] <0.001 WC (cm) 90.9±10.5 90.9±10.6 90.8±10.4 0.544 89.7±10.2 93.2±10.8 <0.001 HC (cm) 97.9±8.26 98.2±8.31 97.4±8.12 0.001 97.2±7.92 99.3±8.74 <0.001 HBP, n (%) 3231 (61.8%) 2133 (59.8%) 1098 (66.1%) <0.001 1904 (54.6%) 1327 (76.2%) <0.001 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 38 SBP (mmHg) 132±16.9 131±16.0 136±18.4 <0.001 130±15.6 138±18.2 <0.001 DBP (mmHg) 80.0±9.50 79.7±9.32 80.5±9.87 0.006 79.3±9.08 81.3±10.2 <0.001 FPG (mmol/L) 8.09±2.86 8.02±2.71 8.26±3.16 0.007 8.01±2.74 8.27±3.09 0.003 HbA1c, % (mmol/mol) 8.78±2.11 (7.22±2.30) 8.70±2.16 (7.13±2.35) 8.97±1.98 (7.43±2.16) <0.001 8.72±2.14 (7.15±2.33) 8.91±2.06 (7.36±2.25) 0.003 TC (mmol/L) 4.76±1.20 4.74±1.17 4.80±1.26 0.096 4.70±1.12 4.86±1.34 <0.001 TG (mmol/L) 1.44 [1.00, 2.10] 1.46 [1.02, 2.12] 1.38 [0.95, 2.06] 0.003 1.37 [0.96, 1.99] 1.59 [1.09, 2.38] <0.001 HDL (mmol/L) 1.10±0.30 1.09±0.30 1.11±0.31 0.026 1.12±0.31 1.06±0.29 <0.001 LDL (mmol/L) 2.99±0.96 3.00±0.94 2.98±1.00 0.488 2.98±0.92 3.01±1.04 0.404 Serum creatinine (µmol/L) 66.0 [55.0, 79.0] 67.0 [56.0, 79.0] 66.0 [54.0, 79.0] 0.447 63.0 [53.0, 73.0] 78.0 [61.0, 99.0] <0.001 eGFR (mL/min/1.73 m2) 95.0 [75.6, 106] 94.9 [76.8, 106] 95.3 [72.9, 106] 0.450 98.1 [87.8, 108] 75.8 [53.4, 99.2] <0.001 UACR (mg/g) 11.5 [6.42, 31.0] 9.79 [5.92, 21.7] 18.1 [8.24, 96.1] <0.001 8.19 [5.47, 12.8] 64.2 [30.3, 246] <0.001 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 39 The data are summarized as the mean ± SD or median [interquartile range] for continuous variables or as n (%) for categorical variables. DKD diabetic kidney disease, DR diabetic retinopathy, CHD coronary heart disease. BMI body mass index, WHR waist-hip ratio, WHtR waist-height ratio, WC waist circumference, HC hip circumference. HBP high blood pressure, SBP systolic blood pressure, DBP diastolic blood pressure, FPG fasting plasma glucose, HbA1c glycated hemoglobin, TC total cholesterol, TG triglycerides, HDL high-density lipoprotein, LDL low-density lipoprotein. UACR, urinary albumin:creatinine ratio, eGFR estimated glomerular filtration rate, FBP fast blood glucose, PBP postload blood glucose, GA glycated albumin. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 40 Table 2. Odds ratios (95% CI) of DR according to anthropometric indices. Prevalence, n (%) OR (95% CI) Model1 Model2 Model3 Fifth of BMI 1 (lowest) 355 (33.8%) 1 [reference] 1 [reference] 1 [reference] 2 345 (32.5%) 0.952 (0.794-1.142) 0.914 (0.753-1.108) 0.892 (0.735-1.084)† 3 283 (27.5%) 0.748 (0.620-0.902) 0.752 (0.615-0.920) 0.722 (0.589-0.886)† 4 332 (32.1%) 0.940 (0.783-1.130) 0.942 (0.772-1.148) 0.898 (0.732-1.100)† 5 (highest) 347 (33.0%) 0.970 (0.808-1.163) 0.924 (0.756-1.129) 0.863 (0.699-1.065)† P for trend 0.711 0.617 0.276† BMI per SD increase 1662 (31.8%) 1.002 (0.945-1.063) 0.988 (0.926-1.054) 0.966 (0.902-1.035)† Fifth of WHR 1 (lowest) 324 (31.1%) 1 [reference] 1 [reference] 1 [reference] 2 307 (29.0%) 0.925 (0.767-1.116) 0.926 (0.761-1.127) 0.936 (0.768-1.141) 3 330 (31.7%) 1.064 (0.883-1.283) 1.046 (0.859-1.274) 1.063 (0.870-1.298) 4 337 (32.5%) 1.107 (0.918-1.335) 1.072 (0.879-1.307) 1.093 (0.892-1.340) 5 (highest) 364 (34.8%) 1.232 (1.023-1.483) 1.102 (0.904-1.344) 1.137 (0.921-1.404) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 41 P for trend 0.005 0.131 0.083 WHR per SD increase 1662 (31.8%) 1.098 (1.035-1.166) 1.056 (0.992-1.125) 1.068 (1.000-1.142) Fifth of WHtR 1 (lowest) 340 (32.7%) 1 [reference] 1 [reference] 1 [reference] 2 320 (31.0%) 0.921 (0.766-1.109) 0.934 (0.768-1.135) 0.952 (0.779-1.164) 3 318 (29.8%) 0.869 (0.722-1.044) 0.863 (0.709-1.050) 0.892 (0.721-1.103) 4 297 (29.4%) 0.845 (0.699-1.019) 0.832 (0.680-1.019) 0.874 (0.691-1.106) 5 (highest) 387 (36.0%) 1.118 (0.930-1.343) 1.035 (0.847-1.265) 1.121 (0.848-1.481) P for trend 0.492 0.902 0.854 WHtR per SD 1662 (31.8%) 1.030 (0.970-1.093) 1.001 (0.938-1.068) 1.026 (0.926-1.136) Fifth of WC 1 (lowest) 310 (33.8%) 1 [reference] 1 [reference] 1 [reference] 2 340 (31.6%) 0.916 (0.759-1.106) 0.870 (0.712-1.063) 0.857 (0.698-1.053) 3 327 (30.8%) 0.890 (0.736-1.077) 0.881 (0.718-1.080) 0.856 (0.685-1.070) 4 336 (30.1%) 0.865 (0.716-1.044) 0.816 (0.666-0.999) 0.785 (0.620-0.994) 5 (highest) 349 (33.2%) 0.998 (0.826-1.206) 0.877 (0.712-1.080) 0.825 (0.621-1.096) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 42 P for trend 0.829 0.198 0.146 WC per SD increase 1662 (31.8%) 0.990 (0.934-1.050) 0.950 (0.891-1.013) 0.903 (0.816-0.998) Fifth of HC 1 (lowest) 333 (34.9%) 1 [reference] 1 [reference] 1 [reference] 2 360 (31.5%) 0.865 (0.721-1.039) 0.871 (0.718-1.056) 0.832 (0.683-1.012) 3 320 (32.4%) 0.905 (0.749-1.094) 0.902 (0.738-1.103) 0.830 (0.671-1.027) 4 300 (32.0%) 0.889 (0.733-1.076) 0.859 (0.700-1.053) 0.761 (0.605-0.956) 5 (highest) 349 (29.0%) 0.767 (0.639-0.921) 0.730 (0.600-0.889) 0.600 (0.463-0.776) P for trend 0.015 0.004 <0.001 HC per SD increase 1662 (31.8%) 0.912 (0.859-0.967) 0.891 (0.836-0.949) 0.79 (0.719-0.868) Model 1: adjusted for age and sex. Model 2: adjusted for age, sex, duration of DM, SBP, TG, LDL, HDL, HbA1c. Model 3: adjusted for age, sex, duration of DM, SBP, TG, LDL, HDL, HbA1c, BMI. (Model 3† adjusted for age, sex, duration of DM, SBP, TG, LDL, HDL, HbA1c, WHR) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 43 Table 3. Odds ratio (95% CI) of DKD according to anthropometric indices. Prevalence, n (%) OR (95% CI) Model 1 Model 2 Model 3 Fifth of BMI 1 (lowest) 276 (26.3%) 1 [reference] 1 [reference] 1 [reference] 2 305 (28.8%) 1.119 (0.920-1.362) 0.936 (0.762-1.150) 0.915 (0.743-1.125) † 3 333 (32.3%) 1.372 (1.130-1.667) 1.119 (0.910-1.375) 1.077 (0.874-1.329)† 4 358 (34.7%) 1.575 (1.298-1.911) 1.228 (1.000-1.510) 1.173 (0.950-1.449)† 5 (highest) 470 (44.7%) 2.657 (2.198-3.217) 1.927 (1.572-2.366) 1.806 (1.459-2.239)† P for trend <0.001 <0.001 <0.001† BMI per SD increase 1742 (33.3%) 1.436 (1.351-1.528) 1.300 (1.218-1.389) 1.277 (1.192-1.370)† Fifth of WHR 1 (lowest) 248 (23.8%) 1 [reference] 1 [reference] 1 [reference] 2 309 (29.2%) 1.239 (1.015-1.512) 1.109 (0.902-1.365) 1.028 (0.834-1.268) 3 361 (34.6%) 1.586 (1.303-1.931) 1.383 (1.128-1.698) 1.239 (1.007-1.527) 4 389 (37.5%) 1.786 (1.469-2.174) 1.476 (1.204-1.812) 1.284 (1.041-1.584) 5 (highest) 435 (41.6%) 2.065 (1.702-2.509) 1.566 (1.277-1.923) 1.243 (1.002-1.544) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 44 P for trend <0.001 <0.001 0.008 WHR per SD increase 1742 (33.3%) 1.255 (1.179-1.338) 1.140 (1.070-1.217) 1.056 (0.985-1.130) Fifth of WHtR 1 (lowest) 262 (25.2%) 1 [reference] 1 [reference] 1 [reference] 2 299 (28.9%) 1.221 (1.001-1.489) 1.054 (0.857-1.297) 0.961 (0.777-1.189) 3 338 (31.6%) 1.379 (1.135-1.676) 1.138 (0.928-1.396) 0.970 (0.779-1.208) 4 371 (36.7%) 1.781 (1.466-2.166) 1.415 (1.151-1.741) 1.113 (0.877-1.414) 5 (highest) 472 (43.9%) 2.619 (2.158-3.184) 1.910 (1.554-2.351) 1.293 (0.974-1.717) P for trend <0.001 <0.001 0.052 WHtR per SD 1742 (33.3%) 1.418 (1.333-1.510) 1.281 (1.198-1.369) 1.112 (1.002-1.233) Fifth of WC 1 (lowest) 227 (24.7%) 1 [reference] 1 [reference] 1 [reference] 2 298 (27.7%) 1.107 (0.902-1.360) 0.897 (0.723-1.113) 0.797 (0.639-0.995) 3 337 (31.7%) 1.322 (1.079-1.620) 1.046 (0.844-1.297) 0.833 (0.660-1.052) 4 409 (36.6%) 1.665 (1.367-2.032) 1.266 (1.026-1.564) 0.936 (0.735-1.192) 5 (highest) 471 (44.8%) 2.377 (1.951-2.903) 1.624 (1.312-2.012) 0.995 (0.746-1.326) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 45 P for trend <0.001 <0.001 0.444 WC per SD increase 1742 (33.3%) 1.404 (1.321-1.493) 1.262 (1.182-1.348) 1.083 (0.978-1.199) Fifth of HC 1 (lowest) 257 (27.0%) 1 [reference] 1 [reference] 1 [reference] 2 333 (29.2%) 1.094 (0.900-1.331) 0.969 (0.790-1.190) 0.870 (0.706-1.073) 3 321 (32.5%) 1.230 (1.007-1.503) 1.023 (0.829-1.264) 0.843 (0.675-1.053) 4 324 (34.5%) 1.391 (1.138-1.702) 1.133 (0.917-1.401) 0.853 (0.674-1.079) 5 (highest) 507 (42.1%) 2.033 (1.686-2.456) 1.585 (1.300-1.937) 1.005 (0.777-1.301) P for trend <0.001 <0.001 0.899 HC per SD increase 1742 (33.3%) 1.306 (1.230-1.387) 1.207 (1.132-1.286) 1.000 (0.911-1.099) Model 1: adjusted for age and sex. Model 2: adjusted for age, sex, duration of diabetes, SBP, TG, LDL, HDL, HbA1c. Model 3: adjusted for age, sex, duration of diabetes, SBP, TG, LDL, HDL, HbA1c, BMI. (Model 3† adjusted for age, sex, duration of diabetes, SBP, TG, LDL, HDL, HbA1c, WHR) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 46 Table 4. Odds ratios (95% CI) of DR and DKD according to BMI, WHR and HC stratified by sex. Male Female Prevalence, n (%) Model2 Model3+ Prevalence, n (%) Model2 Model3 DR Fifth of HC 1 (lowest) 195 (37.4%) 1 [reference] 1 [reference] 113 (31.7%) 1 [reference] 1 [reference] 2 186 (32.7%) 0.851 (0.655- 1.107) 0.823 (0.630-1.074) 160 (31.4%) 0.991 (0.729-1.348) 0.947 (0.691-1.298) 3 179 (29.5%) 0.711 (0.546- 0.927) 0.666 (0.504-0.88) 156 (33.6%) 1.116 (0.816-1.528) 1.033 (0.739-1.444) 4 155 (27.7%) 0.632 (0.481- 0.830) 0.572 (0.423-0.774) 194 (38.8%) 1.311 (0.967-1.782) 1.167 (0.821-1.663) 5 (highest) 176 (26.6%) 0.582 (0.444- 0.762) 0.495 (0.350-0.697) 148 (31.0%) 0.965 (0.704-1.325) 0.798 (0.520-1.225) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 47 P for trend <0.001 <0.001 0.538 0.864 Per SD increase 0.800 (0.732- 0.874) 0.715 (0.632-0.808) 0.996 (0.907-1.093) 0.898 (0.775-1.040) DKD Fifth of BMI 1 (lowest) 172 (29.9%) 1 [reference] 1 [reference] 108 (23.5%) 1 [reference] 1 [reference] 2 197 (32.6%) 1.035 (0.792- 1.352) 1.011 (0.772-1.324)† 102 (22.4%) 0.707 (0.509-0.982) 0.690 (0.496-0.959)† 3 216 (38.0%) 1.298 (0.988- 1.707) 1.249 (0.946-1.650)† 119 (24.9%) 0.857 (0.624-1.177) 0.819 (0.593-1.131)† 4 215 (38.0%) 1.337 (1.016- 1.760) 1.273 (0.960-1.689)† 132 (29.5%) 1.022 (0.745-1.404) 0.972 (0.704-1.343)† 5 (highest) 304 (50.5%) 2.404 (1.828- 3.169) 2.229 (1.666-2.990)† 177 (37.7%) 1.386 (1.019-1.888) 1.300 (0.946-1.789)† P for trend <0.001 <0.001† 0.001 0.007† . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint 48 Per SD increase 1.428 (1.306- 1.563) 1.405 (1.275-1.548)† 638 (27.6%) 1.192 (1.080-1.315) 1.171 (1.058-1.297)† Model 2: adjusted for age, duration of DM, SBP, TG, LDL, HDL, HbA1c. Model 3: adjusted for age, duration of DM, SBP, TG, LDL, HDL, HbA1c, BMI. (Model 3†: adjusted for age, duration of DM, SBP, TG, LDL, HDL, HbA1c, WHR) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted October 8, 2022. ; https://doi.org/10.1101/2022.10.07.22280304doi: medRxiv preprint