The present quantitative cross-sectional study is part of the longitudinal LatDiane study, which commenced in 2013 (18,19). Recruitment of participants took place between January 15 and August 31, 2021, at the Clinical Research Center of the University of Latvia (Riga, Latvia). Details about the present study were shared via the website and social media channels of the University of Latvia. The protocols of the general LatDiane study, and the present sub-study devoted to nutrition, inflammation and gut health, were approved by the Latvian Central Ethics Committee [approval no. 01-29.1/3 (July 10, 2013)] and its further amendments (clearance no. 1/19-10-01, issued on October 1, 2019; no. 01-29.1/2226, dated April 14, 2020). Written informed consent was obtained from all subjects.

The individuals were split into two research groups: Patients with T1D (n=74) and a control group (n=33), which consisted of otherwise healthy individuals with normal glucose metabolism according to the American Diabetes Association criteria (20). The exclusion criteria for both groups were inflammatory bowel disease (Crohn's disease or ulcerative colitis), celiac disease, pregnancy, acute gastrointestinal infection within 2 months of the planned fecal collection, clinical signs of acute inflammation and fever. Demographic data (age and sex), anthropometric data (weight, height, hip and waist circumference) and nutritional data were collected. In addition, blood serum and fecal samples were obtained for the analysis of inflammatory and endotoxemia markers. Information about the history of medication use and self-reported gastrointestinal symptoms was also collected.

Three methods were used to obtain information about the dietary habits of the participants: A food frequency questionnaire (FFQ), a 24-h recall and a 3-day dietary record.

Data about participant intake of energy, macronutrients and fiber were obtained from the 24-h recall and 3-day dietary record. Participants were asked to fill in the food diaries during 2 workdays and 1 weekend day. For accurate data assessment, participants were asked to use a validated portion size picture book or food weighing. The average amount of macronutrients and energy for 4 days (from 24-h recall and 3-day dietary records together) was calculated. Considering that the recommended amount of energy, as well as protein, carbohydrate and fat, is different for women and men, the obtained data on the nutrient intake of the participants were analyzed by sex. The nutritional data and participants' eating habits were compared with the adult nutritional recommendations provided by the Latvian Ministry of Health, which are based on the World Health Organisation’s and Nordic Nutrition recommendations (16,21). Specifically, collected data were compared with the recommended daily energy and macronutrient amounts for adults in the age group 31-60 years, as the median study participant age was 44 (34.93-52.0) years in the T1D group and 37 (32.17-45.79) years in the control group.

The FFQ included 118 different food items (in standard portions) and participants were asked to mark the frequency of their consumption during the past 12 months by choosing one of 10 answer options: ‘Don't use’, ‘A few times a year (1-4 times)’, ‘Once a month’, ‘2-3 times a month’, ‘Once a week’, ‘2-3 times a week’, ‘4-6 times a week’, ‘Once a day’, ‘2-3 times a day’ or ‘4 times a day or more’ (Data S1). For more convenient data analysis, all products were grouped into 27 food groups (Table SI). In the subsequent data analysis, an ‘annual consumption coefficient’ was used, calculated based on the obtained FFQ data. For example, if a product was consumed ‘once a day’, a coefficient of 365 was assigned; if the answer was ‘once a month’, 12 was used, and so on. Subsequently, comparing these responses with dietary recommendations, all answers were converted to times per week, assuming there are 52 weeks in a year.

The dietary data analysis was conducted by the Institute of Food Safety, Animal Health, and Environment ‘BIOR’ (Riga, Latvia) using a custom program based on the Microsoft Dynamics AX 2009 platform (Microsoft Corporation). This program utilizes the Latvian Food Composition Database, which was developed to analyze food consumption data from the ‘Food Consumption Study of the Latvian Population (2012-2013)’ (22) and continues to provide data for other studies on food consumption within the Latvian population. The program includes algorithms and software designed to automate the calculation of daily nutrient intake and/or food group consumption using the national food consumption database, with results presented as average intake values (22).

Data from the FFQ regarding consumption of the 27 aforementioned product groups were categorized as either ‘low consumption’ or ‘moderate-to-high consumption’ based on the median consumption values derived from the present study participants' FFQ data (Table SII). The categorized food consumption data were analyzed for both study groups, the T1D and control groups, in relation to inflammatory and endotoxemia markers.

Information was collected on the medications used by participants, including antihypertensive and cholesterol-lowering drugs, immunosuppressive medications or corticosteroids, as well as medications related to gastrointestinal symptoms, such as acid-reducing agents, antispasmodics and medications used to relieve flatulence, diarrhea or constipation, as well as the use of insulin among participants with T1D (Table I). Participants were asked how often they used each medication (ranging from ‘never’ to ‘every day’). For medications related to gastrointestinal symptoms, participants were divided into two groups: Those who used the medication at least 1-4 times per month or more frequently, and those who used it less than once per month. For the other medications (such as cholesterol-lowering medication), participants were categorized simply as users or non-users.

Information on gastrointestinal symptoms was assessed using a self-reported questionnaire consisting of 18 items that addressed abdominal pain, discomfort and bowel movement disturbances, including diarrhea and constipation. Participants were asked how often and to what extent they experienced these symptoms. The mean value of the gastrointestinal symptom score was then calculated, as previously described (8).

Blood samples were collected via venous puncture and were then assayed to assess clinical markers (for example, blood count and clinical chemistry, including HbA1c) in a certified clinical laboratory. For further analysis of endotoxemia markers and hs-CRP, the blood samples were incubated undisturbed for 30 min at room temperature and then centrifuged at 3,220 x g at 4˚C for 15 min. The serum was then removed from the pellet, transferred into 2-ml tubes, frozen and stored at -20˚C until analysis. Biobanking and sample storage were performed in agreement with the procedures of the Genome Database of the Latvian Population (23).

Stool samples were collected within 2 weeks of blood collection. Participants collected their fecal samples at home using the FecesCatcher (Zymo Research Corp.) to avoid contact of the stool with toilet water, urine and disinfectants, and with sterile collection tubes without buffer (the collection date and time were marked). Within 24 h, the samples were delivered to the laboratory for calprotectin measurement in unfrozen samples.

LPS levels in serum were assessed using the Hycult LAL chromogenic endpoint assay (cat. no. HIT302; Hycult Biotech); LBP levels in serum were measured with the Hycult LBP Human ELISA kit (cat. no. HK315-02; Hycult Biotech); and serum EndoCAb IgG and EndoCAb IgM levels were analyzed using Hycult EndoCAb IgG and IgM ELISA kits (cat. nos. HK504-IGG and HK504-IGM; Hycult Biotech). Hs-CRP levels in serum were determined using the Hycult Human hs-CRP kit (cat no. HK369; Hycult Biotech). All assays were performed according to the manufacturer's protocols. Fecal calprotectin was measured with the Alegria^® Calprotectin ELISA kit (cat. no. ORG280; Orgentec Diagnostika; Sebia) in a certified clinical laboratory (7).

Serum inflammatory and endotoxemia markers were assessed across three plates. Data were examined for consistency across the plates and only LBP required normalization. The normalization procedure was performed using a location-scale transformation, as described in previous studies (7,24). Fecal calprotectin levels varied in the study group, with only six patients exceeding 50 µg/g and five surpassing 200 µg/g. To minimize variability, values >200 µg/g were treated as outliers for further analysis.

Data processing and statistical analysis were performed using the statistical software JASP (version 0.18.3.0) (25) and IBM SPSS Statistics (version 22; IBM Corp.). The normality of continuous data was assessed using the Shapiro-Wilk and Kolmogorov-Smirnov tests, supplemented by visual inspection of histograms. Most variables did not meet the assumption of normality; therefore, data are presented as the median and interquartile range (1st and 3rd quartiles). The proportions of categorical variables between the groups were compared using the χ² proportion test. Comparisons between the T1D and control groups were performed using the Mann-Whitney U test. Spearman's rank correlation coefficient was used to assess correlations between two continuous variables. Associations between categorized product consumption (divided by the median into two groups: Low vs. moderate-to-high) and categorized inflammatory and endotoxemia markers (dichotomized by their medians) were assessed using binary logistic regression, with the latter serving as the outcome variable (Table SII). Five models were analyzed in the T1D group and four in the control group, including univariate and multivariate models adjusted for covariates in various combinations. In both studied groups, Model 1 was univariate, Model 2 was adjusted for body mass index (BMI), Model 3 was adjusted for BMI and diabetes duration in the T1D group, and for BMI and glycated hemoglobin (HbA1c) in the control group. In the T1D group, Model 4 was adjusted for BMI, diabetes duration and HbA1c, and Model 5 was also adjusted for gastrointestinal symptom score. In the control group, Model 4 was adjusted for BMI, HbA1c and gastrointestinal symptom score. Multicollinearity in the regression models was assessed using the variance inflation factor. P<0.05 was considered to indicate a statistically significant difference.

The present study comprised two research groups. In the group of patients with T1D (n=74), 38% (n=28) were male and 62% (n=46) were female. In the control group, or participants with normal glucose metabolism (n=33), 42% (n=14) were male and 58% (n=19) were female. The median age of participants in the T1D group was 44.0 (34.9-52.0) years, with a minimum age of 22 years and a maximum of 80 years, whereas the median age in the control group was 37.0 (32.2-45.8) years. Anthropometric parameters did not differ significantly between the studied groups (Table I).

The median duration of diabetes in the T1D group was 21.0 (13.0-31.8) years and some patients had developed complications of diabetes. HbA1c, HDL cholesterol and hs-CRP levels were statistically significantly higher in the T1D group compared with those in the control group (Table I).

Data from the FFQ were used to analyze the eating habits of the participants during the past 12 months. Study participants consumed cereals and pasta approximately four times a week, potatoes two times a week and bread eight to 10 times a week (at least once daily). Fish and seafood were eaten two to three times a month. Milk and milk products were consumed approximately twice daily by patients with T1D and once daily by the control group. In both studied groups, white and red meat, as well as processed meat and fish products, were consumed two to three times a week, while eggs were consumed approximately three times a week. Patients with T1D consumed vegetable oils and fats, butter and other animal fats equally often (five times a week), whereas the control group consumed more vegetable oils and fats (five times a week) and less butter and other animal fats (once a week). Cooked and fresh vegetables were consumed twice a day in both groups. The participants ate fruits and berries approximately once a day (7.5 times a week in the T1D group and 5.9 times a week in the control group). Legumes, mushrooms (fresh and pickled), pickled and salted vegetables were consumed several times a month, and nuts and seeds were eaten twice a week. The consumption of the mentioned products did not differ significantly in both studied groups (Table II).

Participants drank water ~35.1 times a week, equivalent to approximately five 220-ml glasses of water daily. Analysis of beverage consumption (alcohol, sweetened beverages, water, coffee, tea and cocoa) showed that only sweetened beverage consumption was statistically significantly higher in the control group (P=0.049; Table II), while no significant differences were observed for the other beverage consumptions (all P>0.005; Table II).

Participants ate sweets and/or pastries daily. The control group reported statistically significantly higher consumption of sweets and pastries (P=0.003), while products with sweeteners (artificial and other sugar substitutes) were consumed more frequently by the T1D group (P=0.049) (Table II).

Fast food and snack consumption was recorded on average once a week in the T1D group, and twice a week in the control group. However, the consumption of manufactured sauces, such as mayonnaise and soy sauce, was statistically significantly higher in the control group (P=0.026; Table II).

Data from the 24-h recall and 3-day food records were used to analyze the average macronutrient and energy consumption of participants. As shown in Table III, participants in both studied groups had insufficient energy intake; they consumed enough protein [17-20 percentage of total energy intake (E%)], whereas the amount of carbohydrates consumed (32-41 E%) was lower than recommended and fat intake (40-42 E%) was higher than recommended. In addition, fiber intake was insufficient (14.0-23.9 g). The intakes of protein, fat and energy were statistically significantly lower in female patients with T1D compared with in women in the control group (P=0.025, P=0.003, P=0.003, respectively). There were no statistically significant differences between male participants in both studied groups regarding macronutrient and energy consumption.

The levels of inflammatory and endotoxemia markers were compared, according to product consumption patterns, between the high- and low-consumption groups using Mann-Whitney U test (Table SIII). Statistically significant differences were then further analyzed using binary logistic regression with different models to determine the strength and direction of the associations (Tables IV, V, SIV and SV). All analyses were performed separately for the two studied groups.

According to the results of the Mann-Whitney U test, in the T1D group, significantly higher LPS/HDL ratio levels were found in subjects consuming moderate-to-high amounts of cereals and pasta, red meat, and nuts and seeds (P=0.023, P=0.022 and P=0.036, respectively). Patients with T1D consuming higher amounts of red meat and water had higher LPS levels (P=0.046 and P=0.043, respectively), whereas those who consumed more sweets and pastries had lower LPS levels (P=0.023). Patients with T1D and a higher consumption of processed meat and fish had higher fecal calprotectin levels (P=0.017), and those that consumed more legumes had higher hs-CRP levels (P=0.019). Patients with T1D and a higher consumption of sweeteners and products containing them had lower fecal calprotectin and EndoCAb IgM levels (P=0.04 and P=0.008, respectively).

In the control group, subjects consuming more red meat, white meat, vegetables (fresh and cooked), potatoes and sauces had higher LPS levels (P=0.029, P=0.009, P=0.049, P=0.024 and P=0.026, respectively). In addition, the LPS/HDL ratio and LBP levels were higher in those consuming more white meat (P=0.025 and P=0.002, respectively). In the control group, LBP was higher in those consuming more sweetened beverages (P=0.048), and higher hs-CRP levels were observed in control group participants with a higher consumption of mushrooms (pickled and fresh) (P=0.038) and water (P=0.033). EndoCAb IgM levels were higher in those control group participants consuming more water and less hot drinks, such as coffee, tea or cocoa (P=0.015 and P=0.049, respectively). Control group participants with a higher consumption of alcohol (P=0.027), and a lower consumption of milk and milk products, potatoes, butter and other animal fats had higher EndoCAb IgG levels (P=0.009, P=0.043 and P=0.010, respectively).

A total of 18 food groups that demonstrated statistically significant between-group differences in inflammatory markers, based on the Mann-Whitney U test, were further analyzed using binary logistic regression models (Tables IV and SIV).

In the T1D group, higher consumptions of red meat [LPS, Model 2: OR, 2.85 (95% CI, 1.08-7.56), P=0.035; LPS/HDL ratio, Model 2: OR, 2.63 (95% CI, 1.01-6.80), P=0.047] and cereals and pasta [LPS/HDL ratio, Model 2: OR, 2.68 (95% CI, 1.04-6.87), P=0.041] were associated with higher endotoxemia markers (LPS and LPS/HDL ratio) in the models adjusted for BMI. However, after adjusting for diabetes duration and HbA1c, the association became marginally significant (Table IV). Higher consumption of legumes in the T1D group was associated with higher hs-CRP levels in the fully adjusted model 5 [OR, 4.91 (95% CI, 1.38-17.41), P=0.014]. In the T1D group, a higher consumption of sweeteners and products containing them was statistically significantly inversely associated with EndoCAb IgM and calprotectin levels in all applied regression models. In the model 4, adjusted for BMI, Hb1Ac and duration of diabetes, individuals with T1D who reported a higher consumption of sweeteners and products containing them exhibited a 64% reduction in the odds of elevated EndoCAb IgM levels and a 68% reduction in the odds of elevated calprotectin levels (Table IV).

Notably, in the T1D group, the consumption of coffee, tea and cocoa exhibited a pro-inflammatory effect. Higher consumption of these drinks was associated with an approximately four times higher chance of elevated hs-CRP levels (Model 5, P=0.009; Table IV). In the control group, the opposite effect was observed, and a higher consumption of coffee, tea and cocoa was associated with a 92% lower risk odds of high EndoCAb IgM levels (Model 4, P=0.027; Table SIV).

In the control group, a higher consumption of potatoes [Model 4, OR, 0.11 (95% CI, 0.02-0.69), P=0.018] and butter or other animal fats [Model 4, OR, 0.08 (95% CI, 0.08-0.82), P=0.033] in the fully adjusted model was inversely associated with EndoCAb IgG, suggesting an anti-inflammatory tendency (Table SIV). In addition, a higher consumption of manufactured sauces was associated with a five times higher chance of high LPS levels (Model 3, P=0.047, Table SIV). In the control group, a higher consumption of nuts and seeds was associated with six times higher odds of elevated EndoCAb IgM in multivariate logistic regression model 3 adjusted for BMI and HbA1c [OR, 6.65 (95% CI, 1.25-35.37), P=0.026], suggesting a pro-inflammatory tendency (Table SIV).

Analyzing associations between energy, macronutrients and fiber, and inflammatory and endotoxemia markers in the T1D group, higher energy [Model 5: OR, 0.26 (95% CI, 0.09-0.77), P=0.015], protein [Model 5: OR, 0.27 (95% CI, 0.08-0.87), P=0.028) and fat [Model 5: OR, 0.29 (95% CI, 0.09-0.93), P=0.037] intake were associated with lower hs-CRP levels in fully adjusted models (Table V). No statistically significant associations were observed between carbohydrate intake and inflammatory or endotoxemia markers in the T1D group (data not shown).

Higher fiber intake was associated with 66% lower hs-CRP levels when adjusted for BMI and duration of diabetes [Model 4: OR, 0.34 (95% CI, 0.12-0.99), P=0.047), but after adjustment for HbA1c and gastrointestinal symptom score, the significance was lost. Higher protein [Model 5: OR, 3.37 (95% CI, 1.14-9.95), P=0.028] intake was associated with higher fecal calprotectin levels in the T1D group in the fully adjusted model (Table V).

In the control group, higher carbohydrate consumption was associated with 92% lower odds of elevated EndoCAb IgM levels in fully adjusted model 4 [OR, 0.08 (95% CI, 0.01-0.75), P=0.027; Table SV). In the control group, a higher fat intake was associated with higher LPS and calprotectin levels in several multivariate regression models. In addition, higher fiber intake was associated with higher LPS and hs-CRP levels in multivariate models in this group (Table SV).