Abstract
Background
Albuminuria is an important marker of diabetic kidney disease (DKD) and is associated with adverse renal and cardiovascular outcomes in people with type 2 diabetes mellitus (T2DM). Although sustained hyperglycemia is a recognized contributor to DKD, glycated hemoglobin (HbA1c) does not fully capture short-term glycemic variability. Emerging evidence suggests that hypomagnesemia and altered thyroid function may be associated with albuminuria, but their combined relationships with albuminuria and the possible statistical mediating role of thyroid hormones remain unclear. This study was conducted to identify the associations of albuminuria with glycemic variability, serum magnesium, and thyroid function in T2DM patients.
Materials and methods
A cross-sectional study included 330 adults aged 30-84 years with T2DM of more than five years' duration. Continuous glucose monitoring (CGM)-derived measures, including mean glucose, glucose standard deviation (SD), coefficient of variation (CV), time in range (TIR), and mean amplitude of glycemic excursions (MAGE), were assessed alongside HbA1c, serum magnesium, free triiodothyronine (FT3), free thyroxine (FT4), thyroid-stimulating hormone (TSH), and urinary albumin-to-creatinine ratio (UACR). Associations with albuminuria were examined using univariable and multivariable logistic regression, with additional correlation and exploratory regression-based mediation analysis. Ethical approval and written informed consent were obtained.
Results
Among 330 participants, 208 (63%) had normal UACR, 98 (29.7%) had microalbuminuria, and 24 (7.3%) had macroalbuminuria. Hypomagnesemia was associated with more than twofold higher odds of albuminuria in univariable analysis (odds ratio (OR)=2.26; 95% confidence interval (CI): 1.38-3.70; p=0.001). In Model IV, which was adjusted for demographic variables, higher HbA1c (OR=1.66; 95% CI: 1.34-2.05; p<0.001), SD (OR=1.63; 95% CI: 1.18-2.26; p=0.003), CV (OR=1.06; 95% CI: 1.01-1.10; p=0.016), and MAGE (OR=1.25; 95% CI: 1.07-1.45; p=0.005) were independently associated with higher odds of albuminuria. In contrast, higher FT3 was inversely associated with albuminuria (OR=0.46; 95% CI: 0.28-0.74; p=0.001). TIR, FT4, and TSH were not significantly associated with albuminuria in univariable analysis. HbA1c was negatively correlated with FT3 (ρ=-0.133; p=0.016). Similarly, SD was negatively correlated with FT3 (ρ=-0.227; p<0.001). Exploratory mediation analysis indicated that FT3 statistically accounted for 6.9% of the association between HbA1c and albuminuria and 21.4% of the association between glucose SD and albuminuria.
Conclusion
Greater glycemic variability and lower FT3 levels were independently associated with albuminuria in patients with T2DM, while hypomagnesemia was associated with albuminuria in univariable analysis. Exploratory mediation analysis suggested that FT3 statistically accounted for a modest proportion of the associations of HbA1c and glucose SD with albuminuria. These findings indicate that CGM-derived variability measures, alongside magnesium status and thyroid function, may provide complementary information in the assessment of DKD. Prospective multicenter studies are required to clarify the temporal relationships and clinical significance of these findings.
Keywords: continuous glucose monitoring (cgm), diabetic kidney disease (dkd), free thyroxine (ft4), free triiodothyronine (ft3), glucose time in range (tir), glycated hemoglobin (hba1c), mean amplitude of glycemic excursions (mage), thyroid-stimulating hormone (tsh), type 2 diabetes mellitus (t2dm), urinary albumin-to-creatinine ratio (uacr)
Introduction
Diabetic kidney disease (DKD) or diabetic nephropathy (DN) is one of the most frequent microvascular complications of diabetes mellitus and has been reported in nearly 40% of patients with type 2 diabetes mellitus (T2DM). As a major cause of end-stage renal disease (ESRD), DKD represents a significant global burden in both developed and developing countries by reducing quality of life and life expectancy [1]. The complex mechanism of DKD remains underexplored, as several pathways and mediators are involved in its development. Albuminuria, a primary indicator of DKD, is also a significant risk factor for cardiovascular morbidity and mortality. Therefore, exploring the complex interaction between metabolic and hemodynamic factors is vital to reduce the prevalence and progression of DKD [1,2].
Optimal glycemic control is important in reducing microvascular risk and preventing the onset and progression of DKD, while chronic hyperglycemia leads to renal injury through multiple metabolic and inflammatory mechanisms [1-3]. The continuous glucose monitoring (CGM) system is a valuable emerging tool for assessing not only blood glucose variability but also collective glycemic status, furnishing a more comprehensive exploration of glucose control than traditional computations alone. Limited and conflicting literature related to the investigation of the relationship between DKD and CGM-derived glycemic indices is available so far. Some studies have reported a significant association of severity of albuminuria with glucose variability, while other researchers have found insufficient evidence to conclude such a relationship. These inconsistent findings emphasize the call for further investigations to accurately identify the function of glycemic variability in the development of DKD [4].
Abnormal serum magnesium levels are a common concern among patients with T2DM. The presence of microvascular or macrovascular complications, longer duration of diabetes, and poor glycemic control may increase the likelihood of hypomagnesemia [5]. Previous studies have also reported a significant association between hypomagnesemia, poor glycemic control, and diabetes-related complications. The overall prevalence percentage of hypomagnesemia in T2DM patients ranges from 13.5% to 47.7%, with a prevalence ratio of 32-34 per 100 T2DM patients residing in the South Asian region [5-7].
Thyroid hormones play an important role in regulating metabolic processes and maintaining the physiological functions of several organs. Recent literature have suggested an association between altered thyroid hormone levels and albuminuria in patients with T2DM [8]. Additionally, a significant association between glycemic status, thyroid function, and albuminuria has also been reported [4]. Although separate associations of glycemic variability, serum magnesium, thyroid hormones, and albuminuria have been described in the literature, studies examining these factors together remain limited, particularly in South Asian populations [8]. These findings also raised a hypothesis that thyroid hormones may mediate the associations of albuminuria with serum magnesium and glycemic levels.
Therefore, the primary objective of this study was to investigate the associations of albuminuria with CGM-derived glycemic metrics, serum magnesium, and thyroid hormones in patients with T2DM. Secondary analyses examined correlations of thyroid hormones with CGM-derived metrics and serum magnesium. Exploratory mediation analysis evaluated the statistical mediating role of thyroid hormones in the associations between CGM-derived metrics or serum magnesium and albuminuria.
By combining CGM-derived measures, thyroid hormone indices, serum magnesium levels, and albuminuria, the present study contributes to the existing literature on factors associated with DKD in T2DM.
Materials and methods
Study design
A cross-sectional observational study was conducted among 330 adults aged 30-84 years with T2DM who attended the Department of Medicine and the outpatient diabetes clinic of Sheikh Zayed Hospital, Rahim Yar Khan, Pakistan, between July 2024 and July 2025. Ethical approval was obtained from the Institutional Review Board (IRB) of Sheikh Zayed Medical College/Hospital (approval number: 275/IRB/SZMC/SZH), and written informed consent was obtained from all participants. The research proforma used for data collection is shown in the Appendices.
Sample size
The sample size was estimated prospectively using an expected albuminuria prevalence of approximately 31% based on previous Pakistani studies [9], with a 95% confidence level and 5% absolute precision, giving a required sample of approximately 330 participants. A total of 657 consecutive patients were screened during the study period, of whom 330 fulfilled the inclusion and exclusion criteria and had the required study measurements available for analysis.
Selection criteria
The inclusion of patients in this study was based on the following criteria: (1) age ≥30 and ≤84 years, (2) confirmed T2DM for more than five years' duration, (3) availability of CGM-derived data, and (4) availability of thyroid function tests, serum magnesium, and urinary albumin-to-creatinine ratio (UACR) results. Patients with type 1 diabetes mellitus, with severe liver disease, with dialysis-dependent renal failure, with previous thyroid surgery or active thyroid disease treatment, and who were pregnant or lactating were excluded from this study.
There were no missing values among the 330 participants included in the final analysis; patients without the required study measurements were excluded during eligibility screening.
Covariates
Based on theoretical relevance, covariate selection of demographic variables included age, gender, body mass index (BMI), smoking status, systolic blood pressure (SBP), diastolic blood pressure (DBP), and hypertension history. Duration of diabetes was considered as a disease-related covariate. Hypomagnesemia was considered as a binary covariate and was defined as serum magnesium <1.70 mg/dL; values of 1.70-2.20 mg/dL were classified as the normal reference range. Laboratory investigations including glycated hemoglobin (HbA1c), serum magnesium, serum creatinine, blood urea nitrogen (BUN), estimated glomerular filtration rate (eGFR), cholesterol, triglycerides, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), uric acid, free triiodothyronine (FT3), free thyroxine (FT4), and thyroid-stimulating hormone (TSH) were recorded.
CGM data were obtained using the FreeStyle Libre/Libre 3 system by Abbott Diabetes Care (Witney, United Kingdom). The CGM-based metrics, including mean glucose, standard deviation (SD) glucose, coefficient of variation (CV) glucose, time in range (TIR), and mean amplitude of glycemic excursions (MAGE), were collected. Medication information including use of insulin, sulfonylureas, benzoic acid derivative, biguanide, dipeptidyl peptidase-4 (DPP-4) inhibitor, glycosidase inhibitor, thiazolidinedione (TZD), glucagon-like peptide-1 (GLP-1) receptor agonist, sodium-glucose cotransporter 2 (SGLT-2) inhibitor, β-receptor blocker, calcium channel blockers (CCBs), angiotensin-converting enzyme (ACE) inhibitor, angiotensin II receptor blockers (ARBs), mineralocorticoid receptor antagonist (MRA), diuretic, and α-receptor blocker was also gathered and included in the analysis (research proforma attached in the Appendices).
Dependent variable
Albuminuria was classified according to UACR as follows: normal albuminuria as <30 mg/g, microalbuminuria as 30-299 mg/g, and macroalbuminuria as ≥300 mg/g. For UACR assessment, patients provided a clean-catch, mid-stream urine sample from the first morning void, following the routine collection protocol used at the study center. UACR was used to classify cross-sectional albuminuria status at the study assessment; the study was not designed to establish longitudinal persistence of albuminuria.
For univariate and multivariate binary logistic regression analyses examining factors associated with albuminuria, participants were categorized into two classes: Class I with albuminuria, defined as microalbuminuria or macroalbuminuria with a UACR ≥30 mg/g, and Class II featuring normal albumin excretion, defined as a UACR <30 mg/g. For analysis examining factors associated with macroalbuminuria, participants were categorized as having macroalbuminuria, defined as a UACR ≥300 mg/g, or having microalbuminuria or normal albumin excretion, defined as a UACR <300 mg/g.
Statistical methods
IBM SPSS Statistics for Windows, Version 23.0 (IBM Corp., Armonk, New York, United States), was used for the analysis of study data. The Shapiro-Wilk test was used to assess the normality of continuous variables. Comparison analysis among three classes of albuminuria was performed by using the chi-squared test for categorical variables, the Kruskal-Wallis H rank-sum test for non-normal continuous covariates, and one-way analysis of variance (ANOVA) for normally distributed continuous covariates.
Number with percentage (n (%)), mean±standard deviation (SD), and median with interquartile range (IQR) were calculated for categorical variables, normally distributed variables, and non-normally distributed variables, respectively. Bivariate correlation analysis of thyroid hormone indices with CGM-based metrics and magnesium level was conducted through Spearman's rank correlation test.
To examine and identify the significant risk factors of albuminuria and macroalbuminuria, a univariate binary logistic regression modeling technique was executed. The explanatory variables included were demographic characteristics, disease- and medication-related variables, biochemical factors, CGM-based metrics, serum magnesium, and thyroid hormones.
Variables from serum magnesium, thyroid function, or CGM-based metrics that were significant in univariable analyses were then evaluated in four separate domain-specific multivariable logistic regression models: Model I adjusted for biochemical variables, Model II for medication-related variables, Model III for disease-related variables, and Model IV for demographic variables. These models were used as separate adjustment domains rather than as sequential cumulative models. Each significant CGM-derived metric, FT3, and serum magnesium was evaluated separately within each adjustment model rather than being entered simultaneously into a single exposure model.
Regression-based mediation analysis was performed to examine the potential mediation produced by thyroid hormones in the association of albuminuria with serum magnesium and glycemic level. Albuminuria was considered as a binary outcome variable, HbA1c, mean, SD, and CV of glucose, MAGE, and serum magnesium were taken as exposure variables, and FT3, FT4, and TSH were regarded as mediators. A simple mediation model with a 95% confidence interval (CI) and without adjustment of any covariate was executed. A probability value (p-value) of <0.05 is considered statistically significant. Given the cross-sectional design, the mediation analysis was treated as exploratory statistical mediation and was not interpreted as establishing temporal or causal pathways.
Results
Basic descriptive statistics, including mean, median, and SD, were calculated. The findings showed the significance of explanatory variables based on the chi-squared test, Kruskal-Wallis H rank-sum test, and ANOVA. Univariate and multivariate binary logistic regression models identified the significant explanatory variables. Mediation analysis examined indirect statistical associations through thyroid hormones in the association of albuminuria with serum magnesium and glycemic measures.
Descriptive characteristics
Based on UACR, the total sample of 330 T2DM patients was classified into a macroalbuminuria class (n=24), a microalbuminuria class (n=98), and a normal albuminuria class (n=208). The distribution of basic descriptive characteristics for the three classes of T2DM patients is shown in Table 1.
Table 1. Clinical descriptive statistics.
F: one-way ANOVA test statistic; χ²: chi-squared test statistic; H: Kruskal-Wallis H test statistic
T2DM: type 2 diabetes mellitus; SD: standard deviation; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; BUN: blood urea nitrogen; eGFR: estimated glomerular filtration rate; UACR: urinary albumin-to-creatinine ratio; DPP-4: dipeptidyl peptidase-4; TZD: thiazolidinedione; GLP-1: glucagon-like peptide-1; SGLT-2: sodium-glucose cotransporter 2; CCB: calcium channel blocker; ACE: angiotensin-converting enzyme; ARB: angiotensin II receptor blocker; MRA: mineralocorticoid receptor antagonist; IQR: interquartile range; ANOVA: analysis of variance
| Variables | Normal albuminuria (n=208) | Microalbuminuria (n=98) | Macroalbuminuria (n=24) | Total T2DM patients (N=330) | Test statistic | P-value |
| Age (years), mean±SD | 57.9±10.9 | 58.3±9.5 | 59.1±11.2 | 58.1±10.5 | F=0.16 | 0.850 |
| Sex, n (%) | - | - | - | - | χ²=0.53 | 0.769 |
| Female sex | 99 (47.6) | 45 (45.9) | 13 (54.2) | 157 (47.6) | χ²=72.20 | <0.001 |
| Male sex | 109 (52.4) | 53 (54.1) | 11 (45.8) | 173 (52.4) | χ²=83.83 | <0.001 |
| BMI (kg/m2), mean±SD | 26.1±3.5 | 25.3±3.3 | 25.9±2.9 | 25.8±3.4 | F=2.07 | 0.127 |
| SBP (mmHg), mean±SD | 132.4±13.5 | 132.6±13.2 | 127.3±10.9 | 132.1±13.3 | F=1.69 | 0.185 |
| DBP (mmHg), mean±SD | 78.8±8.9 | 79.9±9.1 | 84.1±10.2 | 79.4±9.1 | F=3.85 | 0.022 |
| Smoking, n (%) | - | - | - | - | χ²=7.55 | 0.110 |
| Smoking: never | 126 (60.6) | 59 (60.2) | 13 (54.2) | 198 (60) | χ²=97.85 | <0.001 |
| Smoking: ex-smoker | 41 (19.7) | 15 (15.3) | 9 (37.5) | 65 (19.7) | χ²=26.71 | <0.001 |
| Smoking: current smoker | 41 (19.7) | 24 (24.5) | 2 (8.3) | 67 (20.3) | χ²=34.24 | <0.001 |
| History of hypertension, n (%) | - | - | - | - | χ²=6.19 | 0.045 |
| Hypertension: no | 113 (54.3) | 39 (39.8) | 10 (41.7) | 162 (49.1) | χ²=104.48 | <0.001 |
| Hypertension: yes | 95 (45.7) | 59 (60.2) | 14 (58.3) | 168 (50.9) | χ²=58.82 | <0.001 |
| Duration of T2DM (years), median (IQR) | 8.3 (5.6-11.8) | 9.05 (4.2-11.8) | 8.0 (5.2-9.2) | 8.4 (5.3-11.7) | H=2.45 | 0.294 |
| Cholesterol (mmol/L), mean±SD | 4.71±0.82 | 4.68±0.84 | 4.80±0.91 | 4.71±0.83 | F=0.21 | 0.811 |
| HDL-C (mmol/L), mean±SD | 1.03±0.21 | 1.02±0.21 | 0.90±0.16 | 1.02±0.21 | F=4.32 | 0.014 |
| LDL-C (mmol/L), mean±SD | 2.44±0.65 | 2.67±0.68 | 2.41±0.84 | 2.51±0.68 | F=4.26 | 0.015 |
| Triglycerides (mmol/L), mean±SD | 1.63±0.59 | 1.78±0.63 | 1.93±0.61 | 1.70±0.61 | F=4.01 | 0.019 |
| Uric acid (μmol/L), mean±SD | 305.68±91.32 | 323.68±77.72 | 356.99±91.00 | 314.76±88.39 | F=4.43 | 0.013 |
| BUN (mmol/L), mean±SD | 5.82±1.46 | 6.17±1.67 | 6.66±1.70 | 5.98±1.56 | F=4.23 | 0.015 |
| Creatinine (μmol/L), median (IQR) | 76.7 (65.9-93.7) | 89.8 (75.8-104.3) | 102.2 (70.4-113.6) | 84.0 (67.5-99.4) | H=5.53 | 0.063 |
| eGFR (mL/min/1.73 m2), median (IQR) | 92.8 (77.9-100.6) | 85.5 (72.1-94.6) | 82.9 (60.1-111.1) | 89.9 (75.0-99.7) | H=0.24 | 0.887 |
| UACR (mg/g), median (IQR) | 15.4 (10.2-21.7) | 144.9 (71.1-214.3) | 581.85 (418.2-846.5) | 22.20 (13.0-113.8) | H=57.37 | <0.001 |
| Hypomagnesemia (mg/dl) (yes), n (%) | 44 (21.2) | 40 (40.8) | 6 (25) | 90 (27.3) | χ²=13.05 | 0.001 |
| Insulin use, n (%) | 112 (53.8) | 50 (51) | 16 (66.7) | 178 (53.9) | χ²=1.90 | 0.386 |
| Sulfonylureas use, n (%) | 33 (15.9) | 18 (18.4) | 6 (25) | 57 (17.3) | χ²=1.37 | 0.503 |
| Benzoic acid derivative use, n (%) | 23 (11.1) | 12 (12.2) | 5 (20.8) | 40 (12.1) | χ²=1.93 | 0.381 |
| Biguanide use, n (%) | 135 (64.9) | 65 (66.3) | 14 (58.3) | 214 (64.8) | χ²=0.54 | 0.763 |
| DPP-4 inhibitor use, n (%) | 50 (24) | 28 (28.6) | 6 (25) | 84 (25.5) | χ²=0.72 | 0.696 |
| Glycosidase inhibitor use, n (%) | 72 (34.6) | 33 (33.7) | 10 (41.7) | 115 (34.8) | χ²=0.56 | 0.757 |
| TZD use, n (%) | 16 (7.7) | 11 (11.2) | 5 (20.8) | 32 (9.7) | χ²=4.62 | 0.100 |
| GLP-1 receptor agonist use, n (%) | 16 (7.7) | 8 (8.2) | 5 (20.8) | 29 (8.8) | χ²=4.70 | 0.095 |
| SGLT-2 inhibitor use, n (%) | 19 (9.1) | 6 (6.1) | 5 (20.8) | 30 (9.1) | χ²=5.05 | 0.080 |
| β-Receptor blocker use, n (%) | 25 (12) | 9 (9.2) | 6 (25) | 40 (12.1) | χ²=4.53 | 0.104 |
| CCB use, n (%) | 50 (24) | 22 (22.4) | 7 (29.2) | 79 (23.9) | χ²=0.481 | 0.786 |
| ACE inhibitor use, n (%) | 20 (9.6) | 11 (11.2) | 5 (20.8) | 36 (10.9) | χ²=2.80 | 0.247 |
| ARB use, n (%) | 51 (24.4) | 31 (31.6) | 7 (29.2) | 89 (27) | χ²=1.78 | 0.412 |
| MRA use, n (%) | 22 (10.6) | 14 (14.3) | 6 (25) | 42 (12.7) | χ²=4.34 | 0.114 |
| Diuretic use, n (%) | 19 (9.1) | 10 (10.2) | 6 (25) | 35 (10.6) | χ²=5.74 | 0.057 |
| α-Receptor blocker use, n (%) | 21 (10.1) | 11 (11.2) | 5 (20.8) | 37 (11.2) | χ²=2.49 | 0.288 |
The mean±SD age of patients with T2DM was 58.13±10.53 including 173 (52.4%) males and 157 (47.6%) females. Statistical tests of comparison showed that history of hypertension, DBP, HDL-C, LDL-C, triglycerides, uric acid, BUN, UACR, and hypomagnesemia (p≤0.05) were significantly different among the three classes. Other variables, including age, sex, smoking status, BMI, SBP, duration of T2DM, cholesterol, creatinine, eGFR, and use of insulin, sulfonylureas, benzoic acid derivative, biguanide, DPP-4 inhibitor, glycosidase inhibitor, TZD, GLP-1 receptor agonist, SGLT-2 inhibitor, β-receptor blocker, CCB, ACE, ARB, MRA, diuretic, and α-receptor blocker, were not significantly different (p>0.05) among the three classes.
Comparative analysis of CGM-based metrics, magnesium level, and thyroid function
Table 2 presented the descriptive statistics of CGM-based metrics, magnesium level, and thyroid function for the whole sample of T2DM patients and each class of albuminuria separately. The analysis showed the deviation of HbA1c (p=0.021), mean of glucose (p=0.001), SD of glucose (p=0.018), CV of glucose (p=0.002), MAGE (p=0.015), FT3 (p=0.007), and magnesium level (p=0.025) among macroalbuminuria, microalbuminuria, and normal albuminuria classes. The findings further found that the distribution of TIR, FT4, and TSH was similar (p>0.05) among all three albuminuria classes.
Table 2. Distribution of CGM-based metrics, magnesium level, and thyroid function.
F: one-way ANOVA test statistic; χ²: chi-squared test statistic; H: Kruskal-Wallis H test statistic
CGM: continuous glucose monitoring; T2DM: type 2 diabetes mellitus; HbA1c: glycated hemoglobin; SD: standard deviation; IQR: interquartile range; TIR: time in range; CV: coefficient of variation; MAGE: mean amplitude of glycemic excursions; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone; ANOVA: analysis of variance
| Variables | Normal (n=208) | Microalbuminuria (n=98) | Macroalbuminuria (n=24) | Total T2DM patients (N=330) | Test statistic | P-value |
| HbA1c (%), median (IQR) | 8.7 (7.9-9.6) | 9.7 (8.9-10.3) | 8.9 (8.5-9.5) | 9.0 (8.1-9.9) | H=7.73 | 0.021 |
| Mean of glucose (mmol/L), mean±SD | 9.2±1.8 | 9.9±1.7 | 9.0±1.5 | 9.4±1.8 | F=7.14 | 0.001 |
| SD of glucose (mmol/L), median (IQR) | 2.1 (1.5-2.6) | 2.4 (1.8-2.8) | 2.5 (1.3-2.8) | 2.2 (1.6-2.7) | H=8.03 | 0.018 |
| TIR (%), median (IQR) | 60.5 (49.2-75.5) | 59.8 (49.0-72.4) | 57.1 (50.2-68.4) | 59.9 (49.3-75.1) | H=1.56 | 0.459 |
| CV of glucose (%), mean±SD | 22.3±5.7 | 24.5±4.7 | 21.5±5.0 | 22.9±5.4 | F=6.25 | 0.002 |
| MAGE (mmol/L), mean±SD | 5.0±1.6 | 5.5±1.4 | 5.6±1.1 | 5.2±1.5 | F=4.28 | 0.015 |
| FT3 (pmol/L), median (IQR) | 4.3 (3.9-4.6) | 4.1 (3.7-4.5) | 4.0 (3.8-4.3) | 4.2 (3.9-4.5) | H=9.92 | 0.007 |
| FT4 (pmol/L), mean±SD | 15.9±2.1 | 15.7±1.9 | 15.5±2.6 | 15.8±2.1 | F=0.33 | 0.720 |
| TSH (mIU/L), mean±SD | 2.1±0.8 | 2.0±0.8 | 1.8±0.9 | 2.1±0.8 | F=1.53 | 0.219 |
| Serum magnesium (mg/dL), mean±SD | 1.9±0.2 | 1.8±0.2 | 1.9±0.2 | 1.8±0.2 | F=3.73 | 0.025 |
Bivariate correlation analysis between CGM-based metrics, magnesium level, and thyroid hormones
Spearman's rank correlation measure (ρ) of thyroid hormone indices with CGM-based metrics and magnesium level is presented in Table 3. The analysis showed that HbA1c (p=0.016) and SD of glucose (p<0.001) were negatively correlated with FT3. Moreover, MAGE (p=0.038) showed a significant negative correlation with TSH. All other CGM-based metrics and serum magnesium were uncorrelated with FT3 for the observed data. Apart from the significant inverse correlation between MAGE and TSH, no other significant correlations involving FT4 or TSH were observed.
Table 3. Correlation between CGM-based metrics, magnesium level, and thyroid hormone level.
HbA1c: glycated hemoglobin; SD: standard deviation; TIR: time in range; CV: coefficient of variation; MAGE: mean amplitude of glycemic excursions; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone; rho (ρ): Spearman's rank correlation coefficient
| Variables | FT3 (pmol/L) | FT4 (pmol/L) | TSH (mIU/L) | |||
| ρ | P-value | ρ | P-value | ρ | P-value | |
| HbA1c (%) | -0.133 | 0.016 | -0.072 | 0.192 | -0.076 | 0.171 |
| Mean of glucose (mmol/L) | 0.012 | 0.832 | -0.043 | 0.434 | -0.048 | 0.383 |
| SD of glucose (mmol/L) | -0.227 | <0.001 | -0.043 | 0.440 | 0.041 | 0.459 |
| TIR (%) | 0.074 | 0.178 | -0.001 | 0.990 | -0.017 | 0.763 |
| CV of glucose (%) | -0.022 | 0.684 | -0.068 | 0.220 | 0.077 | 0.161 |
| MAGE (mmol/L) | 0.009 | 0.867 | -0.075 | 0.173 | -0.114 | 0.038 |
| Serum magnesium (mg/dL) | -0.012 | 0.826 | 0.003 | 0.956 | -0.011 | 0.839 |
Identification of factors associated with albuminuria and macroalbuminuria using univariate binary logistic regression
The univariate binary logistic regression analysis presented in Table 4 was performed to examine the effect of demographic, disease-related, biochemical, and medication information variables on albuminuria and macroalbuminuria. The investigation was further extended to identify the significant associations among CGM-based metrics, magnesium level, and thyroid hormones. The results revealed a significant association of history of hypertension (p=0.013), LDL-C (p=0.021), triglycerides (p=0.011), uric acid (p=0.016), BUN (p=0.012), creatinine (p<0.001), eGFR (p=0.042), and hypomagnesemia (p=0.001) with the risk of albuminuria. The present study found no significant association of albuminuria with any medication-related covariate. Among CGM-based metrics, mean of glucose (p=0.004), SD of glucose (p=0.006), CV of glucose (p=0.013), and MAGE (p=0.004) were associated with the odds of albuminuria. Additionally, HbA1c (p<0.001) was positively while FT3 (p=0.002) and serum magnesium (p=0.033) were inversely associated with the odds of albuminuria.
Table 4. The univariate logistic regression analysis to examine the association of albuminuria with CGM-based metrics, thyroid hormone indices, and magnesium level.
Statistical test: univariate binary logistic regression. Odds ratios are reported with 95% confidence intervals.
OR: odds ratio; CI: confidence interval; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; T2DM: type 2 diabetes mellitus; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; BUN: blood urea nitrogen; eGFR: estimated glomerular filtration rate; DPP-4: dipeptidyl peptidase-4; TZD: thiazolidinedione; GLP-1: glucagon-like peptide-1; SGLT-2: sodium-glucose cotransporter 2; CCB: calcium channel blocker; ACE: angiotensin-converting enzyme; ARB: angiotensin II receptor blocker; MRA: mineralocorticoid receptor antagonist; HbA1c: glycated hemoglobin; TIR: time in range; CV: coefficient of variation; MAGE: mean amplitude of glycemic excursions; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone
| Variables | Odds of albuminuria | Odds of macroalbuminuria | ||
| OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Age (years) | 1.01 (0.98-1.03) | 0.628 | 1.01 (0.97-1.05) | 0.65 |
| Gender (male) | 1.00 (0.64-1.57) | 0.992 | 1.33 (0.58-3.06) | 0.503 |
| BMI (kg/m2) | 0.94 (0.88-1.00) | 0.062 | 1.00 (0.89-1.13) | 0.97 |
| SBP (mmHg) | 0.99 (0.98-1.01) | 0.576 | 0.97 (0.94-1.00) | 0.068 |
| DBP (mmHg) | 1.02 (0.99-1.05) | 0.097 | 1.06 (1.02-1.11) | 0.009 |
| Smoking (yes) | 1.05 (0.79-1.39) | 0.730 | 0.90 (0.53-1.54) | 0.698 |
| History of hypertension (yes) | 1.77 (1.13-2.79) | 0.013 | 1.38 (0.59-3.21) | 0.452 |
| Duration of T2DM (years) | 0.98 (0.94-1.03) | 0.527 | 0.94 (0.85-1.03) | 0.202 |
| Cholesterol (mmol/L) | 0.99 (0.75-1.29) | 0.911 | 1.15 (0.70-1.90) | 0.579 |
| HDL-C (mmol/L) | 0.48 (0.16-1.41) | 0.182 | 0.04 (0.01-0.39) | 0.004 |
| LDL-C (mmol/L) | 1.49 (1.06-2.09) | 0.021 | 0.79 (0.43-1.46) | 0.454 |
| Triglycerides (mmol/L) | 1.63 (1.12-2.37) | 0.011 | 1.95 (1.00-3.80) | 0.049 |
| Uric acid (μmol/L) | 1.01 (1.00-1.02) | 0.016 | 1.01 (1.00-1.02) | 0.016 |
| BUN (mmol/L) | 1.21 (1.04-1.39) | 0.012 | 1.35 (1.03-1.77) | 0.029 |
| Creatinine (μmol/L) | 1.02 (1.01-1.03) | <0.001 | 1.03 (1.01-1.05) | 0.004 |
| eGFR (mL/min/1.73 m2) | 0.99 (0.98-1.00) | 0.042 | 0.99 (0.97-1.02) | 0.620 |
| Hypomagnesemia (yes) | 2.26 (1.38-3.70) | 0.001 | 0.88 (0.34-2.29) | 0.795 |
| Insulin use (yes) | 1.01 (0.65-1.58) | 0.965 | 1.78 (0.74-4.28) | 0.199 |
| Sulfonylurea use (yes) | 1.29 (0.73-2.32) | 0.378 | 1.67 (0.63-4.40) | 0.303 |
| Benzoic acid derivative use (yes) | 1.30 (0.67-2.55) | 0.440 | 2.04 (0.72-5.80) | 0.182 |
| Biguanide use (yes) | 0.99 (0.62-1.59) | 0.978 | 0.74 (0.32-1.73) | 0.489 |
| DPP-4 inhibitor use (yes) | 1.22 (0.74-2.03) | 0.441 | 0.97 (0.37-2.54) | 0.958 |
| Glycosidase inhibitor use(yes) | 1.03 (0.64-1.64) | 0.908 | 1.37 (0.59-3.18) | 0.468 |
| TZD use (yes) | 1.81 (0.87-3.77) | 0.112 | 2.72 (0.94-7.86) | 0.065 |
| GLP-1 receptor agonist use (yes) | 1.43 (0.66-3.09) | 0.361 | 3.09 (1.06-9.01) | 0.039 |
| SGLT-2 inhibitor use (yes) | 0.99 (0.45-2.15) | 0.971 | 2.96 (1.02-8.59) | 0.046 |
| β-Receptor blocker use (yes) | 1.03 (0.52-2.03) | 0.941 | 2.67 (0.99-7.18) | 0.052 |
| CCB use (yes) | 0.99 (0.58-1.67) | 0.956 | 1.34 (0.53-3.36) | 0.534 |
| ACE inhibitor use (yes) | 1.42 (0.71-2.86) | 0.327 | 2.33 (0.82-6.69) | 0.114 |
| ARB use (yes) | 1.39 (0.85-2.29) | 0.191 | 1.13 (0.45-2.81) | 0.801 |
| MRA use (yes) | 1.66 (0.86-3.18) | 0.129 | 2.50 (0.93-6.71) | 0.069 |
| Diuretic use (yes) | 1.50 (0.74-3.04) | 0.259 | 3.18 (1.17-8.66) | 0.023 |
| α-Receptor blocker use (yes) | 1.34 (0.67-2.69) | 0.403 | 2.25 (0.79-6.45) | 0.130 |
| HbA1c (%) | 1.70 (1.38-2.09) | <0.001 | 1.06 (0.76-1.47) | 0.752 |
| Mean of glucose (mmol/L) | 1.21 (1.06-1.37) | 0.004 | 0.88 (0.69-1.11) | 0.272 |
| SD of glucose (mmol/L) | 1.55 (1.14-2.11) | 0.006 | 1.17 (0.69-2.04) | 0.588 |
| TIR (%) | 0.99 (0.98-1.01) | 0.786 | 0.99 (0.97-1.01) | 0.388 |
| CV of glucose (%) | 1.06 (1.01-1.11) | 0.013 | 0.95 (0.88-1.03) | 0.179 |
| MAGE (mmol/L) | 1.25 (1.03-1.46) | 0.004 | 1.18 (0.89-1.55) | 0.179 |
| FT3 (pmol/L) | 0.49 (0.31-0.77) | 0.002 | 0.66 (0.29-1.49) | 0.138 |
| FT4 (pmol/L) | 0.96 (0.86-1.07) | 0.476 | 0.94 (0.77-1.14) | 0.531 |
| TSH (mIU/L) | 0.85 (0.65-1.11) | 0.224 | 0.66 (0.40-1.10) | 0.109 |
| Serum magnesium (mg/dL) | 0.29 (0.09-0.90) | 0.033 | 2.28 (0.27-19.17) | 0.449 |
Concerning macroalbuminuria, the study findings showed that DBP (p=0.009) was a highly significant factor among demographic characteristics. Based on biochemical characteristics, HDL-C (p=0.004) showed an inverse association with the odds of macroalbuminuria, whereas triglycerides (p=0.049), uric acid (p=0.016), BUN (p=0.029), and creatinine (p=0.004) showed positive associations with the odds of macroalbuminuria. Medical treatment, including GLP-1 receptor agonist (p=0.039), SGLT-2 inhibitor (p=0.046), and diuretic (p=0.023), was also significantly associated with the odds of macroalbuminuria.
Multivariate logistic regression adjusted models
Multivariable logistic regression analysis examined four separate adjustment models. Model I adjusted for biochemical variables, Model II for medication-related variables, Model III for duration of T2DM, and Model IV for demographic variables. Across these models, HbA1c, mean glucose, glucose SD, CV, and MAGE remained associated with higher odds of albuminuria, while FT3 remained inversely associated with albuminuria. Serum magnesium was inversely associated with albuminuria in Models I-III, but the association was attenuated and was no longer statistically significant in Model IV (p=0.060). No significant associations with macroalbuminuria were observed for these variables in the adjusted models (Table 5).
Table 5. Association of albuminuria and macroalbuminuria with significant CGM-based metrics, thyroid hormone indices, and magnesium level through multivariate logistic regression model.
Statistical test: multivariable binary logistic regression. Odds ratios are reported with 95% confidence intervals.
Model I is adjusted for eight biochemical variables including serum creatinine, BUN, eGFR, cholesterol, triglycerides, HDL-C, LDL-C, and uric acid. Model II is adjusted for 16 medication information variables including use of insulin, sulfonylureas, benzoic acid derivative, biguanide, DPP-4 inhibitor, glycosidase inhibitor, TZD, GLP-1 receptor agonist, SGLT-2 inhibitor, β-receptor blocker, CCBs, ACE inhibitor, ARBs, MRA, diuretic, and α-receptor blocker. Model III is adjusted for one disease-related variable including duration of T2DM. Model IV is adjusted for seven demographic variables including age, gender, BMI, smoking status, SBP, DBP, and hypertension history.
HbA1c: glycated hemoglobin; SD: standard deviation; CV: coefficient of variation; MAGE: mean amplitude of glycemic excursions; FT3: free triiodothyronine; BUN: blood urea nitrogen; eGFR: estimated glomerular filtration rate; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; PP-4: dipeptidyl peptidase-4; TZD: thiazolidinedione; GLP-1: glucagon-like peptide-1; SGLT-2: sodium-glucose cotransporter 2; CCB: calcium channel blocker; ACE: angiotensin-converting enzyme; ARB: angiotensin II receptor blocker; MRA: mineralocorticoid receptor antagonist; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; T2DM: type 2 diabetes mellitus
| Variables | Odds of albuminuria | Odds of macroalbuminuria | ||
| OR (95% CI) | P-value | OR (95% CI) | P-value | |
| HbA1c (per %) | ||||
| Model I | 1.62 (1.31-2.01) | <0.001 | 0.91 (0.62-1.33) | 0.622 |
| Model II | 1.68 (1.36-2.07) | <0.001 | 1.04 (0.71-1.50) | 0.859 |
| Model III | 1.69 (1.38-2.09) | <0.001 | 1.04 (0.74-1.45) | 0.834 |
| Model IV | 1.66 (1.34-2.05) | <0.001 | 0.97 (0.68-1.38) | 0.857 |
| Mean of glucose (mmol/L) | ||||
| Model I | 1.16 (1.01-1.33) | 0.039 | 0.79 (0.61-1.05) | 0.101 |
| Model II | 1.23 (1.07-1.40) | 0.003 | 0.86 (0.67-1.12) | 0.271 |
| Model III | 1.21 (1.06-1.37) | 0.005 | 0.87 (0.69-1.09) | 0.240 |
| Model IV | 1.19 (1.04-1.36) | 0.010 | 0.88 (0.69-1.12) | 0.281 |
| SD of glucose (mmol/L) | ||||
| Model I | 1.47 (1.06-2.04) | 0.021 | 1.21 (0.66-2.23) | 0.541 |
| Model II | 1.58 (1.15-2.18) | 0.005 | 1.22 (0.66-2.25) | 0.532 |
| Model III | 1.56 (1.14-2.12) | 0.005 | 1.19 (0.68-2.08) | 0.546 |
| Model IV | 1.63 (1.18-2.26) | 0.003 | 1.22 (0.67-2.22) | 0.516 |
| CV of glucose (%) | ||||
| Model I | 1.06 (1.01-1.11) | 0.008 | 0.95 (0.87-1.04) | 0.276 |
| Model II | 1.07 (1.02-1.12) | 0.005 | 0.96 (0.88-1.04) | 0.330 |
| Model III | 1.06 (1.01-1.10) | 0.012 | 0.95 (0.88-1.03) | 0.199 |
| Model IV | 1.06 (1.01-1.10) | 0.016 | 0.92 (0.84-1.00) | 0.055 |
| MAGE (mmol/L) | ||||
| Model I | 1.24 (1.06-1.46) | 0.009 | 1.09 (0.79-1.49) | 0.600 |
| Model II | 1.27 (1.08-1.48) | 0.004 | 1.26 (0.94-1.70) | 0.123 |
| Model III | 1.26 (1.08-1.46) | 0.004 | 1.19 (0.90-1.56) | 0.219 |
| Model IV | 1.25 (1.07-1.45) | 0.005 | 1.17 (0.88-1.54) | 0.282 |
| FT3 (pmol/L) | ||||
| Model I | 0.56 (0.34-0.91) | 0.019 | 0.76 (0.31-1.85) | 0.541 |
| Model II | 0.46 (0.29-0.74) | 0.001 | 0.57 (0.22-1.43) | 0.227 |
| Model III | 0.48 (0.31-0.76) | 0.002 | 0.65 (0.29-1.46) | 0.294 |
| Model IV | 0.46 (0.28-0.74) | 0.001 | 0.58 (0.23-1.44) | 0.239 |
| Serum magnesium (mg/dL) | ||||
| Model I | 0.27 (0.08-0.92) | 0.037 | 3.71 (0.36-38.06) | 0.270 |
| Model II | 0.21 (0.06-0.72) | 0.013 | 1.69 (0.15-19.44) | 0.676 |
| Model III | 0.29 (0.09-0.91) | 0.034 | 2.38 (0.28-19.91) | 0.425 |
| Model IV | 0.32 (0.09-1.05) | 0.060 | 2.47 (0.28-21.63) | 0.413 |
Exploratory mediation analysis
Exploratory regression-based mediation analysis showed statistically significant indirect associations through FT3 for HbA1c and glucose SD, accounting for 6.9% and 21.4% of their respective associations with albuminuria. No statistically significant indirect associations through FT3 were observed for mean glucose, CV, MAGE, or serum magnesium, and no significant indirect associations through FT4 or TSH were identified. These findings are interpreted as exploratory statistical mediation rather than evidence of a temporal or causal pathway (Table 6).
Table 6. Mediating effects of thyroid hormones.
Statistical method: regression-based mediation analysis
CI: confidence interval; HbA1c: glycated hemoglobin; SD: standard deviation; CV: coefficient of variation; MAGE: mean amplitude of glycemic excursions; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone
| Exposure variable | Mediator | Effect | Estimate | 95% CI | P-value | Proportion mediated (%) |
| HbA1c | FT3 | Average indirect | 0.000219 | 0.00001 to 0.0014 | 0.018 | 6.9 |
| HbA1c | FT3 | Average direct | 0.0029 | 0.0005 to 0.0093 | <0.001 | 6.9 |
| HbA1c | FT3 | Total | 0.0032 | 0.0006 to 0.0106 | <0.001 | 6.9 |
| Mean of glucose | FT3 | Average indirect | -0.0006 | -0.0044 to 0.0014 | 0.578 | -3.57 |
| Mean of glucose | FT3 | Average direct | 0.0176 | 0.0092 to 0.02159 | 0.0012 | -3.57 |
| Mean of glucose | FT3 | Total | 0.0170 | 0.0089 to 0.0196 | 0.0024 | -3.57 |
| SD of glucose | FT3 | Average indirect | 0.0159 | 0.0033 to 0.0346 | 0.0112 | 21.4 |
| SD of glucose | FT3 | Average direct | 0.0585 | 0.0047 to 0.087 | 0.0384 | 21.4 |
| SD of glucose | FT3 | Total | 0.0745 | 0.0263 to 0.0981 | 0.0080 | 21.4 |
| CV of glucose | FT3 | Average indirect | 0.0002 | -0.0009 to 0.0011 | 0.6836 | 2.74 |
| CV of glucose | FT3 | Average direct | 0.0067 | 0.0027 to 0.0078 | 0.0100 | 2.74 |
| CV of glucose | FT3 | Total | 0.0069 | 0.0028 to 0.0078 | 0.0100 | 2.74 |
| MAGE | FT3 | Average indirect | 0.00004 | -0.0039 to 0.0038 | 0.9904 | 0.156 |
| MAGE | FT3 | Average direct | 0.0315 | 0.0159 to 0.0376 | 0.0024 | 0.156 |
| MAGE | FT3 | Total | 0.0315 | 0.0159 to 0.0378 | 0.0032 | 0.156 |
| Serum magnesium | FT3 | Average indirect | 0.0090 | -0.0288 to 0.0538 | 0.6412 | -3.94 |
| Serum magnesium | FT3 | Average direct | -0.2384 | -0.2715 to -0.0359 | 0.0280 | -3.94 |
| Serum magnesium | FT3 | Total | -0.2294 | -0.2527450 to -0.01742 | 0.0356 | -3.94 |
| HbA1c | FT4 | Average indirect | 0.00001 | -0.0001 to 0.0002 | 0.7652 | 0.40 |
| HbA1c | FT4 | Average direct | 0.0033 | 0.0006 to 0.0106 | <0.001 | 0.40 |
| HbA1c | FT4 | Total | 0.0033 | 0.0006 to 0.0106 | <0.001 | 0.40 |
| Mean of glucose | FT4 | Average indirect | 0.0001 | -0.0009 to 0.0013 | 0.8112 | 0.86 |
| Mean of glucose | FT4 | Average direct | 0.0164 | 0.0085 to 0.0194 | 0.0020 | 0.86 |
| Mean of glucose | FT4 | Total | 0.0166 | 0.0089 to 0.0194 | 0.0016 | 0.86 |
| SD of glucose | FT4 | Average indirect | 0.0006 | -0.0030 to 0.0053 | 0.7804 | 0.78 |
| SD of glucose | FT4 | Average direct | 0.0741 | 0.0260 to 0.0972 | 0.0076 | 0.78 |
| SD of glucose | FT4 | Total | 0.0746 | 0.0266 to 0.0978 | 0.0080 | 0.78 |
| CV of glucose | FT4 | Average indirect | 0.00008 | -0.0003 to 0.0006 | 0.7164 | 1.20 |
| CV of glucose | FT4 | Average direct | 0.0067 | 0.0026 to 0.0077 | 0.0104 | 1.20 |
| CV of glucose | FT4 | Total | 0.0068 | 0.0028 to 0.0078 | 0.0100 | 1.20 |
| MAGE | FT4 | Average indirect | 0.0004 | -0.0018 to 0.003 | 0.6788 | 1.42 |
| MAGE | FT4 | Average direct | 0.0308 | 0.0148 to 0.0366 | 0.0032 | 1.42 |
| MAGE | FT4 | Total | 0.0312 | 0.0158 to 0.0367 | 0.0028 | 1.42 |
| Serum magnesium | FT4 | Average indirect | -0.0004 | -0.0142 to 0.0145 | 0.9780 | 0.18 |
| Serum magnesium | FT4 | Average direct | -0.2275 | -0.2515 to -0.0194 | 0.0352 | 0.18 |
| Serum magnesium | FT4 | Total | -0.2279 | -0.2512 to -0.0177 | 0.0356 | 0.18 |
| HbA1c | TSH | Average indirect | 0.00004 | -0.0001 to 0.0004 | 0.488 | 1.38 |
| HbA1c | TSH | Average direct | 0.0032 | 0.0006 to 0.0104 | <0.001 | 1.38 |
| HbA1c | TSH | Total | 0.0032 | 0.0006 to 0.0106 | <0.001 | 1.38 |
| Mean of glucose | TSH | Average indirect | 0.0003 | -0.0005 to 0.0020 | 0.5684 | 1.58 |
| Mean of glucose | TSH | Average direct | 0.0163 | 0.0083 to 0.0192 | 0.0024 | 1.58 |
| Mean of glucose | TSH | Total | 0.0166 | 0.0085 to 0.0194 | 0.0016 | 1.58 |
| SD of glucose | TSH | Average indirect | -0.0015 | -0.0079 to 0.0026 | 0.5484 | -1.96 |
| SD of glucose | TSH | Average direct | 0.0761 | 0.0282 to 0.0997 | 0.0068 | -1.96 |
| SD of glucose | TSH | Total | 0.0746 | 0.0266 to 0.0978 | 0.0080 | -1.96 |
| CV of glucose | TSH | Average indirect | -0.0003 | -0.0014 to 0.0002 | 0.2604 | -5.03 |
| CV of glucose | TSH | Average direct | 0.0072 | 0.0032 to 0.0083 | 0.0072 | -5.03 |
| CV of glucose | TSH | Total | 0.0068 | 0.0028 to 0.0078 | 0.0100 | -5.03 |
| MAGE | TSH | Average indirect | 0.0012 | -0.0016 to 0.005 | 0.4176 | 3.78 |
| MAGE | TSH | Average direct | 0.0300 | 0.0134 to 0.0359 | 0.0032 | 3.78 |
| MAGE | TSH | Total | 0.0312 | 0.0158 to 0.0367 | 0.0024 | 3.78 |
| Serum magnesium | TSH | Average indirect | 0.0031 | -0.0172 to 0.0220 | 0.7532 | -1.35 |
| Serum magnesium | TSH | Average direct | -0.2310 | -0.2579 to -0.0200 | 0.0376 | -1.35 |
| Serum magnesium | TSH | Total | -0.2279 | -0.2510 to -0.0176 | 0.0356 | -1.35 |
The direct effect analysis showed that HbA1c, mean, SD, and CV of glucose, MAGE, and serum magnesium were associated with albuminuria, independent of FT3, FT4, and TSH.
Discussion
This cross-sectional study investigated the individual and collective relationships between glycemic variability, hypomagnesemia, thyroid hormones, and albuminuria in T2DM patients. The findings demonstrated that CGM-based metrics, including mean of glucose, SD of glucose, CV of glucose, and MAGE, were associated with higher odds of albuminuria. These relationships persisted in multivariate models adjusted for biochemical factors, medication-related information, disease-related variables, and demographic characteristics.
Higher variation in these CGM-derived metrics was associated with albuminuria in the present study. Previous work has established the clinical value of CGM-derived metrics for characterizing glycemic exposure [3,10-14], while studies specifically examining renal outcomes have reported associations between glycemic variability and albuminuria [4,11]. Regarding TIR, the present study found no significant association with albuminuria or macroalbuminuria. Luo et al. similarly reported a non-significant association of TIR with albuminuria [4].
HbA1c remains an established marker of long-term glycemic exposure, although it does not fully capture short-term glycemic variability. In the present analysis, higher HbA1c was associated with albuminuria. Long-term glycemic exposure has also been linked to microvascular complications in major longitudinal diabetes studies [15,16]. A past study presented a comprehensive discussion about the complications, prevention, and treatment of diabetes. The author stated that a 5.5-9.5% increase in HbA1c causes an increment of approximately 10-fold higher risk of microvascular complications and nearly twofold additional risk of macrovascular disease [2]. Another study conducted in 2017 addressed susceptible, initiatory, and progressive risk factors of kidney-specific complications in diabetic patients. The authors specified target limits of HbA1c according to age, duration of diabetes, and presence of comorbidities to prevent microvascular diseases in diabetic patients [1].
Hypomagnesemia is considered the most frequent disorder associated with diabetes mellitus due to various causes. Several studies have identified that magnesium deficiency is related to increased probability of developing T2DM and progression of DN. Experiments even evidenced that higher magnesium intake is associated with lower fasting measures of insulin for both diabetic and non-diabetic populations [17-20]. The univariate analysis of the present study evidenced that hypomagnesemia is significantly associated with higher odds of albuminuria. Serum magnesium was inversely associated with albuminuria in Models I-III, but the association was attenuated and no longer statistically significant after demographic adjustment in Model IV (p=0.060).
Integrated with CGM-based metrics and magnesium level, the present analysis also investigated the relationship of thyroid hormones with albuminuria. FT3 showed a significant association with lower odds of albuminuria in univariate and multivariate analyses even after adjustment of biochemical, medication-related, disease-related, and demographic factors. The literature reported a higher prevalence rate of abnormal thyroid hormones in diabetic patients compared to non-diabetic individuals [21]. Additionally, thyroid hormones are reported as a significant determinant of DKD [22]. Consistent with the present study, several studies conducted in different countries with diverse populations evidenced that lower FT3 levels are associated with the progression of DKD [23-25]. The present analysis found no significant association of FT4 and TSH with the odds of albuminuria or macroalbuminuria. Several studies investigated the relationship between FT4 and TSH, with the prevalence of albuminuria showing a non-significant effect of these thyroid hormones [4,26]. However, some studies showed a significant association of TSH with a higher risk of albuminuria regardless of gender [27]. Consistent with past studies, the present analysis observed no significant association of HbA1c, mean of glucose, SD of glucose, CV of glucose, MAGE, and serum magnesium with macroalbuminuria [4].
Exploratory mediation analysis identified statistically significant indirect associations through FT3 for HbA1c and glucose SD. FT3 statistically accounted for 6.9% of the HbA1c-albuminuria association and 21.4% of the glucose SD-albuminuria association. No significant indirect associations through FT4 or TSH were observed. Given the cross-sectional design, these findings should be interpreted as exploratory statistical mediation rather than evidence of a temporal or causal pathway.
In 2026, Luo et al. reported non-significant statistical mediation through FT3, FT4, and TSH in the relationships between several glycemic indices and albuminuria [4]. In contrast, the present study identified modest statistically significant indirect associations through FT3 for HbA1c and glucose SD, whereas FT4 and TSH showed no significant indirect associations. Evidence evaluating thyroid hormones within these glycemic-albuminuria relationships remains limited [4,28-30].
Some limitations should be acknowledged about this research study. First, the cross-sectional design does not allow determination of temporal relationships between albuminuria, thyroid function, and glycemic control. The mediation findings should therefore be interpreted as exploratory statistical associations rather than evidence of a temporal or causal pathway. Second, the generalizability of the findings may be limited by the single-center design. Third, residual confounding from unmeasured variables, such as medication dose or duration, lifestyle characteristics, and dietary factors, cannot be excluded despite adjustment for measured covariates. UACR was used to classify albuminuria status at the study assessment and was not intended to establish longitudinal persistence of albuminuria. The macroalbuminuria subgroup was relatively small (n=24), and findings from this subgroup should therefore be considered exploratory. Prospective multicenter studies including larger and more diverse populations are required to clarify the temporal relationships and clinical significance of these findings.
Conclusions
Greater glycemic variability and lower FT3 levels were independently associated with albuminuria in patients with T2DM, while hypomagnesemia was associated with albuminuria in univariable analysis. Exploratory mediation analysis suggested that FT3 statistically accounted for a modest proportion of the associations of HbA1c and glucose SD with albuminuria. These findings indicate that CGM-derived variability measures, alongside magnesium status and thyroid function, may provide complementary information in the assessment of DKD. Prospective multicenter studies are required to clarify the temporal relationships and clinical significance of these findings.
Acknowledgments
All authors contributed equally to the work and should be considered joint-first authors.
Appendices
Figure 1. Research proforma.
Disclosures
Human subjects: Informed consent for treatment and open access publication was obtained or waived by all participants in this study. Institutional Review Board (IRB) of Sheikh Zayed Medical College/Hospital issued approval 275/IRB/SZMC/SZH.
Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.
Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following:
Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work.
Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work.
Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.
Author Contributions
Concept and design: Shoaib Asghar, Kashif Shaikh, Sameera Aftab, Ahmed Nisar, Abdul Wahab, Rabia Qadeer
Acquisition, analysis, or interpretation of data: Shoaib Asghar, Kashif Shaikh, Sameera Aftab, Ahmed Nisar, Abdul Wahab, Rabia Qadeer
Drafting of the manuscript: Shoaib Asghar, Kashif Shaikh, Sameera Aftab, Ahmed Nisar, Abdul Wahab, Rabia Qadeer
Critical review of the manuscript for important intellectual content: Shoaib Asghar, Kashif Shaikh, Sameera Aftab, Ahmed Nisar, Abdul Wahab, Rabia Qadeer
Supervision: Shoaib Asghar
References
- 1.Diabetic kidney disease: challenges, progress, and possibilities. Alicic RZ, Rooney MT, Tuttle KR. Clin J Am Soc Nephrol. 2017;12:2032–2045. doi: 10.2215/CJN.11491116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Diabetic nephropathy: challenges in pathogenesis, diagnosis, and treatment. Samsu N. Biomed Res Int. 2021;2021:1497449. doi: 10.1155/2021/1497449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Clinical targets for continuous glucose monitoring data interpretation: recommendations from the International Consensus on Time in Range. Battelino T, Danne T, Bergenstal RM, et al. Diabetes Care. 2019;42:1593–1603. doi: 10.2337/dci19-0028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Correlations among glycemia and glucose variability, thyroid hormones, and albuminuria in patients with type 2 diabetes mellitus. Luo N, Feng SY, Chen H, et al. Kaohsiung J Med Sci. 2026:0. doi: 10.1002/kjm2.70221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hypomagnesemia in patients with type 2 diabetes. Pham PC, Pham PM, Pham SV, Miller JM, Pham PT. Clin J Am Soc Nephrol. 2007;2:366–373. doi: 10.2215/CJN.02960906. [DOI] [PubMed] [Google Scholar]
- 6.Hypomagnesemia in type 2 diabetes mellitus. Dasgupta A, Sarma D, Saikia UK. Indian J Endocrinol Metab. 2012;16:1000–1003. doi: 10.4103/2230-8210.103020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Magnesium and type 2 diabetes. Barbagallo M, Dominguez LJ. World J Diabetes. 2015;6:1152–1157. doi: 10.4239/wjd.v6.i10.1152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Subclinical hypothyroidism is a risk factor for nephropathy and cardiovascular diseases in type 2 diabetic patients. Chen HS, Wu TE, Jap TS, Lu RA, Wang ML, Chen RL, Lin HD. Diabet Med. 2007;24:1336–1344. doi: 10.1111/j.1464-5491.2007.02270.x. [DOI] [PubMed] [Google Scholar]
- 9.Prevalence of microalbuminuria in type 2 diabetes mellitus. Sana MA, Chaudhry M, Malik A, Iqbal N, Zakiuddin A, Abdullah M. Cureus. 2020;12:0. doi: 10.7759/cureus.12318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Optimal sampling duration for continuous glucose monitoring to determine long-term glycemic control. Riddlesworth TD, Beck RW, Gal RL, Connor CG, Bergenstal RM, Lee S, Willi SM. Diabetes Technol Ther. 2018;20:314–316. doi: 10.1089/dia.2017.0455. [DOI] [PubMed] [Google Scholar]
- 11.Associations between continuous glucose monitoring-derived metrics and diabetic retinopathy and albuminuria in patients with type 2 diabetes. Wakasugi S, Mita T, Katakami N, et al. BMJ Open Diabetes Res Care. 2021;9:0. doi: 10.1136/bmjdrc-2020-001923. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Continuous glucose monitoring versus usual care in patients with type 2 diabetes receiving multiple daily insulin injections: a randomized trial. Beck RW, Riddlesworth TD, Ruedy K, et al. Ann Intern Med. 2017;167:365–374. doi: 10.7326/M16-2855. [DOI] [PubMed] [Google Scholar]
- 13.Continuous glucose monitoring: a consensus conference of the American Association of Clinical Endocrinologists and American College of Endocrinology. Fonseca VA, Grunberger G, Anhalt H, et al. Endocr Pract. 2016;22:1008–1021. doi: 10.4158/EP161392.CS. [DOI] [PubMed] [Google Scholar]
- 14.International consensus on use of continuous glucose monitoring. Danne T, Nimri R, Battelino T, et al. Diabetes Care. 2017;40:1631–1640. doi: 10.2337/dc17-1600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. Stratton IM, Adler AI, Neil HA, et al. BMJ. 2000;321:405–412. doi: 10.1136/bmj.321.7258.405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.The relationship of glycemic exposure (HbA1c) to the risk of development and progression of retinopathy in the diabetes control and complications trial. https://pubmed.ncbi.nlm.nih.gov/7622004/ Diabetes. 1995;44:968–983. [PubMed] [Google Scholar]
- 17.Diabetes mellitus and magnesium [Article in Japanese] Yokota K. https://pubmed.ncbi.nlm.nih.gov/15692158/ Clin Calcium. 2005;15:203–212. [PubMed] [Google Scholar]
- 18.A potential link between magnesium intake and diabetes in Indigenous Australians. Longstreet DA, Heath DL, Vink R. Med J Aust. 2005;183:219–220. doi: 10.5694/j.1326-5377.2005.tb07007.x. [DOI] [PubMed] [Google Scholar]
- 19.Results from the Atherosclerosis Risk in Communities study suggest that low serum magnesium is associated with incident kidney disease. Tin A, Grams ME, Maruthur NM, et al. Kidney Int. 2015;87:820–827. doi: 10.1038/ki.2014.331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.The association between magnesium intake and fasting insulin concentration in healthy middle-aged women. Fung TT, Manson JE, Solomon CG, Liu S, Willett WC, Hu FB. J Am Coll Nutr. 2003;22:533–538. doi: 10.1080/07315724.2003.10719332. [DOI] [PubMed] [Google Scholar]
- 21.Prevalence of thyroid dysfunction among Greek type 2 diabetic patients attending an outpatient clinic. Papazafiropoulou A, Sotiropoulos A, Kokolaki A, Kardara M, Stamataki P, Pappas S. J Clin Med Res. 2010;2:75–78. doi: 10.4021/jocmr2010.03.281w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Thyroid dysfunction and kidney disease: an update. Iglesias P, Bajo MA, Selgas R, Díez JJ. Rev Endocr Metab Disord. 2017;18:131–144. doi: 10.1007/s11154-016-9395-7. [DOI] [PubMed] [Google Scholar]
- 23.Free triiodothyronine levels are associated with diabetic nephropathy in euthyroid patients with type 2 diabetes. Wu J, Li X, Tao Y, Wang Y, Peng Y. Int J Endocrinol. 2015;2015:204893. doi: 10.1155/2015/204893. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Thyrotropin as an independent factor of renal function and chronic kidney disease in normoglycemic euthyroid adults. Sun MT, Hsiao FC, Su SC, Pei D, Hung YJ. Endocr Res. 2012;37:110–116. doi: 10.3109/07435800.2011.640374. [DOI] [PubMed] [Google Scholar]
- 25.Thyroid parameters and kidney disorder in type 2 diabetes: results from the METAL study. Chen Y, Zhang W, Wang N, Wang Y, Wang C, Wan H, Lu Y. J Diabetes Res. 2020;2020:4798947. doi: 10.1155/2020/4798947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.The correlation between thyroid hormone levels and the kidney disease progression risk in patients with type 2 diabetes. Yang Z, Duan P, Li W, et al. Diabetes Metab Syndr Obes. 2022;15:59–67. doi: 10.2147/DMSO.S347862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Correlation of serum thyrotropin and thyroid hormone levels with diabetic kidney disease: a cross-sectional study. Gao J, Liu J. BMC Endocr Disord. 2024;24:170. doi: 10.1186/s12902-024-01699-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Thyroid dysfunction and diabetes mellitus: two closely associated disorders. Biondi B, Kahaly GJ, Robertson RP. Endocr Rev. 2019;40:789–824. doi: 10.1210/er.2018-00163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Association between thyroid hormones and diabetic kidney disease in euthyroid type 2 diabetes mellitus patients. Liu J, Jiang W, Liang J, et al. Front Endocrinol (Lausanne) 2026;17:1674977. doi: 10.3389/fendo.2026.1674977. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.The role of advanced glycation end products between thyroid function and diabetic nephropathy and metabolic disorders. Zhang Y, Wang Y, Kang Q, et al. Sci Rep. 2025;15:7202. doi: 10.1038/s41598-025-88806-7. [DOI] [PMC free article] [PubMed] [Google Scholar]