Results from the Effects of MEtformin on cardiovasculaR function in AdoLescents with type 1 Diabetes (EMERALD) Study: A Brief Report of Kidney and Inflammatory Outcomes

Abstract

Youth with type 1 diabetes (T1D) demonstrate insulin resistance (IR), independent of glycemia, when compared to normoglycemic peers. IR increases the risk for cardiovascular disease (CVD) and diabetic kidney disease (DKD), factors also associated with systemic inflammation. We evaluated the effect of metformin on markers of inflammation and DKD in youth with T1D. EMERALD, a double-blind, randomized, placebo-controlled trial of 3 months of metformin in 48 youth aged 12-21 years with T1D, included baseline and follow-up assessments of serum creatinine and cystatin C to estimate glomerular filtration rate (eGFR), AST, ALT, highly sensitive c-reactive protein, white blood count, platelets, adiponectin, leptin, and urine albumin:creatinine ratio (UACR). Metformin was associated with a 13.9 mL/min/1.73m² (95% CI 4.7-23.1 mL/min/1.73m²) increase in eGFR by serum creatinine vs. placebo (p≤0.01), with a significant difference remaining after multivariable adjustments (p=0.03). Whereas eGFR measured by serum creatinine increased significantly following metformin, no differences were observed with cystatin C, UACR, or systemic inflammatory markers. Additional studies with directly measured GFR in response to metformin in T1D are needed.

Introduction:

Youth with type 1 diabetes (T1D) demonstrate insulin resistance (IR), independent of glycemia, relative to normoglycemic peers ¹. IR increases cardiovascular disease (CVD) and diabetic kidney disease (DKD) risk ^{1, 2}, the leading causes of morbidity and mortality in the T1D population. DKD is also associated with increased local and systemic inflammation, factors that are inversely related to glomerular filtration rate (GFR) ³. We have previously demonstrated that metformin improved IR, body composition, and markers of cardiovascular health in youth with T1D ⁴. Human, animal, and cellular models of type 2 diabetes have also shown improvements in systemic inflammation with metformin ⁵. Additionally, the “Cardiovascular and Metabolic Effects of Metformin in Patients with Type 1 Diabetes (REMOVAL)” study, a double-blind, randomized, placebo-controlled trial of a 3-year course of metformin in 428 adults with T1D and multiple cardiovascular risk factors demonstrated that metformin may have a kidney protective effect vs. a direct effect on creatinine metabolism as estimated GFR (eGFR) by serum creatinine was better maintained over time in the metformin vs. placebo group (p<0.001) ⁶. Yet, similar investigations of the effects of metformin on GFR in youth with T1D are still needed. We hypothesized that metformin would improve inflammation and kidney function in youth with T1D and present a double-blind, randomized, placebo-controlled trial assessing the effects of 3 months of metformin on markers of inflammation and DKD in youth with T1D.

Methods:

Our study was approved by the Colorado Multiple Institutional Review Board (COMIRB) and monitored by an independent Safety Officer. Participants and their guardians provided informed consent and/or assent, as appropriate.

Participants:

Forty-eight adolescents, ages 12-21 years, with T1D from the Effects of MEtformin on cardiovasculaR function in AdoLescents with t1D (EMERALD, ClinicalTrials.gov Identifier: NCT01808690) were included. Study cohort details, including exclusions and study retention, were previously published ⁴. Inclusion criteria included a known diagnosis of T1D by the American Diabetes Association (ADA) criteria plus persistent insulin requirement since diagnosis, the presence of ≥ 1 diabetes-associated autoantibody, diabetes duration ≥ 1 year, Tanner stage >1, and sedentary status (<3 hours of self-reported physical activity per week). A board-certified pediatric endocrinologist completed pubertal staging assessments per the standards of Tanner and Marshall for breast development (girls) and pubic hair (boys and girls). Testicular volumes were also documented (boys). Exclusion criteria included a hemoglobin A1c (HbA1c) >108 mmol/mol, weight >136 kg, BMI <5^th percentile, resting blood pressure (BP) >140/90 mmHg, serum creatinine >0.13 mmol/L, hemoglobin <5.6 mmol/L, smoking, pregnancy, breast feeding, implanted metal, medications impacting insulin sensitivity or blood pressure (i.e. glucocorticoids, non-insulin antihyperglycemic agents, immunosuppressants, and atypical antipsychotics), and severe hypoglycemia, illness, or DKA in the preceding 60 days. Visits were preceded by 3 days without caffeine or strenuous physical activity and participants ate a pre-prepared, fixed macronutrient (i.e. 55% calories from carbohydrates, 30% from fat, 15% from protein) and weight-based diet. Assessments were performed in the morning after a 10 hour fast. Visits were rescheduled for hyperglycemia with large ketones or significant hypoglycemia in the past 24 hours. Blood glucose concentrations the morning of the visit were treated as needed with an insulin bolus if they were above target. No participants reported use of concurrent blood pressure modifying agents during the study. One participant remained on a stable dosage lipid-modifying regimen (Gemfibrozil and Simvastatin) throughout the study.

After an initial screening visit, participants underwent baseline laboratory studies to estimate GFR and evaluate systemic inflammation (see methods below). Participants were then randomized to 3 months of 1000 mg metformin BID or an identical-appearing placebo, after which all study outcomes were repeated.

Laboratory studies:

The following serum and urine studies were drawn fasting in the morning after an overnight insulin drip to normalize glycemia and sent for routine evaluation in the quality-assured Clinical Translational Research Center (CTRC) Core Laboratory: serum creatinine, adiponectin, leptin, highly sensitive c-reactive protein (hs-CRP), aspartate aminotransferase (AST), alanine aminotransferase (ALT), white blood count (WBC), and platelets and urine creatinine. HbA1c was assessed via DCCT-calibrated ion-exchange high-performance liquid chromatography (inter-assay coefficient of variation [CV] 0.81%, Bio-Rad Laboratories, Hercules, CA). Urinary microalbumin for urine albumin to creatinine ratio (UACR) was measured by immunoturbimetric assay (inter-assay CV 1.9-2.9%, Ortho Clinical Diagnostics, Raritan, NJ) in the Children’s Hospital Colorado laboratory. Cystatin C was batched and run at study completion on stored samples via an immunoturbidimetric assay (inter-assay CV 6.5%, Kamiya Biomedical, Seattle, WA) such that the duration of storage varied based on time of enrollment in the study but did not vary by treatment group due to the randomization protocol. Serum creatinine and urine creatinine were measured by the kinetic alkaline picrate method (inter-assay CV 1.3% at 0.22 mmol/L, 1.2% at 0.52 mmol/L, and 1.6% at 1.11 mmol/L and 0.9% at 0.22 mmol/L, 2.9% at 0.35 mmol/L, and 3.7% at 1.06 mmol/L, respectively, Thermo Fisher Scientific, Waltham, MA). Adiponectin was measured via radioimmunoassay (RIA) (inter-assay CV 8.5%, Millipore, Darmstadt, Germany), leptin via RIA (inter-assay CV 5.8%, Millipore, Darmstadt, Germany), and hsCRP via immunorubidimetric assay (inter-assay CV 1.83%, Beckman Coulter, Loveland, CO). All other laboratory evaluations including AST, ALT, WBC, and platelets were performed by standard methods in the CTRC laboratory.

eGFR (mL/min/1.73m²) was evaluated by the serum creatinine, cystatin C, and combined serum creatinine-cystatin C full age spectrum (FAS) equations, eGFR (FAS_Scr) = 107.3/(S_cr/Q_cr), eGFR (FAS_cysC) = 107.3/(S_cysC/Q_CysC), and eGFR (FAS_combined) = 107.3/[α×(S_crQ_cr)+(1–α)×(S_cysC/Q_cysC)] respectively, equations that are validated in both pediatrics and adults, and the Zappitelli equation (eGFR = (507.76×e^{(0.003×height)})/(serum cystatin C^0.635×serum creatinine^0.547). For the FAS equations, it has been previously described that Q_cr represents the median serum creatinine (S_cr) in a healthy population to account for age and sex ⁷. Q_cysC is 0.82 mg/L for all individuals <70 years old. The coefficient α is a weighting factor for the normalized renal biomarkers and α=0.5 was used here (i.e. the denominator = the average of both normalized biomarkers) ⁸. Hyperfiltration was defined conservatively a priori as an eGFR ≥126.8 mL/min/1.73m² per the Zappitelli equation, the 95^th percentile for healthy adolescents in the National Health and Nutrition Examination Survey (NHANES) ⁹.

Statistical Analysis:

Variable distributions were examined prior to analysis. Baseline characteristics were compared between the groups using t-tests or the Mann-Whitney test for continuous variables, and the chi-square or Fisher’s exact test for categorical variables. Changes in study outcomes were calculated as the value at follow-up minus baseline; and changes were compared between the groups using t-tests or Mann-Whitney tests. Linear models compared differences in changes in outcomes while adjusting for baseline values, BMI, systolic blood pressure, and insulin sensitivity (glucose infusion rate/insulin concentration at steady state per lean kg).

Results:

Baseline demographic data are summarized in Table 1. The placebo and metformin groups were similar in age, sex, race/ethnicity, Tanner stage, BMI, diabetes duration, Q_cr, and mean eGFR by FAS-Cr, FAS-Cr and Cystatin C, and Zapitelli-Cr and Cystatin C equations at baseline. Baseline eGFR by FAS-Cystatin C differed significantly between the metformin and placebo groups, placebo: 103.8±17.7 mL/min/1.73m² vs. metformin: 115.6±15.7 mL/min/1.73m² (p=0.02). Changes in laboratory data [final – baseline] over the 3-month treatment period in the metformin vs. placebo treatment groups appear in Table 2. Metformin increased eGFR by serum creatinine by 13.9 mL/min/1.73m² (95% CI 4.7-23.1 mL/min/1.73m²) vs. placebo (p≤0.01) and remained significant after multivariable adjustment for baseline eGFR values, BMI, systolic blood pressure, and insulin sensitivity at 12.14 mL/min/1.73m² (95% CI 1.42-22.87 mL/min/1.73m² (p=0.03) [Supplemental Table 1] without a significant difference in Q_cr between groups. Change in serum creatinine alone, cystatin C, UACR, hs-CRP, WBC, platelets, AST, ALT, adiponectin, and leptin did not differ between groups (Table 2). Group differences (metformin – placebo) in all reported outcomes other than eGFR by serum creatinine were nonsignificant (Supplemental Table 1).

Table 1:

Baseline clinical characteristics of adolescents with type 1 diabetes receiving treatment with either placebo or metformin.

	Placebo	Metformin	p-value
Number	23	25	N/A
Age (years)	16.1 ± 2.7	17.3 ± 2.3	0.06
Sex, N (% female)	11 (48)	14 (56)	0.67
Ethnicity, N (%)
Non-Hispanic White	20 (87)	22 (88)	0.90
Hispanic	2 (9)	2 (8)
Non-Hispanic Black	1 (4)	0 (0)
Other	0 (0)	1 (4)
Tanner Stage
2	1 (4)	1 (4)	0.41
3	2 (9)	0 (0)
4	3 (13)	4 (16)
5	17 (74)	20 (80)
BMI Percentile (%ile)	74.8 ± 23.6	75.1 ± 25.1	0.91
Diabetes Duration (years)	7.7 ± 4.4	8.0 ± 3.7	0.87
HbA1c (mmol/mol)	69 ± 14	72 ± 16	0.62
Serum Creatinine (mmol/L)	0.06 ± 0.02	0.06 ± 0.01	0.82
Qcr	0.67 ± 0.10	0.72 ± 0.09	0.11
Mean eGFR by FAS-Cr (mL/min/1.73m2)	103.2 ± 16.5	112.5 ± 15.4	0.06
Mean eGFR by FAS-Cystatin C (mL/min/1.73m2)	103.8 ± 17.7	115.6 ± 15.7	0.02
Mean eGFR by FAS-Cr and Cystatin C (mL/min/1.73m2)	104.0 ± 16.6	112.5 ± 15.4	0.09
Mean eGFR by Zapitelli-Cr and Cystatin C (mL/min/1.73m2)	99.0 ± 18.2	105.3 ± 17.7	0.25

Data are expressed as mean ± SD unless otherwise specified.

Abbreviations: BMI: body mass index; HbA1c: hemoglobin A1c; Q_cr: median serum creatinine in a healthy population accounting for age and sex; eGFR: estimated glomerular filtration rate; FAS: Full-Age Spectrum; Cr: serum creatinine.

Table 2:

Comparison of changes in laboratory measures over time in the metformin and placebo treatment groups.

Laboratory Measures	Placebo (n=23)	Metformin (n=25)	p-value
Δ AST (IU/L)	−4.55 ± 9.22	−6.92 ± 25.78	0.67
Δ ALT (IU/L)	−0.45 ± 8.36	1.21 ± 8.02	0.50
Δ cystatin C (mg/L)	−0.01 ± 0.09	0.02 ± 0.06	0.17
Δ hs-CRP (mg/L)	1.6 (−0.8, 18.1)	−0.5 (−6.1, 2.7)	0.18
Δ WBC (109 cells/L)	0.37 ± 1.89	0.02 ± 0.92	0.44
Δ platelets (109 cells/L)	1.57 ± 83.32	21.46 ± 36.00	0.32
Δ adiponectin (ng/mL)	1.04 ± 3.91	1.40 ± 5.26	0.80
Δ leptin (ng/mL)	1.11 ± 6.99	−2.75 ± 11.18	0.17
Δ serum creatinine (mmol/L)	0.001 ± 0.010	−0.004 ± 0.006	0.06
Δ UACR (mg/mmol)	−4.65 ± 16.77	−2.61 ± 14.15	0.66
Δ Qcr	0 (0,0)	0 (0,0)	0.33
Δ eGFR (FAS-Cr) (mL/min/1.73m2)	−2.09 ± 15.03	11.21 ± 13.15	<0.01
Δ eGFR by FAS-Cystatin C (mL/min/1.73m2)	1.75 ± 9.24	−2.88 ± 8.69	0.11
Δ eGFR by FAS-Cr and Cystatin C (mL/min/1.73m2)	9.60 ± 14.51	17.72 ± 18.56	0.12
Δ eGFR by Zappitelli-Cr and Cystatin C (mL/min/1.73m2)	−0.30 ± 11.12	3.89 ± 7.76	0.19

Data are expressed as mean ± SD or median (25^th %ile, 75^th %ile).

Abbreviations: Δ: change in variable over time (i.e. final – baseline value); AST: aspartate aminotransferase; ALT: alanine aminotransferase; hs-CRP: highly sensitive c-reactive protein; WBC: white blood count; UACR: urine albumin-to-creatinine ratio; Q_cr: median serum creatinine in a healthy population accounting for age and sex; eGFR: estimated glomerular filtration rate; FAS: Full-Age Spectrum; Cr: serum creatinine.

Discussion:

In this double-blind, randomized, placebo-controlled study of 3 months of metformin in youth with T1D, we demonstrate a significant increase in eGFR by serum creatinine, but no change in eGFR by cystatin C, UACR, or the studied markers of systemic inflammation. Our pediatric cohort eGFR findings are consistent with those of the adult REMOVAL study, a double-blind, randomized, placebo-controlled trial of a 3-year course of metformin in adults with T1D and multiple cardiovascular risk factors which found that metformin was associated with a 4.0 mL/min/1.73m² higher eGFR by serum creatinine than placebo (2.19-5.82 mL/min/1.73m², p<0.0001) ⁶.

In DKD, GFR decline is a clear marker of nephropathy; yet, in adolescents and young adults there are frequently initial elevations in GFR, a phenomenon known as hyperfiltration. Hyperfiltration occurs secondary to increased intraglomerular pressure and renal plasma flow and may lead to progressive nephron injury and albuminuria in the setting of diabetes ^10-12. Studies evaluating whole-kidney hyperfiltration have demonstrated a prevalence of 10-67% in the T1D population ¹³, and, for many, changes in GFR begin shortly after diagnosis ¹⁴. While metformin has classically been contraindicated in severe chronic kidney disease (CKD) (i.e. GFR <30 mL/min) due a possible risk for lactic acidosis ¹⁵, its effect on mild to moderate CKD, and the associated pro-inflammatory state, are of continued interest. The longer-term clinical implications of the increased eGFR observed with metformin in this study are unclear, as increased eGFR could potentially result in earlier clinical decline, which could be outweighed by metformin’s known positive effects on BMI, insulin sensitivity, and vascular function, or could be protective in youth with T1D. Outcomes could also differ based on the population being studied or the starting GFR. Furthermore, the evidence of a significant change in eGFR by FAS-Cr alone but not by FAS-Cystatin C, FAS-Cr and Cystatin C or Zapitelli-Cr and Cystatin C also presents the question of whether metformin is directly impacting GFR or creatinine metabolism, although it is notable that changes in serum creatinine over time were not significant between groups. If metformin does, in fact, alter serum creatinine concentrations, this could mask a decrease in true GFR, an important clinical distinction given the high incidence of metformin usage in the diabetes population and frequent utilization of serum creatinine to estimate GFR. These remaining questions support the need for future longer-term studies of the impact of metformin on GFR in youth and adults with T1D.

Our study has several notable limitations including the markers of kidney function and inflammation being secondary outcomes, the use of calculated rather than measured GFR, and a relatively short treatment duration of 3 months. Additionally, the difference in baseline serum creatinine concentrations approached statistical significance between the metformin and placebo groups (p=0.06) which could theoretically confound a baseline analysis of eGFR by FAS-Cr. However, our study benefitted from a longitudinal study design wherein each individual served as their own control and baseline values were controlled for, minimizing the potential impact of baseline differences, and the change with metformin actually moved away from potential regression to the mean. Lastly, the cystatin C assay was run on stored samples with varied length of storage which could have caused the cystatin C assay to be less accurate than serum creatinine. Limitations presented here are also lessened by the concordance of our findings with the REMOVAL study which demonstrated persistent elevations in GFR after a 3-year treatment duration of metformin therapy in adults with type 1 diabetes ⁶. Consequently, our results argue for future studies to analyze the long-term effects of metformin on GFR and diabetic kidney disease progression in youth with T1D.

In conclusion, whereas eGFR by serum creatinine increased significantly following 3 months of metformin, no differences were observed with cystatin C-based equations, UACR, or markers of systemic inflammation. Additional studies with gold standard GFR measurements and longer-term follow up are now needed to further evaluate the effects of metformin on kidney function and inflammatory outcomes in T1D, particularly in youth.