A low-risk intervention with potentially high rewards.
With the increasing incidence of diabetes in the pediatric population, and type 1 diabetes accounting for 90% of childhood cases, tight glycemic control with insulin remains the standard of care in this population. Not surprising is the ongoing concern over the occurrence of hypoglycemic events associated with insulin use, a risk coincident with the known benefits of insulin therapy, as demonstrated in the Diabetes Control and Complications trial presented in 2015. It is understood that with puberty in diabetes patients comes weight gain and increased insulin resistance due to elevations in circulating sex hormones, which manifests as worsening type 1 diabetes. Based on past experience in improving metabolic states in type 2 diabetes patients using metformin, many providers have attempted off-label metformin use in their adolescent type 1 patients. Metformin works to improve insulin sensitivity by amplifying activity of glucose transporters at the skeletal muscle cells, and directly lowers blood glucose by reducing hepatic production, imparting its effects without causing hypoglycemia. A previous review summarized 2 randomized controlled trials (RCT), which included 60 children and adolescents with type 1 diabetes.
There was a trend toward improvements in metabolic control when metformin was added to insulin, but due to high heterogeneity, a meta-analysis could not be performed. In the time following, new RCTs have been published, adding information to the approach of managing children with type 1 diabetes, and providing more data for review and analysis. This current meta-analysis evaluates the safety and efficacy of adding metformin to standard insulin therapy in type 1 children ages 6 to 19 years.
The primary outcome of this analysis was HbA1c as a surrogate marker for long-term complications, with secondary outcomes being BMI (kg/m2), BMIz (z-score for age and gender), total daily dose of insulin (TIDD, U/kg/d), occurrence of severe hypoglycemia requiring outside intervention from a second person, and diabetic ketoacidosis (DKA) events. Included studies were required to evaluate the effect of the addition of metformin to standard insulin therapy, with any dose and any duration, comparing to placebo or no intervention. Use of co-interventions was permitted if included in both treatment arms. Studies were parallel RCTs following patients for at least 3 months during intervention. Crossover and quasi-randomized trials were excluded.
Effect estimates and their 95% confidence intervals were presented as mean differences for continuous variables. Risk ratios (RR) were used for dichotomous variables. Data were pooled using fixed-effect model, while the random effects model was used when I2 exceeded 50% and heterogeneity could not be explained by previous sub-group analysis.
Of 736 potentially relevant citations identified, 6 studies were included in the review and meta-analysis. The population had 325 subjects with a mean age of 15 years. All studies used a placebo comparator, thus no study compared metformin to a control group with no intervention or a co-intervention. All included studies were determined to be at a low risk-of-bias (RoB) score. The comparative effect on HbA1c from metformin vs placebo was examined in all six trials, displaying no significant mean difference between the groups (MD -0.05, 95% CI -0.19 to 0.29; p=0.7) with moderate heterogeneity detected (I2 = 33%). Effects of metformin on BMI vs placebo were reported in 4 studies, with a statistically significant mean difference (MD -0.11 kg/m2, 95% CI -2.54 to -0.38; p<0.01). It was noted that BMIz was in only 2 studies, where metformin showed a reduction in score difference (SD) (MD -0.11, 95% CI -0.21 to -0.01).
Neither finding demonstrated heterogeneity. With regards to insulin requirements, metformin reduced TIDD (MD -0.15 U/kg/d, 95% CI -0.24 to -0.06; p< 0.01) in all six trials. DKA was studied in 4 trials, but seen in only 2 of them, with no significant difference seen. However, there was a trend toward developing DKA in the metformin treated group, although the absolute numbers were small (5 vs 2). Severe hypoglycemia was reported in 5 studies, without a significant difference vs placebo (RR 4.21; 95% CI 0.95-18.7; p=0.06). As with DKA, there was a trend toward metformin treatment and hypoglycemia, but again, the numbers were small (8 vs 1).
Overall, this review of a small sample size concluded that addition of metformin to insulin in type 1 children and adolescents contributed to the reduction of TIDD by 0.15 U/kg/day, BMI by 1.46 kg/m2 (but not in a statistically significant manner), and BMI z-score by 0.11 SD, but had no effect on HbA1c. The reported quality of the body of evidence was high to moderate for each outcome. A comparison to a similar analysis of adult studies showed the findings to be quite similar, with reductions in TIDD and BMI, with no reduction in HbA1c, and no increase in risk of hypoglycemia and DKA.
However, the same trend toward both hypoglycemia and DKA in metformin treatment groups was also present, suggesting the need for future larger RCTs looking at these effects. In the end, it is up to providers and their patients to determine if the modest gains achieved from metformin on insulin and weight reduction are worth the risk for hypoglycemia and DKA.
- Metformin has been shown to be a valuable tool in treating type 2 diabetes, mostly due to its effectiveness without inducing hypoglycemia.
- Systematic reviews and meta-analyses have yet to provide sufficient numbers of type 1 diabetic subjects who have metformin added to insulin therapy.
- The limited data suggest metformin added to these patients can induce modest reductions in weight and insulin requirements but not in HbA1c, and there does exist a trend toward hypoglycemia and DKA.
International Diabetes Federation. IDF Diabetes Atlas. 7th ed. Brussels, Belgium: International Diabetes Federation. 2015.
Al Khalifah RA, Alnhdi A, Alghar H, et al. The effect of adding metformin to insulin therapy for type 1 diabetes mellitus children: A systematic review and meta-analysis. Pediatr Diabetes. 2017. Epub 2017/02/01. doi: 10.1111/pedi.12493. PubMed PMID: 28145083.
Mark T. Lawrence, University of Colorado-Denver, School of Pharmacy NTPD