Insulin Therapy Does Not Itself Induce Weight Gain in Patients With Type 2 Diabetes

Of obese individuals who lose weight by dieting, >90% eventually return to their original weight; one- to two-thirds of the lost weight is regained within 1 year, and almost all is regained within 5 years. This suggests that body weight is physiologically controlled and that weight changes, in either direction, elicit a potent counter response that resists the change (1). This should also hold true when weight loss is attributable to high blood glucose levels and/or insulin deficiency. Insulin is often first prescribed to type 2 diabetic patients after a period of poor metabolic control by oral agents; this period may be accompanied by weight loss due to insulin deficiency and/or the poor metabolic control itself. We postulated that the weight gain observed during insulin therapy in patients with type 2 diabetes may simply correspond to re-expression of their physiologically controlled body weight. We tested the hypothesis that insulin-treated type 2 diabetic patients might return to the previous maximal weight they reached before the onset of diabetes- and insulin deficiency–induced weight loss.

We conducted a retrospective cohort study of 58 patients with type 2 diabetes who required insulin because of poor metabolic control, despite dietary measures and maximal-dose oral agents (e.g., glibenclamide 15 mg/day and metformin 1,700 mg/day). Insulin-treated type 2 diabetic patients were included in the study if they met the following criteria: 1) diagnosis of diabetes after 30 years of age (such patients were excluded if they had a family history suggestive of maturity-onset diabetes of the young, known islet autoimmunity, or diabetes secondary to endocrine or chronic pancreatic disease), 2) an interval of at least 1 year between diagnosis and insulin therapy, 3) maximal previous lifetime weight mentioned in the file, 4) at least 2 years of follow-up after the beginning of insulin therapy, 5) no use of oral antidiabetic drugs after insulin introduction, and 6) no cancer or other progressive chronic disease and a serum creatinine level <150 μmol/l.

As HbA_1c levels increased during follow-up, the best metabolic control obtained with insulin was expressed as the lowest HbA_1c value recorded during follow-up (at 6 months or 1, 2, or 3 years after insulin introduction). In the same way, the maximal daily insulin dose was the highest value recorded at 6 months or 1, 2, or 3 years.

There were 25 women and 33 men. Age at diagnosis was 52 ± 9 years and age at insulin introduction was 65 ± 9 years. BMI at diagnosis was 29.3 ± 5.6 kg/m², and the HbA_1c level at insulin introduction was 10.9 ± 1.8% (median 10.6%). A total of 11, 42, and 5 patients required 1, 2, and 3 daily insulin injections, respectively, at the end of follow-up. Follow-up after insulin introduction was 2 years in 10 patients and 3 years in 48 patients.

The HbA_1c level fell from 10.9 ± 1.8% at insulin introduction to 8.2 ± 1.8% at 6 months. It then increased at 1 year and subsequently stabilized (8.9 ± 2.1, 8.9 ± 1.5, and 8.3 ± 1.8% at 1, 2, and 3 years, respectively). The minimal HbA_1c level during follow-up was 7.5 ± 1.2%. The initial daily insulin dose was 0.6 ± 0.3 units · kg⁻¹ · day⁻¹, and requirements remained stable during follow-up (0.5 ± 0.2, 0.5 ± 0.2, 0.5 ± 0.2, and 0.6 ± 0.2 units · kg⁻¹ · day⁻¹ at 6 months and 1, 2, and 3 years, respectively). The maximal daily insulin dose reached during follow-up was 0.7 ± 0.3 units · kg⁻¹ · day⁻¹.

In 71% of patients, weight at diagnosis was below the maximal previous weight. Mean weight at diagnosis (80.0 ± 16.6 kg) was 6.1 ± 6.8 kg below the previous maximal weight (86.0 ± 17.0 kg) (t = −6.8, P < 10⁻⁴). All of the patients lost weight between diagnosis and insulin introduction. Weight at insulin introduction (73.8 ± 13.5 kg) was 6.2 ± 9.7 kg below weight at diagnosis (t = −4.8, P = 10⁻⁴) (Fig. 1).

Patients gained 8.0 ± 5.3 kg during the first 2 years of insulin therapy (t = 11.1, P < 10⁻⁶) as follows: 5.0 ± 4.6 kg during the first 6 months, 1.3 ± 3.1 kg between 6 months and 1 year, and 1.3 ± 3.0 kg during the second year. Weight stabilized during the third year (+0.15 ± 3.1 kg, t = 0.3, P = 0.74). Maximal weight during insulin therapy remained 2.6 ± 7.8 kg below maximal weight before insulin therapy (t = 2.6, P = 0.012) (Fig. 1) but was higher than weight at diagnosis (t = −2.94, P = 0.005). Maximal weight during insulin therapy correlated strongly with maximal weight before insulin: maximal weight during insulin = 0.96 × maximal weight before (95% CI 0.94–0.98, t = 13, P < 10⁻⁶, adjusted R² 99.2%). Weight gain correlated with the maximal daily insulin dose (Student’s t test = 2.86, P = 0.03, adjusted R² = 6.6%) but not with metabolic control in terms of the minimal HbA_1c level (Student’s t test = 1.62, P > 0.10, adjusted R² = 4.4%).

In this cohort, weight loss had already started at the time of diagnosis and continued until insulin was introduced. The weight patients reached after the introduction of insulin was highly correlated with their maximal weight before diabetes. Onset of weight loss before the diagnosis of diabetes has rarely been observed, even in obese type 2 diabetic patients and even though type 2 diabetes may remain undiagnosed for as long as 9–12 years (2). Long-term studies of insulin-treated patients with type 2 diabetes suggest that the weight such patients reach is asymptotic, and that most weight gain occurs during the first 3 years (3,4,5). The results of the University Group Diabetes Program (6) were unusual because the patients treated with insulin did not gain weight. In that study, the patients in the placebo group lost weight.

Our results need to be confirmed in a prospective study of early insulin introduction, i.e., before onset of weight loss, in patients with type 2 diabetes. Unfortunately, such a study may prove difficult, as half the weight loss in our population occurred before the diagnosis of diabetes. Our results have important practical implications, as they suggest that maximal previous weight may be predictive of the degree of subsequent weight gain.