Insulin edema was first described by Aaron Leifer in 1928 . Although minor insulin edema might be underrecognized, the incidence of notable or severe insulin edema seems to be a rare event . Several clinical features have been described as risk factors for developing insulin edema including poor glycemic control, new-onset diabetes, type 1 diabetes, low body weight, poor nutritional status, and higher doses of insulin therapy [1, 5]. Many reported cases of insulin edema have been in substantially malnourished or underweight individuals [3, 5, 6]. Our patient has type 1 diabetes, many years of poor glycemic control, and underwent recent intensification of insulin therapy resulting in relatively high daily insulin doses. However, unlike many reported cases, our patient had baseline obesity and presumably adequate nutritional status before developing edema. This bears some resemblance to a case described in 2015 in which the patient had obesity, was not insulin naïve, and developed insulin edema a few days after commencing insulin pump therapy, which resulted in a dramatic and abrupt improvement of glycemic control . We also recently reported insulin edema in an adult with type 2 diabetes using automated basal insulin delivery of U-500 insulin (total daily insulin delivery of > 500 units daily; A1c 8.3%) . In contrast, the patient in this report has type 1 diabetes, had significantly lower insulin requirements and achieved better glycemic control. The 2 cases have in common use of a concentrated insulin in their pump and experienced a rapid significant improvement in glycemic control.
As noted above, insulin edema is a diagnosed of exclusion. In our patient the most common causes of edema, congestive heart failure and hypoalbuminunemia, were excluded leading to our diagnosis of insulin edema. The pathogenesis of insulin edema is not known. The most commonly proposed mechanisms relevant to our patient are the antinatriuretic effect of insulin and increased capillary permeability associated with chronic hyperglycemia [1, 2]. Additional proposed mechanisms relate to fluid resuscitation (during treatment of severe hyperglycemia or diabetic ketoacidosis) and increased glucocorticoid production due to insulin-induced hypoglycemia. Neither of these would apply to our patient; she did not undergo any aggressive fluid treatment and the use of CGM allowed us to confirm that hypoglycemia was not a contributing factor.
Most cases of insulin edema have been relatively benign and self-limited. Even uncomplicated edema is an undesirable occurrence and in one case report, recurrent insulin-induced edema seems to have contributed to nonadherence to insulin regimen in a young patient with cystic fibrosis-related diabetes . However, cases have been reported in which insulin edema precipitated more serious events such as pulmonary edema, congestive heart failure, and acute renal failure [10,11,12,13] even in the absence of predisposing medical conditions.
There is no clear standard for treatment of insulin edema. Our patient’s condition was treated with diuretic therapy and appears to have been responsive, although the literature suggests that the edema would likely have responded to dietary sodium restriction and might also have simply resolved on its own over time . Ephedrine has been used with success in severe recurrent insulin edema .
In this case, insulin delivery increased significantly after initiation of hybrid closed-loop therapy with an approximately 75% increase in total daily insulin dose. Although basal insulin delivery often decreases with use of hybrid closed-loop systems compared to traditional insulin pump therapy , in the presence of elevated HbA1c, initiation of these systems can result in higher basal insulin delivery and higher insulin delivery overall. It would be expected that adults with poor glycemic control, who are underinsulinized will experience an increase in overall dose with hybrid closed-loop systems.
The weight gain experienced by this woman remained even after resolution of edema. Weight gain is a common side effect of initiation or intensification of insulin treatment and although the exact mechanism is not known, both fat accumulation and fluid retention are thought to be involved.
In our case, the use of concentrated insulin U-200 lispro in the insulin pump is off-label. We do not believe this contributed substantially to the development of her insulin edema. First, the patient had already been using U-200 lispro for several years in her insulin pump prior to transitioning to the hybrid closed-loop system and second, U-200 lispro has been determined to be bioequivalent with pharmacokinetics similar to rapid-acting U-100 insulins [17, 18]. Presumably, her insulin requirements would be the same with either concentration of insulin, although it is possible that higher doses were reached more rapidly using the U-200 formulation. The use of concentrated insulins in hybrid closed loop systems warrants further study.
To our knowledge, this is the first report of insulin edema in a person with type 1 diabetes using CGM and a hybrid closed-loop system. This allows a closer inspection of the glycemic variables present with the development of this rare complication and makes clear that hypoglycemia was not a factor in her development of insulin edema. This case also points to a potential concern about initiation of these systems in patients with poor glycemic control at baseline. Overly-rapid improvement in glycemic control has long been known to increase the risk of decompensating diabetic retinopathy. This rare case of insulin edema reminds us of the potential risks of rapid reductions in HbA1c that can be associated with the use of these remarkable new technologies. Consideration might be given to adjusting blood glucose targets when initiating hybrid closed-loop insulin pump therapy in people with baseline poor glycemic control. Given the increasing use of automated insulin delivery systems in type 1 diabetes, future studies examining the best approaches for initiating this therapy in people with high insulin requirements should be considered.