Liraglutide Compromises Pancreatic β Cell Function in a Humanized Mouse Model

Abstract

Incretin mimetics are frequently used in the treatment of type 2 diabetes because they potentiate β cell response to glucose. Clinical evidence showing short-term benefits of such therapeutics (e.g., liraglutide) is abundant; however, there have been several recent reports of unexpected complications in association with incretin mimetic therapy. Importantly, clinical evidence on the potential effects of such agents on the β cell and islet function during long-term, multiyear use remains lacking. We now show that prolonged daily liraglutide treatment of >200 days in humanized mice, transplanted with human pancreatic islets in the anterior chamber of the eye, is associated with compromised release of human insulin and deranged overall glucose homeostasis. These findings raise concern about the chronic potentiation of β cell function through incretin mimetic therapy in diabetes.

Fig. 1. Humanized mouse model to study pancreatic islet function in vivo

Human islets transplanted into the anterior chamber of the eye of streptozotocin (STZ)-induced diabetic nude mice engraft on top of the iris and restore normoglycemia. (a) Representative fluorescence micrograph of a cross-section of a recipient nude mouse eye showing (left) the transplanted human islets on top of the iris and (right) a typical distribution of alpha (glucagon; red) and beta (insulin; green) cells (inset) in the intraocular human islet grafts (blue: DAPI nuclear counterstain; scale bar = 200μm). (b) Ratio of alpha and beta cells in immunostained cross-sections of the intraocular human islets. Ratios (shown as means ± SD) were acquired by dividing the total glucagon or insulin-positive area (including nuclei) by the total area of the corresponding islet(s) in immunostained transplanted-eye sections obtained at the conclusion of the studies ≥250 days after transplantation/treatment initiation. Data based on 19 and 29 analyzed eye sections derived from ≥three liraglutide-treated and saline-treated recipients, respectively. (c) Nonfasting glycemia before and ~35 days after human islet intraocular transplantation in diabetic nude mice treated with liraglutide (300 μg/kg/day s.c.) (Merani et al., 2008) or saline (control). Daily liraglutide or saline treatments were started two days prior to transplantation (n = 14 mice/treatment; data shown as means ± SEM). (d) Kaplan-Meyer curves showing % of normoglycemic animals following human islet transplantation in recipients treated with liraglutide or saline (n = 14 mice/treatment). Median diabetes reversal time was 2 days and 17.5 days in liraglutide vs. saline-treated animals, respectively (p<0.05).

Fig. 2. Effects of long-term daily liraglutide treatment on glucose homeostasis

Longitudinal in vivo follow up in “humanized” mice revealed compromised glycemic control in association with long-term daily treatment with liraglutide. (a) Nonfasting glycemia in originally diabetic nude mice that were transplanted with human islets and treated daily with liraglutide (300 μg/kg/day s.c.) or saline (n = 6 mice/treatment) (red: liraglutide; black: saline). Either treatment was initiated two days prior to transplantation and continued for ≥250 days. The data were binned for each treatment group for indicated time points on the X axis. Green lines represent the geometric mean (Asterisks indicate significance; p=0.045 for [90–120], p=0.724 for [130 – 160], p=0.00012 for [170 – 200]). (b) Kaplan-Meyer curves showing a comprehensive record of % normoglycemic animals during the extended follow up of ≥250 days after treatment initiation (n = 17 – 18 mice/treatment). Liraglutide-driven improvement in human islet function was initially evident based on the higher number of normoglycemic mice during the first ~80 days after treatment initiation (day 0). However, islet function started to deteriorate in the liraglutide-treated mice at ~150 days of treatment as evidenced by the lower % of normoglycemic mice. The overall median “survival” (i.e., normoglycemia) time did not differ significantly between the liraglutide vs. saline-treated mice (107 vs. 132, receptively; p=0.4861). (c) Blood glucose excursion curves during high glucose challenges (intraperitoneal glucose challenge test) performed 64 ± 2, 96 ± 4, 134 ± 7, and 200 days post-treatment initiation (red: liraglutide; black: saline). The challenges were performed only in mice with normoglycemia at the time of the test. These data showed the progressive deterioration of glycemic control by the human islets in the challenged mice (n = 6 – 7 mice/treatment; data shown as means ± SEM). (d) Area under the curve (AUC; shown as means ± SD) values of the corresponding glucose challenge tests shown in 2c (Asterisks indicate significance; p=0.132 for POD64, p=0.00071 for POD96, p=0.0764 for POD134, p=0.326 for POD200). (e) Plasma levels of human insulin measured in blood samples collected from liraglutide- and saline-treated mice during two glucose challenges performed 176 ± 15 days after treatment initiation (shown as means ± SEM). (f) Plasma levels of human C-peptide measured on POD168 under fed conditions (shown as means ± SD; p=0.073). (g) Blood glucose values (shown as means ± SD) during an insulin tolerance test performed on day 243 post-treatment initiation (asterisks indicate significance; p=0.092 for T0, p=0.0347 for T2, p=0.578 for T5, p=0.391 for T10, p=0.791 for T20, p=0.597 for T30, p=0.675 for T40, p=0.9 for T50).