The Effect of Rosiglitazone on Overweight Subjects with Type 1 Diabetes

Clinical Care/Education/Nutrition ORIGINAL ARTICLE

The Effect of Rosiglitazone on Overweight Subjects With Type 1 Diabetes

SUZANNE M. STROWIG, MSN, RN the potentially serious consequences of PHILIP RASKIN, MD excessive weight gain and insulin- induced hypoglycemia, investigators continue to search for treatments that address both the treatment of insulin deﬁciency as well as other metabolic abnormalities that OBJECTIVE — To evaluate the safety and effectiveness of rosiglitazone in the treatment of are associated with diabetes (5). overweight subjects with type 1 diabetes. Insulin resistance, a metabolic abnor- mality common in type 2 diabetes, ap- RESEARCH DESIGN AND METHODS — A total of 50 adult type 1 diabetic subjects 2 pears to be present in individuals with with a baseline BMI Ն27 kg/m were randomly assigned in a double-blind fashion to take insulin and placebo (n ϭ 25) or insulin and rosiglitazone 4 mg twice daily (n ϭ 25) for a period of 8 type 1 diabetes, as well. Although insulin months. Insulin regimen and dosage were modiﬁed in all subjects to achieve near-normal resistance in type 2 diabetes is generally glycemic control. associated with obesity, hypertension, dyslipidemia, and other metabolic disor-

RESULTS — Both groups experienced a significant reduction in HbA1c (A1C) level (rosigli- ders, studies have shown that overweight tazone: 7.9 Ϯ 1.3 to 6.9 Ϯ 0.7%, P Ͻ 0.0001; placebo: 7.7 Ϯ 0.8 to 7.0 Ϯ 0.9%, P ϭ 0.002) and as well as normal-weight adults with type a significant increase in weight (rosiglitazone: 97.2 Ϯ 11.8 to 100.6 Ϯ 16.0 kg, P ϭ 0.008; 1 diabetes can have peripheral and he- Ϯ Ϯ ϭ ϭ placebo: 96.4 12.2 to 99.1 15.0, P 0.016). Baseline measures of BMI (P 0.001), total patic insulin resistance. Insulin clamp daily insulin dose (P ϭ 0.002), total cholesterol (P ϭ 0.005), HDL cholesterol (P ϭ 0.00l), and ϭ procedures performed on nonobese type LDL cholesterol (P 0.02) were predictors of improvement in A1C level only in the group 1 diabetic subjects under a variety of gly- treated with rosiglitazone. Total daily insulin dose increased in subjects taking placebo (74.0 Ϯ 33.8 to 82.0 Ϯ 48.9 units, P Ͻ 0.05 baseline vs. week 32), but it decreased slightly in subjects cemic conditions demonstrated increased taking rosiglitazone (77.5 Ϯ 28.6 to 75.3 Ϯ 33.1 units). Both systolic blood pressure (137.4 Ϯ hepatic glucose production and reduced 15.6 vs. 128.8 Ϯ 14.8 mmHg, baseline vs. week 32, P Ͻ 0.02) and diastolic blood pressure insulin clearance compared with nondia- (87.2 Ϯ 9.4 vs. 79.4 Ϯ 7.2 mmHg, P Ͻ 0.0001) improved in the group treated with rosiglita- betic subjects (6–10). Others have shown zone. The total incidence of hypoglycemia did not differ between groups. that type 1 diabetic subjects with a family history of type 2 diabetes who have un- CONCLUSIONS — Rosiglitazone in combination with insulin resulted in improved glyce- dergone intensive insulin therapy have a mic control and blood pressure without an increase in insulin requirements, compared with greater tendency toward expressing the insulin- and placebo-treated subjects, whose improved glycemic control required an 11% in- markers of insulin resistance, such as crease in insulin dose. Weight gain and hypoglycemia were similar in both groups at the end of the study. The greatest effect of rosiglitazone occurred in subjects with more pronounced mark- weight gain, increased insulin require- ers of insulin resistance. ments, and dyslipidemia (11). Insulin sensitizers such as biguanide, Diabetes Care 28:1562–1567, 2005 metformin, and the thiazolidinedione compounds troglitazone, rosiglitazone, and pioglitazone have significantly im- esearch in the past decade has con- proving glycemic control, these treat- proved blood glucose levels in type 2 di- clusively demonstrated that im- ments involve considerable effort and abetic individuals by suppressing hepatic proved blood glucose control is of cost (3) and can result in weight gain and glucose output and enhancing insulin- R mediated glucose disposal (12,13). Be- benefit in preventing the microvascular increased risk for severe and asymptom- complications of diabetes (1,2). Although atic hypoglycemia (4). Faced with the cause insulin sensitizers have been new insulin analogs and intensive insulin challenge of balancing aggressive insulin effective in improving blood glucose con- treatment regimens are effective in im- replacement therapy against the risk for trol in type 2 diabetic individuals, for ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● whom insulin resistance is a prevailing feature, it seems reasonable to consider From the University of Texas Southwestern Medical Center at Dallas, Dallas, Texas. the possibility that these agents would be Address correspondence and reprint requests to Suzanne M. Strowig, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd., Dallas, TX 75390-8858. E-mail: suzanne.strowig@ of benefit in overweight subjects with utsouthwestern.edu. type 1 diabetes, who are more likely to Received for publication 15 February 2005 and accepted in revised form 14 April 2005. have a degree of insulin resistance. Stud- A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion ies have shown that the insulin sensitizer factors for many substances. metformin significantly reduces the © 2005 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby amount of insulin required to improve marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. blood glucose levels in both adult and ad-

1562 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 Strowig and Raskin olescent subjects with type 1 diabetes ical activity throughout the study. In- one-way ANOVA assessed between- (14–17). No study, however, has investi- formed consent was obtained from all group differences, and a paired t test de- gated the potential benefit of a thiazo- subjects after approval by the university’s termined within-group differences. The lidinedione in the management of type 1 institutional review board. Mann-Whitney U test was used to analyze diabetes. Thus, this study was designed to categorical variables. A regression analy- examine the safety and efficacy of the thia- Intervention sis was performed to identify predictors zolidinedione rosiglitazone in the treat- At the randomization visit, subjects were for change in A1C level and incidence of ment of overweight adults with type 1 given either placebo or rosiglitazone 4 mg adverse events within each group. Pear- diabetes. twice daily. The insulin dose was modi- son correlations were performed for all fied as needed in all subjects in an attempt variables. RESEARCH DESIGN AND to achieve normal glycemic control. At Two-tailed tests were performed for METHODS — Subjects entered the each clinic visit, plasma glucose levels the analyses. A P value Ͻ0.05 was consid- study based on the following criteria: age were downloaded from the patient’s ered statistically significant. All analyses Ն19 years, medical history consistent meter to the computer; all patients used a were conducted using SPSS software, ver- with type 1 diabetes, HbAlc (A1C) level glucose meter that stored plasma glucose sion 12.0. Results are reported as the Ն6.5%, BMI Ն27 kg/m2, insulin dose readings as well as the date and time the means Ϯ SD, unless otherwise indicated. Ͼ35 units per day, and normal hepatic, readings were obtained. renal, and cardiac function. Subjects who The frequency of hypoglycemia was RESULTS — A total of 52 subjects met met the inclusion criteria were randomly determined by the number of plasma glu- the baseline criteria and were randomized assigned in a double-blind fashion to ros- cose readings stored in the patient’s meter to treatment. Two female subjects chose iglitazone 4 mg twice daily or placebo one that were Ͻ65 mg/dl. Mild hypoglycemia not to continue study participation after tablet twice daily. Tablets of rosiglitazone was defined as a reading between 45 and Ͻ4 weeks of treatment. One of these sub- and placebo, identical in appearance, 65 mg/dl. Moderate hypoglycemia was jects had been assigned to rosiglitazone, were dispensed by the pharmacist in the defined as the number of plasma glucose and one had been assigned to placebo. Clinical Trials Office at the University of readings Յ45 mg/dl. Severe hypoglyce- The remaining 50 subjects completed the Texas Southwestern Medical Center. The mia, as in the DCCT (Diabetes Control study. At baseline, the two groups were randomization codes were held in the and Complications Trial), was defined as comparable in age, sex, ethnicity, dura- Clinical Trials Office until the conclusion any low plasma glucose level that patients tion of diabetes, weight, BMI, waist-to- of the study. The insulin dose and/or reg- were unable to treat themselves, and the hip ratio, total daily insulin dose, imen were changed as needed in all sub- patient’s symptoms were reversed with C-peptide, and A1C level. The group ran- jects in an attempt to achieve normal oral carbohydrate, glucagon, or intrave- domized to treatment with rosiglitazone plasma glucose and A1C levels and to nous glucose. Edema was determined to had a higher baseline systolic blood pres- minimize episodes of hypoglycemia. be absent or present based on the patient’s sure (P ϭ 0.003) and total cholesterol Subjects were seen biweekly the first report of swelling combined with a phys- level (P ϭ 0.02) than the group random- month of treatment, once the second ical examination. ized to placebo (Tables 1 and 2). month of treatment, and bimonthly there- A1C improved to comparable levels after, for a total of 32 weeks of observa- Analytical determinations in both groups after 32 weeks of treat- tion. Between-visit contacts with the A1C levels were measured by high- ment (rosiglitazone: 7.9 Ϯ 1.3 to 6.9 Ϯ health care team rarely occurred, and they pressure liquid chromatography (upper 0.7%, P Ͻ 0.0001; placebo: 7.7 Ϯ 0.8 were initiated by the subjects. Complete limit for nondiabetic individuals in this to 7.0 Ϯ 0.9%, P Ͻ 0.0001). An A1C level medical histories, physical examinations, assay was 5.6%). An automated glucose Յ7.0% was achieved by 68% of subjects waist and hip measurements, 3-day food oxidase method (Glucose Analyzer 2; assigned to rosiglitazone vs. 52% of sub- records, and urine microalbumin levels Beckman Instruments, Fullerton, CA) jects assigned to placebo (P ϭ 0.3). An were determined at the beginning and was used to measure plasma glucose con- A1C level between 6.0 and 6.5% was end of the study. Fasting lipid and li- centrations. C-peptide concentrations achieved by 36% of subjects assigned to poprotein levels, fasting plasma glucose were measured by radioimmunoassay us- rosiglitazone vs. 16% assigned to placebo levels, serum chemistries, hemogloblin ing polyclonal antisera. Fasting lipid and (P Ͻ 0.05). and hematocrit levels, and C-peptide con- lipoprotein concentrations were assessed Regression analysis showed that the centrations were obtained at baseline, by standard laboratory methods. only predictors of improvement in A1C week 16, and at the end of the study level in the placebo-treated group were (week 32). Liver function tests, body Statistical analysis baseline A1C level (P Ͻ 0.0001) and fre- weight, blood pressure, and A1C levels Demographic and outcome variables quency of self–glucose testing (P ϭ were obtained at each visit. were checked for normality across 0.003, r2 ϭ 0.703). Baseline A1C level Subjects were asked to check their groups. A log transformation improved was also the most significant predictor of plasma glucose levels at least four times normality for BMI, total daily insulin improvement in A1C in the rosiglitazone- daily. Glycemic control, tolerance to the dose, A1C, VLDL cholesterol, LDL cho- treated group. However, when baseline assigned treatment, and frequency of hy- lesterol, HDL cholesterol, triglycerides, A1C level was removed from the model, poglycemia were assessed at each visit. VLDL triglyceride, alanine aminotransfer- the baseline BMI (P ϭ 0.001), total daily Subjects were encouraged to maintain ase, aspartate aminotransferase, C- insulin dose (P ϭ 0.001), total cholesterol baseline levels of dietary intake and phys- peptide, and urine albumin levels. A (P ϭ 0.003), HDL cholesterol (P ϭ 0.00l),

DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 1563 Rosiglitazone and type 1 diabetes

Table 1 — Baseline characteristics week 16; placebo: 185.8 Ϯ 90.6 to 177.6 Ϯ 66.1 mg/dl; P ϭ 0.51). However, Rosiglitazone ϩ insulin Placebo ϩ insulin by the end of the study, fasting plasma glucose levels were not significantly dif- n 25 25 ferent from baseline, nor were they differ- Age (years) 43.7 Ϯ 13.3 41.1 Ϯ 9.2 ent between groups. Sex (M/F) 16/9 18/7 The insulin regimens used were not Ethnic group (n) 22 Caucasian, 23 Caucasian, different between treatment groups at the 3 African American 1 African American, beginning or the end of the study. At base- 1 Hispanic line, the rosiglitazone-treated group con- Duration of diabetes (years) 20.7 Ϯ 13.3 18.1 Ϯ 9.3 sisted of 10 subjects who used an insulin Waist-to-hip ratio 0.91 Ϯ 0.07 0.93 Ϯ 0.06 pump, 10 subjects who used three or four Ϯ Ϯ Fasting C-peptide (ng/ml) 0.3 0.3 0.3 0.3 daily insulin injections of long- or inter- Data are means Ϯ SD or n. mediate-acting insulin in combination with rapid- or short-acting insulin, and 5 and LDL cholesterol (P ϭ 0.03) were sig- vs. rosiglitazone subjects with a BMI Ն30 subjects who used twice-daily insulin. nificant predictors of improvement in kg/m2) (Fig. 1). This did not occur in sub- The group assigned to placebo consisted A1C level in subjects treated with rosigli- jects taking placebo, even though baseline of 7 subjects who used an insulin pump, tazone (r2 ϭ 0.730). This was not the case A1C levels were not different between 15 subjects who used three to four daily in the placebo group. Neither baseline groups (rosiglitazone 8.1 Ϯ 1.5% vs. pla- insulin injections, and 3 who used twice- waist-to-hip ratio nor systolic or diastolic cebo 7.9 Ϯ 0.9% when BMI Ն30 kg/m2). daily insulin. By the end of the study, blood pressures were associated with The A1C level improved by 0.64 Ϯ 1.2% none of the subjects used twice-daily in- change in A1C level in either group. Fre- in placebo-treated subjects with a BMI sulin, and twice as many subjects in both quency of self–glucose testing was also Ն30 kg/m2 (n ϭ 12; mean BMI 33.9 Ϯ groups were using insulin glargine to not a significant predictor for improved 1.8 kg/m2) compared with 0.67 Ϯ 0.7% meet their basal insulin needs (32% of A1C in those treated with rosiglitazone. in those with a BMI Ͻ30 kg/m2 (mean subjects). Also, ϳ60% of all subjects were Rosiglitazone-treated subjects with a BMI 28.5 Ϯ 0.18 kg/m2). Cholesterol lev- using rapid-acting insulin at baseline BMI Ն30 kg/m2 (n ϭ 16, mean BMI els were essentially unchanged in place- (versus short-acting insulin), and ϳ80% 35.2 Ϯ 5.4 kg/m2) experienced signifi- bo-treated subjects, irrespective of were using rapid-acting insulin by the end cantly greater improvements in A1C baseline BMI. of the study. Subjects using insulin pump (Ϫ1.4 Ϯ 1.2%) and total cholesterol Mean fasting blood glucose levels therapy or multiple daily insulin injec- (Ϫ18.2 Ϯ 31.6 mg/dl) levels compared were somewhat more improved in the tions at baseline remained on that treat- with rosiglitazone-treated subjects with a group treated with rosiglitazone midway ment throughout the course of the study. BMI Ͻ30 kg/m2 (mean BMI 28.4 Ϯ 0.7 through the study (week 16), approach- The total daily insulin dose increased kg/m2; A1C Ϫ0.4 Ϯ 0.5% and total cho- ing significance (rosiglitazone: 172.6 Ϯ in the group assigned to placebo (74.0 Ϯ lesterol ϩ13.8 Ϯ 31.6 mg/dl, P Ͻ 0.05, 68.3 to 139.3 Ϯ 67.9 mg/dl baseline vs. 33.8 units at baseline vs. 82.0 Ϯ 48.9 at

Table 2—Results before and 32 weeks after treatment with rosiglitazone and insulin or placebo and insulin

Rosiglitazone and insulin Placebo and insulin Baseline Week 32 Baseline Week 32 Weight (kg) 97.2 Ϯ 11.8 100.6 Ϯ 16.0* 96.4 Ϯ 12.2 99.1 Ϯ 15.0* BMI (kg/m2) 32.7 Ϯ 5.4 34.0 Ϯ 7.4* 31.1 Ϯ 3.1 32.0 Ϯ 4.2* Systolic blood pressure (mmHg) 137.4 Ϯ 15.6† 128.8 Ϯ 14.8* 125.8 Ϯ 10.0 127.5 Ϯ 14.0 Diastolic blood pressure (mmHg) 87.2 Ϯ 9.4 79.4 Ϯ 7.2‡ 84.7 Ϯ 7.0 83.2 Ϯ 7.1 Insulin dose (units/day) 77.5 Ϯ 28.6 75.3 Ϯ 33.1 74.0 Ϯ 33.8 82.0 Ϯ 48.9* Fasting plasma glucose (mg/dl) 172.6 Ϯ 68.3 153.3 Ϯ 66.4 185.8 Ϯ 90.6 173.4 Ϯ 64.8 A1C (%) 7.9 Ϯ 1.3 6.9 Ϯ 0.7‡ 7.7 Ϯ 0.8 7.0 Ϯ 0.9‡ Total cholesterol (mg/dl) 201.9 Ϯ 37.6† 195.2 Ϯ 37.5§ 177.9 Ϯ 33.1 175.1 Ϯ 31.7 LDL cholesterol (mg/dl) 117.8 Ϯ 32.0 120.1 Ϯ 35.0 103.4 Ϯ 25.2 102.6 Ϯ 25.4 HDL cholesterol (mg/dl) 53.9 Ϯ 20.5 53.6 Ϯ 19.4 47.8 Ϯ 15.0 49.1 Ϯ 15.8 VLDL cholesterol (mg/dl) 15.8 Ϯ 13.2 16.4 Ϯ 9.8 13.8 Ϯ 9.4 14.6 Ϯ 15.6 Triglycerides (mg/dl) 108.1 Ϯ 64.6 101.8 Ϯ 53.8 92.8 Ϯ 38.9 85.4 Ϯ 39.7 VLDL triglycerides (mg/dl) 66.8 Ϯ 57.6 61.9 Ϯ 46.6 55.8 Ϯ 31.6 48.8 Ϯ 33.0 Hemoglobin (gm/dl) 14.4 Ϯ 1.5 13.6 Ϯ 1.9‡§ 14.5 Ϯ 1.0 14.5 Ϯ 1.0 Hematocrit (%) 42.0 Ϯ 3.9 39.8 Ϯ 5.1‡§ 42.3 Ϯ 3.0 42.2 Ϯ 2.9 Data are means Ϯ SD. *P Ͻ 0.05 vs. baseline; †P Ͻ 0.05 vs. placebo and insulin at baseline; ‡P Ͻ 0.0001 vs. baseline; §P Ͻ 0.05 vs. placebo and insulin at week 32.

1564 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 Strowig and Raskin

group was signiﬁcantly greater than in the placebo-treated group at the end of the study (P ϭ 0.009). Both groups were similar in terms of baseline history and treatment of hypertension and the number of subjects who were given additional doses and/or classes of antihypertensive medication during the course of the study.

Adverse events Rosiglitazone treatment was associated with an increased incidence of edema re- quiring diuretics compared with placebo (eight subjects versus one, P ϭ 0.01). One female subject reduced her study medication (rosiglitazone) to one tablet daily at week 18 because of significant edema in her extremities, despite the use of diuretic medication. Another female subject dis- continued study medication at week 30 Figure 1—Change in mean A1C level (A) and total daily insulin dose (B) based on baseline BMI because of severe edema that resulted in in type 1 diabetic subjects after 8 months of treatment with insulin and rosiglitazone or insulin and hospitalization for congestive heart fail- placebo. *P Ͻ 0.05 vs. rosiglitazone-treated subjects with a BMI Ͻ30 kg/m2;†P Ͻ 0.05 vs. ure. There is a possibility, however, that Ն 2 rosiglitazone-treated subjects with a BMI 30 kg/m . this patient had exceeded the prescribed dose of rosiglitazone by mistake because the subject’s bottle of diuretic medication week 32, P Ͻ 0.05), compared with a cantly increased in both treatment groups was found to also contain study medica- modest decline in the group treated with (rosiglitazone: ϩ3.3 Ϯ 5.8 kg, P ϭ 0.008; tion (rosiglitazone). None of the subjects rosiglitazone (77.5 Ϯ 28.6 units at base- placebo: ϩ2.7 Ϯ 5.2 kg; P Ͻ 0.02 vs. treated with placebo reduced their dose of line vs. 75.3 Ϯ 33.1 at week 32). Al- baseline), and an increase in weight was study medication. The incidence of mild though the absolute values were not strongly associated with improvement in anemia, as reflected in a significant de- significantly different between groups, A1C level in both groups (rosiglitazone: crease in hemoglobin (rosiglitazone: the change in insulin dose was signifi- r ϭϪ0.511; placebo: r ϭϪ0.607; P Յ 14.4 Ϯ 1.5 vs. 13.6 Ϯ 1.9 g/dl) and he- cantly greater in the group treated with 0.009). An increase in weight was not as- Ϯ placebo (rosiglitazone: Ϫ2.2 Ϯ 17.8 sociated with the presence of edema in the matocrit (rosiglitazone: 42.0 3.9 vs. ϩ Ϯ Ͻ ϭ 39.8 Ϯ 5.1%; P Ͻ 0.0001 baseline vs. units; placebo: 8.1 17.9; P 0.05). group treated with rosiglitazone (r Ͻ The significant change in total daily insu- Ϫ0.244, P ϭ 0.20). Mean caloric intake week 32; P 0.05 rosiglitazone vs. pla- lin dose in the placebo group was largely based on food records did not differ be- cebo at week 32), was also associated with caused by the increase in insulin required tween groups (rosiglitazone: 2,066.4 Ϯ rosiglitazone. Regression analysis showed by subjects with a baseline BMI Ն30 589.2 and 1,994.9 Ϯ 726.5 calories/day that female sex was the only significant kg/m2 (ϩ12.3 Ϯ 24.2 units/day), which for baseline and week 32, respectively; predictor of the incidence of edema re- ϭ Ͻ significantly differed from the decrease in placebo: 2,313.0 Ϯ 582.5 and 2,152.5 Ϯ quiring diuretics (r 0.736, P 0.0001) ϭ insulin requirements in rosiglitazone- 551.9). and the incidence of anemia (r 0.510, ϭ treated subjects with a baseline BMI Ն30 Lipid levels did not change in either P 0.009). There were no incidents of kg/m2 (Ϫ5.7 Ϯ 21.4 units/day; rosiglita- group. Urine albumin levels significantly elevated alanine aminotransferase or as- zone vs. placebo P Ͻ 0.05) (Fig. 1). When improved in both groups (rosiglitazone: partate aminotransferase levels in either examined as units per kilogram per day, 54.7 Ϯ 127.7 to 23.3 Ϯ 42.7 mg/24 h, group. rosiglitazone-treated subjects with a base- P Ͻ 0.03; placebo: 37.7 Ϯ 68.1 to 25.2.Ϯ The total incidence of mild (P ϭ 0.9) line BMI Ն30 kg/m2 experienced a signif- 35.2 mg/24 h, P ϭ 0.003, baseline vs. and severe hypoglycemic episodes (ros- icant reduction in insulin requirements at week 32). Both systolic blood pressure iglitazone: n ϭ 12; placebo: n ϭ 8; P ϭ Ϫ week 32 (Ϫ0.10 Ϯ 0.16 units kg 1 (137.4 Ϯ 15.6 to 128.8 Ϯ 14.8 mmHg, 0.3) was not different between groups. Ϫ day 1) compared with rosiglitazone- P Ͻ 0.02, baseline vs. week 32) and dia- However, midway through the study, treated subjects with a baseline BMI Ͻ30 stolic blood pressure (87.2 Ϯ 9.4 to subjects treated with rosiglitazone had Ϫ kg/m2 (ϩ0.03 Ϯ 0.05 units kg 1 79.4 Ϯ 7.2 mmHg, P Ͻ 0.0001) im- significantly more self-obtained plasma Ϫ day 1, P Ͻ 0.04) and compared with pla- proved in the group treated with rosigli- glucose readings Յ45 mg/dl (P Ͻ 0.05). cebo-treated subjects with a BMI Ն30 tazone. No change in blood pressure One male subject had to decrease his dose Ϫ kg/m2 (ϩ0.07 Ϯ 0.19 units kg 1 occurred in the placebo group. The of study medication (rosiglitazone) to one Ϫ day 1, P ϭ 0.02). change in systolic and diastolic blood tablet daily at week 8 because of frequent Mean body weight and BMI signifi- pressure in the rosiglitazone-treated episodes of hypoglycemia.

DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 1565 Rosiglitazone and type 1 diabetes

CONCLUSIONS — These results enced the greatest improvement in A1C study, may be what is necessary to achieve demonstrated that overweight type 1 dia- and total cholesterol level while lowering target glycemic levels. betic subjects taking rosiglitazone experi- their insulin dose. In addition, when Our subjects gained comparable enced significant and comparable baseline A1C level was removed from the amounts of weight gain irrespective of improvements in A1C levels (Ϫ1.0%) as model, baseline levels of BMI, total daily treatment with rosiglitazone or placebo, subjects taking insulin alone (Ϫ0.7%) insulin dose, total cholesterol, HDL cho- reflecting the increase in weight that is when insulin therapy was adjusted in lesterol, and LDL cholesterol (markers of associated with improved levels of glyce- both groups in an attempt to achieve insulin resistance) were significant pre- mic control. Our subjects were encour- near-normal blood glucose levels. How- dictors of improvement in A1C level in aged to maintain baseline levels of dietary ever, the placebo-treated group required subjects treated with rosiglitazone. In intake, and food records revealed no 11% more insulin to achieve these results. contrast, baseline A1C level and fre- change in dietary intake at the end of the Both groups also gained significant but quency of blood glucose testing were the study. Other studies have shown, how- comparable amounts of weight (ϳ3 kg). only predictors of a decrease in A1C in the ever, that with caloric restriction, weight This weight gain was not related to the group treated with placebo, suggesting gain can be avoided when glycemic con- presence of edema or change in insulin that self-management behaviors were more trol is improved with thiazolidinediones dose, and it was not associated with any important in improving A1C level in sub- (23,24). increase in caloric intake based on 3-day jects who were not taking rosiglitazone. The insulin sensitizer metformin has food records. Rather, weight gain was Although the overweight type 1 dia- resulted in improved glycemic control strongly associated with improved A1C betic subjects treated with rosiglitazone without weight gain in type 2 diabetic levels in both groups. By the end of the experienced a significant improvement in subjects, both as dual insulin and met- study, neither group had experienced any their mean A1C level while lowering their formin therapy (18,25) and as triple ther- significant change in lipid levels, and the total daily insulin dose, the reductions in apy when metformin was added to insulin overall frequency of hypoglycemia was A1C level (from 7.9 to 6.9%) and total therapy and later combined with a thiazo- the same. The group treated with rosigli- daily insulin dose (Ϫ3.0%) were not as lidinedione (26). Studies of normal- tazone experienced significantly more pronounced as we have seen in previous weight type 1 diabetic subjects (BMI Յ26 edema and anemia, with female sex being studies involving type 2 diabetes. When kg/m2) given metformin resulted in no the only significant predictor of these we combined troglitazone with insulin weight gain, but the subjects also experi- events. therapy in type 2 diabetic subjects, A1C enced only small if any improvement in Differences between the groups levels improved from 8.5 to 6.4% with a glycemic control (15–17). Therefore, it is emerged, however, when the subjects’ 13% reduction in insulin dose (18). The not known whether metformin in combi- baseline BMI was taken into consider- increased risk for hypoglycemia in type 1 nation with insulin in overweight (BMI ation. Rosiglitazone-treated subjects with diabetes may be the main obstacle pre- Ն30 kg/m2) type 1 diabetic subjects a baseline BMI Ն30 kg/m2 experienced a venting a more profound blood glucose– would result in improved glycemic con- significantly greater improvement in A1C lowering effect of rosiglitazone in these trol without weight gain. (Ϫ1.4%) and total cholesterol level individuals. In this study, type 1 diabetic Rosiglitazone-treated subjects experi- (Ϫ18.2 mg/dl) and a significant decrease subjects taking rosiglitazone had signifi- enced a significant improvement in sys- in insulin (Ϫ1.0 units/kg) compared with cantly more plasma glucose readings Ͻ45 tolic and diastolic blood pressure over the rosiglitazone-treated subjects whose mg/dl midway through the study (week course of the study. No change in blood baseline BMI was Ͻ30 kg/m2 (A1C: 16) than subjects taking placebo. This pressure was observed in subjects taking Ϫ0.4%; total cholesterol: ϩ13.8 mg/dl; may have precluded our taking a more placebo. Although the group assigned to insulin dose: ϩ0.3 units/kg). These out- aggressive approach to insulin therapy, or treatment with rosiglitazone had a higher comes were not influenced by BMI in sub- it may have prompted us to reduce the baseline systolic blood pressure than the jects treated with placebo. In fact, insulin dose. In addition, the patients may group assigned to placebo, baseline dia- placebo-treated subjects with a BMI Ն30 have altered their self-management be- stolic blood pressures were similar be- kg/m2 accounted for the greatest increase haviors in their effort to reduce the fre- tween the groups. Thus, even though the in insulin dose in the placebo group quency of hypoglycemia and avoid study was not designed to evaluate the (ϩ12.3 units/day), whereas subjects with episodes of severe hypoglycemia. impact of rosiglitazone on blood pressure, a BMI Ն30 kg/m2 taking rosiglitazone ac- On the other hand, 68% of our ros- it appears that rosiglitazone had a benefi- tually decreased their insulin require- iglitazone-treated subjects achieved an cial effect. Hypertension was aggressively ments (Ϫ5.7 units/day). This is more A1C level Յ7.0%. This is in contrast to treated in all subjects, and there were no remarkable considering that the rosiglita- results obtained in studies of type 2 dia- differences between groups regarding his- zone-treated subjects with a BMI Ն30 betic subjects in whom insulin and thia- tory of hypertension at baseline, the num- kg/m2 experienced a reduction in A1C of zolidinedione treatment resulted in ber of subjects who were being treated for 1.4% compared with placebo-treated significantly lower A1C levels, but whose hypertension at baseline, or the number subjects with a BMI Ն30 kg/m2, whose average A1C levels at the end of the study of subjects who were prescribed in- A1C level improved by 0.6%. These re- exceeded 7.8% (19–22). Because these creased doses and/or additional classes of sults suggest that rosiglitazone had the type 2 diabetic studies did not allow for an antihypertensive medication. greatest effect on subjects who may have increase in insulin dose, the simultaneous In summary, rosiglitazone may be an had the most insulin resistance (BMI Ն30 use of intensive insulin regimens with a effective adjunct to insulin therapy in type kg/m2) because these subjects experi- thiazolidinedione, as was done in this 1 diabetic subjects with a BMI Ն30 kg/m2

1566 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 Strowig and Raskin who have markers of insulin resistance, tions Trial. JAMA 276:1409–1415, 1996 as additional therapy in adolescents with such as high insulin requirements, hyper- 4. The Diabetes Control and Complications poorly controlled type 1 diabetes: ran- tension, and hypercholesterolemia. De- Trial Research Group: Hypoglycemia in domised placebo-controlled trial with as- spite the variability seen among subjects the Diabetes Control and Complications pects on insulin sensitivity. Eur J with a BMI Ն30 kg/m2 who took rosigli- Trial. Diabetes 46:271–286, 1997 Endocrinol 149:323–329, 2003 5. Baron AD, Kim D, Weyer C: Novel pep- 17. Hamilton J, Cummings E, Zdravkovic V, tazone, lower overall total daily insulin tides under development for the treat- Finegood D, Daneman D: Metformin as dosage and significant improvements in ment of type 1 and type 2 diabetes an adjunct therapy in adolescents with A1C, cholesterol, and blood pressure lev- mellitus (Review). Curr Drug Targets Im- type 1 diabetes and insulin resistance: a els were observed. However, the potential mune Endocr Metabol Disord 2:63–82, randomized controlled trial. Diabetes benefit of rosiglitazone must be balanced 2002 Care 26:138–143, 2003 against the potential risk for hypoglyce- 6. DeFronzo RA, Simonson D, Ferrannini E: 18. Strowig SM, Avile´s-Santa ML, Raskin P: mia, edema, and anemia, especially in Hepatic and peripheral insulin resistance: Comparison of insulin monotherapy and women who appear to experience edema a common feature of type 2 (non-insulin- combination therapy with insulin and and anemia more profoundly with this dependent) and type 1 (insulin-depen- metformin or insulin and troglitazone in drug. Because aggressive insulin therapy dent) diabetes mellitus. Diabetologia 23: type 2 diabetes. Diabetes Care 25:1691– 313–319, 1982 1698, 2002 using increased doses of insulin led to 7. DeFronzo RA, Hendler R, Simonson D: 19. Schwartz S, Raskin P, Fonseca V, Grav- overall improvements in glycemic control Insulin resistance is a prominent feature eline JF, for the Troglitazone and Exoge- that were comparable to that which was of insulin-dependent diabetes. Diabetes nous Insulin Study Group: Effect of achieved in the group treated with rosigli- 31:795–801, 1982 troglitazone in insulin-treated patients tazone, treatment with rosiglitazone 8. Del Prato S, Nosadini R, Tiengo A, Tessari with type II diabetes mellitus. N Engl J Med should only be considered after efforts at P, Avogaro A, Trevisan R, Valerio A, Mug- 338:861–866, 1998 improving glycemic control with inten- geo M, Cobelli C, Toffolo G: Insulin-me- 20. Buse JB, Gumbiner B, Mathias NP, Nelson sive insulin treatment, increased blood diated glucose disposal in type 1 diabetes: DM, Faja BW, Whitcomb RW: Troglita- glucose self-monitoring, and dietary evidence for insulin resistance. J Clin En- zone use in insulin-treated type 2 diabetic management do not result in desired gly- docrinol Metab 57:904–910, 1983 patients: the Troglitazone Insulin Study 9. Ma¨kimattila S, Virkama¨ki A, Malmstro¨m Group. Diabetes Care 21:1455–1461, cemic goals. More investigation into the R, Utriainen T, Yki-Jarvinen H: Insulin re- 1998 long-term effects of rosiglitazone may sistance in type 1 diabetes mellitus: a ma- 21. Raskin P, Rendell M, Riddle MC, Dole JF, provide more definitive evidence regard- jor role for reduced glucose extraction. Freed MI, Rosenstock J; Rosiglitazone ing the beneficial impact of rosiglitazone J Clin Endocrinol Metab 81:707–712, 1996 Clinical Trials Study Group: A random- on glycemic control and cardiovascular 10. Greenfield JR, Samaras K, Chisholm DJ: ized trial of rosiglitazone therapy in pa- risk factors in type 1 diabetic individuals Insulin resistance, intra-abdominal fat, tients with inadequately controlled who have features of insulin resistance. cardiovascular risk factors, and andro- insulin-treated type 2 diabetes. Diabetes gens in healthy young women with type 1 Care 24:1226–1232, 2001 diabetes mellitus. J Clin Endocrinol Metab 22. Rosenstock J, Einhorn D, Hershon K, Acknowledgments— This study was sup- 87:1036–1040, 2002 Glazer NB, Yu S; Pioglitazone 014 Study ported in part by a grant from GlaxoSmith 11. Purnell JQ, Dev RK, Steffes MW, Cleary Group: Efficacy and safety of pioglitazone Kline. PA, Palmer JP, Hirsch IB, Hokanson JE, in type 2 diabetes: a randomised, placebo- The authors are indebted to Kelly Sherrill Brunzell JD: Relationship of family history controlled study in patients receiving sta- for her technical assistance in conducting the of type 2 diabetes, hypoglycemia, and au- ble insulin therapy. Int J Clin Pract 56: study. toantibodies to weight gain and lipids 251–257, 2002 with intensive and conventional therapy 23. Reynolds LR, Konz EC, Frederich RC, in the Diabetes Control and Complica- Anderson JW: Rosiglitazone amplifies the References tions Trial. Diabetes 52:2623–2629, 2003 benefits of lifestyle intervention measures 1. The Diabetes Control and Complications 12. DeFronzo RA: Pharmacologic therapy for in long-standing type 2 diabetes mellitus. Trial Research Group: The effect of inten- type 2 diabetes mellitus (Review). Ann In- Diabetes Obes Metab 4:270–275, 2002 sive treatment of diabetes on the develop- tern Med 131:281–303, 1999 24. Asnani S, Richard BC, Desouza C, Fon- ment and progression of long-term 13. Inzucchi SE: Oral antihyperglycemic seca V: Is weight loss possible in patients complications in insulin-dependent dia- therapy for type 2 diabetes: scientific re- treated with thiazolidinediones? Experi- betes mellitus. N Engl J Med 329:977–986, view. JAMA 287:360–372, 2002 ence with a low-calorie diet. Curr Med Res 1993 14. Pagano G, Tagliaferro V, Carta Q, Caselle Opin 19:609–613, 2003 2. UK Prospective Diabetes Study (UKPDS) MT, Bozzo C, Vitelli F, Trovati M, 25. Avile´s-Santa L, Sinding J, Raskin P: Effects Group: Intensive blood-glucose control Cocuzza E: Metformin reduces insulin re- of metformin in patients with poorly con- with sulphonylureas or insulin compared quirement in type 1 (insulin-dependent) trolled, insulin-treated type 2 diabetes with conventional treatment and risk of diabetes. Diabetologia 24:351–354, 1983 mellitus: a randomized, double-blind, complications in patients with type 2 di- 15. Meyer L, Bohme P, Delbachian I, Lehert P, placebo-controlled trial. Ann Intern Med abetes (UKPDS 33). Lancet 352:837–853, Cugnardey N, Drouin P, Guerci B: The 131:182–188, 1999 1998 benefits of metformin therapy during 26. Strowig SM, Avile´s-Santa ML, Raskin P: 3. The Diabetes Control and Complications continuous subcutaneous insulin infu- Improved glycemic control without Trial Research Group: Lifetime benefits sion treatment of type 1 diabetic patients. weight gain using triple therapy in type 2 and costs of intensive therapy as practiced Diabetes Care 25:2153–2158, 2002 diabetes. Diabetes Care 27:1577–1583, in the Diabetes Control and Complica- 16. Sa¨rnblad S, Kroon M, Åman J: Metformin 2004

DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 1567