Effect of Metformin and Rosiglitazone Combination Therapy in Patients with Type 2 Diabetes Mellitus a Randomized Controlled Trial

ORIGINAL CONTRIBUTION

Effect of Metformin and Rosiglitazone Combination Therapy in Patients With Type 2 Diabetes Mellitus A Randomized Controlled Trial

Vivian Fonseca, MD Context Most antidiabetic agents target only 1 of several underlying causes of dia- Julio Rosenstock, MD betes. The complementary actions of the antidiabetic agents metformin hydrochlo- Rita Patwardhan, PhD ride and rosiglitazone maleate may maintain optimal glycemic control in patients with type 2 diabetes; therefore, their combined use may be indicated for patients whose Alan Salzman, MD, PhD diabetes is poorly controlled by metformin alone. Objective To evaluate the efficacy of metformin-rosiglitazone therapy in patients YPE 2 DIABETES IS CHARACTER- whose type 2 diabetes is inadequately controlled with metformin alone. ized by decreased insulin secre- Design Randomized, double-blind, placebo-controlled trial from April 1997 and March tion1,2 and insulin sensitivity in 1998. liver, adipose tissue, and skel- Setting Thirty-six outpatient centers in the United States. etalT muscle. Together these abnormali- Patients Three hundred forty-eight patients aged 40 to 80 years with a mean fast- ties confound efforts to treat diabetes ing plasma glucose level of 12.0 mmol/L (216 mg/dL), a mean glycosylated hemo- because most antidiabetic agents tar- globin level of 8.8%, and a mean body mass index of 30.1 kg/m2 were randomized. get only 1 underlying cause of the dis- Interventions Patients were assigned to receive 2.5 g/d of metformin plus placebo ease. Approximately 50% of patients (n = 116); 2.5 g/d of metformin plus 4 mg/d of rosiglitazone (n = 119); or 2.5 g/d of treated with monotherapy require addi- metformin and 8 mg/d of rosiglitazone (n = 113) for 26 weeks. tional therapy to achieve target glyco- Main Outcome Measures Glycosylated hemoglobin levels, fasting plasma glu- sylated hemoglobin (HbA1c) levels 3 3 cose levels, insulin sensitivity, and ␤-cell function, compared between baseline and years after diagnosis. week 26, by treatment group. Rosiglitazone maleate, a member of the thiazolidinedione class of antidia- Results Glycosylated hemoglobin levels, fasting plasma glucose levels, insulin sensitivity, and ␤-cell function improved significantly with metformin-rosiglitazone therapy betic agents that was recently approved in a dose-dependent manner. The mean levels of glycosylated hemoglobin decreased by the US Food and Drug Administra- by 1.0% in the 4 mg/d metformin-rosiglitazone group and by 1.2% in the 8 mg/d tion, targets insulin resistance by bind- metformin-rosiglitazone group and fasting plasma glucose levels by 2.2 mmol/L (39.8 ing to the transcription factor peroxi- mg/dL) and 2.9 mmol/L (52.9 mg/dL) compared with the metformin-placebo group some proliferator-activated receptor-␥, (PϽ.001 for all). Of patients receiving 8 mg/d of metformin-rosiglitazone, 28.1% promoting synthesis of glucose trans- achieved a glycosylated hemoglobin level of 7% or less. Dose-dependent increases in porters and activating adipocyte differ- body weight and total and low-density lipoprotein cholesterol levels were observed entiation.4-6 In contrast, metformin (PϽ.001 for both rosiglitazone groups vs placebo). The proportion of patients report- hydrochloride promotes glucose low- ing adverse experiences was comparable across all groups. ering by reducing hepatic glucose pro- Conclusions Our data suggest that combination treatment with once-daily metformin- duction and gluconeogenesis and by rosiglitazone improves glycemic control, insulin sensitivity, and ␤-cell function more enhancing peripheral glucose uptake.7-10 effectively than treatment with metformin alone. Because metformin and rosiglit- JAMA. 2000;283:1695-1702 www.jama.com azone act through different mechanisms, their combined use may be in- 4 mg/d and 8 mg/d of rosiglitazone ma- Authors Affiliations and Financial Disclosure are listed at the end of this article. dicated in patients whose disease is leate to maximal-dosage of metformin Corresponding Author and Reprints: Vivian Fon- poorly controlled with a maintenance in patients with poorly controlled type seca, MD, Department of Medicine, Endocrinology Sec- tion, Tulane University, 1430 Tulane Ave, PO Box dose of metformin. This study evalu- 2 diabetes. Combined efficacy was as- SL53, New Orleans, LA 70112 (e-mail: vfonseca ated the efficacy and safety of adding sessed by comparing the level changes @mailhost.tcs.tulane.edu).

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

in HbA1c, fasting plasma glucose (FPG), ing C-peptide concentration of 0.27 eride levels, or lipid metabolism fructosamine, serum insulin, free fatty nmol/L (0.8 ng/mL) or more at screen- disorders were eligible; lipid-lowering acids (FFA), lipids, lactate, and esti- ing. Subjects were required to have a agents were maintained at the same dos- mates of insulin sensitivity and ␤-cell body mass index, calculated as weight age level throughout the study. function (BCF) between combined met- in kilograms divided by the square of formin-rosiglitazone treatment and height in meters, of 22 to 38 and a Study Design metformin-placebo alone.11 weight change of no more than 10% be- This multicenter, randomized, double- tween screening and baseline. blind, placebo-controlled trial was con- METHODS Patients were excluded if they had ducted at 36 sites in the United States be- Study Subjects clinically significant renal or hepatic dis- tween April 1997 and March 1998. To detect a 0.75% absolute difference ease, angina, New York Heart Associa- Before the study, patients discontinued in HbA1c between treatment groups, 65 tion Classification class III or IV car- all antihyperglycemic medications, with evaluable patients per group would be diac insufficiency, symptomatic diabetic the exception of metformin. Metformin required to achieve a power of 95%. neuropathy, significant clinical abnor- dose tolerability was determined dur- Planned enrollment was 280 patients mality on electrocardiogram, abnor- ing a 3-week period in which metfor- (approximately 93 per group). Per- mal laboratory test results (blood chem- min was titrated to 2.5 g/d; afterward, pa- sons between the ages of 40 and 80 istry, hematology, or urinalysis), use of tients entered a 4-week, single-blind years with type 2 diabetes as defined by chronic insulin therapy, participated in metformin-placebo maintenance pe- the National Diabetes Data Group12 any rosiglitazone-related study, or used riod with a weight-maintenance diet. with FPG concentrations of between any investigational drug (excluding During this maintenance period, only in- 7.8 and 16.7 mmol/L (140 and 300 metformin) within 30 days of study (or vestigators were aware that patients were mg/dL) at screening and during the pla- 5 half-lives of the investigational drug, receiving the metformin-placebo treat- cebo-maintenance period while tak- if longer than 30 days). Anorectic agents ment. Patients previously treated with ing 2.5 g/d of metformin were eligible. were discontinued at least 30 days be- metformin at 2.5 g/d proceeded di- All patients demonstrated insulin se- fore screening. Patients with hyperli- rectly to maintenance; thus, with the ex- cretory capacity as determined by a fast- pemia, elevated cholesterol or triglyc- ception of metformin, patients re- frained from medication for a minimum Figure 1. Study Profile of Patients Randomized to Receive Metformin Hydrochloride Alone or of 4 weeks and a maximum of 7 weeks. With Rosiglitazone Maleate At the end of the maintenance period, patients with inadequate glycemic con- 443 Patients Screened trol (FPG concentration range, 7.7-16.7 mmol/L[140-300mg/dL])wererandomly 6 Patients Not Registered assigned (1:1:1 ratio) to receive double- blind metformin treatment in 1 of 3 com- 437 Patients Registered binations: placebo (control), 4 mg of rosi-

89 Not Randomized glitazone, or 8 mg of rosiglitazone once 62 Did Not Meet Criteria daily for 26 weeks. Randomization was 13 Adverse Experience 14 Other computer generated with a fixed block size. No patient, investigator, or sponsor

348 Randomized was aware of treatment allocation until study completion (FIGURE 1). This study was conducted in accor- 116 Metformin and Placebo 119 Metformin and 113 Metformin and Rosiglitazone, 4 mg/d Rosiglitazone, 8 mg/d dance with the Declaration of Hel- sinki (as amended, 1989), Title 21 of Follow-up Follow-up Follow-up the US Code of Federal Regulations, and 109 Week 4 116 Week 4 105 Week 4 105 Week 8 114 Week 8 102 Week 8 Good Clinical Practice guidelines. The 103 Week 12 106 Week 12 99 Week 12 99 Week 18 104 Week 18 97 Week 18 institutional review board at each cen- 94 Week 26 101 Week 26 95 Week 26 ter approved the protocol, and subjects provided informed consent be- 22 Withdrew 18 Withdrew 18 Withdrew fore enrollment. 8 Lack of Efficacy 4 Lack of Efficacy 4 Lack of Efficacy 5 Adverse Experience 7 Adverse Experience 6 Adverse Experience 0 Lost to Follow-up 3 Lost to Follow-up 4 Lost to Follow-up Efficacy and Safety Measurements 9 Other 4 Other 4 Other Laboratorymeasurementsforefficacyand

94 Completed Trial 101 Completed Trial 95 Completed Trial safety were performed by SmithKline BeechamClinicalLaboratories(VanNuys,

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

Calif) on blood collected in the fasting Safety monitoring included physi- ric analysis were not met, the Wilcoxon state. Fasting plasma glucose concentra- cal examination, vital sign assess- rank sum test was used. Pairwise com- tions, total cholesterol, high-density ment, weight measurement, electrocar- parisons to placebo used Dunnett mul- lipoprotein cholesterol (HDL-C), and tri- diogram, adverse experience query, and tiple comparison procedure to main- glyceride levels were measured by an laboratory tests. tain a 2-sided .05 significance level within Olympus analyzer (Olympus Clinical In- each parameter. The statistical signifi- struments Division, Lake Success, NY); Statistical Methods cance of the within-group change from levelsofHbA1c weremeasuredbythehigh- The primary population for efficacy baseline was tested by a paired t test or performance liquid chromatography analysis was the intention-to-treat popu- a signed rank test. Safety parameters, in- method (Variant, Bio-Rad, Hercules, lation, those with at least 1 value while cluding clinical laboratory tests, vital Calif); C-peptide by radioimmunoassay receiving therapy (last observation was signs, and body weight, were examined (DiagnosticProducts,LosAngeles,Calif); carried forward in the case of missing using 1-way analysis of variance. Statis- insulin by radioimmunoassay (Linco Re- data or early withdrawals). Efficacy and tical analyses were performed using sta- search Inc, St Charles, Mo); fructosamine safety parameters were measured at base- tistical software (SAS/STAT Software, Re- by colorimetric analysis (RoTAG fructos- line and after 26 weeks of treatment. lease 6.12, SAS Institute Inc, Cary, NC). amine assay, Roche Diagnostic Systems, Safety parameters were assessed based Indianapolis,Ind);andFFAbyenzymatic/ on week 26 data (without the last ob- RESULTS colorimetric analysis (Wako Diagnostic, servation carried forward). Of 443 patients screened, 437 entered the Richmond, Va) using a COBAS analyzer Treatment groups were compared us- titration and maintenance period and 348 (RocheDiagnosticSystems).Low-density ing analysis of covariance with terms for were randomized to treatment (Figure lipoprotein cholesterol (LDL-C) concen- baseline, treatment, and center. The as- 1). Most withdrawals were due to fail- trations were estimated from total choles- sumptions of the statistical model were ing to meet inclusion criteria (69.7%). terolandHDL-Cdeterminationsusingthe tested before application. The Levene test Baseline characteristics were similar Friedewaldcalculation.13 Lactatewasmea- of heterogeneity across treatments was among treatment groups (TABLE 1). sured by enzymatic spectrophotometric applied at a significance level of ␣ = .01. Fifty-eight patients withdrew before analysis using an Olympus analyzer If significant, the Shapiro-Wilk test of completion of the double-blind phase: (OlympusClinicalInstrumentsDivision). nonnormality (␣ = .01) was examined. 22 from the placebo group and 18 from Estimates of insulin sensitivity de- Parametric analysis or nonparametric the 4-mg/d and 18 from the 8-mg/d termined by homeostasis model assess- analysis was used, depending on re- rosiglitazone groups. Most participants ment (HOMA-S) and BCF (HOMA-B) sults of test assumptions. If prospec- withdrew because of adverse experi- were calculated using FPG and immu- tively defined assumptions for paramet- ences or lack of efficacy (Figure 1). noreactive insulin values, or C-peptide levels. HOMA is a mathematical Table 1. Baseline Demographic and Metabolic Characteristics of Randomized Patients model based on glucose and insulin in- (Intention-to-Treat Population) teraction in different organs, includ- Metformin Hydrochloride Metformin and Rosiglitazone Maleate ing the pancreas, liver, and peripheral Hydrochloride tissues.11 HOMA estimates of BCF and and Placebo 4 mg/d 8 mg/d insulin sensitivity were calculated for Baseline Characteristics (n = 113) (n = 116) (n = 110) each participant’s FPG and insulin, or Age, mean (SD), y 58.8 (9.2) 57.5 (10.5) 58.3 (8.8) Sex, % C-peptide levels, and expressed rela- Men 74.3 62.1 68.2 tive to values in a lean, nondiabetic Women 25.7 37.9 31.8 reference population aged 18 to 25 Duration of diabetes, mean (SD), y 7.3 (5.7) 7.5 (6.3) 8.3 (6.3) 14-16 years. HOMA-S determinations of Prior treatment, % insulin sensitivity or insulin resis- Diet and exercise only 4.4 6.0 4.5 tance have been validated by compari- Oral monotherapy 48.7 39.7 43.6 son with results of glucose clamp stud- Oral combination therapy 46.9 54.3 51.8 ies,11,14 intravenous glucose tolerance Baseline HbA1c, mean (SD), %* 8.6 (1.3) 8.9 (1.3) 8.9 (1.5) 11,15 Baseline fasting plasma glucose, 11.87 (2.91) 11.90 (3.17) 12.19 (3.05) tests, and continuous infusion of mean (SD), mmol/L† 15 glucose with model assessment. The Body mass index, mean (SD), kg/m2 30.3 (4.4) 30.2 (4.2) 29.8 (3.9) HOMA-B method has been validated by Race, % comparison with the intravenous glu- White 81.4 80.2 77.3 cose tolerance test and continuous in- Black 3.5 6.9 10.0 fusion of glucose model assessment.17 Other 15.0 12.9 12.7 Application of HOMA has also been *HbA1c indicates glycosylated hemoglobin. †To convert from millimoles per liter to milligrams per deciliter divide by 0.0555. used in epidemiological studies.18,19

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

Glycemic Control 4-mg/d group and by 36.8 µmol/L from units in the 4-mg/d and by 3.8 units in

ThemeanHbA1c levelsdecreasedsignifi- 351.8 µmol/L in the 8-mg/d group (ref- the 8-mg/d rosiglitazone groups com- cantly from baseline in a dose-dependent erence range, 200-278 µmol/L). pared with the control group. fashion in both rosiglitazone groups by Although the mean FPG concentra- The metformin-rosiglitazone combi- 0.56% in the 4-mg/d and by 0.78% in the tions did not change significantly in the nation increased HOMA-B in a dose- 8-mg/d rosiglitazone groups. But the con- control group, they significantly de- dependent fashion. The median baseline trol group experienced a significant increased in a dose-dependent order from HOMA-B values ranged from 32.5 to 35.8 creaseinHbA1c levels(0.45%)(FIGURE 2). baseline in both rosiglitazone groups (1.8 units and were significantly increased by Furthermore, both rosiglitazone groups mmol/L [–33.0 mg/dL], 4-mg/d rosigl- 10.3 to 13.7 units in the rosiglitazone had HbA1c levels lower than those in the itazone; –2.7 mmol/L [–48.4 mg/dL], groups compared with the control group. control group by 1.0% in the 4-mg/d and 8-mg/d-rosiglitazone; PϽ.0001). The 1.2% in the 8-mg/d rosiglitazone groups. mean FPG concentrations in both rosi- Other Metabolic Effects In contrast to results observed in the glitazone groups also had decreased com- In the control group, the insulin value control group, the mean HbA1c levels pared with the control group (−2.2 decreased by 11.05 pmol/L from a base- in the rosiglitazone groups decreased mmol/L [−39.8 mg/dL], 4-mg/d rosigl- line of 118.56 pmol/L after treatment after week 4 and plateaued by week 18 itazone; –2.9 mmol/L [–52.9 mg/dL], (P = .03) and in the 4-mg/d and 8-mg/d (FIGURE 3).Thepercentagewhoachieved 8-mg/d rosiglitazone; PϽ.0001) rosiglitazone groups the insulin values a1.0%reductioninHbA1c concentrations (FIGURE 4). Furthermore, FPG concen- respectively decreased by 12.98 pmol/L was 32.8% in the 4-mg/d and 37.3% in trations in both rosiglitazone groups de- from 124.55 pmol/L (P = .01) and by the8-mg/drosiglitazonegroupsand7.1% creased during the first 4 weeks, pla- 31.07 pmol/L from 136.73 pmol/L in the control group. teaued at 12 to 18 weeks, and remained (P = .14). The C-peptide values respec- Twenty-five (28.1%) of 89 patients stable thereafter (FIGURE 5). Nine pa- tively decreased by 0.10 nmol/L from taking 8 mg/d of rosiglitazone achieved tients (7.9%) in the control group, 25 0.93 nmol/L (PϽ.001), by 0.07 nmol/L the target HbA1c control levels of 7.0%, (21.6%) in the 4-mg/d and 33 (30.0%) from 0.92 nmol/L (P = .01), and by 0.12 and 51 patients (57.3%) in the same in the 8-mg/d rosiglitazone groups nmol/L from 0.93 nmol/L (PϽ.001). group achieved HbA1c levels of 8.0%, achieved FPG concentrations of less Mean total cholesterol-HDL-C, and or below the American Diabetes Asso- than7.8 mmol/L (140 mg/dL). LDL-C levels from baseline in both rosi- ciation action point. Yet only 7.6% of glitazone groups achieved statistically the patients in the control group Effects on Insulin Sensitivity significant increases in all treatment achieved HbA1c levels of 7.0% and and BCF groups compared with the control 35.9% achieved an HbA1c level of 8.0%. Adding rosiglitazone to maximum group (TABLE 2). Total cholesterol– The mean baseline fructosamine lev- doses of metformin significantly in- HDL-C ratios in the rosiglitazone els of 341.73 µmol/L in the control group creased HOMA-S values. The median groups were not significantly differ- increased by 12.3 µmol/L. But in the rosi- baseline HOMA-S values ranged from ent from those in the control group. glitazone groups the levels decreased by 46.6 to 49.0 units. The HOMA-S val- 27.9 µmol/L from 340.9 µmol/L in the ues increased dose-dependently by 1.7 Figure 3. Mean Change in Glycosylated Hemoglobin (HbA1c) Levels Over Time in Patients Taking Metformin Hydrochloride Figure 2. Change in Glycosylated Hemoglobin (HbA1c) Levels at Week 26 in Patients Taking Alone Compared With Patients Taking Metformin Hydrochloride and Rosiglitazone Maleate Compared With Taking Metformin Alone Metformin and Rosiglitazone Maleate Combined 1.0 +0.45 Comparison 1.0 Comparison With P<.001 With Baseline Control Group at 26 Weeks , %

1c Metformin and Placebo 0.5 0.5 Metformin and 4 mg/d of Rosiglitazone Metformin and 8 mg/d of Rosiglitazone 0 0 10.5

, % 10.0 –0.5 1c –0.5 9.5 ,% 1c –0.56 in HbA 9.0 –1.0 –1.0 P<.001 8.5 –0.78 –1.5 P<.001 –1.5 –1.0 8.0

< Mean HbA Mean Change From Control Group Control Mean Change From P .001 –1.2 P<.001 7.5 Mean Change From Baseline in HbA Mean Change From –2.0 –2.0 Metformin Metformin Metformin Metformin Metformin 7.0 and and and and and Placebo 4 mg/d of 8 mg/d of 4 mg/d of 8 mg/d of –7 –6–5 –4–2 04 812 18 26 Rosiglitazone Rosiglitazone Rosiglitazone Rosiglitazone Treatment Week

Error bars indicate 95% confidence interval. Error bars indicate SE.

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

Changes in LDL-C levels were evalu- mmol/L (13 mg/dL) from 4.20 mmol/L Mean fasting lactate levels de- ated based on those at baseline. In that (162 mg/dL) in 20 patients taking creased significantly in patients tak- analysis, we identified 2 subgroups: 8-mg/d. ing both dose levels of rosiglitazone those with levels lower than 3.37 Changes in triglyceride levels also compared with those in the control mmol/L (Ͻ130 mg/dL) and those at were evaluated based on baseline val- group (4-mg/d rosiglitazone, P = .012; that level or higher. We did not pro- ues, using 2 subgroups: those with lev- 8-mg/d rosiglitazone, P = .002). Free vide P values for any of the subgroups els lower than 2.26 mmol/L (Ͻ200 fatty acids concentrations decreased sig- because the values were not large mg/dL) and those with that level or nificantly from baseline in both rosigl- enough for statistical analyses and be- higher. In the lower subgroup, the me- itazone groups. (TABLE 3). cause the subgroups were not random- dian baseline triglyceride values in- ized, so significance could not be es- creased by 0.15 mmol/L (13 mg/dL) Safety tablished. In the lower subgroup, the from 1.44-mmol/L (128-mg/dL) in 52 The percentage of patients with at least median baseline LDL-C value in- patients in the control group. In the 1 adverse event were comparable among creased by 0.13 mmol/L (5 mg/dL) from rosiglitazone groups, the median base- each group (75.2%, 4-mg/d rosigl- 2.59 mmol/L (100 mg/dL) in 51 pa- line triglyceride value increased by 0.16 itazone; 78.2%, 8-mg/d rosiglitazone; tients in the control group. In both rosi- mmol/L (15 mg/dL) from 1.67 mmol/L 76.7%, control). The most frequently re- glitazone groups, the LDL-C values in- (148-mg/dL) in 56 patients taking ported adverse events were upper respi- creased by 0.54-mmol/L (21 mg/dL) 4-mg/d and by 0.07 mmol/L (6 mg/dL) ratory tract infection, diarrhea, and head- from a median baseline value of 2.69 from 1.34-mmol/L (119 mg/dL) in 55 ache. One death due to acute myocardial mmol/L (104-mg/dL) in 57 patients tak- patients taking 8-mg/d. The treatment infarction occurred in the 4-mg/d rosi- ing 4-mg/d and from 2.64-mmol/L (102 values in all groups remained less than glitazone group but was judged to be un- mg/dL) in 60 patients taking 8-mg/d, 2.25 mmol/L (200 mg/dL). related to study medication. Serious non- resulting in medians that remained be- In the higher subgroup, the median fatal adverse events occurred in 5 (4.3%) low 3.37 (Ͻ130 mg/dL) for all 3 treat- baseline triglyceride values decreased of 116 patients in the control group and ment groups. by 0.12 mmol/L (11 mg/dL) from 3.24- in 5 (4.2%) of 119 patients in the 4-mg/d In the higher subgroup, the median mmol/L (287 mg/dL) in 41 patients in and 5 (4.4%) of 113 patients in the baseline LDL-C value increased by 0.07 the control group. In the rosiglitazone 8-mg/d rosiglitazone groups, none con- mmol/L (3 mg/dL) from 3.78-mmol/L groups, the baseline median triglycer- sidered related to study medication. (146 mg/dL) in 30 patients in the con- ide value increased by 0.15 mmol/L (13 Symptomatic mild or moderate hy- trol group. In the rosiglitazone groups, mg/dL) from 3.50 mmol/L (310 mg/dL) poglycemia was reported by 2 patients the median baseline LDL-C value in- in 43 patients taking 4-mg/d and de- in the control group and by 3 patients creased by 0.31 mmol/L (12 mg/dL) from creased by 0.72 mmol/L (64 mg/dL) in the 4-mg/d and by 5 patients in the 3.72 mmol/L (144-mg mg/dL) in 27 pa- from 3.16-mmol/L (280 mg/dL) in the 8-mg/d rosiglitazone groups. No patients taking 4-mg/d and by 0.34- 8-mg/d rosiglitazone group. tient required third-party interven-

Figure 4. Change in Fasting Plasma Glucose (FPG) Concentrations at Week 26 in Patients Figure 5. Mean Fasting Plasma Glucose Taking Metformin Hydrochloride and Rosiglitazone Maleate Compared With Patients Taking (FPG) Concentrations Over Time in Patients Metformin Alone Taking Metformin Hydrochloride Alone Compared With Patients Taking Metformin 20 +5.9 Comparison 20 Comparison With and Rosiglitazone Maleate P<.18 With Baseline Control Group at 26 Weeks Metformin and Placebo 0 0 Metformin and 4 mg/d of Rosiglitazone Metformin and 8 mg/d of Rosiglitazone 260 –20 –20 240

–40 –40 220

–33.0 in FPG, mg/dL P<.001 200 –39.8 –60 –60 –48.4 P<.001 180 < FPG, mg/dL

P .001 Group Control Mean Change From –52.9 P<.001 160

Mean Change From Baseline in FPG, mg/dL Mean Change From –80 –80 Metformin Metformin Metformin Metformin Metformin 140 and and and and and Placebo 4 mg/d of 8 mg/d of 4 mg/d of 8 mg/d of –7 –6–5 –4–2 04 812 18 26 Rosiglitazone Rosiglitazone Rosiglitazone Rosiglitazone Treatment Week

To convert from milligrams per deciliter to millimoles per liter multiply by .0555. Error bars indicate 95% con- To convert from milligrams per deciliter to millimoles fidence interval. per liter multiply by .0555. Error bars indicate SE. . ©2001 American Medical Association. All rights reserved. (Reprinted) JAMA, April 5, 2001—Vol 285, No. 13 1699

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

tion or hospitalization, but the metfor- Both rosiglitazone groups experi- decreases in hemoglobin levels were min dose was reduced from 2.5 g/d to enced small but statistically signifi- −5.0 g/L in the 4-mg/d and –8.0 g/L in 2.0 g/d in 2 patients. No one withdrew cant decreases in hemoglobin and the 8-mg/d rosiglitazone groups because of hypoglycemia, and there hematocrit levels, which occurred pri- (PϽ.0001 for both groups), and mean were no biochemically documented in- marily during the first 12 to 18 weeks decreases in hematocrit were –1.8% in stances of FPG levels of less than 2.78- of treatment, after which values for both the 4-mg/d and –2.5% in the 8-mg/d mmol (Ͻ50 mg/dL). parameters increased slightly. The mean rosiglitazone groups (PϽ.0001 for both groups). There were no significant changes in these parameters in the con- Table 2. Change in Lipid Parameters From Baseline at Week 26* trol group. One patient in each rosi- Metformin Hydrochloride Metformin and Rosiglitazone Maleate glitazone group withdrew because of Hydrochloride anemia, and 1 patient in the 4-mg/d rosi- Lipid Parameter and Placebo 4 mg/d 8 mg/d glitazone group with low hemoglobin Total cholesterol, mmol/L and hematocrit levels was withdrawn No. of patients 113 116 110 from the study after week 8 because of Baseline, mean (SD) 5.32 (1.02) 5.25 (0.92) 5.19 (1.22) Week 26, mean (SD) 5.50 (1.03) 5.98 (1.05) 6.01 (1.41) evidence of gastrointestinal tract bleed- Mean (SD) change from baseline† 0.18 (0.61) 0.72 (0.74) 0.82 (1.07) ing, considered by the investigator to P value‡ .0018 Ͻ.0001 Ͻ.0001 be unrelated to the study medication. Mean difference from placebo . . . 0.53 0.60 There were no significant changes P value§ . . . Ͻ.0001 Ͻ.0001 from baseline in vital signs or electro- HDL cholesterol, mmol/L cardiogram parameters in the rosi- No. of patients 112 116 110 glitazone groups compared with the Baseline, mean (SD) 1.14 (0.28) 1.18 (0.29) 1.20 (0.37) control group. Although infrequent, Week 26, mean (SD) 1.20 (0.29) 1.32 (0.34) 1.36 (0.42) edema was observed with greater fre- Mean (SD) change from baseline† 0.06 (0.14) 0.13 (0.19) 0.16 (0.24) quency in the rosiglitazone groups P value‡ Ͻ.0001 Ͻ.0001 Ͻ.0001 (2.5%, 4-mg/d; 3.5%, 8-mg/d) than in Mean difference from placebo . . . 0.08 0.10 the control group (0.9%). No one with- P value§ . . . .0002 .0002 drew due to edema. Total cholesterol–HDL ratio Those in the control group experi- No. of patients 112 116 110 enced a mean decrease in body mass of Baseline, mean (SD) 4.86 (1.37) 4.62 (1.13) 4.58 (1.40) 1.2 kg from baseline, but those in the Week 26, mean (SD) 4.79 (1.27) 4.80 (1.48) 4.77 (1.75) rosiglitazone groups experienced a mean Median change from baseline† −0.015 0.115 0.130 body mass increase of 0.7 kg in the P value‡ .73 .12 .18 4-mg/d and 1.9 kg in the 8-mg/d rosi- Median difference from placebo . . . 0.150 0.16 P value† . . . .13 .16 glitazone groups (P = .0001 for both LDL cholesterol, mmol/L groups). There were no significant dif- No. of patients 104 108 102 ferences in waist-to-hip ratios among Baseline, mean (SD) 3.03 (0.88) 2.99 (0.78) 2.91 (0.84) groups. Week 26, mean (SD) 3.13 (0.97) 3.46 (0.86) 3.45 (1.04) No one in the rosiglitazone groups Mean (SD) change from baseline† 0.10 (0.44) 0.46 (0.58) 0.53 (0.76) experienced elevations of alanine ami- P value‡ .02 Ͻ.0001 Ͻ.0001 notransferase (ALT) levels greater than Mean difference from placebo . . . 0.36 0.40 3 times the upper limit of the refer- P value§ . . . Ͻ.0001 Ͻ.0001 ence range. Mean changes in aspartate Triglycerides, mmol/L࿣ aminotransferase (AST), ALT, and to- No. of patients 113 116 110 tal bilirubin levels were similar in all Baseline, mean (SD) 2.77 (2.19) 2.54 (1.56) 2.57 (2.07) groups, with a slight decrease ob- Week 26, mean (SD) 2.78 (1.79) 2.62 (1.57) 2.57 (1.87) served in mean ALT (–1.9 U/L, con- Mean (SD) change from baseline† 0.008 (1.32) 0.08 (1.35) −0.003 (1.72) trol; –1.9 U/L, 4-mg/d rosiglitazone; P value‡ .95 .53 .98 –3.4 U/L, 8-mg/d rosiglitazone). Mean Mean difference from placebo . . . −0.06 −0.10 alkaline phosphatase decreased in all P value§ . . . .73 .56 groups (–3.5 U/L, control; –12.0 U/L, *To convert cholesterol levels from millimoles per liter to milligrams per deciliter divide by 0.0259. HDL indicates high- density lipoprotein; LDL, low-density lipoprotein; and ellipses, not applicable. Data are rounded. 4 mg/d rosiglitazone; –14.7 U/L, 8-mg/d †Calculated only for patients with both a baseline and a week 26 value. ‡Significance level is .05. rosiglitazone); the mean value for all §Significance level is .0259. groups was within the reference range. ࿣To convert triglyceride levels from millimoles per liter to milligrams per deciliter divide by 0.0113. Two patients in the control group were

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

noted to have liver function tests for po- Table 3. Change in Free Fatty Acid Levels at Week 26 (Compared With Baseline and tential clinical concern (Ͼ3 times the Metformin Hydrochloride and Placebo)* upper limit of the reference range) while Free Fatty Acids, mg/dL in treatment. Both completed the study Metformin and with elevated transaminase values. Rosiglitazone Maleate Metformin COMMENT and Placebo 4 mg/d 8 mg/d (n = 113) (n = 116) (n = 110) This is the first large, multicenter, clini- Baseline, mean (SD) 18.26 (7.75) 18.39 (7.56) 18.44 (8.00) cal trial demonstrating the efficacy and Week 26, mean (SD) 18.17 (6.06) 15.78 (6.05) 14.15 (6.13) safety of combined rosiglitazone and Change from baseline, mean (SD)† −0.09 (7.66) −2.61 (6.69) −4.30 (7.88) metformin treatment in patients with P value‡ .90 Ͻ.0001 Ͻ.0001 type 2 diabetes. The combination treat- Mean difference from control . . . −2.62 −4.22 ment of metformin and rosiglitazone sig- P value§ . . . .0003 Ͻ.0001 nificantly reduced HbA1c and FPG con- *Ellipses indicate not applicable †Calculated only for patients with both a baseline and a week 26 value. centrations, in a dose-ordered fashion ‡Significance level is .05. compared with baseline and with met- §Significance level is .0259. formin alone. Conversely, treatment with metformin was associated with significant increases in HbA1c concentra- tent with glucose-clamp studies using into subgroups revealed nonsignificant tions, indicating that these agents other thiazolidinedione drugs.23,24 increases in patients with baseline tri- complement each other to achieve op- The improvements in HOMA-B with glyceride levels lower than 2.26 mmol/L timal glycemic control and confirming metformin-rosiglitazone treatment (not (Ͻ200 mg/dL). Among patients in the the clinical utility of metformin in com- observed with metformin alone) were 8-mg/d rosiglitazone group whose base- bination with a thiazolidinedione drug.20 unexpected and introduce an impor- line was higher than 2.26 mmol/L (Ͼ200 Consistent with the mechanisms of tant potential therapeutic benefit of rosi- mg/dL), there was a significant statisti- action of metformin and rosiglitazone, glitazone. Although the exact mecha- cal decrease observed (64 mg/dL). The the reductions in FPG concentrations nism underlying this improvement clinical significance of lipid level changes were proportionately smaller than those remains to be determined, rosiglitazone- may be minimal, because lipid- 25 observed in HbA1c concentrations. mediated reductions in glucotoxicity lowering therapy may be often admin- Maximum doses of metformin decrease and lipotoxicity secondary to elevated istered to patients with diabetes irre- hepatic gluconeogenesis, which prin- concentrations of circulating FFA or spective of prior heart disease history.31,32 cipally affects FPG concentrations, both26,27 are candidate mechanisms by Elevated FFA may play a role in the whereas rosiglitazone enhances insu- which rosiglitazone may improve BCF. development of insulin resistance, be- lin sensitivity at the peripheral level and The effects of rosiglitazone on BCF and cause it is associated with increased he- affects overall glucose disposal, includ- insulin sensitivity are consistent with its patic glucose output33,34 and may con- ing postprandial excursions. Because the effects on long-term glycemic control and tribute to ␤-cell dysfunction via a lipo- relative contribution of postprandial suggest that it may possibly delay or pre- toxic effect.26,27 Elevated FFA has also glucose on glycemic control depends on vent disease progression. been linked to endothelial dysfunction the magnitude of FPG concentra- Despite significant increases in total and hypertension35,36 and enhanced tions,21 rosiglitazone may have an effect cholesterol, HDL-C, and LDL-C with the platelet aggregation and coagulation,37,38 on postprandial hyperglycemia, as dem- metformin-rosiglitazone treatments, the which may increase cardiovascular risk. onstrated directly in a rosiglitazone trial total cholesterol–HDL-C ratio, which did Thereforemetformin-rosiglitazonetreat- that showed significant improvements not change significantly, may be a bet- ment was significantly more effective in in fasting and postprandial glucose con- ter predictor of cardiovascular out- lowering FFA than the metformin alone. centrations and excursions.22 come than either total cholesterol or The weight gain observed in those re- The complementary actions of com- HDL-C levels alone.28-30 Since this study ceiving metformin-rosiglitazone treat- bined metformin and rosiglitazone is was not designed to assess long-term ment may be attributed to increased further supported by the effects of rosi- lipid effects, the long-term significance adipocyte differentiation,39,40 fluid reten- glitazone on insulin sensitivity de- of these changes is unknown; however, tion,39,41 or increased appetite.42 De- spite maximum doses of metformin. patients with baseline plasma LDL-C lev- spite weight increases, no significant dif- Rosiglitazone may provide added thera- els lower than 3.37 mmol/L (Ͻ130 ferences in waist-to-hip ratio among peutic value by reducing peripheral in- mg/dL) remained less than that level after groups were observed, suggesting that sulin resistance. While HOMA-S is an therapy. No significant changes in tri- rosiglitazone treatment leads to in- indirect method for determining insu- glyceride levels were noted in any treat- creased energy storage in subcutaneous lin sensitivity, these results are consis- ment group, and segregation of patients adipose sites that are not associated with

Downloaded From: https://jamanetwork.com/ on 09/28/2021 COMBINATION THERAPY IN DIABETES

43 increased cardiovascular risk. The small REFERENCES patients with type 2 diabetes. Diabetologia. In press. 23. Yamasaki Y, Kawamori R, Wasada T, et al. Pio- decreases in hemoglobin and hemato- 1. DeFronzo R. Lilly Lecture 1987: the triumvirate: glitazone (AD-4833) ameliorates insulin resistance in crit levels associated with metformin- ␤-cell, muscle, liver. Diabetes. 1988;37:667-687. patients with NIDDM. Tohoku J Exp Med. 1997;183: 2. The Expert Committee on the Diagnosis and Clas- 173-183. rosiglitazone therapy may relate to sification of Diabetes Mellitus. Report of the Expert Com- 24. Mimura K, Umeda F, Hiramatsu S, et al. Effects plasma volume expansion derived from mittee on the Diagnosis and Classification of Diabetes of a new oral hypoglycaemic agent (CS-045) on meta- 44 Mellitus. Diabetes Care. 1999;22(suppl 1):S5-S19. bolic abnormalities and insulin resistance in type 2 dia- fluid retention and hemodilution. 3. Turner RC, Cull CA, Frighi V, Holman RR, for the betes. Diabet Med. 1994;11:685-691. Metformin-rosiglitazone therapy may UK Prospective Diabetes Study (UKPDS) Group. Gly- 25. Marshak S, Leibowitz G, Bertuzzi F, et al. Im- cemic control with diet, sulfonylurea, metformin, or paired beta-cell functions induced by chronic expo- be a safe alternative therapy to attain insulin in patients with type 2 diabetes mellitus. JAMA. sure of cultured human pancreatic islets to high glu- optimal glycemic control where mono- 1999;281:2005-2012. cose. Diabetes. 1999;48:1230-1236. therapy has failed because the statis- 4. Lehmann JM, Moore LB, Smith-Oliver TA. An an- 26. Paolisso G, Howard BV. Role of non-esterified fatty tidiabetic thiazolidinedione is a high affinity ligand for acids in the pathogenesis of type 2 diabetes mellitus. tically significant decreases in lactate peroxisome proliferator-activated receptor gamma. Diabet Med. 1998;15:360-366. levels associated with metformin- J Biol Chem. 1995;270:12953-12956. 27. McGarry JD, Dobbins RL. Fatty acids, lipotoxicity 5. Saltiel AR, Olefsky JM. Thiazolidinediones in the and insulin secretion. Diabetologia. 1999;42:128-138. rosiglitazone treatment indicate that treatment of insulin resistance and type 2 diabetes. 28. Rubins HB, Robins SJ, Collins D, et al. Gemfibro- rosiglitazone may correct metabolic ab- Diabetes. 1996;45:1661-1669. zil for the secondary prevention of coronary heart dis- 6. Smith SA, Cawthorne MA, Coyle PJ, et al. BRL ease in men with low levels of high-density lipopro- normalities beyond reducing hypergly- 49653C normalizes glycemic control in Zucker fatty fa/fa tein cholesterol. N Engl J Med. 1999;341:410-418. cemia, and further suggest differing and rats by improving hepatic and peripheral tissue sensi- 29. Criqui MH, Golomb BA. Epidemiologic aspects of tivity to insulin. Diabetologia. 1993;36(suppl 1):A184. lipid abnormalities. Am J Med. 1998;105:48S-57S. complementary actions of metformin 7. Consoli A, Nurihan N, Capani F, Gerich J. Predomi- 30. Kinosian B, Glick H, Garland G. Cholesterol and and rosiglitazone; and ALT elevations nant role of gluconeogenesis in increased hepatic glu- coronary heart disease. Ann Intern Med. 1994;121: cose production in NIDDM. Diabetes. 1989;38:550- 641-647. greater than 3 times the upper limit of 557. 31. Haffner SM, Lehto S, Ronnemaa T, Pyo¨ra¨la¨K, the reference range were not observed 8. Stumvoll M, Nurjhan N, Perriello G. Metabolic ef- Laakso M. Mortality from coronary heart disease in in either of the rosiglitazone groups. fects of metformin in NIDDM. N Engl J Med. 1995; subjects with type 2 diabetes and in nondiabetic sub- 333:550-554. jects with and without prior myocardial infarction. In summary, combination metformin- 9. Cusi K, Consoli A, DeFronzo RA. Metabolic ef- N Engl J Med. 1998;339:229-234. rosiglitazone treatment is effective and fects of metformin on glucose and lactate metabo- 32. Pyo¨ra¨la¨ K, Pedersen TR, Kjekshus J, Faergeman lism in noninsulin-dependent diabetes mellitus. J Clin O, Olsson AG, Thorgeirsson G. Cholesterol lowering safeinreducinghyperglycemiainpatients Endocrinol Metab. 1996;81:4059-4067. with simvastatin improves prognosis of diabetic pa- with type 2 diabetes. In patients whose 10. Bailey CJ, Turner RC. Drug therapy: metformin. tients with coronary heart disease. Diabetes Care. 1997; N Engl J Med. 1996;334:574-579. 20:614-620. fundamental abnormality is insulin re- 11. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, 33. Hermann LS, Melander A. Biguanides: basic as- sistance, such a combination raises the Treacher DF, Turner RC. Homeostasis model assess- pects and clinical uses. In: Alberti KGMM, DeFronzo ment: insulin resistance and ␤-cell function from fast- RA, eds. International Textbook of Diabetes Melli- exciting possibility of treating diabetes ing plasma glucose and insulin concentrations in man. tus. Vol 1. Chichester, England: John Wiley & Sons Inc; by targeting the underlying cause of the Diabetologia. 1985;28:412-419. 1992:773-795. 12. National Diabetes Data Group. Classification and 34. Bergman RN, Mittelman SD. Central role of the disease, rather than the traditional ap- diagnosis of diabetes mellitus and other categories of adipocyte in insulin resistance. J Basic Clin Physiol Phar- proach of stimulating insulin secretion. glucose intolerance. Diabetes. 1979;28:1039-1057. macol. 1998;9:205-221. Nearly 30% of patients taking the com- 13. Friedewald WT, Levy RI, Fredrickson DS. Estima- 35. Steinberg HO, Tershoby M, Monestel R, et al. El- tion of the low-density lipoprotein cholesterol in plasma evated circulating free fatty acid levels impair endo- bination therapy achieved HbA1c levels without use of the preparative ultracentrifuge. Clin thelium-dependent vasodilation. J Clin Invest. 1997; of 7% or less. This level of glycemic con- Chem. 1972;18:499-502. 100:1230-1239. 14. Emoto M, Nishizawa Y, Maekawa K, et al. Ho- 36. Fagot-Campagna A, Balkau B, Simon D, et al. High trol is 3-fold greater than what was meostasis model assessment as a clinical index of in- free fatty acid concentration: an independent risk fac- achieved among those taking metformin sulin resistance in type 2 diabetic patients treated with tor for hypertension in the Paris Prospective Study. Int sulfonylureas. Diabetes Care. 1999;22:818-822. J Epidemiol. 1998;27:808-813. alone. Additional investigation is needed 15. Hermans MP, Levy JC, Morris RJ, Turner RC. Com- 37. Mitropoulos KA, Miller GJ, Watts GF, Durrington to determine whether this combination parison of insulin sensitivity tests across a range of glu- PN. Lipolysis of triglyceride-rich lipoproteins activates co- cose tolerance from normal to diabetes. Diabetolo- agulant factor XII: a study in familial lipoprotein-lipase will alter the long-term risk of cardiovas- gia. 1999;42:678-687. deficiency. Atherosclerosis. 1992;95:119-125. culardiseaseordelaydiseaseprogression. 16. Matsuda M, DeFronzo RA. Insulin sensitivity in- 38. Mikhailidis DP, Mahadevaiah S, Hutton RA, Geor- dices obtained from oral glucose tolerance testing: com- giadis E, Cramp DG, Ginsburg J. Plasma non- parison with the euglycemic insulin clamp. Diabetes esterified fatty acids and platelet aggregation. Thromb Author Affiliations: Department of Medicine, Endocri- Care. 1999;22:1462-1470. Res. 1983;32:641-643. nology Section, Tulane University, New Orleans, La (Dr 17. Hermans MP, Levy JC, Morris RJ, Turner RC. Com- 39. Day C. Thiazolidinediones: a new class of anti- Fonseca); Dallas Diabetes Center, Dallas, Tex (Dr Ro- parison of tests of ␤-cell function across a range of diabetic drugs. Diabet Med. 1999;16:179-192. senstock); Departments of Biometrics (Dr Patwardhan) glucose tolerance from normal to diabetes. Diabetes. 40. Hallakou S, Doare L, Foufelle F, et al. Pioglitazone and Clinical Research and Development (Dr Salzman), 1999;48:1779-1786. induces in vivo adipocyte differentiation in the obese SmithKline Beecham Pharmaceuticals, Collegeville, Pa. 18. United Kingdom Prospective Diabetes Study Zucker fa/fa rat. Diabetes. 1997;46:1393-1399. Financial Disclosure: Dr Rosenstock has received re- Group. United Kingdom Prospective Diabetes Study 41. Young MM, Squassante L, Wemer J, van Marle search grants from or has served as a consultant for or 24. Ann Intern Med. 1998;128:165-175. SP, Dogterom P, Johnkman JH. Troglitazone has no has been on the speaker’s bureau of Eli Lilly and Co, Novo 19. Haffner SM, Miettenen H, Stern MP. The ho- effect on red cell mass or other erythropoietic param- Nordisk Pharmaceuticals Inc, Hoechst Marion Roussel meostasis model in the San Antonio Heart Study. Dia- eters. Eur J Clin Pharmacol. 1999;55:101-104. Pharmaceuticals Inc, Bristol-Myers Squibb Co, Smith- betes Care. 1997;20:1087-1092. 42. Shimizu H, Tsuchiya T, Sato N, Shimomura Y, Ko- Kline Beecham Pharmaceuticals, Novartis, Parke-Davis, 20. Inzucchi SE, Maggs DG, Spollett GR, et al. Effi- bayashi I, Mori M. Troglitazone reduces plasma lep- Bayer Takeda, Roche Laboratories, Astra Zeneca, and cacy and metabolic effects of metformin and tro- tin concentration but increases hunger in NIDDM pa- Johnson & Johnson. He holds stock in Eli Lilly and Co, glitazone in type II diabetes mellitus. N Engl J Med. tients. Diabetes Care. 1998;21:1470-1474. SmithKline Beecham Pharmaceuticals, and Inhale. 1998;338:867-872. 43. Seidell JC, Hautvast JG, Deurenberg P. Over- Funding/Support: Support for this clinical trial was re- 21. Mooradian AD, Thurman JE. Drug therapy of post- weight: fat distribution and health risks. Infusions- ceived from SmithKline Beecham Pharmaceuticals. prandial hyperglycaemia. Drugs. 1999;57:19-29. therapie. 1989;16:276-281. Acknowledgment: We thank Sylvia K. Chai, PhD, for 22. Raskin P, Rappaport EB, Cole ST, Yan Y, Patward- 44. Dogterom P, Jonkman JHG, Vallance SE. Rosi- her significant contributions to the preparation and re- han R, Freed MI. Rosiglitazone short-term mono- glitazone: no effect on erythropoiesis or premature red view of the article. therapy lowers fasting and post-prandial glucose in cell destruction. Diabetes. 1999;48(suppl 1):A98.

Downloaded From: https://jamanetwork.com/ on 09/28/2021 LETTERS

Figure. Dose-Related Changes in Oral-Cecal Transit Times

180

150

120

90 Oral-Cecal Transit Time, min Oral-Cecal Transit 60

30 Placebo Methylnaltrexone, Placebo Methylnaltrexone, Placebo Methylnaltrexone, 0.3 mg/kg 1.0 mg/kg 3.0 mg/kg

Changes in individual oral-cecal transit times of 12 patients receiving long-term methadone treatment after placebo and 3 oral methylnaltrexone doses (4 patients in each dose group). Squares represent mean values.

methylnaltrexone levels. Mean (SD [range]) peak plasma level antagonizes loperamide-induced delay of orocecal transit. Dig Dis Sci. 1987;32: 829-832. for the other 4 patients (1 from the 1.0 mg/kg group and 3 from 5. Sykes NP. Oral naloxone in opioid-associated constipation. Lancet. 1991;337: the 3.0 mg/kg group) was 17.8 (6.6 [10-26]) ng/mL. 1475. Comment. Tertiary opioid antagonists, such as naloxone, cross the blood-brain barrier and reverse both the pain-relieving ben- efits and the adverse effects of opiates. Although oral naloxone CORRECTIONS

may relieve opioid-induced constipation, the therapeutic index Incorrect Wording and Footnote Symbol: In The Rational Clinical Examination en- is very narrow,5 and naloxone may induce opioid withdrawal titled “Does This Patient Have Carpal Tunnel Syndrome?” published in the June symptoms. Many patients receiving opioid pain medications face 21, 2000, issue of THE JOURNAL (2000;283:3110-3117), there was incorrect wording in the abstract. In the Conclusion paragraph on page 3110, the first sentence a difficult choice between burdensome adverse effects or inef- should have read, “Hand symptom diagrams, hypalgesia, and thumb abduction fective analgesia. Methylnaltrexone may allow for more aggres- strength testing are helpful in establishing the electrodiagnosis of CTS.” Also, in Table 2 on pages 3114 and 3115, the “No. of Hands” columns should include sive use of opioid analgesics with fewer adverse effects. The low dagger symbols (†) instead of asterisks (*) down the column to indicate which stud- methylnaltrexone plasma levels observed in our study suggest that ies used subjects instead of individual hands. Incorrect Wording and Data Presentation and Omitted Acknowledgment: In the this charged compound acts directly in the gut. Oral methylnal- Original Contribution entitled “Effect of Metformin and Rosiglitazone Combina- trexone has potential clinical utility in managing opioid- tion Therapy in Patients With Type 2 Diabetes Mellitus: A Randomized Controlled induced constipation with minimal adverse effects. Trial” published in the April 5, 2000, issue of THE JOURNAL (2000;283:1695- 1702), incorrect wording and incorrect data presentation were printed. On page Chun-Su Yuan, MD, PhD 1695, in the “Results” section of the Abstract, the sentence that read “28.1% achieved a glycosylated hemoglobin of less than 7%” should have read “7% or Joseph F. Foss, MD less.” On page 1698, the last sentence in the “Glycemic Control” section should Departments of Anesthesia and Critical Care have read “Nine patients (7.9%) in the control group, 25 (21.6%) in the 4-mg/d, University of Chicago and 33 (30.0%) in the 8-mg/d rosiglitazone groups achieved FPG concentrations Chicago, Ill of less than 7.8 mmol/L (140 mg/dL).” On page 1699 in the “Other Metabolic Effects” section, in the penultimate paragraph, which reports triglyceride find- Funding/Support: This work was supported in part by a grant from the Interna- ings, the phrase that read “in the rosiglitazone groups, the median baseline tri- tional Anesthesia Research Society, National Institutes of Health grants R0I CA79042 glyceride value increased...by0.07 mmol/L (6 mg/dL) from 1.34-mmol/L (119 and M01 RR00055. Methylnaltrexone was originally formulated and subse- mg/dL) in 55 patients taking 8-mg/d,” should have read “decreased by 0.72 mmol/L quently modified by faculty at the University of Chicago. The University of Chi- (64 mg/dL) from 9.16 mmol/L (280 mg/dL) in 37 patients taking 8 mg/d.” The cago and Drs Yuan and Foss stand to benefit financially from the further devel- last sentence of the penultimate paragraph of the “Other Metabolic Effects” sec- opment of methylnaltrexone. tion was repeated from the prior paragraph and should be deleted. In Table 2, the “Total cholesterol–HDL ratio” section should not have been converted to mmol/L. 1. Levy MH. Pharmacologic treatment of cancer pain. N Engl J Med. 1996;335: To calculate the proper ratio, divide the values in that section by 0.0259. In the 1124-1131. footnote of Table 2, the cholesterol conversion factor should have read “0.0259.” 2. Glare P, Lickiss JN. Unrecognized constipation in patients with advanced can- On page 1701 in the “Comment” section in the third column, the line that read cer: a recipe for therapeutic disaster. J Pain Symptom Manage. 1992;7:369-371. “Among patients in the 8-mg/d rosiglitazone group. . . there was a significant sta- 3. Yuan CS, Foss JF, O’Connor M, et al. Methylnaltrexone for reversal of consti- tistical decrease observed (6.4-mg/dL)” should have read “(64 mg/dL).” In ad- pation due to chronic methadone use. JAMA. 2000;283:367-372. dition, Sylvia K. Chai, PhD, should have been included in the acknowledgment for 4. Basilisco G, Camboni G, Bozzani A, Paravicini M, Bianchi PA. Oral naloxone her significant contributions to the preparation and review of the article.

Downloaded From: https://jamanetwork.com/ on 09/28/2021