This is the second in a series of articles on presentations at the American Diabetes Association Annual Meeting, San Diego, California, 10–14 June 2005.
Alan Cherrington (Nashville, TN) gave the President’s Address at the American Diabetes Association (ADA) Annual Meeting, reviewing some of the progress, and lack thereof, occurring over the past 2 decades. In 1985, there were 6.4 million adults with diagnosed diabetes. The National Institutes of Health invested <$200 million in diabetes research. Human insulin had just become available, and the only oral agents were the sulfonylureas. Home blood glucose monitoring and insulin pump technology were in their “infancy.” Pancreas transplantation had met with little success, with a 25% surgical mortality. Only 30% of patients were insulin independent for ≥1 year, and islet transplants were unsuccessful. The genetic bases of type 1 and type 2 diabetes were unknown, triggers for disease onset were unknown, mechanisms of insulin action were not understood, diabetes prevention was nonexistent, and the team approach to diabetes treatment had just begun.
Now there has been an alarming increase in diabetes across the country, with the U.S. in 2003 having 13.8 million people with diagnosed and 5 million with undiagnosed diabetes, as well as 41 million pre-diabetic individuals. The prevalence of diabetes increased virtually every year since 1990, due to population changes, particularly with an increase in diabetes in minorities, increased awareness and detection of diabetes, a change in diagnostic criteria, a reduction in macro- and microvascular complications with attendant increase in life expectancy of diabetes, further increasing the number of individuals and, most important, a tremendous increase in obesity, causing insulin resistance and hence diabetes. The cost of the disease was $92 billion, and the investment in research was somewhat under $800 million in 2002; Cherrington pointed out that spending ∼1% on research is far below the level needed to address this important public health issue.
The Diabetes Control and Complications Trial, U.K. Prospective Diabetes Study, Diabetes Prevention Program, and numerous further studies have increased our understanding of diabetes treatment, with the pharmaceutical industry developing new therapeutic approaches, including insulin analogs, metformin, thiazolidinediones (TZDs), α glucosidase inhibitors, nonsulfonylurea secretagogues, and agents to facilitate weight loss, as well as drugs addressing blood pressure and lipid abnormalities. Insulin pumps have improved, continuous glucose monitoring devices are being developed, and “we will soon be able to close the loop” with integrated pumps and continuous glucose monitors. Pancreas transplantation has evolved to a stable approach, and islet transplantation has proven feasible, with promising research in islet/β-cell replacements, perhaps using embryonic or adult stem cell sources, gene transfer to turn cells already existing in the body into glucose-sensing and -secreting cells, and xenotransplantation using encapsulation technology. Tissue-selective insulin analogs may become important for type 1 diabetes, with inhaled, oral, and buccal insulin preparations rapidly coming closer to feasibility. For type 2 diabetes, glucagon-like peptide-1 analogs and dipeptidyl peptidase-IV inhibitors, adipokine modulators, canabinoid receptor antagonists, appetite regulators, and gluconeogenic and glycogenolytic regulators may offer important new approaches. Finally, the role of nonphysician health care providers has become greater, with there now being almost 15,000 certified diabetes educators in the U.S., although to realize the full value of these approaches, research into health outcomes will be necessary.
There is, however, no evidence that the quality of care of diabetes has improved. The population’s BMI is increasing, and there has been no improvement among people with diabetes in HbA1c (A1C) levels. Many diabetic individuals receive substandard glycemic care, although blood pressure and cholesterol levels have improved (1). Problems with quality of care are present across a wide array of diseases, so that, Cherrington noted, only 45–50% of people with diabetes receive recommended care. There are only ∼5,000 adult and 1,000 pediatric endocrinologists in the U.S., with general practitioners treating the majority of people with diabetes, although these providers only adhere to 20–80% of recommended treatment approaches (2). The incidence of diabetes could realistically be decreased by ∼20% if lifestyle interventions were applied to subjects with impaired glucose tolerance (IGT) (3), suggesting the importance of prevention. Yet current predictions suggest that by the year 2030 there will be 23 million individuals with diagnosed and 7 million with undiagnosed diabetes and that an additional 70 million people will have impaired fasting or postprandial glucose (4). Direct diabetes costs will be ∼$175 billion/year, with an additional $75 billion/year in indirect doses, and “the economic and personal burden of diabetes will be almost overwhelming.” Cherrington suggested four approaches to avoiding this outcome, by continuing to invest in research, by adopting a chronic care treatment model, by focusing on early treatment and prevention, and by finding a way to limit obesity. He pointed out the need to abandon the acute care model, which is not effective for diabetes, and moving to a patient-centered approach, in which “patients and their families, not health care professionals, are responsible for the management of the disease.” The chronic care model, he suggested, allows development of “productive interactions.” This focus on population-based approaches must be incorporated into the teaching of health care professionals, and it is necessary to create incentives for good performance, with focus on prevention and early treatment. Furthermore, Cherrington stated, one must decrease the prevalence of obesity. Too often, physicians fail to advise obese individuals to diet and exercise (5, 6) or promote the use of healthy foods. Schools need to restore physical education programs and remove vending machine foods. A separate organization is being formed by the ADA to address these issues, the Association for Weight Management and Obesity Prevention. Quoting Goethe, Cherrington concluded, “Knowing is not enough: we must apply; willing is not enough: we must do.”
Early identification of type 2 diabetes
A number of research presentations at the ADA meeting addressed aspects of the development of type 2 diabetes. Nichols and Brown (abstract 117) studied 28,335 individuals in the Kaiser Permanente Northwest Region with fasting glucose <100, 100–109, and 110–125 mg/dl on two occasions, finding annual health costs of $4,357, $4,580, and $4,960/person, respectively, with annual costs of $3,799 for people with normal fasting glucose that did not subsequently worsen. Pharmacy, outpatient, and inpatient costs increased similarly across the three groups, suggesting validity to the lowering of the fasting glucose criterion from 110 to 100 mg/dl in the U.S. Kang et al. (abstract 1019) studied 1,166 nondiabetic Koreans aged 30–65 years, 10% of whom had developed diabetes by 3-year follow-up. The fasting glucose was as sensitive as the 2-h glucose for predicting diabetes, with receiver operator characteristics curve analysis showing that 103.5 mg/dl gave the best combination of sensitivity and specificity. Dong et al. (abstract 1023), however, reported differing conclusions from glucose tolerance testing in 2,033 subjects in Qingdao, China. The lowering of the fasting glucose criterion for impaired fasting glucose (IFG) and IGT from >110 to >100 mg/dl increased the prevalence of IFG from 11 to 28% and that of IGT from 29 to 52%, although the new criteria decreased the proportion of IFG with IGT from 22 to 17% and resulted in lower prevalences of overweight, obesity, hypertension, hypertriglyceridemia, low HDL cholesterol, and cigarette smoking among subjects with IFG than were obtained using the current World Health Organization criteria. The authors suggest the need for prospective studies to determine whether the criteria are appropriate in the Chinese population. Vaccaro et al. (abstract 1069) similarly reported that among 43,114 nondiabetic Italian telephone company employees aged 35–65 years, lowering the diagnostic threshold for IFG increased its prevalence from 12 to 41%, with particularly great increase among women <45 years, from 4 to 22% of the studied population, so that we must be cautious in generally applying the new criterion.
The fasting glucose may not be optimal for assessment of risk in children. Libman et al. (abstract 1874) performed 2-h glucose tolerance tests on 40 overweight children (mean age 12 years), showing greater correlation of the 2-h than the fasting glucose with triglyceride, fasting insulin, HDL cholesterol, and adiponectin. Hirschler et al. (abstract 1869) studied 73, 41, and 53 children (mean age 7 years) with BMI above the 95th, between the 85th and 95th, and below the 85th percentile for age and sex. Using three or more of waist circumference at or above the 75th percentile, blood pressure above the 95th percentile, HDL below the 5th percentile, triglyceride above the 95th percentile, and glucose ≥100 mg/dl to define the metabolic syndrome, 19 met the criteria, showing mean BMI and waist circumference 2 and 2.6 SDs above normal versus 1.7 and 1.3 SDs above normal for overweight and obese children without metabolic syndrome. Triglyceride was 128 vs. 73 mg/dl, HDL cholesterol 35 vs. 52 mg/dl, systolic blood pressure 117 vs. 101 mmHg, adiponectin 22 vs. 29, and the homeostasis model assessment (HOMA) of insulin resistance (HOMA-IR) measure 2.6 vs. 1.5; the latter was the only independent risk factor in multivariate analysis. Hannon and Arslanian (abstract 209) performed hyperinsulinemic-euglycemic clamps on 35 overweight 14-year-old subjects. Comparing 18 with triglyceride—to—HDL cholesterol ratio<3 with 17 whose ratio was ≥3, there was no significant difference in BMI, but visceral fat was 53% greater, adiponectin 34% lower, and insulin sensitivity 59% lower in those with the higher ratio, suggesting this to be a useful measure of insulin resistance in adolescents, as has been shown in adults (7).
Williams et al. (abstract 367) presented data validating a decision tree approach based on data from the San Antonio Heart Study predicting individuals to be at risk of diabetes either with fasting glucose ≥100 mg/dl and HDL cholesterol <50 mg/dl or fasting glucose ≥95 mg/dl and a parent/sibling with diabetes. This approach was slightly more accurate in predicting diabetes risk than was the presence of IGT or of the Adult Treatment Panel III (ATP III) metabolic syndrome. Wilson et al. (abstract 368) presented a similar decision rule approach based on parental diabetes, obesity, and metabolic syndrome traits. Chen et al. (abstract 984) performed a factor analysis of BMI, waist circumference, systolic and diastolic blood pressure, triglycerides, HDL cholesterol, fasting and 2-h postchallenge plasma glucose, and the HOMA-IR in 4,572 subjects, aged 20–74 years, in an epidemiological survey in Shanghai. Four factors were identified in men: body size, glucose/insulin resistance, blood pressure, and lipids, with the glucose/insulin resistance and lipid components appearing as one factor in women, with the authors suggesting that the “metabolic syndrome” may actually represent a number of related but essentially separate processes.
Festa et al. (abstract 1011) reported the relationship between change in glucose tolerance and change in insulin secretion among 802 persons aged 40–69 years followed for 5.2 years, all showing reduction in insulin sensitivity, those with diabetes and with IGT starting from insulin sensitivity approximately half that of those with normal glucose tolerance (NGT). Those with NGT or IGT at baseline and NGT at follow-up had increase in insulin secretion, while worsening glucose tolerance or continued IGT or diabetes was associated with decreased insulin secretion, confirming the importance of both factors and confirming earlier studies suggesting that the development of diabetes occurs in those persons with insulin resistance whose insulin secretion worsens. Meigs et al. (abstract 365) reported that C-reactive protein (CRP) and, to a greater extent, plasminogen activator inhibitor-1 and von Willebrand factor predicted risk of development of diabetes among 2,924 Framingham offsprings, correcting for age; sex; physical activity; HDL level; smoking; parental history of diabetes; use of alcohol, nonsteroidal anti-inflammatory drugs, estrogen, or hypertension therapy; IFG/IGT, triglyceride; waist circumference; and insulin sensitivity, suggesting that markers of inflammation and endothelial dysfunction may be of pathogenic importance. Hansis-Diarte et al. (abstract 1064) reported that among 1,697 individuals <50 years of age, those whose parents both smoked cigarettes during their childhood had a 2.8-fold greater likelihood of having diabetes, adjusting for age, sex, ethnicity, socioeconomic status, and personal smoking status.
Several studies explored aspects of the relationship of diabetes development to obesity as well as the concept of the metabolic syndrome. In analysis of visceral and subcutaneous fat, BMI, and waist circumference as predictors of diabetes in 947 participants in the Diabetes Prevention Program (abstract 1002), each SD increase in visceral fat (roughly a one-third increase) was roughly associated with a one-third increase in likelihood of diabetes development. For subcutaneous fat, at the fourth to fifth lumbar vertebral level, only among men was there an association with diabetes risk, which was approximately equal to that for visceral fat. Each 12-cm increase in waist circumference was associated with a 53% increase in diabetes risk for men, but there was no significant association of either waist circumference or of BMI with diabetes risk for women. Geiss et al. (abstract 997) analyzed the 1997–2003 National Health Interview Surveys. The incidence of diabetes among U.S. adults aged 18–79 years increased from 4.9 to 6.9 · 1,000 people−1 · year−1, with incidences in 2003 of 16.8 per 1,000 individuals aged 65–79 years and of 18.3 per 1,000 obese individuals. In 2003, 59% of new cases were obese and an additional 30% overweight. In a study of 750 subjects without diabetes, Dabella et al. (abstract 726) found that coronary artery calcification correlated as well with increased waist circumference alone as with ATP III–defined metabolic syndrome, “suggesting no effect on coronary artery calcification of a risk factor grouping beyond that predicted by the cumulative effects of individual components.” Blackburn et al. (abstract 966) compared the ATP III metabolic syndrome with the hypertriglyceridemic waist phenotype, defined as having both waist circumference ≥80 cm and fasting triglyceride ≥133 mg/dl, in 254 women undergoing coronary angiography, showing similar coronary artery disease predictive power, suggesting this alternative (and potentially simpler) approach to allow identification of a similar group of individuals. Brixner et al. (abstract 1044) studied 9,394 subjects receiving prescriptions for antipsychotic drugs, showing that 1,514 had an increase in BMI to either ≥7% above baseline or ≥25 kg/m2. The likelihood of an increase in BMI was 1.39-fold greater with risperidone than with a conventional antipsychotic drugs, 1.36-fold greater with quetiapine, and 1.76-fold greater with olanzapine, although not increased with aripiprazole, ziprasidone, or clozapine.
Aso et al. (abstract 678) reported an association of plasma thrombin—activatable fibrinolysis inhibitor, a newly discovered inhibitor of fibrinolysis, with LDL cholesterol but not metabolic syndrome components, while plasminogen activator inhibitor-1 correlated with metabolic syndrome and related parameters, including BMI, triglyceride, alanine aminotransferase, HOMA-IR, and CRP. Kazaks et al. (abstract 700) studied 43 subjects with mild to moderate asthma and 26 control subjects with normal pulmonary function, showing a negative relationship between both forced vital capacity and the forced expiratory volume in 1 s and serum insulin in individuals with asthma but not in control subjects, while neither BMI nor CRP correlated with parameters of lung function, suggesting that asthma may be part of the extended insulin resistance syndrome. Kahn et al. (abstract 1045) studied 903 subjects with recently diagnosed diabetes, finding that although CRP levels increased with the number of metabolic syndrome components, this was explained by adjustment for BMI and not by adjustment for A1C, suggesting a “relationship … determined by body adiposity, not by glucose control.”
Peroxisome proliferator—activated receptor—directed therapy
The mechanism of action of therapeutic agents acting at peroxisome proliferator—activated receptor (PPAR) nuclear receptors, as well as concepts of the development of insulin resistance, were explored in several presentations at the ADA meeting. Graham et al. (abstract 17) studied mice not expressing and overexpressing GLUT4, showing reduced and increased insulin sensitivity, respectively, with the adipocyte secretory product retinol-binding protein (RBP)-4 increased and decreased in the two models. RBP-4 levels were decreased by rosiglitazone, and RBP-4 administration decreased insulin sensitivity, while mice not expressing RBP4 have increased insulin sensitivity, suggesting this to be an adipocyte factor involved in the development of insulin resistance.
In a study of the mechanism of TZD action in humans, Yu et al. (abstract 43) measured insulin sensitivity during lipid plus heparin infusion to raise free fatty acid levels, showing 36 and 33% reductions in six obese and six lean individuals without diabetes, with the latter having approximately twice as great baseline insulin sensitivity. After 12 weeks of 45 mg pioglitazone daily, baseline and lipid plus heparin—induced insulin resistance improved in the obese but not the lean group. Miyazaki et al. (abstract 455) administered 8 mg rosiglitazone daily for 3 months versus placebo to 29 type 2 diabetic subjects not receiving pharmacologic treatment. Fasting glucose decreased 45 mg/dl, body fat mass increased 3.1 kg, FFA decreased 156 mEq/l, tumor necrosis factor α decreased 0.79 pg/ml, adiponectin increased 9.9 mg/ml, and urine albumin decreased 8.7 μg/mg creatinine; in multivariate analysis, the latter correlated with the changes in tumor necrosis factor α and adiponectin, suggesting either a direct or an indirect effect of the TZD on renal function.
John Buse (Chapel Hill, NC) discussed the weight gain associated with TZD treatment, noting that these agents change fat cell biology and fat metabolism, as evidenced by the increase in small adipocytes in the Zucker diabetic fatty rat model. The higher the baseline weight, the greater the risk of development of diabetes, cardiovascular disease, and malignancy; therefore, weight gain has been considered a disincentive to TZD use. Humans with type 2 diabetes show enlargement of a number of fat depots within liver and muscle, intra-abdominal, and intravascular, as well as subcutaneous fat, with evidence that abnormal distribution of fat from subcutaneous depots leads to potentially harmful effects, and with the TZDs redistributing fat to sites where it lacks adverse effect.
Troglitazone studies in both monotherapy and combination therapy appeared to show that weight loss was associated with an increase in A1C and weight gain with a decrease in A1C, with the greatest weight gain seen in patients also treated with insulin or a sulfonylurea. The suggestion therefore was made that weight gain actually may be desirable with the class of agents. In a study comparing pioglitazone with rosiglitazone, however, both were associated with close to a 2-kg weight gain over 6 months, but this was not associated with the initial BMI and there was no relationship between weight gain and change in A1C or triglyceride or HDL cholesterol levels. Furthermore, Buse showed studies with the non-TZD insulin sensitizer MBX-102, a mixed PPARγ agonist/antagonist, in which there was no greater weight gain, fluid retention, or adipogenesis than seen with placebo, even in combination with insulin, leading to the suggestion that “perhaps weight gain is not a sine qua non of PPARγ agonists.” Rosenstock et al. (abstract 44) presented these data in greater detail, noting that MBX-102 is the single enantiomer of halofenate and reporting 3-month studies of 217 individuals with type 2 diabetes receiving insulin with or without oral agents. Comparing MBX-102 in a dose of 200 or 400 mg daily with placebo, weight gain was 0.3, 0.7, and 0.8 kg and edema was seen in 11, 6, and 16% of patients, respectively. A1C decreased 1% with the 400-mg dose, but 0.3% with placebo, and triglyceride decreased 15% versus a 23% increase, potentially suggesting that the agent is a better-tolerated PPARγ agonist than the TZDs.
Buse reviewed a study comparing 45 mg pioglitazone daily with 1 g metformin twice daily in stable overweight individuals with IGT and fasting blood glucose <110 mg/dl. Blood glucose did not change with metformin but decreased with pioglitazone. With metformin there was no change in subcutaneous or visceral fat or in the visceral-subcutaneous fat ratio, while with pioglitazone, subcutaneous fat increased, and visceral fat decreased slightly, leading to a substantial decrease in the visceral-subcutaneous fat ratio. Muscle biopsy showed a 40% decrease in intramyocellular fat with pioglitazone but not with metformin. In a study of subjects with nonalcoholic steatohepatitis treated with 4 mg rosiglitazone twice daily for 48 weeks, liver biopsy showed reduction in hepatic fat in 14 of 26 subjects (8). Thus, the TZDs are associated with weight gain but reduce liver, muscle, and visceral fat. Buse mentioned that the TZDs also might decrease β-cell lipids and reduce lipotoxicity, improving β-cell function.
Meta-analysis suggests a typical weight gain of 2.5–5 kg at 1 year when these agents are used clinically, with evidence of reduction in A1C and blood pressure, of lipid benefits, of reduction in inflammatory markers and coagulation factors, and of improvement in endothelial function and vascular biology, suggesting that these agents may be particularly beneficial, although Buse cautioned, “as of yet we do not have proof.” Furthermore, he pointed out that weight gain is seen with insulin, noting that among intensively treated subjects in the Diabetes Control and Complications Trial, the highest quartile of weight gain averaged 15 kg and that weight gain is also seen with sulfonylurea therapy, so that weight gain in association with A1C reduction cannot simply be considered a negative effect of this class of therapies. Nevertheless, it is desirable to minimize weight gain, and Buse suggested explaining to patients beginning treatment with a TZD that there is potential for weight gain. As weight gain is seen particularly in younger women with shorter duration of diabetes and higher baseline A1C, he recommended that lifestyle intervention be particularly considered for individuals with these characteristics. Explaining to patients the need to monitor weight at home as well as in clinic may be useful, and it may be appropriate to reduce TZD dosages for patients with more than a 4- to 5-lb weight gain. Other potential approaches include the use of a very-low-calorie diet on initiation of TZD treatment (9) and, perhaps, use of exenatide, pramlintide, or some of the newer weight loss medications. Buse also suggested that earlier use of TZDs might be another approach allowing limitation of weight gain.
Several studies reported at the ADA meeting addressed this topic. Nichols and Gomez (abstract 13) reported 12-month weight change after initiation of treatment in 9,546 individuals with type 2 diabetes in the Kaiser Permanente Northwest population, showing 3.9-lb weight gain with a sulfonylurea, 5.3-lb loss with metformin, 7.3-lb gain with insulin, and 10.4-lb gain with a TZD. Younger age, male sex, higher A1C, and use of selective serotonin reuptake inhibitor antidepressants predicted greater weight gain among TZD-treated subjects. Hollander et al. (abstract 12) randomized 568 insulin-treated type 2 diabetic subjects to 2 mg rosiglitazone once versus twice daily versus placebo for 24 weeks, with a decrease in A1C from 8.9 to 8.3%, from 9.0 to 8.2%, and from 9.1 to 8.7%, respectively. Edema was reported in 11 vs. 6 vs. 11%, and the frequency of hypoglycemia was similar in all groups, suggesting potential usefulness of very-low-dose treatment.
Several additional studies reported approaches to TZD treatment. Dabiri et al. (abstract 1904) randomized 195 type 2 diabetic children (age 14 years) to 2 mg rosiglitazone versus 500 mg metformin twice daily, with subsequent titration to 4 and 1,000 mg twice daily to achieve fasting glucose <126 mg/dl. A1C at screening was 8.2 vs. 8.1%, decreasing after 24 weeks to 7.5 vs. 7.2% in the drug-naïve subgroup. The rosiglitazone-treated group gained 3 kg, suggesting that metformin may be more advantageous in the treatment of children. Schöndorf et al. (abstract 606) administered placebo, 4 mg rosiglitazone, or 8 mg rosiglitazone in addition to glimepiride in 104 subjects with type 2 diabetes. Adiponectin increased 42, 72, and 100%, respectively, with improvement in insulin sensitivity only in the groups receiving rosiglitazone.
A number of studies compared the lipid effects of rosiglitazone and pioglitazone. Plotkin et al. (abstract 951) studied the effect of simvastatin in TZD-treated patients with type 2 diabetes. Of 24 vs. 27 receiving 45 mg pioglitazone daily, 40 mg simvastatin versus placebo led to a 37% decrease vs. a 2% increase in LDL cholesterol, a 12 vs. 5% decrease in triglyceride, and no significant difference in HDL cholesterol. Of 38 vs. 45 receiving 8 mg rosiglitazone daily, treatment with 40 mg simvastatin versus placebo similarly led to a 34% fall vs. 3% increase in LDL cholesterol, a 17% fall vs. 4% increase in triglyceride, and a 4% increase vs. 2% decrease in HDL cholesterol. Brackenridge et al. (abstract 957) treated 24 individuals with type 2 diabetes with 30 mg pioglitazone, 8 mg rosiglitazone, or placebo for 3 months, with decrease in A1C from 7.5 to 6.8% with pioglitazone and from 6.9 to 6.5% with rosiglitazone. FFA decreased similarly by 27% with both agents, and there was no change in LDL, HDL, or triglyceride levels. The VLDL apolipoprotein (apo)B secretion rate decreased 23% with rosiglitazone but did not change with pioglitazone, and the VLDL apoB catabolic rate was unchanged in all the groups. Deeg et al. (abstract 960) compared the effects of 4 mg rosiglitazone twice daily versus 45 mg pioglitazone daily on lipid subfractions in 325 vs. 333 individuals with type 2 diabetes, finding a 13 vs. 5% decrease in small LDL cholesterol levels, with pioglitazone decreasing triglyceride primarily via reduction in large VLDL.
King et al. (abstract 467) reported retrospective analysis of 18 type 2 diabetic subjects treated with pioglitazone plus nicotinic acid, with a 61% increase seen in HDL cholesterol and a 24% decrease in triglyceride in nine individuals characterized as having low HDL cholesterol, and moderate triglyceride elevation, while nine individuals with high HDL cholesterol and high triglyceride levels had 2 and 50% changes in the respective parameters; the authors commented that the magnitude of the HDL increase in the former group suggests the potential for additive benefit. Lautamäki et al. (abstract 498) studied 58 subjects with type 2 diabetes and coronary disease treated with 8 mg rosiglitazone daily versus placebo, showing an increase in LDL cholesterol from 89 to 104 mg/dl, but with an increase in large, buoyant LDL from 50 to 58 mg/dl, without a significant change in small, dense LDL, in average LDL particle size, or in HDL or triglyceride concentrations. Wulff et al. (abstract 978) reported studies of the PPARδ agonist GW501516 and a new selective, partial PPARδ agonist, NNC, showing glucose-lowering effects in mouse models of diabetes and pre-diabetes, with reduction in triglyceride and increase in HDL cholesterol levels.
A well-recognized dilemma is the evidence of an antiatherosclerotic benefit of PPARγ treatment but with simultaneous fluid retention that increases the risk of congestive heart failure. Collins et al. (abstract 733) showed similar atherosclerosis regression with either rosiglitazone or enalapril, and He et al. (abstract 738) observed decreased intimal lesion progression and matrix metalloproteinase expression with pioglitazone in mice not expressing the LDL receptor. In mice expressing human apoE3-Leiden, Havekes et al. (abstract 956) reported that with both fat and cholesterol feeding, tesaglitazar, a dual PPARα/γ agonist, decreased plasma cholesterol, with reduction in atherosclerotic lesion area even in comparison to that of mice whose cholesterol was lowered to the same extent with dietary change.
Forst et al. (abstract 86) and Hohberg et al. (abstract 605) treated 173 type 2 diabetic subjects with 45 mg pioglitazone versus 1–6 mg glimepiride daily for 6 months and showed a similar fall in A1C. Carotid IMT decreased with pioglitazone but not with glimepiride, with the degree of reduction correlating with improvement in HOMA of insulin sensitivity, decrease in intact proinsulin, and increase in adiponectin; these effects only occurred with sensitizer treatment. Evidence was presented that cardiovascular risk improvement with pioglitazone was independent of glucose lowering. Kim et al. (abstract 613) treated 210 type 2 diabetic subjects receiving sulfonylurea and/or metformin with 15 mg pioglitazone daily, showing that CRP and carotid IMT decreased over a mean of 8 months. Higher baseline CRP was associated with both greater anti-inflammatory and insulin-sensitizing effects of pioglitazone.
Wilding et al. (abstract 80) randomized 224 individuals with class 1–2 congestive heart failure to placebo versus 4–8 mg/day rosiglitazone. Nine versus 26% showed worsening dyspnea and 17 vs. 26% worsening edema, leading to increased diuretic requirement in 14 vs. 26% during the 52-week period of observation. The optimal approach for people with TZD-related edema remains to be determined. In an intriguing exploration, Karalliedde et al. (abstract 81) described 381 individuals treated with 4 mg rosiglitazone b.i.d. for 12 weeks with 260 people showing evidence of volume expansion as defined by hematocrit decrease of at least 0.5%, randomized to five approaches: continued rosiglitazone alone, discontinuation of rosiglitazone, or continuation of rosiglitazone with a 7-day period of treatment with 40 mg furosemide, 25 mg hydrochlorothiazide, or 50 mg spironolactone, with dosages doubled on day 2 if urine output failed to increase 30%. The hematocrit fell an additional 0.89% with continuation of rosiglitazone alone and by 0.12% with discontinuation of rosiglitazone, by 0.7% with furosemide, and by 0.02% with hydrochlorothiazide, while increasing 0.24% with spironolactone, with body weight decreasing 1.2 kg with spironolactone and 1.0 kg with hydrochlorothiazide The authors interpret the response to spironolactone (despite decreased plasma aldosterone before initiation of the diuretic) to suggest that PPARγ agonists activate aldosterone-sensitive distal nephron sodium transport.
There are important TZD effects on blood pressure. Pahler et al. (abstract 533) reported a prospective observational study of 1,393 hypertensive type 2 diabetic subjects treated with pioglitazone. At 20 weeks, 292 subjects not receiving hypotensive medications had a decrease in blood pressure from 143/85 to 138/81 mmHg, 727 on unchanged doses of hypotensive medications had a decrease from 148/86 to 140/83 mmHg, and 374 whose hypotensive medications were decreased nevertheless had a decrease in blood pressure from 149/86 to 143/82 mmHg. Home et al. (abstract 542) reported a 12-month study with 24-h ambulatory blood pressure measurement. Of 379 individuals receiving a sulfonylurea, treatment with rosiglitazone resulted in 3/3 mmHg lower blood pressure than did treatment with metformin, while of 380 individuals receiving metformin as baseline therapy, addition of rosiglitazone led to significantly lower blood pressure by 3/2 mmHg. Bakris et al. (abstract 543) randomized 389 type 2 diabetic subjects with microalbuminuria treated with metformin to 4 mg rosiglitazone versus 5 mg glyburide daily, showing 3/3 mmHg lower systolic/diastolic blood pressure and 16% lower albumin-to-creatinine ratio. Miura et al. (abstract 45) found that over 12 weeks, the angiotensin receptor blocker telmisartan increased adiponectin from 7.2 to 7.9 and decreased CRP from 1.2 to 0.9 mg/l in 30 individuals with type 2 diabetes and hypertension previously receiving a different angiotensin-directed antihypertensive agent. The fasting insulin decreased from 11 to 9, and HOMA-IR decreased from 3.5 to 2.6, without significant change in fasting glucose or A1C, supporting evidence of PPARγ agonist activity for this angiotensin receptor blocker.
Buffon et al. (abstract 655) studied 12 metformin-treated subjects with type 2 diabetes given pioglitazone 30 mg daily versus placebo, showing a 54 vs. 0% reduction in monocyte interleukin-6 production, a potential mechanism of anti-inflammatory effect. However, Basu et al. (abstract 707) treated 9 and 10 people with type 2 diabetes with pioglitazone and glyburide, respectively, reporting that neither agent reduced lipoprotein-associated phospholipase A2 or vascular cell adhesion molecule, pioglitazone reduced E-selectin ∼10%, and glyburide reduced intercellular cell adhesion molecule ∼12%, so that the TZD effects on inflammatory markers may be inconsistent.
In an important caution regarding TZD effects, Schwartz et al. (abstract 163) presented 4-year follow-up data of 666 subjects with diabetes, aged 70–79 years at baseline. Sixty-nine subjects reported TZD use, which was associated with a 0.76% annual reduction in bone mineral density at the hip, adjusted for age, race, baseline bone mineral density, weight, weight change, A1C, and use of calcium and vitamin D supplements, other hypoglycemic drugs, osteoporosis drugs, oral estrogen, oral steroids, and thiazides. Reversal of the anabolic effects of insulin on bone may be responsible.
PPARγ treatment of individuals without diabetes
Shadid et al. (abstract 975) treated 37 obese nondiabetic individuals with pioglitazone versus diet and exercise, with both interventions improving insulin sensitivity. Fat mass increased 1.4 vs. decreased 9.5 kg, with no change vs. a 40-mg/dl decrease in triglyceride, a 4- vs. 20-mg/dl decrease in LDL cholesterol, and 20 vs. 23% decrease in small LDL particle number, respectively, suggesting that lifestyle modification may be the preferred approach. Buchanan et al. (abstract 157) enrolled 89 women with prior gestational diabetes who had completed the Troglitazone in Prevention of Diabetes study of 45 mg pioglitazone daily, showing 4.6% annual diabetes incidence, comparable to that seen during troglitazone treatment. Independent diabetes predictors were greater baseline glycemic response to oral glucose, an expected marker of diabetes risk, and lesser reduction in insulin from enrollment to 1 year, a marker of insulin-sensitizing response to treatment. Samaha et al. (abstract 609) treated 60 nondiabetic individuals with metabolic syndrome and low HDL cholesterol with 8 mg rosiglitazone daily for 12 weeks, showing a 34% decrease in CRP but an increase in non-HDL cholesterol. Szapary et al. (abstract 963) randomized 60 nondiabetic subjects with HDL <40 mg/dl in men and <50 mg/dl in women and metabolic syndrome to 30 mg pioglitazone increasing to 45 mg daily versus placebo for 12 weeks. They found an HDL increase of 14 vs. 0% and a small LDL decrease of 18 vs. 0%, without significant weight gain or difference in triglyceride, LDL cholesterol, or apoB, although with an increase in non-HDL cholesterol of 5% vs. a fall of 2%. Lucidi et al. (abstract 84) treated 16 women with polycystic ovary syndrome with 45 mg pioglitazone daily for 12 months, showing improvement in insulin sensitivity and decrease in glucose and insulin levels during glucose tolerance testing, with reduction in testosterone, androstenedione, CRP, and triglyceride and an increase in LDL size.
PPARα agonist treatment
Altomonte et al. (abstracts 8 and 1375) reported that hamsters receiving a high-fructose diet showed stimulation of hepatic Foxo1 expression, correlating with increased hepatic apoC-III production, glucose intolerance, and hypertriglyceridemia. Administration of fenofibrate reduced triglyceride and glucose levels, decreased apoC-III, and reduced hepatic Foxo1 expression. In an animal model of hepatic Foxo1 overexpression, hepatic gluconeogenesis increased with IGT, increased liver fat, and decreased liver glycogen, and insulin receptor substrate-1 and -2 were downregulated. With a dominant-negative Foxo1 mutant, hepatic gluconeogenesis decreased and insulin sensitivity improved. The authors suggested that hepatic Foxo1 acts to lessen insulin responsiveness, that dysregulation of Foxo1 links decreased insulin response to increased apoC-III production and hypertriglyceridemia, and that part of the effect of fibrates (in this species) may involve a decrease in Foxo1 signaling. Armoni et al. (abstract 212) reported that Foxo1 decreased adipocyte PPARγ1 promoter activity in a dose-dependent fashion, potentially explaining some of its effect on insulin responsiveness.
Bajaj et al. (abstract 611) compared the effects of fenofibrate alone or in combination with pioglitazone in six individuals with type 2 diabetes, showing the former to lower triglycerides from 206 to 136 mg/dl, while addition of pioglitazone further lowered triglyceride to 88 mg/dl, also improving glycemic control, decreasing triglycerides, enhancing insulin sensitivity, and increasing muscle AMP kinase activity in association with an increase in plasma adiponectin. Asano et al. (abstract 965) studied 10 subjects with type 2 diabetes treated with 150 mg fenofibrate daily, showing a decrease in triglyceride from 250 to 160 mg/dl, increase in HDL cholesterol from 44 to 51 mg/dl, increase in A1C from 6.4 to 6.8%, and increase in plasma ubiquinol-10 from 687 to 850 nmol/l, suggesting a potential antioxidant benefit.
Côté et al. (abstract 976) measured adiponectin at month 18 in 1,928 men in the Veterans Affairs High Density Lipoprotein Cholesterol Intervention Trial with gemfibrozil versus placebo. There was the expected negative correlation of adiponectin with waist circumference, insulin level, and triglyceride level, positive correlation of adiponectin with HDL cholesterol, and association of both diabetes and IFG with reduction in adiponectin. Gemfibrozil had no effect on adiponectin, but the adipokines appeared as an important risk marker. Trial participants both with adiponectin below the 50th percentile and with insulin above the 75th percentile had a 42% increase in cardiovascular disease events.
Rosenson et al. (abstracts 590 and 979) administered 160 mg fenofibrate daily versus placebo to 45 nondiabetic individuals with triglyceride>150 mg/dl and at least two other ATP III metabolic syndrome criteria, reporting falls in intracellular adhesion molecule-1, vascular adhesion molecule-1, monocyte chemoattractant protein-1, and CRP with fenofibrate but not with placebo, suggesting an anti-inflammatory effect. Triglyceride decreased 43% and LDL cholesterol 6%, with a 15% increase in HDL cholesterol, a 45% reduction in postfat load triglyceride levels, and reduction in postload small LDL and LDL particle number.
PPAR α/γ
In the db/db mouse before development of diabetes, Tozzo et al. (abstract 269) reported that the dual PPAR α/γ agonist muraglitazar prevented development of diabetes, lowered circulating insulin levels, and prevented the loss of pancreatic insulin content seen in untreated animals after 12 weeks. Kwon et al. (abstract 501) reported studies of another dual PPARα/γ agonist, CKD-501, showing both glycemic and lipid effects in a diabetic rodent model.
David Kendall (Minneapolis, MN) discussed clinical studies of muraglitazar effects on glucose and lipids in type 2 diabetic patients. He discussed both a long-term phase 2 monotherapy trial and a 24-week active-control combination trial with metformin, with a 26-week extension. In a 104 week study of 88 subjects treated with 5 mg muraglitazar, A1C fell from 7.9 to 6.4% at 16 weeks, with this level maintained through the study period, with triglyceride levels decreasing 22% from a baseline mean of 156 mg/dl and HDL increasing 29% from a baseline of 40 mg/dl; the open-label study format, with a large number of subjects withdrawing, in part because of lack of efficacy, does lessen our assurance that the effects reported will generally apply to individuals receiving the treatment.
The phase 2 study compared 0.5–20 mg muraglitazar or 15 mg pioglitazone in 1,477 patients for 24 weeks, with additional 26-week treatment of 985 patients. At 24 weeks, patients treated with pioglitazone reduced A1C by an average of 0.6%, while A1C decreased up to 1.8% with the 10- and 20-mg muraglitazar doses. Muraglitazar decreased triglycerides 13–39%, increased HDL >20%, and led to modest decreases in non-HDL and LDL cholesterol over 104 weeks. The 15-mg pioglitazone dose, in contrast, decreased triglyceride by 12% and increased HDL by 18%. Body weight increased 5.86 kg with 5 mg, 8.94 kg with 10 mg, and 9.01 kg with 20 mg muraglitazar, while increasing 1.91 kg at 104 weeks with 15 mg pioglitazone. Kendall stated, “When you improve glycemic control [with PPARγ agonists], the price you pay is weight gain,” but the very great degree of weight gain with the 10- and 20-mg doses is a concern, and the drug will only be studied/developed at the 5-mg dosage level. Twenty-five percent of subjects treated with 5 mg muraglitazar and 30.8% of those receiving pioglitazone reported peripheral edema by week 104. Congestive heart failure, however, was seen in 3, 6, and 6 patients at the 5-, 10-, and 20-mg muraglitazar doses, respectively, and in none receiving pioglitazone; 8–11% of those treated with muraglitazar compared with 5% of those receiving pioglitazone withdrew from the long-term study.
Kendall described an active comparator study of 1,159 patients with type 2 diabetes receiving metformin at a dose of at least 1,500 mg (mean 1,850 mg) daily receiving 5 mg muraglitazar versus 30 mg pioglitazone daily for 24 weeks; the study was also presented by Rubin et al. (abstracts 14 and 2113). A1C decreased from 8.1 to 7.0 vs. 7.3%, triglyceride decreased 28 vs. 13%, HDL increased 19 vs. 14%, apoB decreased 12 vs. 6%, and non-HDL cholesterol decreased 6 vs. 1%. Of those with baseline triglyceride >150 mg/dl (mean 265), there was a 35 vs. 19% decrease in triglyceride. Edema occurred in 9 vs. 7%, weight gain was 1.4 vs. 0.6 kg, and heart failure—related events occurred in three versus one of the patients, respectively. Preliminary analysis of 50-week data for this study shows that A1C decreased 1.12 vs. 0.7%, triglyceride decreased 25% vs. 11%, and weight increased 2.5 vs. 1.5 kg. Frederich et al. (abstract 967) reported a 2-year study of 1.5 mg muraglitazar daily in 75 type 2 diabetic subjects, 5 mg muraglitazar daily in 108 type 2 diabetic subjects, and 15 mg pioglitazone daily in 65 type 2 diabetic subjects. Triglycerides decreased 13, 22, and 12%; non-HDL cholesterol decreased 11, 11, and 9%; and HDL cholesterol increased 17, 29, and 18%, respectively. It will, of course, be important to compare 5 mg muraglitazar with pioglitazone at the 45-mg dose, and such a study is planned.
Mohideen et al. (abstracts 518 and 968) presented results of a study of 583 people with type 2 diabetes and mean A1C ∼8.2% while receiving 15 mg glyburide daily, showing A1C to decrease 1 and 1.2% with 2.5 and 5 mg muraglitazar while increasing 0.2% with placebo over 24 weeks. Triglycerides decreased 14 and 26% and increased 3%, non-HDL cholesterol decreased 1 and 6% and increased 2%, and HDL cholesterol increased 7, 14, and 0%, respectively. Edema occurred in 9, 10, and 7%; weight gain in 3, 7, and 2%; and hypoglycemia in 15, 19, and 8%, respectively. Bristol-Myers Squibb and Merck are developing muraglitazar under the brand name of Pargluva. Note, however, that a recent report called into question the safety of this agent (10).
In a study of another dual PPAR α/γ agonist, Goldstein et al. (abstract 83) administered 0.1, 0.5, 1, 2, or 3 mg tesaglitazar or 45 mg pioglitazone daily for 12 weeks to 418 type 2 diabetic individuals with baseline A1C 7.1%, fasting glucose 170 mg/dl, and BMI 30.7 kg/m2. Glucose fell 9, 30, 41, 55, and 61 mg, respectively, with tesaglitazar and 39 mg/dl with pioglitazone; triglyceride fell 5, 17, 33, 41, 41, and 8%; HDL increased 1, 5, 15, 13, 13, and 6%; and non-HDL cholesterol decreased 3, 9, 13, 22, 25, and 2%. Edema developed in 4–7% of individuals receiving tesaglitazar, 4% of those with pioglitazone, and 3% of a placebo-treated group. Fagerberg et al. (abstract 614) and Schuster et al. (abstracts 615 and 972) presented results of a study of tesaglitazar 0.1, 0.25, 0.5, and 1 mg daily given for 12 weeks to insulin-resistant people without diabetes. ApoB decreased 1, 4, 5, and 12%; apoA-I increased 0, 0, 3, and 4%; and apoC-III decreased 5, 13, 18, and 25%. Postlipid meal triglyceride decreased 41% and insulin 31%, with 45 and 59% reductions in National Cholesterol Education Program—defined metabolic syndrome and IFG, respectively, in individuals receiving the 1-mg dose.