Diabetes Mellitus: A Fundamental and Clinical Text
3rd Edition

Prior to the “molecular revolution” many pharmaceutical agents were identified empirically using a combination of traditional medicinal chemistry and in vivo pharmacology. In the early 1960s this approach to drug discovery yielded a prototypical lipid-lowering agent, the fibric acid derivative p-chlorophenoxyisobutyrate (clofibrate) (1,2), which served as the starting point for a structure activity relationship exercise that has lasted nearly four decades (3). Early simple modifications to the basic fibrate structure gave rise to pharmaceutically superior compounds such as fenofibrate, bezafibrate, gemfibrozil, and ciprofibrate (Fig. 78.1).

Although individual compounds display unique qualities, as a class the fibrates possess similar adverse and beneficial activities. With regard to the former, chronic use of these agents in rodents stimulates peroxisome proliferation, hepatomegaly, and later stage hepatic carcinogenicity (4). Although alterations in liver enzymes have been noted with clinical use of some fibrates, such changes are typically transient in nature and have not been linked to any clear hepatic injury (5), suggesting that the toxicologic profile for these agents varies greatly from species to species (6).

The fibrates all possess similar beneficial effects on lipid metabolism and other aspects of the atherogenic process via multiple mechanisms of action (7,8,9,10). In addition to their antiatherosclerotic activities, the fibrates have shown a modest utility for improving glucose homeostasis in diabetic patients (11). This important observation set the stage for an entirely new class of antidiabetic agent, the TZDs or glitazones.

In early attempts to discover additional fibrate-like antidyslipidemic agents, researchers at Takeda (Osaka, Japan) identified the compound AL-294, which displayed not only potent lipid-lowering activities, but also significantly lowered glucose levels in obese diabetic KK mice (12). Because the active species of the chloro ester was believed to be the corresponding acid, initial modifications to this compound focused on discovery of a suitable acid head group. AL-321, which utilized a thiazolidine-2,4-dione head group, exhibited a slightly improved profile and became the starting point for the antidiabetic TZDs. After additional modifications to the tail portion of the molecule, aimed at improving potency and reducing toxicity, Takeda publicly introduced the first candidate TZD, 5-[4-(1-methylcyclohexylmethoxy)benzyl]thiazolidine-2,4-dione (ciglitazone) in 1982 (13,14) (Fig. 78.1). In the years following the disclosure of ciglitazone, a large effort was undertaken by multiple groups to identify more potent, well-tolerated analogues of the TZDs as possible treatments for type 2 diabetes (15,16). Troglitazone, pioglitazone, and rosiglitazone (Fig. 78.1), however, stand out as
the only three TZDs to progress through the regulatory approval process to clinical practice.

At the time of their discovery, the TZDs represented not only a novel chemical class of antidiabetic agent, but also pioneered a unique mode of action in addressing a primary causative and underlying feature of diabetes, insulin resistance. Whereas individual TZDs display a variety of differences, as a class they all possess a similar ability to reduce hyperglycemia and dyslipidemia in a variety of animal models without stimulating insulin secretion (17). As a result of their ability to alter multiple aspects of metabolism, including insulin sensitivity, hyperglycemia, and dyslipidemia without the risk of hypoglycemia, the TZDs possess ideal pharmacologies for the diabetic phenotype. Despite intensive investigation, however, the mechanisms through which these agents exert their many effects are still being unveiled today.

Despite decades of animal research, clinical experience with the fibrates, and emergence of the TZDs as a novel therapeutic approach in development, the molecular target of these agents remained unknown until 1990, when Issemann and Green cloned a novel member of the steroid hormone receptor superfamily that they termed peroxisome proliferator–activated receptor (18). Several years later the Xenopus laevis orthologue of PPAR and two additional novel highly homologous genes were cloned (19). As a result of the overall similarity among the mouse PPAR receptor and three Xenopus-derived receptors, the nomenclature PPAR-α, β, and γ was adopted. It later became apparent that the term PPAR was a misnomer for the β and γ subtypes because they are not primary receptors for peroxisome proliferating agents nor do they induce peroxisome proliferation upon activation. Shortly after the report of Dreyer et al. (19), mammalian orthologues of PPAR-β and γ were identified by several groups (20,21,22). Whereas PPAR-α and γ are highly conserved across species, PPAR-β is quite divergent, and thus the mammalian nomenclature utilizes the terminology PPAR-δ rather than PPAR-β.

The three mammalian PPAR subtypes share a similar general structure and molecular mechanism of action (23,24) (Fig. 78.2). Like several other nuclear hormone receptors, the functional PPAR exists in a heterodimer complex with the 9-cis retinoic acid receptor (RXR) (25,26). The permissive nature of the PPAR-RXR heterodimer allows the receptor complex to be activated by ligand binding to either obligate partner. X-ray crystallography studies revealed the nature of PPAR-γ–RXR dimerization and demonstrated how ligands for either receptor stabilize the complex in an active state allowing for interaction with a series of cofactors that possess no specific DNA binding affinity, but are nonetheless critical elements in the transcriptional process (27). The activated heterodimer complex interacts with a peroxisome proliferator response element (PPRE) located within the regulatory region of select genes. Specifically, the DNA binding domains of the heterodimer complex bind a direct response-1 (DR-1) motif, which consists of two AGGTCA
sequences separated by a single nucleotide (23,24,26,27).

With the discovery of the PPAR family, a large effort was initiated to identify novel highly potent subtype selective ligands (Table 78.1). Such ligands display not only potential therapeutic utility, but also provide a means for unraveling the biologic functions of the different PPAR subtypes through a process described previously as reverse endocrinology (27).

The PPAR-α receptor is highly expressed in tissues displaying enhanced capacities for fatty acid catabolism such as liver, heart, skeletal muscle, brown adipose tissue, and kidney (28). An array of naturally occurring saturated and unsaturated fatty acids bind and activate PPAR-α with micromolar affinities under experimental conditions (28,29,30). Although fatty acids are routinely present in the plasma at micromolar concentrations, it is unclear if intracellular levels of this magnitude are achieved. The fatty acid derivative eicosanoids such as 8(S)hydroxyeicosapentaenoic acid possess submicromolar affinities for PPAR-α; however, these molecules are typically not generated in the prominent PPAR-α–responsive tissues. Because no appropriate high-affinity naturally occurring ligand for PPAR-α has been identified, it has been suggested that PPAR-α may function as a general lipid sensor capable of responding to local shifts in fatty acid concentrations (23).

Consistent with the need for large clinical doses, early characterization studies determined that the fibrates, in general, possess low affinity for PPAR-α and are mildly subtype selective, at best (31). As a result, many groups have recently sought to identify PPAR-α ligands with enhanced potency and improved selectivity. GlaxoWellcome researchers identified a ureidofibrate series with significantly improved potency for PPAR-α relative to the early fibrates, but lacking subtype selectivity. Optimization of the ureidofibrates yielded a related ureidothioisobutyric acid compound (GW9578), which is highly potent and selective for PPARα. Consistent with the in vitro profile, GW9578 displayed potent lipid-lowering activity in rodents (32). Similarly, Merck researchers identified a tetrazole series possessing properties of enhanced PPAR-α potency, but relatively low subtype selectivity (33). This series was manipulated to provide a range of potent compounds with different subtype selectivities. Compound-12 (33) exemplifies orders of magnitude greater potency than the fibrates for PPAR-α with approximately 10-fold selectivity over PPAR-γ. Scientists at Kyorin (Tochigi, Japan) embarked on a similar approach by utilizing the nonsubtype selective PPAR ligand KRP-297 as the basis for a structural optimization exercise that identified a series of potent and selective phenylpropanoic acid derivatives exemplified by KCL1998001079 (34). At Eli Lilly and Company (Indianapolis, IN, USA), a potent triazolone series of PPAR-α–selective compounds was identified through the screening of directed libraries (35). Optimization of this series yielded the compound LY518674, which despite significantly lower potency for mouse PPAR-α, relative to human PPAR-α, lowered plasma triglycerides with a half-maximal efficacious dose (ED₅₀) of 0.1 ± 0.04 mg/kg and elevated high-density lipoprotein cholesterol (HDL-C) by 50% at a dose of 0.3 ± 0.1 mg/kg in the human apo A-I transgenic mouse.

Whereas dyslipidemia has been the primary end point for PPAR-α activators, additional potential therapeutic value for these agents continues to emerge (36,37,38,39,40). Cell adhesions and vascular inflammation are critical aspects of atherosclerotic development, and PPAR-α ligands have been shown to down-regulate several key mediators of these processes, including vascular cell adhesion molecule-1 (VCAM-1), nuclear factor κB (NFκB), interleukin-6, and tumor necrosis factor-α (TNF-α). As noted earlier, fibrates have also shown a mild ability to improve glucose tolerance. In support of these findings, PPAR-α agonists have been shown to improve insulin sensitivity in both rodent and nonhuman primate models of obesity and insulin resistance (41,42) leading to speculation that new-generation
potent selective PPAR-α ligands may be beneficial for the treatment of metabolic syndrome (41). The pharmacologic impact of highly potent PPAR-α–selective ligands will clearly be an area of intense interest for all metabolic diseases as new-generation molecules advance into clinical development.

As a result of differential promoter usage and alternative splicing, the PPAR-γ gene is expressed as two isoforms, PPAR-γ1 and PPAR-γ2 (23,24,31). The PPAR-γ2 isoform, which possesses an additional 30 amino acids at the N-terminus, is almost exclusively expressed in adipose tissue. The PPAR-γ1 isoform displays a more heterogenous expression pattern with measurable levels in heart, skeletal muscle, intestine, kidney, pancreas, and spleen, in addition to adipose tissue. Similar to PPAR-α, PPAR-γ is activated by an array of naturally occurring fatty acids when examined under experimental conditions (30,43). Distinct from the other PPAR subtypes, however, PPAR-γ preferentially binds polyunsaturated fatty acids. In addition, several fatty acid–derived eicosanoids, such as 9-HODE and 13-HODE, have been shown to bind PPAR-γ in the micromolar range. The most widely studied naturally occurring PPAR-γ ligand, however, is the prostaglandin metabolite 15-deoxy-Δ12,14-prostaglandin J_2.

Although the TZDs were the first compounds identified as high-affinity PPAR-γ ligands (44), they were optimized empirically via rodent in vivo pharmacology. Newer generation PPAR-γ ligands have been discovered and optimized through use of the cloned and expressed human PPAR-γ receptor (45). Utilizing this new methodology, researchers at GlaxoWellcome (Research Triangle Park, NC) identified a series of tyrosine-based PPAR-γ agonists exemplified by the compound GI262570 (farglitazar), which possesses subnanomolar affinity for human PPAR-γ and displays potent glucose- and lipid-lowering activity in diabetic rodent models (46). Similarly, the group at Merck (Rahway, NJ) identified a highly potent phenylacetic acid derivative, L-796449, although this compound is approximately equipotent on all three human PPAR subtypes (47). Through manipulation of the central linear tether, however, Merck researchers were able to devise a
general method for establishing selective PPAR-γ agonists as exemplified by compound-8 (33). Examples such as these, and other recently identified TZD and non-TZD PPAR-γ ligands, have refined understanding of PPAR-γ both in terms of basic ligand-receptor interaction as well as physiologic function and therapeutic utility.

The TZDs and non-TZD PPAR-γ agonists improve insulin sensitivity via multiple mechanisms (48,49). Although alterations have been noted in many tissues, there is little doubt that adipose tissue plays a major role in the overall effects of these agents. At a whole-body level, activation of PPAR-γ promotes a general redistribution of fatty acids away from skeletal muscle, liver, and the central adipose stores toward a peripheral subcutaneous storage location (50). This stealing of fatty acids away from skeletal muscle and liver relieves the fatty acid–induced insulin resistance in these tissues (51,52). At the cellular level, PPAR-γ activation stimulates the remodeling process by inducing the differentiation of preadipocytes into nascent, insulin-sensitive, fat cells (24). In addition, it has been noted that PPAR-γ agonists induce apoptosis in mature lipid-laden non–insulin-sensitive fat cells, suggesting the continual nature of this remodeling process (50).

Within the fat cell, PPAR-γ agonists alter the expression of numerous genes involved in energy homeostasis, with most playing key roles in lipid uptake, storage, and metabolism (23,24,31). Fatty acid uptake into adipocytes is augmented in response to PPAR-γ activation as a result of increased expression of lipoprotein lipase (LPL) and the fatty acid transporters CD36 and FATP-1. Several other genes intimately involved in the synthesis of triglycerides within the fat cell, such as aP2, PEPCK, and acyl-CoA synthetase, are also upregulated in response to PPAR-γ activation. Additional genes upregulated as a result of PPAR-γ activation include the mitochondrial uncoupling proteins (UCP1,2, and 3), c-CBL–associated protein (which is component of the insulin-signaling pathway), and the insulin receptor substrate-2 (IRS-2). In contrast to the positively influenced genes, the expression of 11-β hydroxysteroid dehydrogenase 1, which mediates the formation of intracellular cortisol from cortisone, is inhibited in response to PPAR-γ activation (23).

PPAR-γ agonists also regulate the expression of several secreted proteins that influence growth, lipid metabolism, and insulin sensitivity in a variety of tissues (23,24,31). Levels of the angiogenic protein, vascular endothelial growth factor (VEGF) (53), and adiponectin (ACRP-30) (23), an adipocyte secreted protein that appears to sensitize tissues to insulin, are augmented in response to PPAR-γ activation. In contrast, expression of the proinflammatory cytokine TNF-α (23), which has been linked to insulin resistance in a variety of tissues, as well as the ob gene product leptin (23) and the recently identified novel hormone resistin (54), are all reduced in response to PPAR-γ activation.

In addition to the systemic effects of secreted proteins, PPAR-γ influences a number of other processes within nonadipose tissues. In skeletal muscle, PPAR-γ agonists suppress the expression of PDK-4, thereby enhancing the activity of the pyruvate dehydrogenase complex and allowing for increased flux of glucose-derived intermediates through the oxidative catabolism pathway (27). In endothelial cells, PPAR-γ agonists upregulate the expression of endothelial nitric oxide synthase, allowing for increased NO production and resultant local vasodilation (55). This may partially explain the ability of PPAR-γ agonists to mildly reduce blood pressure in both animals and humans.

As a result of effects on numerous tissues and cellular processes, the therapeutic potential and utility of the PPAR-γ agonists has extended well beyond the metabolic diseases (23,31). PPAR-γ agonists have been shown to reduce certain mediators of vascular inflammation, albeit at exposure levels inconsistent with PPAR-γ affinity. Contrastingly, rosiglitazone has demonstrated beneficial antiinflammatory activities in rodent models of colitis, which may be explained, at least in part, through its ability to inhibit the NFκB system at concentrations consistent with PPAR-γ affinity. As a result of PPAR-γ’s influence on the cell cycle, beneficial effects of PPAR-γ agonists have been observed in several different cancers, including liposarcoma, breast, and colon, although the effects on the latter remain controversial.

PPAR-δ possesses a ubiquitous tissue expression pattern, but similar to the other PPAR subtypes, PPAR-δ binds a range of fatty acids and eicosanoids under experimental conditions (29). The polyunsaturated fatty acids, such as dihomo-γ-linolenic acid, eicosapentaenoic acid, and arachidonic acid, as well palmitic acid and its analogue 2-bromopalmitic acid, bind PPAR-δ with micromolar affinities. PGA₁, PGD₂, and the partially synthetic carbaprostacyclin also bind PPAR-δ with affinities in the micromolar range.

Of the three PPAR subtypes, PPAR-δ is the least understood, in part from a relative lack of selective high-affinity ligands. Several ligands, however, have been investigated as a result of their PPAR-δ activity. Merck (Rahway, NJ) researchers identified several series of PPAR-δ ligands that were either weakly active or nonisoform selective (33). L-165041 was reported to be the most potent and selective PPAR-δ ligand, with nanomolar affinity for human PPAR-δ and approximately 10-fold selectivity (47). Contrastingly, L-165041 is only weakly active on murine PPAR-δ and much less selective. Recently, GlaxoSmithKline (Research Triangle Park, NC) reported on a breakthrough highly potent and selective PPAR-δ compound with their identification of GW501516, which possesses single-digit nanomolar affinity for PPAR-δ and approximately 1,000-fold selectivity over the other PPAR subtypes (56).

Although PPAR-δ has been implicated in a wide variety of processes, interest in this PPAR subtype as a target for metabolic disease was initiated with the finding that L-165041 elevated HDL-C levels in obese diabetic db/db mice (57). Potential therapeutic implications of targeting PPAR-δ were further realized with the finding that GW501516 substantially improved lipid profiles and fasting insulin levels in obese insulin-resistant nonhuman primates (56). At a mechanistic level, GW501516 was shown to induce the expression of the reverse cholesterol transporter ABC-A1 and stimulate apo A-I–meditated reverse cholesterol transport in human THP-1 macrophage cells.

Undoubtedly with the recent emergence of PPAR-δ as an additional target for metabolic abnormalities, additional potent and selective ligands will be brought forward in the near future.

The beneficial effects of PPAR-α and PPAR-γ ligands in atherosclerosis related to their direct vascular effects has been the subject of a number of recent reviews (40,69). From a clinical perspective, PPAR-α ligands such as fibrates have a lengthy history of clinical use for their hypolipidemic effects, and several clinical studies have evaluated the effects of these agents to ameliorate atherosclerosis and reduce cardiovascular events (69). Plasma triglycerides were reduced significantly using fibrates in these trials, and this was accompanied by increases in HDL-C levels. Most importantly, the drug treatments reduced the number of cardiac end points without any adverse effects on total mortality. A number of studies have evaluated the lipid effects of the two clinically available PPAR-γ ligands, pioglitazone and rosiglitazone. Reduction in FFA levels have been noted with both pioglitazone and rosiglitazone therapy. Pioglitazone therapy has led to consistent lowering of triglyceride levels, generally in the range of 9% to 20%, whereas rosiglitazone therapy has resulted in either no change or an increase in mean triglyceride levels (70). Both pioglitazone and rosiglitazone therapy have been associated with significant improvement in HDL-C (up to 20%). Pioglitazone therapy, when compared with placebo, did not have a statistically significant effect on low-density lipoprotein cholesterol (LDL-C) at any dose studied (15,30, or 45 mg/day); however, an increase compared with baseline was reported in one study (58,59). In contrast, rosiglitazone therapy significantly increased LDL-C levels compared with baseline or placebo uniformly, but significant reductions in the levels of small dense LDL-C have been reported (63,70). The different effects of the various TZDs on lipid metabolism need further investigation, but considering its central role in lipid metabolism, pharmacologic modulation of PPAR-γ activity may result in an overall improvement of the dyslipidemic phenotype. Reductions in blood pressure and other key atherogenic risk factors such as PAI-1 have also been noted following PPAR-γ treatment (70). Importantly, treatment of type 2 patients with PPAR-γ agonists has been associated with decreases in carotid intimal medial thickness, a clinical measure of atherosclerotic progression, and a significant reduction in neointimal tissue proliferation after coronary stent placement as assessed by intravascular ultrasonography of the coronary vessels (70).

As mentioned previously, the TZDs are the first class of antidiabetic agents to specifically address the underlying insulin resistance of type 2 diabetes. Utilizing variations of the hyperinsulinemic clamp technique, pioglitazone and rosiglitazone have produced significant improvements in glycemic control and in peripheral (muscle) and hepatic sensitivity to insulin in several clinical studies (71,72). Consistent with a primary effect in adipocytes, improvements in insulin sensitivity and glycemic control have been associated with weight gain in patients. Furthermore, the increase in body weight has been shown to positively correlate with the reduction in HbA_1C (73). Body weight increase with TZD treatment is characterized by a general remodeling of fat tissue in which subcutaneous preadipocytes differentiate into small mature fat cells and differentiated large adipocytes (hypertrophic adipocytes) in visceral or subcutaneous fat depots undergo apoptosis (50). To specifically determine whether the alterations in abdominal fat distribution after TZD treatment are related to the improvement in glycemic control or insulin sensitivity, Miyazaki et al. (74) recently evaluated the effect of pioglitazone on abdominal fat distribution and insulin sensitivity in type 2 diabetic patients. Pioglitazone treatment led to a significant decrease in visceral fat that was associated with improvements in both hepatic and peripheral tissue sensitivity to insulin.

Until recently, adipose tissue was considered to be a passive depot for storing excess energy. It is now clear that adipose tissue is an active endocrine organ elaborating a number of products that are potential regulators of glucose homeostasis or insulin resistance. Recently adiponectin, a novel adipose-specific protein with putative antiatherogenic and antiinflammatory effects, was discovered (75). In humans, obesity and type 2 diabetes are associated with low plasma adiponectin concentrations in different ethnic groups, and the degree of hypoadiponectinemia is more closely related to the degree of insulin resistance and hyperinsulinemia than to the degree of adiposity and glucose intolerance (76). In addition, adiponectin levels were decreased fivefold in patients with severe insulin resistance in association with dominant-negative PPAR-γ mutations (77). Administration of PPAR-γ agonists have increased adiponectin levels in both normal human subjects (77) and in type 2 diabetic patients (78,79). Thus, induction of adipose tissue adiponectin expression and consequent increases in circulating adiponectin levels may represent a novel mechanism for PPAR-γ–mediated enhancement of whole-body insulin sensitivity (77).

Contrastingly, the novel, cysteine-rich, adipocyte-specific secretory factor resistin has been proposed as a potential adipocyte-derived
mediator of the link between obesity and insulin resistance, and suppression of resistin expression by PPAR-γ agonists might explain the beneficial effects of these compounds in insulin-resistant states (54). Relatively low resistin gene expression, however, in human tissues has raised questions about the role of this gene in humans (80). Clearly, further studies of this novel protein are needed to clarify its role in human metabolism.

In addition to alterations in the adipocyte, treatment with PPAR-γ agonists has been associated with improvement in β-cell function as assessed by the homeostasis model assessment (59,65). Because of the ability of PPAR-γ agonists to ameliorate insulin resistance and potentially enhance β-cell function, there has been growing interest in their ability to prevent type 2 diabetes. Recently, Buchanan et al. demonstrated a greater than 50% reduction in the incidence of type 2 diabetes in young Hispanic women with recent gestational diabetes who were treated with a PPAR-γ agonist (81). Protection from diabetes persisted 8 months after study medications were stopped, was associated with the preservation of pancreatic β-cell function, and appeared to be mediated by a reduction in the secretory demands placed on β-cells by chronic insulin resistance.

Weight gain has been consistently reported following monotherapy with pioglitazone and rosiglitazone, generally in the range of 0.5 to 3 kg. Compared with monotherapy, weight gain is somewhat less when these agents are used in combination with metformin, but increased when used in combination with insulin. Weight gain appears to involve mostly peripheral subcutaneous sites, with a reduction in visceral fat depots, the latter being better correlated with insulin resistance.

The first TZD approved for clinical use, troglitazone, was withdrawn from the market because of an increased risk for idiosyncratic hepatic toxicity. Data from double-blind studies of pioglitazone and rosiglitazone in patients with type 2 diabetes suggest they do not adversely affect liver function. Most patients eligible for clinical trials had alanine aminotransferase (ALT) or aspartate aminotransferase (AST) values greater than 2.5 times the upper limit of normal (ULN). During therapy, the proportion of patients with ALT values greater than three times the ULN was similar with pioglitazone or rosiglitazone compared with placebo or active comparator. However, isolated cases of severe hepatotoxicity have been reported with both pioglitazone and rosiglitazone (82,83). Periodic monitoring of liver function tests is recommended by the U.S. Food and Drug Administration. TZDs should not be given to patients with signs of serious hepatic dysfunction.

Pioglitazone and rosiglitazone are also associated with increased fluid retention, and an increased incidence of edema was reported in clinical trials. The highest incidence of edema was reported when these agents were used in combination with insulin. Case reports of pulmonary and peripheral edema associated with rosiglitazone and pioglitazone have been published (84,85). Because of this plasma volume expansion, TZDs are not recommended for patients with heart failure (New York Heart Association class III or IV). The decreases in hemoglobin and hematocrit noted in clinical trials is felt to be related to plasma volume expansion.

TZDs should not be administered to women who are pregnant or breast-feeding since it is not known whether these drugs have teratogenic effects or are secreted in human breast milk.