The most fundamental clinical measure of inflammation is the white blood cell count. Even within the reference interval, higher total leukocyte counts precede and predict the incident risk of type 2 diabetes (1) and coronary heart disease (2), primarily because of the granulocyte subpopulation, rather than the lymphocyte and monocyte subpopulations. More specifically, high concentrations of circulating cytokines, particularly C-reactive protein (CRP)⁴ and interleukin-6 (IL-6), have been associated with the development of type 2 diabetes (3, 4) and cardiovascular disease (5). Indeed, a variety of circulating proinflammatory cytokines and acute-phase reactants are increased in obesity, the metabolic syndrome, hypertension, nonalcoholic steatosis, polycystic ovarian syndrome, type 2 diabetes, and cardiovascular disease. Mechanistically, inflammation marks an acquired cause of both insulin resistance and impaired insulin secretion [reviewed in (6, 7)]. Together, these data raise the question of whether inflammation can be targeted to treat or reduce disease risk, given that the target is mild and chronic low-grade inflammation and is not overt local or systemic acute infections.

At the cellular and molecular level, nuclear factor κB (NF-κB) is the inflammation master switch that controls the synthesis of many proteins critical for the activation and maintenance of the inflamed state. The hypothesis is that obesity stimulates NF-κB activity and additional stress pathways in adipose tissue, liver, and leukocytes, thereby promoting insulin resistance. The first firm evidence to support inflammation, not only as a marker but also as a mediator of disease, was provided more than 17 years ago. Tumor necrosis factor α (TNF-α), a proinflammatory cytokine, was shown to be produced by adipose tissue and to promote insulin resistance. In preclinical rodent studies, blockade of TNF-α was shown to improve insulin resistance, a result that raised hopes that TNF-α blockade would also work in humans (8). In the meantime, antibodies against TNF-α and related TNF-α–binding proteins, such as etanercept and infliximab, were developed for the treatment of rheumatoid arthritis, inflammatory bowel disease, and other inflammatory conditions. Despite their great efficacy in treating these inflammatory conditions, these antibodies have proved unsuccessful in treating obesity-related inflammation and type 2 diabetes (9, 10), perhaps because obesity-induced inflammation in humans is a more complex process that involves cytokines and chemokines other than TNF-α, such as adiponectin, leptin, IL-6, resistin, monocyte chemoattractant protein 1, plasminogen activator inhibitor 1, angiotensinogen, visfatin, retinol-binding protein 4, serum amyloid A, and many others. Although leptin and adiponectin are true adipokines that appear to be produced exclusively by adipocytes, other adipokines are also produced in activated macrophages and/or other cells; however, the relative amounts of each adipokine produced by the adipocyte and associated adipose tissue macrophages remain unknown (Fig. 1). In addition, obesity-mediated subacute chronic inflammation involves organs other than adipose tissue, such as the liver or endothelium, that may have important roles in the pathogenesis of diabetes and atherosclerotic processes. At the tissue level, multiple cellular stresses can modulate both phosphorylation and transcriptional events via activation of c-Jun N-terminal kinases and NF-κB, respectively, to regulate enzymatic functions and the concentrations of multiple cytokines, chemokines, and cellular receptors [reviewed in (6)]. In addition to c-Jun N-terminal kinase signaling, these cytokines and their signaling pathways may be novel targets of pharmacologic therapies, or at least they may be clinically useful as biomarkers of disease risk (11, 12).

Epidemiology-based studies suggest that factors besides obesity, including periodontal disease [reviewed in (13)], gut flora (14), and even exposure to air pollutants (15), may contribute to chronic inflammation and therefore may warrant alternative approaches to treatment.

The known antiinflammatory drugs are generally separated into the steroid-based glucocorticoids, the nonsteroidal antiinflammatory drugs, and the more potent immunosuppressive drugs, such as cyclosporine and methotrexate. Glucocorticoids and most immunosuppressive drugs in current clinical practice lack glucose-lowering efficacy and have many long-term deleterious consequences. Similarly ineffective at insulin sensitization and glucose lowering are the commonly used nonsteroidal antiinflammatory drugs, including the nonselective cyclo-oxygenase (COX) inhibitors (e.g., ibuprofen, naproxen) and the selective COX2 inhibitors (e.g., celecoxib).

A recent series of studies suggests, however, that the nonacetylated salicylates, an older class of antiinflammatory drugs infrequently used today, show promise as potential new treatments for diabetes. Two case reports published well over a century ago demonstrated that sodium salicylate lowered blood glucose (30, 31) No subsequent studies were conducted, however, and this finding was lost in the old medical literature. The effect was rediscovered when high doses of aspirin used for rheumatoid arthritis were noted to lower the fasting glucose concentration in patients with type 2 diabetes (32), although the molecular mechanism underlying the glucose-lowering effects remained unknown. At about this time, sulfonylureas, with their better safety and tolerance profiles, became available as the first oral antidiabetes agents, and the effects of salicylates appeared to have been largely forgotten.

Subsequent preclinical studies identified target tissues and cell types for both NF-κB–mediated insulin resistance and salicylate’s ability to reverse obesity-induced insulin resistance. The tissues commonly considered to have important roles in the development of type 2 diabetes include skeletal muscle, liver, fat, and pancreatic β cells. Muscle is a primary site of glucose disposal and utilization. The liver is a major site of gluconeogenesis, the dysregulation of which in type 2 diabetes contributes to fasting hyperglycemia. Adipose tissue is the primary energy depot and is thought to secrete substances that promote insulin resistance. The β cells of the pancreas produce insulin, the prime regulator of glucose homeostasis. Modern genetic methods of tissue-specific gene activation and inhibition have demonstrated that NF-κB activity is modulated in each of these tissues.

Clinical trials with the prodrug salsalate, which is marketed in the US for the treatment of joint pain, have recently been conducted. Salsalate has fewer side effects than aspirin or salicylate and is inexpensive to manufacture. Pilot clinical trials to test the efficacy of salsalate were conducted with small groups of patients for periods of 2 to 4 weeks. Administration of salsalate to patients with type 2 diabetes or to obese prediabetic patients reduced the blood concentrations of glucose, triglycerides, and nonesterified fatty acids (41, 42). Adiponectin, a potentially cardioprotective protein produced by adipose tissue, was increased by up to 45%, a result consistent with insulin sensitization. NF-κB–binding activity was reduced (41, 43). Higher salsalate doses increased insulin secretion (41), a result that has been observed in other studies of salicylates (44, 45). In humans, reduced metabolic clearance of insulin leads to increased circulating insulin concentrations, which may also participate in glycemic improvement (40, 41). These pilot studies observed no seriously detrimental side effects, although some participants who took higher doses experienced tinnitus (ringing in the ears), a well-known reversible accompaniment to salicylate therapy. Likewise, short-term administration of high salsalate doses also improves endothelial function, a biomarker of early atherosclerosis, in obese patients (43).

The results of these small proof-of-concept studies were sufficiently encouraging that the NIH funded larger clinical trials for both type 2 diabetes and cardiovascular disease. The TINSAL-T2D (Targeting Inflammation Using Salsalate in Type 2 Diabetes) study was conducted as a multicenter, double-masked, multidose randomized trial of placebo or salsalate (3.0, 3.5, or 4.0 g/day in 3 doses) administered for 14 weeks as an add-on to current therapy in patients with type 2 diabetes. All 3 salsalate doses lowered glycohemoglobin concentrations (46). Each dose also significantly decreased fasting blood glucose and triglyceride concentrations and increased adiponectin concentrations. Considerations of renal safety will require further evaluation because the investigators found a small increase in urinary albumin. Although the number of patients studied and the trial duration were insufficient to warrant recommending the use of salsalate for type 2 diabetes at this time, these findings support the search for additional antiinflammatory strategies for treating patients with type 2 diabetes. Fully recruited trials involving more patients and longer exposure times are now under way (http://clinicaltrials.gov/ct2/show/NCT00799643). Given the common soil shared by obesity, chronic subacute inflammation, diabetes, and cardiovascular disease, trials are also under way to examine the effects of targeting inflammation via the effects of salsalate on endothelial function and on the coronary artery plaque burden, as assessed with a new imaging modality (multidetector computed tomography angiography), in patients with atherosclerosis (http://clinicaltrials.gov/ct2/show/NCT00624923). Depending on the results of these ongoing investigations, a large simple trial of this inexpensive approach may be warranted.