Diabetes mellitus and insulin resistance in patients with rheumatoid arthritis: Risk reduction in a chronic inflammatory disease

Authors

  • Mary Chester Wasko,

    Corresponding author
    1. University of Pittsburgh, Pittsburgh, Pennsylvania
    • University of Pittsburgh School of Nursing, Room 453B, 440 Victoria Street, Pittsburgh, PA 15261
    Search for more papers by this author
    • Dr. Wasko has received consultancy fees (less than $10,000 each) from UCB and Centocor.

  • Jonathan Kay,

    1. UMass Memorial Medical Center and University of Massachusetts Medical School, Worcester
    Search for more papers by this author
    • Dr. Kay has received consultancy fees (less than $10,000 each) from Array Biopharma, Bristol-Myers Squibb, Celgene, Centocor, Eisai, Genentech, Roche, Sanofi-Aventis, UCB, and Wyeth.

  • Elizabeth C. Hsia,

    1. Centocor Research and Development, Malvern, Pennsylvania and University of Pennsylvania, Philadelphia
    Search for more papers by this author
    • Drs. Hsia and Rahman own stock and/or hold stock options in Johnson & Johnson.

  • Mahboob U. Rahman

    1. Centocor Research and Development, Malvern, Pennsylvania and University of Pennsylvania, Philadelphia
    Search for more papers by this author
    • Drs. Hsia and Rahman own stock and/or hold stock options in Johnson & Johnson.


Abstract

Objective

To perform a systematic literature review of the potential association among molecular markers of inflammation, alterations in body composition, and insulin resistance (IR), a precursor to type 2 diabetes mellitus (DM), in rheumatoid arthritis (RA) patients. To determine the impact of tumor necrosis factor α (TNFα) as a pivotal proinflammatory cytokine in the pathophysiology of type 2 DM and RA, and the effect of antirheumatic drugs on glycemic control.

Methods

We performed a search of PubMed to identify articles on IR and body habitus in patients with RA.

Results

Patients with RA had characteristics placing them at high risk for IR and type 2 DM. The incidence and prevalence of type 2 DM in RA was not clearly increased compared with the general population; however, studies suggested that patients with RA are likely to have IR and have increased risk of cardiovascular disease (CVD). The prevalence of type 2 DM and IR could be estimated from reports of risk factors for CVD in RA patients. The TNFα antagonists provided rapid and effective control of RA-related inflammation. Evidence indicated that extended use of TNFα antagonists in RA may provide the additional benefit of improving insulin sensitivity. These treatment-related changes may contribute to an overall reduction in the risk of type 2 DM and CVD in RA patients.

Conclusion

Controlling inflammation may improve insulin sensitivity and subsequently reduce the risk of developing type 2 DM in RA patients. This may also reduce the risk of CVD in this high-risk group. Future studies are required to elucidate the relationships between inflammation, body composition, IR, TNFα antagonist use, and the risk of developing type 2 DM in RA patients.

INTRODUCTION

Rheumatoid arthritis (RA) is a systemic, autoimmune disorder that primarily manifests as chronic synovial inflammation of multiple joints. Over the last few decades it has become increasingly apparent that chronic activation of the immune system, as observed in the pathogenesis of RA, is associated with corollary changes in intermediary metabolism, potentially leading to increased risk of cardiovascular disease (CVD) (1, 2). Several reports have discussed the association between chronic inflammatory disease states and disorders in intermediary metabolism (3–7), particularly peripheral insulin resistance (IR). In addition, numerous independent studies have implicated the association of multiple immunoregulatory components (including tumor necrosis factor α [TNFα] and interleukin-6 [IL-6]) in RA, IR, and type 2 diabetes mellitus (DM) (6, 8–11). Disease-associated reduction in lean muscle mass and sedentary lifestyle likely further contribute to IR in patients with RA (12).

Although the prevalence of type 2 DM might be expected to be increased among patients with RA, large cross-sectional studies have not definitively established an association between these two conditions. Because type 2 DM and IR are important risk factors for CVD, a common comorbidity in patients with RA (1, 2), a review of the literature to examine factors associated with glucose regulation and type 2 DM in RA patients is warranted.

MATERIALS AND METHODS

A PubMed search of English-language articles published from 1949 to April 2010 was conducted to identify clinical trials and randomized controlled studies that evaluated IR, and specifically the effects of RA medication, in patients with RA. Search terms included various permutations of the following: adalimumab, adiponectin, anakinra, certolizumab, C-reactive protein (CRP), diabetes, disease-modifying antirheumatic drug (DMARD), etanercept, glucocorticoids, golimumab, infliximab, IL-6, interleukins, IR, insulin sensitivity, leptin, metabolic dysfunction, methotrexate, obesity, body composition, body habitus, pathophysiology, proinflammatory cytokines, RA, risk factors, and TNFα (antagonists). Background information on epidemiology and pathophysiology was garnered from multiple sources, including review articles.

RESULTS

Epidemiology of DM, IR, and CVD in RA.

Several studies have examined the prevalence of type 2 DM in patients with RA, demonstrating a modest increase or no increase in risk compared with the general population. Using the PharMetrics Patient-Centric Database containing integrated US health plan data, Han et al (13) found a significantly higher prevalence of type 2 DM among 28,208 RA patients compared with controls matched 4:1 for age, sex, and geographic region, as well as time in health plan (10.4% versus 7.6%; P = 0.01). However, given the size of the database and the absolute rates, this difference in prevalence is somewhat modest (13). Notably, the cross-sectional nature of the study and the scope of available patient data limited the ability to consider the effects of RA disease severity, activity, duration, and patterns of medication use over time on the risk of developing type 2 DM.

In contrast, no significant association between RA and type 2 DM was established in a large, population-based study among older women participating in the National Health and Nutrition Examination Survey III. Of the 5,302 subjects age ≥60 years, 144 (2.7%) had RA according to the classification by Rasch et al (14), and 24 subjects (17%) with RA had prevalent DM as evidenced by questionnaire response, diagnostic code, or a fasting plasma glucose concentration >125 mg/dl (15). Among the evaluable 5,152 individuals without RA, 815 (16%) had prevalent DM; the DM prevalence rates among those with and without RA were similar (P = 0.46). Because of the cross-sectional study design, disease causality and temporality could not be addressed. Additionally, RA severity, activity, duration, and patterns of medication use were not accounted for in the analyses.

Likewise, other studies have found no relationship between RA and DM. In the report by Gabriel et al of comorbidities in patients with RA and osteoarthritis, prevalence rates of DM were similar (16). Solomon and colleagues found no significant differences in the rate of DM among women with RA (n = 287) and without RA (n = 87,019) participating in the Nurses' Health Study (4.8% versus 4.4%; P = not significant) (17).

Abnormalities in glucose metabolism with a pattern of IR, however, have been well documented in patients with RA and may correlate with RA activity. Svenson et al (18) performed intravenous glucose tolerance testing on 45 patients with active RA and on sex-matched healthy controls, and they observed higher basal serum insulin levels and higher maximum insulin responses to glucose loading consistent with IR in the patients with active RA. Comparing 39 patients with RA and 39 patients with osteoarthritis, Dessein et al (19) noted that RA patients had greater IR, as estimated by the quantitative insulin sensitivity check index (QUICKI; P < 0.05), although this difference disappeared after controlling for CRP level, which was the study's surrogate marker for inflammation. In a subsequent study using the homeostatic model assessment (HOMA) of IR (HOMA-IR) (20) and beta cell function (HOMA-B) calculations in 94 patients with RA, these authors reported that patients with “high-grade inflammation” (high-sensitivity CRP level >1.92 mg/liter) were more likely to be insulin resistant than those with low-grade inflammation, despite similar calculated beta cell function (21). In multivariable mixed regression models, abdominal obesity and patient's assessment of disease activity were the only study variables that predicted IR.

Another study by Dessein and colleagues (22) found that the prevalence of IR, measured by HOMA-IR and QUICKI, in 87 nondiabetic patients with inflammatory arthritis was comparable with that among controls after adjusting for body mass index (BMI) and erythrocyte sedimentation rate. Chung et al (23) reported that HOMA-IR values were higher in patients with RA than those with systemic lupus erythematosus, after adjusting for age, race, sex, BMI, and current corticosteroid use (P = 0.03); however, this study did not include nonrheumatic disease controls. These findings suggest that underlying inflammation, as reflected by disease activity, is important in the development of IR in patients with RA or inflammatory arthritis.

Because glucose intolerance is a commonly used criterion in definitions of metabolic syndrome (24–27), reviewing the literature on RA and metabolic syndrome yields further insight into the relationship between RA and IR. Using the World Health Organization (WHO) criteria for metabolic syndrome, which require patients to have IR, Chung et al (23) reported an increased prevalence of metabolic syndrome among patients with RA (42% with longstanding disease, 31% with early disease) compared with controls (11%; P < 0.001). Using Adult Treatment Panel III criteria for metabolic syndrome, which do not require glucose intolerance as a symptom, Karvournaris et al (28) documented metabolic syndrome in 44% of 200 patients with RA, versus a comparable rate of 41% among 400 controls (P = 0.5). Because not all studies in RA have used metabolic syndrome definitions that include insulin sensitivity, the prevalence of IR in patients with RA compared with subjects without RA or other chronic inflammatory diseases requires further elucidation.

As in the general population (29), both DM and IR appear to be independent risk factors for atherosclerotic CVD in patients with RA (23, 30). In a large study of 631 patients with RA who underwent carotid ultrasonography to detect atherosclerotic plaque, a diagnosis of DM was more common among those patients with evidence of plaque (P < 0.001) (30). Dessein and colleagues (31) were the first to report an association between IR and atherosclerosis in patients with RA, noting that IR as an individual risk factor for identifying individuals with carotid plaque was more useful than either the WHO-defined or the National Cholesterol Education Program–defined metabolic syndrome (32, 33). Chung et al (23) subsequently found that those patients with RA with WHO-defined metabolic syndrome had an increased risk of coronary artery calcification, independent of age and sex. Rho et al recently reported that elevated levels of leptin, an adipocytokine that mediates inflammation, are associated with IR but surprisingly attenuate the relationship between HOMA-IR and coronary calcification in a group of 169 patients with RA (34). Pamuk and colleagues (35) also noted that among 63 patients with RA and 34 controls, calculated HOMA-IR (odds ratio [OR] 15.85, 95% confidence interval [95% CI] 2.23–112.89; P = 0.006) and Health Assessment Questionnaire scores (r = 0.28, P = 0.04) were independently associated with carotid plaque, suggesting that IR could contribute to atherogenesis in the context of active RA. La Montagna et al (36) also reported an association between HOMA-IR and increased carotid intima-media thickness in a cohort of 45 RA patients undergoing carotid ultrasonography.

The lack of a clear association between RA and type 2 DM in published reports emphasizes the need for larger well-controlled studies that use well-defined diagnostic criteria for both RA and type 2 DM in patient populations stratified prospectively by RA severity and disease activity to reflect the systemic burden of chronic inflammation. Given the increased CVD risk among patients with RA and the established relationship between IR and CVD risk, the relationship between RA and IR also warrants further study. Interventions that both effectively treat RA joint disease activity and have a favorable impact on the burden of IR and type 2 DM might also be expected to reduce the likelihood of developing CVD in this high-risk population.

Body composition, disease severity, and IR in patients with RA.

Changes in body composition, which can predispose patients with RA to IR, have been described previously. Specifically, TNFα, a critical inflammatory cytokine in chronic active RA, has detrimental effects on lean muscle mass, resulting in sarcopenia and a relative increase in body adipose tissue, a condition known as rheumatoid cachexia (37). Therefore, an individual may have a normal BMI calculated from height and weight but may at the same time experience an abnormal body composition with decreased lean muscle mass, increased adiposity, and a theoretical propensity toward developing IR. These changes in body habitus have been well described in RA and develop independent of corticosteroid use (12). Roubenoff and colleagues (37) first reported an association between TNFα and loss of lean muscle mass in patients with RA. Walsmith et al (38) later found an inverse association between TNFα production and body cell mass after adjusting for age and physical activity among 20 women with RA who were compared with 20 healthy female controls matched for age, race, and BMI.

Assessment of body composition for total and regional lean muscle mass and body fat by dual x-ray absorptiometry has shown that changes in body composition occur in both male and female patients with early RA (disease duration ≤12 months), although truncal fat deposition was higher than expected only in female patients (39). Giles et al (12) also reported sex differences in body composition, in that women with RA were more likely than controls to have sarcopenia, be overfat, and display sarcopenic obesity. Those not treated with traditional or biologic DMARDs were more likely to have changes in body composition, suggesting that controlling disease activity may have favorable effects on lean body mass and adipose tissue deposition and that these changes occur independently of corticosteroid use.

Levels of adiponectin, an adipocytokine, are inversely related to adiposity and IR, and adiponectin exerts favorable effects on insulin sensitivity and atherosclerosis (40, 41). Adiponectin production by adipose cells is reduced by exposure to TNFα (40), and low adiponectin levels co-occur with characteristics of metabolic syndrome in patients with RA (42). However, in a study of 169 patients with RA, adiponectin levels were not associated with HOMA-IR or coronary calcium scores (34), although leptin levels correlated with HOMA-IR indices.

Future work in this area must take into account the limitations of BMI as an indicator of body composition and use other methods to assess body mass. As RA patients receive more effective therapies and are able to become more physically active, these body composition abnormalities may decrease in prevalence. Improvement in body composition, particularly decreased central adiposity and increased lean muscle mass, may consequently reduce the likelihood of IR and type 2 DM in patients with RA.

Commonalities between the pathophysiologies of RA and IR.

Although the precise pathophysiologic mechanisms remain to be elucidated, the roles of immune activation and proinflammatory mediators, such as TNFα, IL-1, and IL-6, are well documented in the pathogenesis of RA (43). Over the last decade, chronic inflammation and proinflammatory cytokines have been increasingly recognized to play an integral role in the pathogenesis of IR (Figure 1). Adipose tissue, which was once only considered to be passive tissue with the sole function of excess energy storage, is now known to also produce many mediators that regulate metabolism in other tissues and appears to be important in the initiation and promotion of IR (44). In obesity, excess lipid accumulation produces a low-grade inflammatory process in adipose tissue by activating inflammatory signaling pathways that involve transcriptional activation of NF-κB. The induction of chemokines and vascular adhesion molecules results in infiltration of macrophages into adipose tissue. These adipose-tissue macrophages produce inflammatory cytokines, such as TNFα and IL-6, thereby increasing inflammatory activity both locally and systemically. Concurrently, adipocytes also produce mediators, known as adipokines, that not only produce metabolic effects, but also have inflammatory consequences. In response to inflammatory mediators and free fatty acids produced by visceral adipose tissue or resulting from lipid accumulation in the liver, hepatocytes and liver immune cells also produce proinflammatory cytokines. Several of these proinflammatory cytokines and adipokines are also involved in the pathogenesis of RA; some of these shared mediators and the data supporting their involvement in the development of IR (10, 45) will be discussed here.

Figure 1.

Commonalities between the pathophysiologies of insulin resistance (IR) and rheumatoid arthritis (RA). A chronic inflammatory condition is maintained in patients with RA by the activation and persistence of proinflammatory cytokines (see ref.103). Adipose tissue is an important source of proinflammatory cytokines. Activation of these cytokines leads to IR and other extraarticular manifestations of RA. IL-1 = interleukin 1; TNF-α = tumor necrosis factor α.

TNFα plays a central role in the pathogenesis of RA, as demonstrated by the clinical and radiographic improvement achieved by patients who receive anti-TNFα therapy (46, 47). TNFα has also been implicated in the development of IR. In vitro studies have shown that TNFα can obstruct the bioactivity of insulin by inhibiting tyrosine phosphorylation and subsequent activation of both the insulin receptor and insulin receptor substrate 1 (48). In a model of obese mice, knockout mice lacking expression of TNFα or TNF receptors were more sensitive than controls to insulin (49). Studies in both mice (50) and humans (51) have shown that adipocytes overexpress TNFα in obese individuals; however, another study in humans reported contradictory results (52).

Similar to TNFα, inhibition of the interaction between IL-6 and the IL-6 receptor with the monoclonal antibody tocilizumab is effective in treating RA (53). IL-6 also appears to play a role in IR and the development of type 2 DM. As with TNFα, IL-6 appears to affect insulin-signaling pathways by diminishing the effect of insulin (9). Reduction of IL-6 levels by the administration of anti-IL6 antibodies increased insulin sensitivity in 2 murine models of obesity (54); however, IL-6 knockout mice are insulin resistant and become obese when mature (55). Human adipocytes release IL-6 (52), and IL-6 production is increased in obesity, with higher secretion in visceral than subcutaneous adipose tissue (10). In addition, elevated IL-6 levels have been observed in women who later developed type 2 DM (56). Other inflammatory cytokines appear to be associated with both RA and IR or type 2 DM. Elevated levels of IL-2 are observed in the serum (57) and synovial tissues (58), and correlate with insulin sensitivity (57) in patients with RA. IL-17 is elevated in the synovial tissues of patients with RA (59) and inhibits adipocyte differentiation (60).

In addition to proinflammatory cytokines, several other proteins that are secreted by adipose tissue are associated with chronic inflammation. These proteins, called adipokines, may in turn modulate proinflammatory cytokine levels, thereby contributing indirectly to inflammation. Leptin is secreted by adipocytes and is involved in the regulation of appetite and metabolism through its action on the hypothalamus. Leptin levels are increased in chronic inflammatory conditions, including RA and lupus (61). In one study, a relationship between leptin and adiposity was observed in patients with RA; however, leptin was not associated with RA disease activity parameters, and leptin levels did not change after anti-TNFα treatment (62). A recent study of IR in patients with RA found an independent association between leptin levels and IR, such that leptin levels correlated with HOMA-IR indices even after adjustment for BMI, inflammation, and cardiovascular risk factors (34). Because leptin decreases insulin secretion and improves insulin sensitivity, high concentrations of leptin may indicate a state of leptin resistance. Inflammatory states in which CRP production is increased may promote leptin resistance, since CRP can interfere with leptin receptor binding.

Resistin is secreted by adipocytes in mice but predominantly by monocytes and macrophages in humans (63). Murine models suggest that resistin plays a role in IR and hyperglycemia. Neutralization of resistin in insulin-resistant obese rodents improved insulin sensitivity, while infusion of resistin into lean control rodents induced IR (10). Resistin has proinflammatory properties mediated by the up-regulation of IL-6 and TNFα from peripheral blood mononuclear cells (PBMCs) (64). Resistin also induces the expression of vascular cellular adhesion molecule 1 and intracellular adhesion molecule 1: in the hypothalamus, resistin increases hepatic IR associated with increased TNFα and IL-6 production in the liver. Resistin levels are increased in the synovial fluid of patients with RA, compared with healthy controls (9) and osteoarthritis patients (65). A strong association between serum resistin levels and laboratory markers of inflammation, particularly with CRP, was found in patients with RA undergoing anti-TNF therapy because of severe disease refractory to conventional DMARDs (66). In these patients, anti-TNF therapy yielded a rapid reduction of serum resistin levels. However, no correlation between serum resistin and HOMA-IR was observed (66). Consistent with these observations, serum resistin levels were positively associated with levels of inflammatory markers, including soluble TNFα receptor 2, IL-6, and lipoprotein-associated phospholipase A2, but not with measures of IR in asymptomatic individuals with a family history of premature coronary artery disease (67).

Adiponectin, another adipokine, appears to play an important role in maintaining insulin sensitivity through its involvement in metabolic and antiinflammatory regulation (68). Reduced levels of adiponectin increase levels of blood glucose and body fat, thereby resulting in IR (69). Adiponectin levels are inversely related to adiposity and IR, and adiponectin has favorable effects on insulin sensitivity and atherosclerosis (40, 41). Adiponectin production by adipose cells is reduced by exposure to TNFα (40), and circulating adiponectin levels are decreased in obese individuals and in patients with type 2 DM (10). In a recent cross-sectional study, fasting levels of glucose, insulin, adiponectin, resistin, and TNFα were measured in 2,356 patients who did not have prevalent DM or elevated levels of adipokines (70). Overall, the age- and sex-adjusted HOMA-IR levels, a calculated measure of glucose utilization, correlated inversely with adiponectin and directly with TNFα and resistin concentrations (70). Besides its effects on insulin sensitivity, adiponectin produces antiinflammatory effects by reducing TNFα secretion by macrophages through its inhibitory effects on NF-κB activation (71). Some studies have suggested that adiponectin inhibits TNFα-induced adhesion molecule expression as well as NF-κB activation in endothelial cells and stimulates the production of the antiinflammatory cytokine IL-10 by macrophages (71). Because of adiponectin's supposed beneficial metabolic and antiinflammatory effects, increasing adiponectin levels by upregulating production has been considered a therapeutic strategy for metabolic syndrome (72).

Lower levels of adiponectin also have been associated with features of the metabolic syndrome in patients both with and without RA (42). However, although TNFα reduces adiponectin production, patients with RA have increased serum (73) and synovial fluid (74) adiponectin levels. These increased adiponectin levels may reflect a compensatory mechanism counterbalancing elevated levels of the proinflammatory mediators leptin and TNFα. Adiponectin is hypothesized to have insulin-sensitizing and antiatherogenic effects (40, 41). However, in a study of 169 patients with RA, adiponectin levels were not associated with HOMA-IR or coronary calcium stores, but leptin levels were associated with HOMA-IR indices (34). Whether the role of adiponectin in RA patients is similar to that in patients with obesity/metabolic syndrome requires further clarification.

In summary, RA and IR are states of chronic inflammation with shared elements in their pathogeneses. The proinflammatory cytokines TNFα and IL-6, which are key mediators of inflammation in RA, are overproduced in visceral adipose tissue, impair insulin receptor signaling, and are associated with IR. Adipokines that modulate metabolism and insulin sensitivity also have effects on inflammation and are themselves modulated by inflammatory cytokines. The shared pathogenic pathways between RA and IR suggest possible mechanisms whereby patients with chronic inflammatory conditions may be at higher risk for impaired glucose metabolism. However, the complex interactions among cytokines and adipokines, which have both metabolic and inflammatory effects, are not yet fully elucidated.

Effects of RA medications on IR.

Corticosteroids.

Glucocorticoids are commonly used to treat RA and have effects on basic metabolic function (75, 76). However, although the use of high-dose glucocorticoids often results in hyperglycemia, the antiinflammatory effects of low daily doses may improve glycemic control through enhanced pancreatic insulin secretion and peripheral insulin sensitivity. In a cross-sectional study of 398 RA patients grouped by glucocorticoid exposure, Toms et al (77) found no association between glucocorticoid use and metabolic syndrome. In addition, similar proportions of patients in each group studied (no/limited exposure for <3 months; low-dosage long-term exposure at <7.5 mg/day for >6 months; and medium-dosage, long-term exposure at ≥7.5 to 30 mg/day for >6 months) had elevated fasting plasma glucose levels (77). In a mixed regression model analysis of 94 patients with RA, Dessein and Joffe (21) observed that abdominal obesity and patient's assessment of disease activity (by visual analog scale) were predictors of IR. The Disease Activity Score using 28-joint counts for swelling and tenderness and the patient's assessment of disease activity were associated with reduced beta cell function, and a low cumulative glucocorticoid dose was associated with improvements in beta cell function as estimated by HOMA-B (21). An earlier study demonstrated that short-term use of prednisolone (cumulative dose 210 mg) tended to improve insulin sensitivity from baseline among 14 patients with active RA (78), supporting the theory that low-level corticosteroid exposure, at least when used in the short term, may help to correct inflammation-related metabolic abnormalities such as IR in patients with active RA. Alternatively, Dessein et al (79) found that, after adjusting for BMI, higher-dose oral prednisone (cumulative dose 4.8 [range 2.0–8.5 gm]) and more frequent annual pulsed glucocorticoid administration were associated with decreased insulin sensitivity as determined by the QUICKI in the steroid-treated subset of 92 consecutively evaluated patients with RA. These results are more consistent with the conventional expectations for steroid treatment.

Traditional DMARDs.

Data regarding traditional DMARDs and metabolic function are more limited. In one report, the use of DMARDs (chloroquine, methotrexate, minocycline, or azathioprine) along with dietary intervention for 3 months reduced the median HOMA-IR by 36% among 22 patients with inflammatory arthritis and IR and/or dyslipidemia (P = 0.006) (76). Toms et al (80) demonstrated, in a cross-sectional study of 387 patients with RA, that metabolic syndrome was less common among those taking methotrexate (OR 0.517, 95% CI 0.33–0.81; P = 0.004), although prior use of methotrexate and current use of other DMARDs, glucocorticoids, and biologic agents did not affect the risk of metabolic syndrome. Specifically, methotrexate use was associated with lower fasting glucose levels. A negative association between methotrexate use and metabolic syndrome was also observed by Zonana-Nacach et al (81). It is unclear whether this protective relationship results solely from the antiinflammatory effects of methotrexate, as suggested by Zonana-Nacach and colleagues, or from other mechanisms of action of methotrexate, as proposed by Toms et al (80).

Hydroxychloroquine (HCQ) use in patients with RA has been associated with a reduction in the risk of developing DM (82). Among healthy and diabetic subjects, favorable changes in insulin secretion, sensitivity, and clearance have been demonstrated with chloroquine (CQ) use, underscoring its multifactorial effect on glycemic regulation (83). In vitro studies have shown that HCQ and CQ both inhibit TNFα production by PBMCs stimulated by phytohemagglutinin or lipopolysaccharide (LPS) (84). Using murine macrophages, Bondeson and Sundler (85) demonstrated a weaker inhibitory effect of HCQ than CQ or quinacrine, on zymosan-induced expression of TNFα. Building on these findings, Weber and Levitz (86) demonstrated that CQ inhibits LPS-induced transcription of TNFα by blocking mitogen-activated protein kinase signaling pathways and subsequently regulating TNFα production. These effects on TNFα production may provide a mechanism by which HCQ modulates IR, given the pivotal role of this proinflammatory cytokine in glycemic regulation. Other mechanisms of action have been suggested for the effect of antimalarial drugs on glucose metabolism, but they have not been linked directly to glycemic regulation (87–90). However, a recent study indicated that HCQ induces autophagic cell death in fibroblasts, thereby having an “antiproliferative effect” (91). These observations suggest direct regulatory effects of HCQ on cellular energy homeostasis. In a recent review, Jin and White (92) highlighted the importance of autophagy in cellular response to metabolic stress, particularly in malignancies; as such, it is being explored as an adjunct to traditional therapies for various solid tumors (Lotze M: unpublished observations).

Nonsteroidal antiinflammatory drugs (NSAIDs).

NSAIDs are commonly used in the treatment of RA. Based on their inhibition of cyclooxygenase-mediated inflammation, NSAIDs are expected to improve IR and reduce the risk of developing type 2 DM. Conversely, in subjects with type 2 DM, indomethacin administration decreases insulin secretion with a subsequent increase in endogenous intrahepatic glucose production. While data regarding the effect of other traditional NSAIDs on glucose and insulin metabolism are quite limited, a recent 1-month trial of salsalate (4.0 gm/day) in 20 obese nondiabetic adults improved glycemia according to fasting glucose and oral glucose tolerance testing, with a reduction in C-peptide levels, suggesting improvement in insulin sensitivity (93). In a subsequent report, salsalate-treated type 2 DM patients were more likely than placebo-treated patients to have a reduction in glycosylated hemoglobin levels of ≥0.5% (P = 0.009); additionally, the salsalate-treated patients improved in fasting glucose and glycated albumin levels (94). These findings are consistent with sporadic reports, over the past century, of salicylates having beneficial effects on glycemia and insulin requirements in patients with established DM (95). An inhibitory effect of salsalate on NF-κB has been suggested as the mechanism of action in glycemic control (95).

TNFα antagonists.

The efficacy of TNFα antagonists for the treatment of RA is well established. Because TNFα antagonists target immune activation, it has long been surmised that their use would improve long-term metabolic dysfunction resulting from chronic inflammation. Indeed, animal studies have clearly shown that TNFα antagonists effectively improve insulin sensitivity (96). Several recent studies have demonstrated that treatment with TNFα antagonists alters the lipid profile (7) and improves insulin sensitivity in patients with RA (Table 1) (7, 97–102). Tam et al (7) reported favorable changes in measures of insulin sensitivity in 19 patients with RA treated with infliximab. At week 6, HOMA-IR and insulin levels were both reduced; this reduction became statistically significant at week 14 (P < 0.05). The changes in HOMA-IR and fasting insulin levels did not correlate with improvements in disease activity. Changes in CRP level were not associated with changes in any metabolic parameters. Infliximab-treated patients did not experience weight gain. These results are consistent with those from other studies of patients treated with etanercept or infliximab (7, 98–101). Seriolo and colleagues (101) conducted the only published study describing the effects of etanercept on IR, which found no difference in the improvement in insulin sensitivity between 18 patients with RA receiving etanercept and 14 patients with RA receiving infliximab.

Table 1. Summary of studies evaluating the effect of TNFα inhibition on IR in patients with RA*
Author, year (ref.)TNFα inhibitorNo. of RA patientsDuration of followupEvaluation methodEffect of TNFα inhibition on IR
  • *

    TNFα = tumor necrosis factor α; IR = insulin resistance; RA = rheumatoid arthritis; HOMA = homeostasis model assessment; QUICKI = quantitative insulin sensitivity check index.

  • HOMA by fasting serum insulin μU/ml × fasting plasma glucose (mmoles/liter); QUICKI by 1/log insulin (μU/ml) + log glucose (mg/dl).

Seriolo et al, 2007 (101)Etanercept/infliximab3824 weeksHOMA/QUICKINo change at week 12, but improvement at week 24
Oguz et al, 2007 (102)Infliximab79.6 months (mean)HOMAImprovement
Tam et al, 2007 (7)Infliximab1914 weeksHOMAImprovement
Gonzalez-Gay et al, 2006 (99)Infliximab27120 minutes from start of the infusionHOMA/QUICKIImprovement
Kiortsis et al, 2005 (98)Infliximab286 monthsHOMA/QUICKIStudy also included 17 patients with ankylosing spondylitis
Improvement in insulin sensitivity only among patients with the most IR at baseline (n = 14)
Rosenvinge et al, 2007 (97)Adalimumab98 weeksHyperinsulinemic euglycemic clampNo effect on insulin sensitivity
Yazdani-Biuki et al, 2004 (100)Infliximab2Up to 240 daysHOMAStudy also included 3 patients with psoriatic arthritis
Improvement

These favorable effects described above have not been observed with the anti-TNFα agent adalimumab. Specifically, there was no change from baseline following treatment with adalimumab, in hyperinsulinemic euglycemic clamp following treatment with adalimumab (97). However, this study evaluated only 9 patients with RA, and the changes in insulin sensitivity were assessed after only 8 weeks of treatment. The beneficial metabolic effects of anti-TNFα treatment may manifest only after extended treatment. Therefore, 8 weeks of adalimumab therapy may have been insufficient to detect treatment-related changes in insulin sensitivity (7, 100). Alternatively, differences in study patient characteristics or the metabolic effects of various anti-TNF therapies may account for the discrepancies across these studies.

Despite the lack of evidence of metabolic benefit with adalimumab, the benefits of treatment with TNFα antagonists as a class appear to extend beyond control of joint disease activity. The partial restoration of metabolic function and alleviation of IR are becoming increasingly evident with prolonged anti-TNF treatment. Whether this beneficial effect is sustained with long-term treatment varies depending on joint disease activity and response to anti-TNF therapy. Variation in this benefit by the TNF antagonist used is unclear. Moreover, the relationship between body composition and the effect of TNFα antagonists on metabolic function remains unexplored.

DISCUSSION

Patients with chronic inflammatory diseases such as RA are at higher risk of developing impaired glucose metabolism that may eventually progress to type 2 DM. Further translational studies are needed, particularly at the level of molecular determinants, to definitively establish the link between IR and type 2 DM in patients with RA. However, it is clear that chronic systemic inflammation is pivotal in the pathogenesis of both RA and IR. Patients with RA in whom disease activity is effectively controlled may experience the additional clinical benefit of improved insulin sensitivity. Clinical evidence suggests the severity and duration of RA, in addition to visceral/abdominal obesity, are important factors that influence IR and the risk of developing type 2 DM. These factors require prospective consideration in studies that evaluate the risks and benefits of therapeutic strategies on the development of IR and of type 2 DM.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Wasko had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Wasko, Kay, Rahman.

Acquisition of data. Wasko, Hsia, Rahman.

Analysis and interpretation of data. Wasko, Kay, Hsia, Rahman.

ROLE OF THE STUDY SPONSOR

Writing assistance was provided in collaboration with Centocor Ortho Biotech, Inc. Drs. Hsia and Rahman were employees of Centocor Ortho Biotech, Inc. at the time of the study. All authors reviewed and approved the manuscript before submission and jointly agreed to submit the final version of the manuscript.

Acknowledgements

We thank Mary Whitman, PhD, and Kirsten Schuck of Centocor Ortho Biotech Services, LLC, and Scott Newcomer, MS, of Cephalon, Inc. (formerly of Centocor Ortho Biotech Services, LLC) for editorial assistance and writing support on this manuscript.

Ancillary