Salsalate Improves Glycemia and Inflammatory Parameters in Obese

Emerging Treatments and Technologies ORIGINAL ARTICLE

Salsalate Improves Glycemia and Inﬂammatory Parameters in Obese Young Adults

1,2,3 1,2 AMY FLEISCHMAN, MD, MMSC RAQUEL BERNIER, BS been reinvestigated, and the molecular 1,2 1,2 STEVEN E. SHOELSON, MD, PHD ALLISON B. GOLDFINE, MD target was identified to be the I␬B kinase complex ␤ (IKK␤)/nuclear factor ␬B (NF- ␬B) pathway (8,9), a central integrator of OBJECTIVE — Sedentary lifestyle and a western diet promote subacute-chronic inflamma- proinflammatory signals (2). The thera- tion, obesity, and subsequently dysglycemia. The aim of the current study was to evaluate the peutic potential of high-dose aspirin is efficacy of the anti-inflammatory drug salsalate to improve glycemia by reducing systemic in- limited by bleeding risk. Salsalate, a dimer flammation in obese adults at risk for the development of type 2 diabetes. of salicylic acid, has an established safety profile after decades of use for rheumatic RESEARCH DESIGN AND METHODS — In a double-masked, placebo controlled pain. As a nonacetylated salicylate, sal- trial, we evaluated 20 obese nondiabetic adults at baseline and after 1 month of salsalate or salate is an equipotent inhibitor of NF-␬B placebo. but has a lower bleeding risk than aspirin RESULTS — Compared with placebo, salsalate reduced fasting glucose 13% (P Ͻ 0.002), (10,11). glycemic response after an oral glucose challenge 20% (P Ͻ 0.004), and glycated albumin 17% To our knowledge, this is the first (P Ͻ 0.0003). Although insulin levels were unchanged, fasting and oral glucose tolerance test study to assess metabolic changes with C-peptide levels decreased in the salsalate-treated subjects compared with placebo (P Ͻ 0.03), administration of salicylates to obese in- consistent with improved insulin sensitivity and a known effect of salicylates to inhibit insulin dividuals without diabetes. We hypothe- clearance. Adiponectin increased 57% after salsalate compared with placebo (P Ͻ 0.003). Ad- sized that salicylates administered for 1 ditionally, within the group of salsalate-treated subjects, circulating levels of C-reactive protein month would improve glycemia in obese Ͻ were reduced by 34% (P 0.05). young adults. CONCLUSIONS — This proof-of-principle study demonstrates that salsalate reduces glycemia and may improve inflammatory cardiovascular risk indexes in overweight individuals. RESEARCH DESIGN AND These data support the hypothesis that subacute-chronic inflammation contributes to the patho- METHODS — The Joslin Diabetes genesis of obesity-related dysglycemia and that targeting inflammation may provide a therapeu- Center institutional review board ap- tic route for diabetes prevention. proved the double-masked, placebo- controlled study. Written informed Diabetes Care 31:289–294, 2008 consent was obtained. Subjects were Ͻ30 years and obese, with BMI Ն30 kg/m2. besity, occurring at epidemic rates opment of insulin resistance (1,2). Thus, Participants were excluded for recent worldwide, is a major risk factor for targeting the inflammatory pathway may blood donation, change in weight Ͼ5% in O diabetes and cardiovascular dis- be a novel pharmacologic intervention for the preceding 6 months, use of medica- ease. Thus, there is an urgent need for diabetes prevention and treatment. tion known to alter glucose metabolism, effective interventions to prevent diabetes Salicylates are among the most com- acute febrile illness, biochemical evidence in obese populations. The importance of monly used nonsteroidal anti-inflamma- of renal or hepatic dysfunction, aspirin lifestyle modification in obesity and dia- tory drugs. The benefits of salicylates for allergy, history of gastritis or gastrointes- betes is well recognized. However, disap- treatment of diabetes have long been rec- tinal bleeding, or diabetes. Women were pointing long-term results of these ognized (3,4). High doses of the salicylate excluded for pregnancy, lactation, or lack treatments have led to increased interest aspirin (4–7 g/day) improve fasting and of contraception use. in pharmaceutical intervention. Obesity postprandial hyperglycemia in patients Participants were instructed to con- and high-fat western diets activate inflam- with diabetes (5–7). In recent studies, the sume a high-carbohydrate diet (250–300 matory processes, which promote devel- hypoglycemic actions of salicylates have g/day) and abstain from strenuous exer- ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● cise for 3 days before evaluations and not to alter dietary or exercise habits during From the 1Harvard Medical School, Boston, Massachusetts; the 2Joslin Diabetes Center, Boston, Massachu- setts; and the 3Children’s Hospital Boston, Boston, Massachusetts. the study. Blood pressure was measured Address correspondence and reprint requests to Allison B. Goldfine, MD, Joslin Diabetes Center, One twice (DINAMAP PRO-100; General Joslin Place, Boston, MA 02215. E-mail: allison.goldfi[email protected]. Electric Healthcare) with the patient su- Received for publication 5 September 2007 and accepted in revised form 18 October 2007. pine for 10 min. Fasting lipids and cyto- Published ahead of print at http://care.diabetesjournals.org on 24 October 2007. DOI: 10.2337/dc07- 1338. Clinical trial reg. no. NCT00258115, clinicaltrials.gov. kines were measured, and oral glucose Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRP, C-reactive protein; tolerance tests (OGTTs) were performed FFA, free fatty acid; HOMA-IR, homeostasis model assessment of insulin resistance; HOMA-IRC-peptide, with glucose, insulin, and C-peptide lev- homeostasis model assessment of insulin resistance calculated using C-peptide; IKK␤,I␬B kinase complex els measured before and 30, 60, 90, and ␤ ␬ ␬ ; NF- B, nuclear factor- B; OGTT, oral glucose tolerance test. 120 min after a 75-g glucose load. All sub- © 2008 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby jects were nondiabetic on the basis of marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. American Diabetes Association guidelines

DIABETES CARE, VOLUME 31, NUMBER 2, FEBRUARY 2008 289 Salsalate improves glycemia in obesity

(12). Insulin resistance was determined Table 1—Baseline subject characteristics using homeostasis model assessment of insulin resistance (HOMA-IR) for insulin Salsalate therapy Placebo and HOMA-IR calculated using C- peptide (HOMA-IRC-peptide), as described Sex (male/female) 1/8 2/9 by the modified formula HOMA-IR Ethnicity (W/H/B/other) 5/3/1 5/1/4/1 C- Ϯ Ϯ peptide ϭ (fasting C-peptide ϫ fasting glu- Age (years) 23.5 1.1 24.1 1.0 cose)/22.5 (13). BMI (kg/m2) 36.3 Ϯ 2.2 38.9 Ϯ 2.5 Subjects were randomly assigned by a Waist circumference (cm) 106 Ϯ 5 116 Ϯ 6 research pharmacist to receive salsalate, Current smoking 3 of 9 2 of 11 4.0 g/day (Caraco Pharmaceutical Labo- Cholesterol (mmol/l) 4.0 Ϯ 0.3 4.6 Ϯ 0.2 ratories, Detroit, MI) divided in two Triglycerides (mmol/l) 1.0 Ϯ 0.1 1.1 Ϯ 0.2 doses, or identical placebo for 4 weeks. Systolic blood pressure 121 Ϯ 4 123 Ϯ 4 Participants and study personnel were Diastolic blood pressure 67 Ϯ 371Ϯ 3 blinded to the treatment assignment. The Fasting glucose (mmol/l) 5.0 Ϯ 0.2 4.8 Ϯ 0.1 starting dose was selected on the basis of 120-min glucose (mmol/l) 6.7 Ϯ 0.4 6.8 Ϯ 0.4 tolerability data in patients with arthritis Data are means Ϯ SEM. There were no significant differences in the baseline characteristics between the (14). A dose-reduction plan was specified salsalate and placebo groups. W, white, non-Hispanic; H, Hispanic; B, African-American. a priori with stepped reductions of 500 mg for symptoms related to salicylate use, such as tinnitus or headache. Compliance min (a), partial enzymatic digestion of al- pants. There were no significant changes was evaluated by salicylate levels. bumin at glycated sites (b), reassessment in weight, systolic or diastolic blood pres- Twenty-seven subjects were enrolled. of albumin (c), and a Ϫ c. sure, or standard lipid profiles in either One subject became ineligible because of group (Table 2). blood donation after screening. Three sub- Statistical analysis jects withdrew consent because of personal The primary end point was change from Glucose metabolism conflicts and were considered noninforma- baseline in glycemic measures between Fasting glucose decreased 13% in the sal- tive; two had been randomly assigned to salsalate and placebo groups. Secondary salate group compared with the placebo placebo and one to salsalate. Three subjects outcomes were also assessed as change group after 1 month (Ϫ0.4 Ϯ 0.2 vs. were withdrawn because of rash. from baseline within group. Data are pre- ϩ0.2 Ϯ 0.1 mmol/l, respectively, P Ͻ sented as mean Ϯ SEM. Unpaired (sal- 0.002) (Fig. 1A, left). The glucose area Assays salate versus placebo) and paired (before under the curve after a 75-g OGTT was Glucose was measured by glucose oxida- versus after treatment) Student’s t tests also significantly reduced in salsalate tion, cholesterol and HDL were measured were performed. Treatment effects were compared with placebo-treated subjects Ϫ by a cholesterol esterase assay, triglycer- determined by calculating percent change (Ϫ130 Ϯ 53 vs. ϩ38 Ϯ 15 mmol min 1 Ϫ ides were measured via hydrolysis to glyc- for each participant. Repeated measures l 1, P Ͻ 0.004) (Fig. 1A, center). Like- erol and free fatty acids (FFAs) (Beckman and area under the curve analyses were wise, within groups the glycemic re- Synchron CX3 Delta and CX9; Beckman performed to compare response to OGTT sponse to an OGTT improved after Coulter, Brea, CA), and glycohemoglobin before and after treatment. Completer salsalate treatment (repeated-measures was measured by high-performance liq- analysis was prespecified. Multiple re- analysis P Ͻ 0.01) but not placebo (Fig. uid chromatography (Tosho 2.2; Tosoh gression analysis was performed to evalu- 1B). Consistently, we found a 17% reduc- Bioscience, San Francisco, CA). Immuno- ate the contribution of covariates to the tion in the percentage of glycated albumin assays were performed in duplicate using primary end points. Results were consid- between salsalate and placebo (Ϫ2.2 Ϯ commercial assays including radioimmu- ered significant with two-tailed P Ͻ 0.05. 0.5 vs. Ϫ0.1 Ϯ 0.1%, P Ͻ 0.0003) (Fig. noassays for insulin and C-peptide (Diag- 1A, right). Therefore, glycemia, assessed nostic Systems Laboratories, Webster, RESULTS — Twenty participants com- by fasting glucose postglucose load and TX) and adiponectin (Linco Research, St. pleted the protocol. Subject baseline char- glycated albumin, was significantly re- Charles, MO) and enzyme-linked immu- acteristics were similar (Table 1) in both duced after salsalate compared with pla- noabsorbent assays for FFAs (Wako groups. All participants had central obesity cebo. No hypoglycemia was noted in Chemicals, Richmond, VA), and interleu- (15) and normal fasting glucose values. either group. kin-6 and vascular cell adhesion mole- Three subjects (two in the placebo group Fasting insulin and levels during an cule-1 (R&D Systems, Minneapolis, MN). and one in the treatment group) had base- OGTT remained unchanged after sal- C-reactive protein (CRP) was analyzed by line 120-min OGTT glucose values consis- salate or placebo (Fig. 1C). However, the immunoturbidometry (Wako Chemi- tent with impaired glucose tolerance. No reduction in fasting C-peptide comparing cals). Salicylate levels were assessed com- subject had diabetes at baseline. salsalate and placebo-treated groups was mercially by colorimetric assay (Quest In salsalate-treated participants, significant (Ϫ0.53 Ϯ 0.14 vs. ϩ0.06 Ϯ Laboratories, Cambridge, MA). Glycated mean serum salicylate levels were in the 0.1 nmol/l, respectively, P Ͻ 0.01). The albumin was evaluated by a Hitachi 911 therapeutic range established in rheuma- percent difference in C-peptide area un- lipid and protein analyzer and kits from tology practice (0.7–2.2 mmol/l): 1.35 Ϯ der the curve after an OGTT between AsahiKasei (Tokyo, Japan). Glycated al- 0.18 mmol/l at 2 weeks and 1.23 Ϯ 0.25 groups was also significant (Ϫ19 Ϯ 6 vs. bumin was calculated by determining the mmol/l at 4 weeks. Salicylate levels were ϩ2 Ϯ 6%, P Ͻ 0.03) (Fig. 1D). colorimetric measurement of total albu- undetectable in placebo-treated partici- As salicylates have previously been

290 DIABETES CARE, VOLUME 31, NUMBER 2, FEBRUARY 2008 Fleischman and Associates

Table 2—Glycemic, inﬂammatory, lipid, and body composition parameters

Salsalate Placebo P value for comparison of % change % change change between Baseline 1 month from baseline Baseline 1 month from baseline groups Glucose metabolism Glucose (mmol/l) 5.0 Ϯ 0.2 4.6 Ϯ 0.1* Ϫ8 4.8 Ϯ 0.1 5.1 Ϯ 0.1* ϩ5 Ͻ0.002 Glycated albumin (%) 12.4 Ϯ 0.4 10.3 Ϯ 0.3* Ϫ17 12.5 Ϯ 0.3 12.4 Ϯ 0.4 Ϫ1 Ͻ0.0003 Glucose AUC (mmol 905 Ϯ 42 776 Ϯ 50 Ϫ14 801 Ϯ 40 840 Ϯ 42 ϩ6 Ͻ0.004 minϪ1 lϪ1) C-peptide (nmol/l) 1.6 Ϯ 0.2 1.1 Ϯ 0.2* Ϫ24 1.5 Ϯ 0.2 1.5 Ϯ 0.2 ϩ5 Ͻ0.01 Insulin (pmol/l) 112 Ϯ 21 124 Ϯ 18 ϩ27 97 Ϯ 12 98 Ϯ 10 ϩ6 0.5 HOMA IR C-peptide 1.1 Ϯ 0.2 0.7 Ϯ 0.1* Ϫ29 1.0 Ϯ 0.1 1.0 Ϯ 0.1 ϩ10 Ͻ0.004 Inflammatory parameters Adiponectin (mg/l) 10.6 Ϯ 1.7 16.2 Ϯ 2.7* ϩ56 10.8 Ϯ 1.5 10.5 Ϯ 1.5 Ϫ1 Ͻ0.003 FFA (mEq/l) 0.46 Ϯ 0.08 0.25 Ϯ 0.05 Ϫ14† 0.42 Ϯ 0.06 0.37 Ϯ 0.08 Ϫ14 0.9 CRP (nmol/l) 48.2 Ϯ 10.5 30.2 Ϯ 7.4* Ϫ34 45.4 Ϯ 8.4 42.7 Ϯ 9.6 Ϫ8 0.2 Lipid parameters LDL (mmol/l) 2.9 Ϯ 0.3 2.8 Ϯ 0.2 Ϫ5 3.3 Ϯ 0.2 3.2 Ϯ 0.2 Ϫ4 0.8 Triglycerides (mmol/l) 1.0 Ϯ 0.1 0.8 Ϯ 0.1 Ϫ12 1.1 Ϯ 0.2 1.3 Ϯ 0.2 ϩ15 0.08 BMI (kg/m2) 36.3 Ϯ 2.2 36.5 Ϯ 2.2 ϩ0.5 38.9 Ϯ 2.5 38.8 Ϯ 2.5 Ϫ0.3 0.5 Data are means Ϯ SEM. *P Ͻ 0.05 for within-group analysis. †There was one outlier in the treatment group. With the exclusion of this subject, the change in FFA concentrations is a significant reduction of 46% (P ϭ 0.02). demonstrated to reduce insulin clearance mEq/l, P ϭ 0.6) (Fig. 2B), although the groups (P Ͼ 0.1). However, one participant (7), we assessed insulin sensitivity by between-group comparisons did not had an isolated rise in ALT to less than HOMA IRC-peptide, which was signifi- reach statistical significance. Additional twice the upper range of normal, and a cantly lowered by 29 Ϯ 10% after sal- inflammatory markers interleukin-6 and second participant had a similarly mild salate and increased 10 Ϯ 7% after soluble vascular cell adhesion molecule-1 elevation in ALT and AST. Both resolved placebo (between groups, P Ͻ 0.004). did not change significantly after salsalate spontaneously. However, HOMA-IR using insulin was therapy (data not shown). Using multiple not significantly changed in either group regression analysis, salsalate assignment CONCLUSIONS — Obesity is an es- (4.3 Ϯ 0.9 vs. 4.2 Ϯ 0.6 and 3.5 Ϯ 0.5 vs. independently predicted the change in tablished risk factor for development of 3.7 Ϯ 0.4, before vs. after salsalate and fasting blood glucose (P Ͻ 0.009), re- type 2 diabetes and is associated with a placebo, respectively), with no difference sponse to glucose load (P ϭ 0.05), and chronic inflammatory process (16,17). in the change between groups (P ϭ 0.7). glycated albumin (P Ͻ 0.003) when ad- Higher fasting glucose levels within the These findings are consistent with re- justing for the change in adiponectin, normoglycemic range independently pre- duced insulin clearance and improved in- FFAs, and CRP. dict risk for diabetes (18). In this proof- sulin sensitivity contributing to the of-principle study, we evaluated whether improved glycemia. Safety and tolerability the widely prescribed anti-inflammatory A priori, the protocol specified dose re- agent salsalate could have an impact on Cytokines and adipokines duction for signs of salicylism so the max- glycemia in obese young adults. In a ran- In secondary analyses, cytokines and adi- imum tolerable dose was evaluated for domized, double-masked, placebo- pokines were assessed to evaluate the an- efficacy. Three participants required dose controlled study, we demonstrate ti-inflammatory mechanism for reduction attributable to complaints of reductions in glycemia, both fasting and improvement in glycemia. Of impor- tinnitus, headache, or dizziness. Two of after an OGTT, and reduced glycated al- tance, adiponectin increased significantly the three participants were receiving ac- bumin after 1 month of salsalate. As evi- comparing salsalate-treated and placebo tive medication; one tolerated 3.5 g/day, dence of the anti-inflammatory effects of groups (ϩ5.6 Ϯ 1.8 vs. Ϫ0.3 Ϯ 0.4 mg/l, one tolerated 3.0 g/day, and both com- this therapy, we demonstrated increased respectively, P Ͻ 0.003) (Fig. 2a). In pleted the trial without symptoms. adiponectin and a trend toward reduc- change from baseline within-group anal- There were no noted changes in lab- tions in CRP and FFA levels, which may ysis, CRP concentrations decreased 34% oratory analyses of renal function, elec- also be important for reducing risks of di- (Ϫ18.0 Ϯ 7.7 nmol/l, P Ͻ 0.05) after sal- trolytes, or anion gap. Three participants abetes and cardiovascular disease. This is salate, with a nonsignificant decrease after who received active therapy were with- the first human trial to demonstrate that placebo (Ϫ2.8 Ϯ 7.2 nmol/l, P ϭ 0.7), drawn for rash. No respiratory distress salsalate can produce significant improve- and fasting nonesterified FFAs showed a was noted. There was no statistically ments in glycemia and adiponectin in tendency to decline by 14% after salsalate significant change in mean alanine amino- obese individuals. Although our study (Ϫ0.20 Ϯ 0.10 mEq/l, P ϭ 0.07) with no transferase (ALT) or aspartate aminotrans- was limited in sample size and duration, change after placebo (Ϫ0.06 Ϯ 0.05 ferase (AST) in the treatment or placebo the data are consistent with the hypothe-

DIABETES CARE, VOLUME 31, NUMBER 2, FEBRUARY 2008 291 Salsalate improves glycemia in obesity

Figure 2—Changes in inﬂammatory markers and mediators. A: Adiponectin increased significantly in the salsalate-treated group compared with the placebo group. B: CRP and FFA were lower after salsalate (upper panel) but unchanged after placebo (lower panel) by within- group analysis (*P Ͻ 0.05). u, baseline; f, after therapy. Figure 1—Changes in glycemia, insulin, and C-peptide. A: Fasting glucose, AUC after an OGTT, and glycated albumin were reduced in salsalate-treated subjects compared with the placebo group. sis that targeting inﬂammation may pro- B: The salsalate-treated group (left panel) showed improvements in glycemia after an OGTT compared with the placebo group (right panel). Mean and SEM data are shown before and 30, 60, vide a potential treatment for dysglycemia 90, and 120 min after 75 g oral glucose. E, baseline data. C and D: Data for insulin (C) and and a method for diabetes prevention. C-peptide (D) are shown before and 30, 60, 90, and 120 min after a 75-g oral glucose load for both Both sodium salicylate and acetylsal- the salsalate-treated (left panels) and placebo (right panels) groups. AUC, area under the curve. icylic acid (aspirin) have been shown to improve glycemia in humans with type 2

292 DIABETES CARE, VOLUME 31, NUMBER 2, FEBRUARY 2008 Fleischman and Associates diabetes (5,7,19). Previous human stud- atherosclerotic properties (26,27), and is mild elevations in hepatic enzymes may be ies have lasted for 1–2 weeks. The current decreased in obese populations (28). In- consistent with expected findings in the study is the first to suggest that effects creased adiponectin could affect glucose obese population. Salsalate has been widely may be sustained over longer durations. homeostasis. Likewise, although modest, prescribed for more than a half century, and On rare occasions, salicylates have been the reduction in levels of FFAs could also no specific concerns in overweight individ- noted to cause hypoglycemia. Notably, contribute to the improvements in glyce- uals have been noted. However, safety in there was no symptomatic or measured mic parameters and insulin sensitivity obese subjects needs to be monitored and hypoglycemia in our study group. Addi- through the decreased activation of in- established in larger, long-term studies. tionally, our data suggest that improved flammatory mediators (29). Multiple re- This study is limited by the relatively insulin sensitivity may contribute to regression analyses suggest that observed small sample size and short-term dura- duced glucose levels. Evaluation of insu- changes in the measured circulating in- tion. However, the excellent efficacy to lin resistance was limited by the fact that flammatory cytokines do not fully ac- lower glycemia and improve adiponectin insulin clearance has been shown to be count for the glycemic-lowering effects of in obese individuals in this proof-of- reduced by salicylates, as demonstrated salsalate. Although no weight change was principle trial does support potential use during a hyperinsulinemic-euglycemic noted in our current study, the impact of of anti-inflammatory modulators for pre- clamp before and after 2 weeks of high- alterations in food intake or energy vention of insulin resistance and diabetes. dose aspirin (7). Although we did not di- expenditure precipitated by salsalate Metformin (34,35), rosiglitazone (36), rectly assess insulin clearance in this therapy cannot be excluded. Future and troglitazone (37) have previously cohort, altered insulin clearance would mechanistic studies are required to con- been shown to reduce inflammation in confound interpretation of insulin sensi- firm the hypothesis that salsalate im- obesity through the inhibition of NF-␬B tivity based on circulating insulin levels. proves glycemia by affecting the IKK␤/ activity or related cytokines, which may No effect of salicylates on C-peptide clear- NF-␬B signaling pathways in humans. contribute to their impact on diabetes ance has been demonstrated. Thus, the Although efficacious for treatment of prevention and treatment. The reduction finding of an improvement in insulin sen- patients with diabetes, the adverse safety in fasting glucose in this study is consis- sitivity was based on interpretation of profile of high doses of aspirin, particu- tent with or greater in magnitude than the HOMA-IRC-peptide, as described previ- larly the risk of bleeding, limits clinical change demonstrated in the larger, longer ously (13), and was supported by in- therapeutic utility. Nonacetylated salicy- term diabetes prevention trials with met- creased adiponectin. lates, such as salsalate, do not inhibit formin (38), rosiglitazone (39), or trogli- Mechanism(s) by which salicylates platelets and have reduced risk of bleed- tazone (40). Whether these effects will be lower glucose have historically been poorly ing (10,11). Additionally, salsalate is not sustained with continued administration understood and attributed to prostaglandin soluble in the acidic gastric environment of salsalate needs to be determined. suppression (20). However, aspirin doses with a resultant lack of gastrointestinal In summary, our data demonstrate required for both anti-inflammatory and side effects (30). Salsalate has been ad- that targeting inflammation using sal- glucose-lowering effects exceed levels nec- ministered to humans for decades and has salate improves glycemia, insulin resis- essary for inhibition of cyclooxygenase en- an established long-term safety profile. tance, and inflammatory profiles in obese zymes and prostaglandin synthesis. Therefore, the choice of salsalate, a salic- individuals at risk for development of di- Nonacetylated salicylates, such as salsalate, ylate dimer that is an equipotent inhibitor abetes. This proof-of-principle study sug- lack an acetyl group and do not effectively of NF-␬B activity without adverse effects gests a physiologically relevant approach inhibit cyclooxygenase enzymes (21) and on platelet aggregation and bleeding risk, to improving the inflammatory milieu as- thus are unlikely to act via this proposed is a logical choice for study. sociated with obesity. More extensive mechanism. Recently, high doses of salicy- Side effects reported in the current studies are needed to evaluate the use of lates have been recognized to inhibit NF-␬B study were tinnitus, headache, rash, and anti-inflammatory strategies in general activity (9), possibly through inhibition of transient elevations in ALT and AST. Tinni- and salsalate specifically as preventative IKK␤ (22). The direct role of the IKK␤/ tus, an expected side effect, was reported in and therapeutic interventions in obesity NF-␬B pathway in development of diet- two patients receiving active therapy and and diabetes. induced obesity and insulin resistance has one receiving placebo. Symptoms resolved been validated in animal models (23,24). In with dose adjustment as anticipated. Rash addition, insulin resistance in suppressor of occurred at a rate higher than reported by Acknowledgments— This work was sup- cytokine signaling 3 knockout mice is im- registration trials (31). The reason for the ported by the National Institutes of Health proved by sodium salicylate, suggesting an increased rate of this side effect is not appar- (Grants K23-DK02795 to A.B.G., P50- inflammatory mechanism of action (25). ent in this group without a history of allergy HL083813 to A.B.G. and S.E.S., P30-DK36836, Although we did not directly assess NF-␬B to nonsteroidal anti-inflammatory drugs. and General Clinical Research Center Grant or suppressor of cytokine signaling 3 activ- There was no evidence of respiratory dis- M01-RR001032), a Lilly Fellowship Grant, and ity, we hypothesize that the mechanism by tress or anaphylaxis in our participants. The Clinical Investigator Training Program, Harvard- which salsalate improves glycemia is via the transient, mild ALT and AST elevation is a MIT Division of Health Sciences and Technolo- established effect of salicylates on the IKK␤/ rare reported side effect, and attribution to gy-Beth Israel Deaconess Medical Center, with NF-␬B signaling pathways. the medication is unclear. Obesity itself is support from Pfizer and Merck (A.F.). We thank the patients and staff of the Joslin Other mechanisms could contribute associated with mild elevations in serum Diabetes Center General Clinical Reasearch to the altered metabolic parameters dem- ALT and AST and with hepatic steatosis Center, Dr. Laura Coombs, George Washing- onstrated. Adiponectin is an adipocyte- and/or inflammation (32). Recent estimates ton Biostatistical Center, and Drs. Ernie derived cytokine, which has insulin suggest that Ͼ33% of obese individuals Schaefer and Bela Asztalos, New England sensitizing, anti-inflammatory, and anti- may have hepatic steatosis (33). Therefore, Medical Center.

DIABETES CARE, VOLUME 31, NUMBER 2, FEBRUARY 2008 293 Salsalate improves glycemia in obesity

Augsburg cohort study, 1984–1998. Arch 30. Roth S, Bennett R, Caldron P, Hartman R, References Intern Med 163:93–99, 2003 Mitchell C, Doucette M, Ekholm B, Gold- 1. Hansson GK: Inflammation, atherosclero- 17. Duncan BB, Schmidt MI: The epidemiol- lust B, Lee E, Wilson R: Reduced risk of sis, and coronary artery disease. N Engl ogy of low-grade chronic systemic inflam- NSAID gastropathy (GI mucosal toxicity) J Med 352:1685–1695, 2005 mation and type 2 diabetes. Diabetes with nonacetylated salicylate (salsalate): 2. Shoelson SE, Lee J, Goldfine AB: Inflam- Technol Ther 8:7–17, 2006 an endoscopic study. Semin Arthritis mation and insulin resistance. J Clin Invest 18. Tirosh A, Shai I, Tekes-Manova D, Israeli Rheum 19:11–19, 1990 116:1793–1801, 2006 E, Pereg D, Shochat T, Kochba I, Rudich 31. Disalcid package insert. Northridge, CA, 3. Ebstein W: Zur therapie des diabetes mel- A: Normal fasting plasma glucose levels 3M Pharmaceuticals, 1998 litus, insbesondere u¨ ber die anwendeng and type 2 diabetes in young men. N Engl 32. Sheth SG, Gordon FD, Chopra S: Nonal- der salicylauren natron bei demselben. J Med 353:1454–1462, 2005 coholic steatohepatitis. Ann Intern Med Berl Klin Wochenschr 13:337–340, 1876 19. Gilgore SG, Rupp JJ: Response of blood 126:137–145, 1997 4. Williamson R: On the treatment of glyco- glucose to intravenous salicylate. Metabo- 33. Choi JW: Association between elevated suria and diabetes mellitus with sodium lism 10:419–421, 1961 serum hepatic enzyme activity and total salicylate. Br Med J 1:760–762, 1901 20. Baron SH: Salicylates as hypoglycemic body fat in obese humans. Ann Clin Lab Sci 5. Reid J, Macdougall AI, Andrews MM: As- agents. Diabetes Care 5:64–71, 1982 33:257–264, 2003 pirin and diabetes mellitus. Br Med J 33: 21. Loll PJ, Picot D, Garavito RM: The struc- 34. Dandona P, Aljada A, Ghanim H, Mo- 1071–1074, 1957 tural basis of aspirin activity inferred from hanty P, Tripathy C, Hofmeyer D, 6. Gilgore S: The influence of salicylate on hy- the crystal structure of inactivated prosta- Chaudhuri A: Increased plasma concen- perglycemia. Diabetes 9:392–393, 1960 glandin H2 synthase. Nat Struct Biol tration of macrophage migration inhibi- 7. Hundal RS, Petersen KF, Mayerson AB, 2:637–643, 1995 tory factor (MIF) and MIF mRNA in Randhawa PS, Inzucchi S, Shoelson SE, 22. Yin MJ, Yamamoto Y, Gaynor RB: The an- mononuclear cells in the obese and the Shulman GI: Mechanism by which high- ti-inflammatory agents aspirin and salicy- suppressive action of metformin. J Clin dose aspirin improves glucose metabo- late inhibit the activity of I␬B kinase-␤. Endocrinol Metab 89:5043–5047, 2004 lism in type 2 diabetes. J Clin Invest 109: Nature 396:77–80, 1998 35. Isoda K, Young JL, Zirlik A, MacFarlane 1321–1326, 2002 23. Kim JK, Kim YJ, Fillmore JJ, Chen Y, LA, Tsuboi N, Gerdes N, Schonbeck U, 8. Yuan M, Konstantopoulos N, Lee J, Han- Moore I, Lee J, Yuan M, Li ZW, Karin M, Libby P: Metformin inhibits proinflam- sen L, Li ZW, Karin M, Shoelson SE: Re- Perret P, Shoelson SE, Shulman GI: Pre- matory responses and nuclear factor-␬B versal of obesity- and diet-induced insulin vention of fat-induced insulin resistance in human vascular wall cells. Arterioscler resistance with salicylates or targeted dis- by salicylate. J Clin Invest 108:437–446, Thromb Vasc Biol 26:611–617, 2006 ruption of Ikk␤. Science 293:1673–1677, 2001 36. Mohanty P, Aljada A, Ghanim H, Hofm- 2001 24. Arkan MC, Hevener AL, Greten FR, eyer D, Tripathy D, Syed T, Al-Haddad 9. Kopp E, Ghosh S: Inhibition of NF-␬Bby Maeda S, Li ZW, Long JM, Wynshaw-Bo- W, Dhindsa S, Dandona P: Evidence for a sodium salicylate and aspirin. Science 265: ris A, Poli G, Olefsky J, Karin M: IKK-␤ potent antiinflammatory effect of rosigli- 956–959, 1994 links inflammation to obesity-induced in- tazone. J Clin Endocrinol Metab 89:2728– 10. Estes D, Kaplan K: Lack of platelet effect sulin resistance. Nat Med 11:191–198, 2735, 2004 with the aspirin analog, salsalate. Arthritis 2005 37. Aljada A, Garg R, Ghanim H, Mohanty P, Rheum 23:1303–1307, 1980 25. Torisu T, Sato N, Yoshiga D, Kobayashi T, Hamouda W, Assian E, Dandona P: Nuclear 11. Mielants H, Veys EM, Verbruggen G, Yoshioka T, Mori H, Iida M, Yoshimura A: factor-␬B suppressive and inhibitor-␬B Schelstraete: Comparison of serum salic- The dual function of hepatic SOCS3 in stimulatory effects of troglitazone in obese ylate levels and gastro-intestinal blood insulin resistance in vivo. Genes Cells 12: patients with type 2 diabetes: evidence of an loss between salsalate (Disalcid)and other 143–154, 2007 antiinflammatory action? J Clin Endocrinol forms of salicylates. Scand J Rheumatol 10: 26. Berg AH, Combs TP, Du X, Brownlee M, Metab 86:3250–3256, 2001 169–173, 1981 Scherer PE: The adipocyte-secreted pro- 38. Kitabchi AE, Temprosa M, Knowler WC, 12. Diagnosis and classification of diabetes tein Acrp30 enhances hepatic insulin ac- Kahn SE, Fowler SE, Haffner SM, Andres mellitus. Diabetes Care 28:S37–S42, 2005 tion. Nat Med 7:947–953, 2001 R, Saudek C, Edelstein SL, Arakaki R, 13. Wallace TM, Levy JC, Matthews DR: Use 27. Yamauchi T, Kamon J, Waki H, Terauchi Murphy MB, Shamoon H: Role of insulin and abuse of HOMA modeling. Diabetes Y, Kubota N, Hara K, Mori Y, Ide T, Mu- secretion and sensitivity in the evolution Care 27:1487–1495, 2004 rakami K, Tsuboyama-Kasaoka N, Ezaki of type 2 diabetes in the diabetes preven- 14. April P, Abeles M, Baraf H, Cohen S, Cur- O, Akanuma Y, Gavrilova O, Vinson C, tion program: effects of lifestyle interven- ran N, Doucette M, Ekholm B, Goldlust B, Reitman ML, Kagechika H, Shudo K, tion and metformin. Diabetes 54:2404– Knee CM, Lee E, et al.: Does the acetyl Yoda M, Nakano Y, Tobe K, Nagai R, 2414, 2005 group of aspirin contribute to the antiin- Kimura S, Tomita M, Froguel P, Kad- 39. Gerstein HC, Yusuf S, Bosch J, Pogue J, flammatory efficacy of salicylic acid in the owaki T: The fat-derived hormone adi- Sheridan P, Dinccag N, Hanefeld M, Hoog- treatment of rheumatoid arthritis? Semin ponectin reverses insulin resistance werf B, Laakso M, Mohan V, Shaw J, Zin- Arthritis Rheum 19:20–28, 1990 associated with both lipoatrophy and man B, Holman RR: Effect of rosiglitazone 15. Executive Summary of the Third Report obesity. Nat Med 7:941–946, 2001 on the frequency of diabetes in patients with of the National Cholesterol Education 28. You T, Yang R, Lyles MF, Gong D, Nicklas impaired glucose tolerance or impaired fast- Program (NCEP) Expert Panel on Detec- BJ: Abdominal adipose tissue cytokine ing glucose: a randomised controlled trial. tion, Evaluation, and Treatment of High gene expression: relationship to obesity Lancet 368:1096–1105, 2006 Blood Cholesterol in Adults (Adult Treat- and metabolic risk factors. Am J Physiol 40. Buchanan TA, Xiang AH, Peters RK, Kjos ment Panel III). JAMA 285:2486–2497, 288:E741–E747, 2005 SL, Marroquin A, Goico J, Ochoa C, Tan 2001 29. Tripathy D, Mohanty P, Dhindsa S, Syed S, Berkowitz K, Hodis HN, Azen SP: Pres- 16. Thorand B, Lowel H, Schneider A, Kolb T, Ghanim H, Aljada A, Dandona P: Ele- ervation of pancreatic ␤-cell function and H, Meisinger C, Frohlich M, Koenig W: vation of free fatty acids induces inflam- prevention of type 2 diabetes by pharma- C-reactive protein as a predictor for inci- mation and impairs vascular reactivity in cological treatment of insulin resistance in dent diabetes mellitus among middle- healthy subjects. Diabetes 52:2882–2887, high-risk Hispanic women. Diabetes 51: aged men: results from the MONICA 2003 2796–2803, 2002

294 DIABETES CARE, VOLUME 31, NUMBER 2, FEBRUARY 2008