Effects of Gevokizumab on Glycemia and Inflammatory Markers in Type 2

Clinical Care/Education/Nutrition/Psychosocial Research ORIGINAL ARTICLE

Effects of Gevokizumab on Glycemia and Inﬂammatory Markers in Type 2 Diabetes

1 3 CLAUDIA CAVELTI-WEDER, MD HANY ZAYED, PHD (IL-1Ra), was tested in a placebo-controlled 1 4 ANDREA BABIANS-BRUNNER, MD ALAN M. SOLINGER, MD 1 5 study of 70 patients (5). At 13 weeks of CORNELIA KELLER, MD THOMAS MANDRUP-POULSEN, MD, PHD 2 6 treatment, glycated hemoglobin was signif- MARC A. STAHEL, MD CHARLES A. DINARELLO, MD 2 1 icantly improved as a result of enhanced MALAIKA KURZ-LEVIN, MD MARC Y. DONATH, MD b-cell secretory function. Of interest, some improvement promoted by IL-1b blockade lasted for at least 39 weeks follow- OBJECTIVEd b Metabolic activation of the innate immune system governed by interleukin (IL)-1 ing treatment withdrawal (6). More re- contributes to b-cell failure in type 2 diabetes. Gevokizumab is a novel, human-engineered monoclonal anti–IL-1b antibody. We evaluated the safety and biological activity of gevokizumab in cently, IL-1Ra also was shown to improve patients with type 2 diabetes. insulin secretion in prediabetic patients (7). This indicates the disease-modifying po- RESEARCH DESIGN AND METHODSdIn a placebo-controlled, dose-escalation tential of this therapy (8). study, a total of 98 patients were randomly assigned to placebo (17 subjects) or gevokizumab Although the results obtained with (81 subjects) at increasing doses and dosing schedules. The primary objective of the study was to fi IL-1Ra in patients with type 2 diabetes are evaluate the safety pro le of gevokizumab in type 2 diabetes. The secondary objectives were to assess encouraging, IL-1Ra has characteristics pharmacokinetics for different dose levels, routes of administration, and regimens and to assess that make it unsuitable for the treatment biological activity. of type 2 diabetes, such as its short half-life RESULTSdThe study drug was well tolerated with no serious adverse events. There was one and injection-site reactions (9). It would hypoglycemic event whereupon concomitant insulin treatment had to be reduced. Clearance of have to be taken on a daily basis to maintain gevokizumab was consistent with that for a human IgG2, with a half-life of 22 days. In the adequate suppression of IL-1b. Moreover, combined intermediate-dose group (single doses of 0.03 and 0.1 mg/kg), the mean placebo- the amount of IL-1Ra required would make correcteddecreaseinglycatedhemoglobinwas0.11,0.44,and0.85%after1,2(P =0.017),and3 P treatment an expensive option. Of impor- ( = 0.049) months, respectively, along with enhanced C-peptide secretion, increased insulin tance, an anti–IL-1 strategy that spares sensitivity, and a reduction in C-reactive protein and spontaneous and inducible cytokines. IL-1a may offer safety benefits. Therefore, CONCLUSIONSdThis novel IL-1b–neutralizing antibody improved glycemia, possibly via specifically reducing IL-1b activity with restored insulin production and action, and reduced inflammation in patients with type 2 longer-lasting agents would provide a diabetes. This therapeutic agent may be able to be used on a once-every-month or longer therapeutic improvement in terms of sus- schedule. tained inhibition of IL-1b–mediated islet- associated inflammation. Diabetes Care – 35:1654 1662, 2012 Gevokizumab is a recombinant human- engineered monoclonal antibody that nset of type 2 diabetes occurs when to a pathological activation of the innate binds and neutralizes human IL-1b pancreatic b-cells fail to adapt to the immune system by metabolic stress and is with a Kd of 0.3 pmol/L (10). The predicted O – increased insulin demand caused governed by interleukin (IL)-1 signaling circulating half-life of the antibody was 22 by insulin resistance. Morphological and (1–4). 25 days. The Fc of gevokizumab is the IgG2 therapeutic intervention studies have un- Based on the above-described islet isotype, which unlike IgG1 Fc does not ac- covered an inflammatory process in islets inflammation with a predominant role for tivate the FcRIII on macrophages and other of patients with type 2 diabetes character- IL-1b, we initiated clinical trials of IL-1 myeloid cells, thereby reducing the proba- ized by the presence of cytokines, im- antagonism in type 2 diabetes. In a proof- bility of antibody-dependent cell-mediated mune cells, b-cell apoptosis, amyloid of-concept study, the naturally occurring cytotoxicity. In this study, we evaluated the deposits, and fibrosis.Thisisattributed antagonist of IL-1b,IL-1receptoragonist safety, pharmacokinetics, and diabetes- and inflammation-related parameters of ccccccccccccccccccccccccccccccccccccccccccccccccc gevokizumab in patients with type 2 diabetes. From the 1Department of Endocrinology, Diabetes, and Metabolism, University Hospital Basel, Basel, Switzerland; the 2Department of Ophthalmology, University Hospital Zurich, Zurich, Switzerland; the 3Department of Biostatistics, XOMA, Berkeley, California; the 4Department of Clinical Development, XOMA, Berkeley, RESEARCH DESIGN AND 5 California; the Department of Biomedical Sciences, University of Copenhagen, Copenhagen, Denmark, METHODSdThis first-in-humans, and the Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden; and the 6Department of Medicine, University of Colorado Denver, Aurora, Colorado. randomized, placebo-controlled, dose- Corresponding author: Marc Y. Donath, [email protected]. escalation study was carried out concur- Received 15 November 2011 and accepted 27 March 2012. rently in the U.S. and Switzerland. The U.S. DOI: 10.2337/dc11-2219. Clinical trial reg. no. NCT00541983, clinicaltrials.gov. arm of the study included 68 subjects (56 A slide set summarizing this article is available online. © 2012 by the American Diabetes Association. Readers may use this article as long as the work is properly in the gevokizumab group and 12 in the cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ placebo group) and took place between licenses/by-nc-nd/3.0/ for details. September 2007 and June 2009. The Swiss

1654 DIABETES CARE, VOLUME 35, AUGUST 2012 care.diabetesjournals.org Cavelti-Weder and Associates arm of the study included 30 subjects (25 aminotransferase/alanine aminotransferase intravenous stimulation tests of C-peptide in the gevokizumab group and 5 in the .2 3 upper limit of normal; alkaline phos- release were included. A glucose stimula- placebo group) and took place between phatase .2 3 upper limit of normal; tion test was performed at baseline and at November 2007 and July 2009. XOMA positive test for GAD65 or insulinoma- 4 and 13 weeks to assess b-cell function (U.S.), the manufacturer of gevokizumab, associated protein 2 autoantibodies; and insulin production/reserve. Subjects sponsored the study. known HIV antibody positivity, hepatitis B ingested 75 g glucose and were sampled The eligibility criteria of both arms of surface antigens, and/or hepatitis C anti- for plasma glucose and C-peptide mea- the study were similar. Patients were re- bodies; history or evidence of thyroid abnor- surements at –10, 0, 30, 60, 90, and 120 quested to continue stable doses of baseline malities, including active hyperthyroidism min. Immediately at the end of the oral antidiabetes treatments (metformin, sulfo- needing medication; abnormal T3, T4, glucose tolerance test, an intravenous in- nylureas, and/or insulin) and were asked thyroglobulin, or thyroid-stimulating hor- jection of 0.3 g glucose/kg body wt, 0.5 mg not to change their diet and exercise mone levels; infectious disease (C-reactive glucagon, and 4.5 g arginine was adminis- habits during the study. Both arms of the protein [CRP] .30 mg/L), fever, or infec- tered. Blood then was sampled at 0, 3, 6, 9, study tested single intravenous doses of tion requiring treatment with antibiotics and 12 min. The patients were asked to gevokizumab at the same five-dose levels within 3 weeks; history of recurrent infec- forgo their antidiabetes medication and (0.01, 0.03, 0.1, 0.3, and 1.0 mg/kg). In tion or predisposition to infection; (active fast for 8–9 h prior to the test. Whole- addition, there was a three-part study in the leg or foot ulcer); malignancy within 5 body insulin sensitivity index was calculat- U.S. arm that tested a single intravenous years prior to study entry other than car- ed as 10,000/square root of [fasting glucose dose of 3.0 mg/kg (part 1), a single sub- cinoma in situ of the cervix or adequately (mmol/L) 3 fasting insulin (pmol/L) 3 cutaneous dosing at three dose levels (part treated, nonmetastatic squamous or basal (mean glucose {mmol/L} 3 mean insulin 2), and multiple subcutaneous dosing at two cell carcinoma of the skin; history of severe {pmol/L} during the oral glucose tolerance dose levels (part 3) (Fig. 1A). The follow-up allergic or anaphylactic reactions to human- test)] (12). of patients was 2 months in the U.S. arm and ized or murine monoclonal antibodies; his- 3 months in the Swiss arm. Randomization tory of tuberculosis, positive purified Whole-blood cultures for ex vivo was performed with the use of a common, protein derivative test, or active atopic dis- cytokine production central, computer-generated system with ease requiring medication; asthma; immu- Heparinized blood was obtained before permutated blocks of six schedules. nodeficiency; history or symptoms of a and on days 1, 7, and 28 after the admin- Throughout the course of the study, demyelinating disease; clinically significant istration of gevokizumab. At the same time, treatment assignment was blinded to study diabetic macular edema and/or proliferative blood was sent for routine hematology. A subjects and investigators but not to the diabetic retinopathy by history or fundo- total of 250 mLofbloodwereaddedto sponsor (XOMA). The sponsor maintained scopy; use of a therapeutic monoclonal 1.5-mL sterile cryotubes (Fisher) con- blinding in all communications with inves- antibody within 90 days; participation in taining 750 mL of RPMI (Grand Island) tigators and other site personnel. The study an investigational drug or device trial within or RPMI with premeasured stimulants was approved by investigational regulations 30 days; major surgery within 28 days prior as follows: lipopolysaccharides (LPS) committees in Switzerland and the U.S. to day 0; receipt of a live (attenuated) vac- (Sigma), such that the final concentration Written informed consent was obtained cine within 3 months; and female subjects was 100 ng/mL; heat-killed Staphylococcus from each patient prior to randomization. who were pregnant, planning to become epidermidis, such that the final concentra- pregnant during the course of the study, tion was 10 bacteria per WBC (assuming Inclusion and exclusion criteria or who were breastfeeding. the average WBC is 7,500 mm3); the com- The inclusion criteria were type 2 diabetes bination of IL-12 plus IL-18, such that the diagnosed according to the American Di- Study end points final concentration of IL-12 (R&D Systems) abetes Association (11), with a duration of The primary objective of the study was to was 2 ng/mL and the final concentration of .3months(.6 months for the U.S. evaluate the safety of gevokizumab in type IL-18 was 20 ng/mL; or LPS plus muramyl arm); age $18 and #70 years; a glycated 2 diabetic patients. The safety assessments dipeptide (Sigma), such that the final con- hemoglobin level at screening $7.5 and included evaluation of adverse events, clin- centration of LPS was 10 ng/mL and the #12% (#12.5% for the U.S. arm); BMI ically significant changes from baseline in final concentration of muramyl dipeptide $23 and #40 kg/m2; stable disease, de- vital signs, electrocardiograms, and labora- was 10 mg/mL. fined as no recent change in type 2 diabe- tory values (hematology, chemistry, and The tubes were capped tightly and tes medications; and the patient’s agreement urinalysis); infusion reactions; and immune incubated at 378C for 24 h. Thereafter, the not to change his or her diet and exercise responses to gevokizumab. The Swiss arm tubes were inverted several times, 100 mL during the study. additionally included diabetic foot exams, of 5% Triton X (Sigma) were added, and, The exclusion criteria were use of anti- thyroid physical exams, and ophthalmic after thorough mixing, the tubes were fro- inflammatory therapy other than aspirin exams, and the U.S. arm included assess- zen at –208C. After last-patient last-visit, (#100 mg/day); use of immunosuppressive ments of forced expiratory volumes. the tubes were thawed and assayed for therapy; use of b2 and nonselective adrener- Secondary objectives comprised phar- total cytokines. Cytokines assayed were gic blockers; use of thiazolidinediones, macokinetics of gevokizumab, standard IL-1a, tumor necrosis factor (TNF) a, glucagon-like peptide agonists, or dipeptidyl diabetic parameters (glycated hemoglobin, interferon (IFN) g, IL-6, and IL-1Ra peptidase-4 inhibitors; fasting C-peptide fasting glucose, and C-peptide), and inflam- using electrochemiluminesence (BioVeris, ,400pmol/L(,1.20 mg/L); hemoglobin matory markers (CRP, WBC counts, and ex Gaithersburg, MD). The electrochemilumi- ,8.0 g/dL; white blood cells (WBCs) vivo cytokine production from whole- nesence assay for cytokines has been de- ,3.0 3 103/mm3; platelet count ,125 3 blood cultures in response to inflammatory scribed previously (13). The coefficient of 103/mm3; creatinine .1.5 mg/dL; aspartate stimulants). In the Swiss arm, oral and variance for each cytokine assay is ,20%. care.diabetesjournals.org DIABETES CARE, VOLUME 35, AUGUST 2012 1655 Gevokizumab in type 2 diabetes

Figure 1dEnrollment, outcomes, and pharmacokinetics. A: A total of 329 patients with type 2 diabetes underwent physical and biochemical screening; 231 were found to be ineligible according to entry criteria. Ninety-eight patients were randomly assigned to receive either gevokizumab or placebo. In the gevokizumab group, one patient was withdrawn after ,1 week. A total of 98 patients completed the 13-week treatment and evaluation of study end points. B, C,andD: Plasma concentrations following a single intravenous (IV) (n =55)(B), subcutaneous (SC) (n =16)(C), or multiple subcutaneous (n =10)(D) doses of gevokizumab. (A high-quality color representation of this ﬁgure is available in the online issue.)

Cytokine levels in picograms per mil- infusion of gevokizumab at 100% for each Statistical analyses liliter of the lysed blood culture were cytokine. For each time point (days 1, 7, Safety and biological activity analyses in- converted to percentage change in cyto- and 28) following the infusion, the per- cluded all subjects who received any kine levels per million WBCs based on the centage change from the preinfusion level amount of study drug (intent-to-treat anal- WBCs at the time of the blood draw for was calculated for each subject. The mean ysis). No interim analyses were carried out each time point. The data then were percent change (6SEM) for each cytokine in the study. Values are expressed as analyzed by setting the cytokine con- and each dose group or placebo was then means 6 SEM. Because of the small sample centration per million WBCs before the calculated for each time point. size, as a result of the dose-escalation design

1656 DIABETES CARE, VOLUME 35, AUGUST 2012 care.diabetesjournals.org Cavelti-Weder and Associates of the study, patients with the single in- metastasizing carcinoma of the appen- and a mean clearance of 2.5 6 0.1 mL/day/kg travenous or subcutaneous injection of dix ,1 week after receiving 0.1 mg/kg (Fig. 1B). The concentration-time profile gevokizumab were combined to placebo/ gevokizumab. The cancer was discovered for the single subcutaneous dose was char- low-, intermediate-, and high-dose groups, after the patient disclosed a chronic recur- acterized by an absorption phase followed as follows, in order to allow statistical test- rent pain of the upper abdomen for many by a one-compartment elimination phase, ing. Patients with placebo and the extremely months; this information was not stated by with a mean elimination half-life of 23.9 6 low dose of 0.01 mg/kg were combined as the patient at screening. The patient re- 0.9 days and a mean clearance of 3.5 6 0.3 the placebo/low-dose group; patients with mained in the study for safety monitoring mL/day/kg (Fig. 1C). A supplementary 0.03 and 0.1 mg/kg as the intermediate- butdidnotperformtheglucosetolerance population pharmacokinetic analysis de- dose group; and patients with 0.3, 1.0, tests. termined the absorption rate constant to and 3.0 mg/kg as the high-dose group. Sta- The patients did not change their anti- be 0.13 6 0.02 mL/day/kg and the bio- tistical differences were tested with the use diabetes medications throughout the availability to be 69 6 4%. Initial and ter- of a Wilcoxon rank sum test. A P value of study, as reflected by recorded types and minal half-lives, clearance, and volume of ,0.05 was considered to indicate statistical doses of antidiabetes medication, expect for distribution were typical for a human anti- significance. All reported P values are two one patient who developed a hypoglycemic body subject to a nonspecific clearance sided. event 7 days after receiving 0.1 mg/kg mechanism. Gevokizumab showed predict- gevokizumab. In this case, insulin dose able dose linearity with maximum concentra- RESULTS was reduced by 4 units. tions proportional through the observed dose range. Samples from three subjects Baseline characteristics and Primary end point were confirmed positive with antibody compliance Safety. Gevokizumab administered over against gevokizumab. The titer levels for all Ninety-eight of 329 patients initially two logarithmic dose intervals was safe and threesubjectswerelow(,250 pg/L) and did screened were randomly assigned to placebo well tolerated. There were two serious ad- not impact on the half-life of gevokizumab. or treatment with gevokizumab (Fig. 1A). verse events, including a carcinoma of the Glycemia. Glycated hemoglobin decreased The most frequent reason for ineligibility appendix (see above) and a carotid artery in a dose-dependent, U-shaped manner, was an HbA1c level ,7.5%. Table 1 shows occlusion, which occurred 44 days after the with the greatest decline from baseline the baseline characteristics of patients after last injection of the study drug at a dose of glycated hemoglobin seen in the 0.1 mg/kg randomization. Most demographic and 0.03 mg/kg. Both adverse events were not dose group at all time points. After 1 month, baseline characteristics were well balanced judged as related to the study drug. One the mean HbA1c reduction from baseline between the groups. The exceptions were patient experienced hypoglycemia. He also in the intermediate-dose group (0.03 and median glycated hemoglobin in the single- was treated with insulin and remained 0.1 mg/kg) was 20.41 (60.09) and 20.30% dose intravenous subjects, which at 8.6% asymptomatic after reduction of the insulin (60.10) in the placebo/low-dose group was slightly lower than the other groups; dose. (placebo and 0.01 mg/kg), respectively duration of diabetes ranking from 5.8 to (P = 0.46) (Fig. 2A). After 2 months, the 9.7 years; and differences in treatment. Secondary end points intermediate-dose group had a mean reduc- None of these differences were statistically Pharmacokinetics. The pharmacokinetics tion from baseline in HbA1c of 20.49% significant. of intravenous gevokizumab demonstrated (60.12) compared with 20.05% (60.13) All subjects but one completed the a dose-proportional biexponential decline, in the placebo/low-dose group, which study. That patient was diagnosed with with a mean b half-life of 22.4 6 0.9 days was a statistically significant difference

Table 1dBaseline characteristics of the patients

Single Single Subcutaneous Characteristics Placebo intravenous dose subcutaneous dose multidose n 17 55 20 10 Age (years) 50 (34–65) 54 (30–70) 50 (40–67) 52 (34–66) Sex (% male/female)* 82/18 69/31 75/25 40/60 BMI (kg/m2)31(21–41) 30 (24–37) 29 (24–39) 35 (29–41) CRP (mg/L) 2.5 (0.6–11.8) 2.1 (0.5–16.3) 2.1 (0.8–14.1) 7.5 (2.1–11.7) Glycated hemoglobin (%) 9.1 (7.6–11.6) 8.6 (6.9–12.0) 9.0 (7.0–11.3) 8.9 (7.0–11.1) Fasting plasma glucose (mg/dL) 234 (139–333) 190 (94–398) 160 (132–339) 230 (137–267) Fasting C-peptide (ng/mL) 2.5 (1.5–6.1) 2.7 (0.9–9.0) 2.4 (1.1–3.9) 2.5 (1.2–4.2) Disease duration (years) 9.7 (2.7–19) 5.8 (0.5–28) Antidiabetes treatment (%) 18.8 30 Diet only 23.5 9.1 37.5 10 Metformin only 11.8 25.5 Sulfonylurea only 11.8 0 12.5 0 Metformin and sulfonylurea 23.5 21.8 12.5 50 Insulin alone or in combination 29.4 43.6 18.8 10 Data are median (range) unless otherwise indicated. *Data are frequency (percent). care.diabetesjournals.org DIABETES CARE, VOLUME 35, AUGUST 2012 1657 Gevokizumab in type 2 diabetes

(P =0.017)(Fig.2B). This significant re- (P = 0.0063), with strong trends after 1 was not reduced at the dose of gevokizumab P duction in HbA1c levels still was present and 2 months ( = 0.08 and 0.05, respec- most potently reducing CRP, suggesting a after 3 months (decrease of 20.68% tively). This effect was less pronounced in reprogramming effect of 28 days of IL-1b [60.25] from baseline in the intermediate- the high-dose group, reflecting the same inhibition in the myeloid compartment of dose group compared with an increase of dose effect as for HbA1c and area under the bone marrow independent of effects +0.17% [60.29] in the placebo/low-dose the curve of C-peptide. As expected, neu- on systemic inflammation. However, the group; P = 0.049) (Fig. 2C). Thus, the pla- trophil and platelet counts were slightly production of the anti-inflammatory IL- cebo-corrected mean glycated hemoglo- decreased by gevokizumab but remained 1Ra was unaffected by the same doses bin decrease was 20.11, 20.44, and within the normal range (data not of gevokizumab that reduced CRP and 20.85% after 1, 2, and 3 months, respec- shown). glycated hemoglobin. Of importance, the tively, between the intermediate-dose and The anti-inflammatory effect of production of IFNg induced by the com- placebo/low-dose groups. gevokizumab on ex vivo cytokine produc- bination of IL-12 plus IL-18 was unchanged The decrease from baseline in mean tion also was tested in whole-blood cul- in gevokizumab-treated patients (not glycated hemoglobin in the high-dose tures in patients from the Swiss arm of the shown), in accordance with other studies group (0.3, 1.0, and 3.0 mg/kg) did not study. Gevokizumab decreased total TNFa of gevokizumab measuring the produc- reach statistical significance after 1, 2, or 3 production induced by LPS (Fig. 3A–D). tion of IFNg (16) as well as in mice de- months, respectively, consistent with the Moreover, spontaneous production of IL-6 ficient in IL-1b.IFNg is a key cytokine in U-shaped dose-response curve. In addi- from peripheral blood cells of patients protecting the host from opportunistic tion, the small subgroup of patients (n =5 treated with gevokizumab was time depen- infections (17,18). In over 100,000 pa- for each 0.03 and 0.3 mg/kg) receiving three dently decreased, most potently at 0.3 mg/kg tients with rheumatoid arthritis treated consecutive (baseline and days 14 and 28) on day 28 (Fig. 3E–H). In contrast, the with anakinra, there are no reports of in- subcutaneous injections of gevokizumab anti-inflammatory cytokines IL-1Ra as creased opportunistic infection, which is did not reach statistical significant changes well as IFNg remained unaffected by the in sharp contrast to patients treated with in HbA1c. treatment (not shown). TNFa-blocking therapies (18). These b-Cell secretory function. Along with data suggest that the IFNg-sparing effects the changes in glycated hemoglobin, mean CONCLUSIONSdThis single- and of blocking IL-1b is associated with an C-peptide release in the combined oral multiple-dose study with the anti–IL-1b advantageous risk profile of gevokizumab glucose and intravenous stimulation tests antibody, gevokizumab, provides prelim- and other IL-1 antagonists with regard to was significantly increased from baseline in inary data for a disease-modifying ap- opportunistic infections. the intermediate-dose group after 1 month proach to type 2 diabetes by targeting The most beneficial and consistent (+43.4 6 14.8 nmol/L z min) but not in the the master inflammatory cytokine impair- effects on glycemia, C-peptide secretion, placebo/low-dose group (23.6 6 13.4 ing the insulin-producing b-cells. Of and inflammatory parameters appeared nmol/L z min; P =0.045)(Fig.2D). After note, because the IL-1 receptor antagonist at intermediate doses of gevokizumab. 3 months, an increase in mean C-peptide used in an earlier trial in type 2 diabetes Unexpectedly, with increasing doses, a release from baseline was observed in the (5,6) blocks receptor binding and signal- U-shaped dose-response relationship was intermediate-dose group (+25.5 6 28.9 ing in response to both IL-1a and -b, this observed. Although no toxic effects were nmol/L z min) compared with the pla- is the first study to show that specific neu- observed at the higher concentrations of cebo/low-dose group (210.3 6 14.9 tralization of IL-1b yields similar effects. the anti–IL-1b antibody, a decrease in ef- nmol/L z min), although not reaching sta- If confirmed, this observation provides ficacy was apparent. Assuming that this tistical significance (P = 0.39) (Fig. 2E). evidence that IL-1b antagonism alone preliminary observation is confirmed, Insulin sensitivity measures. Whole- may be sufficient to reduce b-cell failure, one explanation may be the physiologic body insulin sensitivity index was slightly, allowing for a therapy that spares the ac- role of inflammation. Indeed, very low but not significantly, increased in the tion of IL-1a, providing a safety advan- concentrations of IL-1b promote insulin intermediate-dose group after 1 month tage over IL-1 receptor blockade. secretion and b-cell proliferation (19). At (mean increase +0.10 6 0.12 compared No evidence of safety or pharmacoki- higher doses, the potency of gevokizumab with baseline; P = 0.2). At the same time, netic issues for gevokizumab was identified. in neutralizing endogenous IL-1b may de- there was a decrease in the high-dose and The effects observed on glycemia, C-peptide prive the b-cell of these physiologic levels placebo/low-dose groups (20.33 6 0.29 secretion, and CRP support previous studies of IL-1b. Likewise, excessive IL-1b and 20.16 6 0.14 decrease compared using IL-1Ra (5–7) and are in line with more plays a pathological role in many inflam- with baseline, respectively). recent studies using other neutralizing anti- matory diseases (9), but low levels serve Systemic and leukocyte inflammation. bodies to IL-1b (14,15). host defense, promote healing, and en- In parallel to HbA1c values, levels of CRP We compared ex vivo cytokine pro- hance growth factor production. Further- were lowered in a dose-dependent, duction from cultured whole blood from more, IL-1b governs a large panel of U-shaped fashion (Fig. 2F–I), with the patients treated with either gevokizumab cytokines and chemokines, some having most pronounced decreases at the earlier or placebo and observed a consistent beneficial effects under specificcondi- time points. When dose groups were reduction in the total TNFa as well as tions: IL-1b causes recruitment of macro- combined as described above and CRP IL-6 synthesized at all doses of gevokizumab. phages to islets, which, depending on the levels .20 mg/dL were excluded from Although such a reduction was antici- level of activation, may be beneficial or the analysis because they suggested ongoing pated since IL-1b can induce TNFa deleterious to the b-cell secretory function infection, there was a statistically different and IL-6, the greatest reduction in IL-6, (20,21). Furthermore, IL-6, which also is change in CRP at day 14 in the intermedi- but not of TNFa, was observed 28 days regulated by IL-1b, promotes b-cell func- ate-dosegroupcomparedwithbaseline after the infusion of gevokizumab, and IL-6 tion via glucagon-like peptide-1 production

1658 DIABETES CARE, VOLUME 35, AUGUST 2012 care.diabetesjournals.org Cavelti-Weder and Associates

Continued on p. 1660 care.diabetesjournals.org DIABETES CARE, VOLUME 35, AUGUST 2012 1659 Gevokizumab in type 2 diabetes

Figure 2dChanges in glycated hemoglobin, b-cell secretory function, and CRP during the study period. Average absolute differences in glycated hemoglobin levels betweenbaselineand1(A), 2 (B), and 3 (C) months in each study group (n = 15 for placebo, n =10for0.01mg/kg,n =15for 0.03 mg/kg, n = 16 for 0.1 mg/kg, n = 15 for 0.3 mg/kg, n = 10 for 1.0 mg/kg, and n = 5 for 3.0 mg/kg at 1 and 2 months and n = 5 for all groups at the 3-month time point). b-Cell secretory function was assessed by a 2-h oral glucose tolerance test, followed by intravenous stimulation with 0.3 g glucose/kg body wt, 0.5 mg glucagon, and 4.5 g arginine. Average absolute differences between baseline and 1 (D) and 3 (E) months are shown (n =5 for placebo and 0.01, 0.03, 0.1, 0.3, and 1.0 mg/kg). Average absolute differences in CRP between baseline and 2 weeks (F) and 1 (G), 2 (H), and 3 (I) months in each study group (n = 15 for placebo, n = 10 for 0.01 mg/kg, n = 15 for 0.03 mg/kg, n = 16 for 0.1 mg/kg, n = 15 for 0.3 mg/kg, n = 10 for 1.0 mg/kg, and n = 5 for 3.0 mg/kg at 2 weeks and 1 and 2 months and n = 5 for groups at the 3-month time point). The data are depicted as means 6 SEM.

(22). We conclude, therefore, that an appro- A growing number of animal studies level and type of inflammation, and blood priate dose and duration of gevokizumab have demonstrated a role for IL-1b in in- flow between islets and insulin-sensitive treatment will be crucial in order to reshape sulin resistance (23–26). In humans, tissues, other doses and duration of treat- the immune system in patients with type 2 studies hint at a role for IL-1b in insulin ment may be necessary to uncover and diabetes. In particular, the very high affinity resistance (23,27), and a study in patients make use of these effects. In addition, patients of gevokizumab with responses observable with prediabetes treated with anakinra were not subjected to hyperinsulinemic- at concentrations as low as femtomolar pointed to improvement in insulin sensi- euglycemic clamps, which may have un- concentrations offers obvious pharmaco- tivity; however, only in patients without coveredsucheffects. logical advantages, such as dosage vol- injection site reactions (7). In the current The improvement of HbA1c could be ume, but its use should be limited at study, a tendency toward such effects also attributed to enhanced insulin secretion or high doses. was observed. Because of differences in action. On the one hand, the improvement

1660 DIABETES CARE, VOLUME 35, AUGUST 2012 care.diabetesjournals.org Cavelti-Weder and Associates

with less cardiovascular risk in type 2 diabetes. Accordingly, a large phase 3 study using an anti–IL-1b antibody in 17,000 patients recently has been launched (28). This should allow uncovering the full potential of IL-1b antagonism in the treatment of cardiovascular disease and diabetes.

AcknowledgmentsdThis work was supported by the Swiss National Science Foundation, Na- tional Institutes of Health Grant AI-15614 (to C.A.D.), the Novo Nordisk Foundation (to T.M.-P.), and XOMA. M.A.S. and H.Z. are employees and stock- holders of XOMA. M.Y.D. has a patent for the use of IL-1 antagonism in type 2 diabetes. No other potential conﬂicts of interest relevant to this article were reported. C.C.-W., T.M.-P., C.A.D., and M.Y.D. de- signed the study. C.C.-W., A.B.-B., C.K., M.A.S., and C.A.D. performed the experiments and researched data. C.C.-W., A.M.S., T.M.-P., C.A.D., and M.Y.D. analyzed data, contributed to the discussion, and wrote, reviewed, and edited the manuscript. M.Y.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors thank Patrizia Zala (University Hospital Basel) for her help in performing the study and to Leonid Reznikov and Tania Azam (University of Colorado Denver) for assistance in the whole-blood assay.

References 1. Donath MY, Böni-Schnetzler M, Ellingsgaard H, Ehses JA. Islet inflammation impairs the pancreatic beta-cell in type 2 diabetes. Phys- iology (Bethesda) 2009;24:325–331 2. Donath MY, Schumann DM, Faulenbach M, Ellingsgaard H, Perren A, Ehses JA. Islet inflammation in type 2 diabetes: from metabolic stress to therapy. Diabetes Care Figure 3dCytokine changes in LPS-induced TNFa and spontaneous IL-6 production in whole- 2008;31(Suppl. 2):S161–S164 blood cultures. A: LPS-induced TNFa in patients receiving placebo (n =5).B: LPS-induced TNFa 3. Donath MY, Shoelson SE. Type 2 diabetes in patients receiving 0.1 mg/kg gevokizumab (n =5).C: LPS-induced TNFa in patients receiving as an inflammatory disease. Nat Rev Im- 0.03 mg/kg gevokizumab (n =5).D: Dose-response of gevokizumab on LPS-induced TNFa on day munol 2011;11:98–107 7 in patients treated with gevokizumab. E: Spontaneous production of IL-6 in patients receiving 4. Maedler K, Sergeev P, Ris F, et al. Glucose- placebo (n =5).F: Spontaneous production of IL-6 in patients receiving 0.1 mg/kg gevokizumab induced beta cell production of IL-1beta (n =5).G: Spontaneous production of IL-6 in patients receiving 0.3 mg/kg gevokizumab (n =5). contributes to glucotoxicity in human H: Dose response of gevokizumab on spontaneous production of IL-6 on day 28. The data are pancreatic islets. J Clin Invest 2002;110: expressed as mean (6SEM) percentage change on days 1, 7, and 28 from baseline levels per 851–860 million WBCs before the infusion of either placebo or gevokizumab. Details on the calculations 5. Larsen CM, Faulenbach M, Vaag A, et al. are described in RESEARCH DESIGN AND METHODS. Interleukin-1-receptor antagonist in type 2 diabetes mellitus. N Engl J Med 2007; 356:1517–1526 6. Larsen CM, Faulenbach M, Vaag A, Ehses JA, of HbA1c did not show a clear correlation mechanism of improvement of glycemic with the increase of C-peptide secretion. parameters. Donath MY, Mandrup-Poulsen T. Sustained effects of interleukin-1 receptor antagonist On the other hand, the improvement in in- Levels of CRP were consistently de- treatment in type 2 diabetes. Diabetes Care sulin sensitivity failed to reach statistical creased by gevokizumab in the current 2009;32:1663–1668 significance. However, the small sample study and consistent with other IL-1 block- 7. van Asseldonk EJ, Stienstra R, Koenen TB, size of our study does not allow for a de- ing agents (9). A reduction in CRP often is Joosten LA, Netea MG, Tack CJ. Treatment finitive conclusion on the underlying interpreted as beneficial and associated with anakinra improves disposition index but care.diabetesjournals.org DIABETES CARE, VOLUME 35, AUGUST 2012 1661 Gevokizumab in type 2 diabetes

not insulin sensitivity in nondiabetic subjects 15. Sloan-Lancaster J, Polzer J, Miller JW, in type 2 diabetes. Diabetes 2007;56: with the metabolic syndrome: a randomized, Scherer JC, Berg JK, Landschulz WH. 2356–2370 double-blind, placebo-controlled study. Safety, tolerability and efficacy of subcuta- 22. Ellingsgaard H, Hauselmann I, Schuler B, J Clin Endocrinol Metab 2011;96:2119–2126 neous LY2189102, a neutralizing IL-1b et al. Interleukin-6 enhances insulin secre- 8. Donath MY, Mandrup-Poulsen T. The use antibody, in patients with type 2 diabetes. tion by increasing glucagon-like peptide-1 of interleukin-1-receptor antagonists in the Late-breaking abstract presented at the secretion from L cells and alpha cells. Nat treatment of diabetes mellitus. Nat Clin 71st Scientific Sessions of the American Med 2011;17:1481–1489 Pract Endocrinol Metab 2008;4:240–241 Diabetes Association, 24–28 June 2011, 23. Koenen TB, Stienstra R, van Tits LJ, et al. 9. Dinarello CA. Interleukin-1 in the patho- San Diego Convention Center, San Diego, Hyperglycemia activates caspase-1 and genesis and treatment of inflammatory California TXNIP-mediated IL-1b transcription in diseases. Blood 2011;117:3720–3732 16. Gül A, Tugal-Tutkun I, Dinarello CA, et al. human adipose tissue. Diabetes 2011;60: 10. Roell MK, Issafras H, Bauer RJ, et al. Kinetic Interleukin-1b-regulating antibody XOMA 517–524 approach to pathway attenuation using 052 (gevokizumab) in the treatment of 24. Stienstra R, Joosten LA, Koenen T, et al. XOMA 052, a regulatory therapeutic anti- acute exacerbations of resistant uveitis The inflammasome-mediated caspase-1 body that modulates interleukin-1beta ac- of Behcet’s disease: an open-label pilot activation controls adipocyte differentia- tivity. J Biol Chem 2010;285:20607–20614 study. Ann Rheum Dis 2012;71:563– tion and insulin sensitivity. Cell Metab 11. American Diabetes Association. Diagnosis 566 2010;12:593–605 and classification of diabetes mellitus. Di- 17. Ottenhoff TH, Verreck FA, Lichtenauer- 25. Vandanmagsar B, Youm YH, Ravussin A, abetes Care 2011;34(Suppl. 1):S62–S69 Kaligis EG, Hoeve MA, Sanal O, van Dissel et al. The NLRP3 inflammasome instigates 12. Matsuda M, DeFronzo RA. Insulin sensi- JT. Genetics, cytokines and human infec- obesity-induced inflammation and insulin tivity indices obtained from oral glucose tious disease: lessons from weakly patho- resistance. Nat Med 2011;17:179–188 tolerance testing: comparison with the genic mycobacteria and salmonellae. Nat 26. Wen H, Gris D, Lei Y, et al. Fatty acid- euglycemic insulin clamp. Diabetes Care Genet 2002;32:97–105 induced NLRP3-ASC inflammasome ac- 1999;22:1462–1470 18. Wallis RS. Reactivation of latent tubercu- tivation interferes with insulin signaling. 13. Puren AJ, Fantuzzi G, Gu Y, Su MS, losis by TNF blockade: the role of in- Nat Immunol 2011;12:408–415 Dinarello CA. Interleukin-18 (IFNgamma- terferon gamma. J Investig Dermatol 27. Ghanim H, Abuaysheh S, Sia CL, et al. In- inducing factor) induces IL-8 and IL-1beta Symp Proc 2007;12:16–21 crease in plasma endotoxin concentrations via TNFalpha production from non-CD14+ 19. Maedler K, Schumann DM, Sauter N, et al. and the expression of Toll-like receptors and human blood mononuclear cells. J Clin Low concentration of interleukin-1b in- suppressor of cytokine signaling-3 in mono- Invest 1998;101:711–721 duces FLICE-inhibitory protein-mediated nuclear cells after a high-fat, high-carbohydrate 14. Rissanen A, Botha J, Thuren T. IL-1b an- b-cell proliferation in human pancreatic meal: implications for insulin resistance. Di- tibody (canakinumab) improves insulin islets. Diabetes 2006;55:2713–2722 abetes Care 2009;32:2281–2287 secretion rates in subjects with impaired 20. Ehses JA, Lacraz G, Giroix MH, et al. IL-1 28. Ridker PM, Thuren T, Zalewski A, Libby P. glucose tolerance (IGT) and type 2 di- antagonism reduces hyperglycemia and Interleukin-1b inhibition and the pre- abetes (T2DM). Late-breaking abstract tissue inflammation in the type 2 diabetic vention of recurrent cardiovascular events: presented at the 71st ScientificSessionsof GK rat. Proc Natl Acad Sci USA 2009;106: rationale and design of the Canakinumab the American Diabetes Association, 24–28 13998–14003 Anti-inflammatory Thrombosis Outcomes June 2011, San Diego Convention Center, 21. Ehses JA, Perren A, Eppler E, et al. Increased Study (CANTOS). Am Heart J 2011;162: San Diego, California number of islet-associated macrophages 597–605

1662 DIABETES CARE, VOLUME 35, AUGUST 2012 care.diabetesjournals.org