2746 Diabetes Care Volume 37, October 2014

Low-Dose Otelixizumab Anti-CD3 Ronnie Aronson,1 Peter A. Gottlieb,2 Jens S. Christiansen,3 Thomas W. Donner,4 Monoclonal DEFEND-1 Emanuele Bosi,5 Bruce W. Bode,6 Paolo Pozzilli,7,8 and the DEFEND Study: Results of the Randomized Investigator Group* Phase III Study in Recent-Onset Human

Diabetes Care 2014;37:2746–2754 | DOI: 10.2337/dc13-0327

OBJECTIVE Previous studies demonstrated that the anti-CD3 otelixizu- mab, administered at a total dose of 48–64 mg, can slow the loss of C-peptide in recent-onset type 1 diabetes patients, with frequent reactivation of Epstein Barr virus (EBV). The DEFEND-1 (Durable Response Therapy Evaluation for Early or New-Onset Type 1 Diabetes) trial was designed to test whether a lower dose of otelixizumab could preserve C-peptide secretion in new-onset type 1 diabetes patients. 1LMC Diabetes & Endocrinology, Toronto, On- RESEARCH DESIGN AND METHODS tario, Canada A multicenter, randomized, placebo-controlled trial was performed in sites in the 2University of Colorado, Denver, CO U.S., Canada, and Europe. Two hundred eighty-one patients were randomized to 3University of Aarhus, Aarhus University Hospi- treatment with 3.1 mg otelixizumab administered over 8 days or placebo. The tal, Aarhus, Denmark 4Johns Hopkins University School of Medicine, primary end point of the study was the change in C-peptide area under the curve Baltimore, MD (AUC) from a 2-h mixed-meal tolerance test at month 12. 5San Raffaele Hospital Scientific Institute, Vita

EMERGING TECHNOLOGIES AND THERAPEUTICS Salute San Raffaele University, Milan, Italy RESULTS 6Emory University, Atlanta Diabetes Associates, The change in 2-h C-peptide AUC was not different between placebo-treated Atlanta, GA 7Universita` Campus Bio-Medico di Roma, Rome, patients and otelixizumab-treated patients (20.20 vs. 20.22 nmol/L, P = 0.81). Italy 8 Secondary end points, including HbA1c, glucose variability, and insulin dose, were Barts and the London School of Medicine and also not statistically different between the two groups. More patients in the Dentistry, Queen Mary, University of London, London, U.K. otelixizumab group than in the placebo group experienced adverse events, mostly grade 1 or grade 2. There was no EBV reactivation (viral load >10,000 copies/106 Corresponding author: Ronnie Aronson, ronnie. [email protected]. peripheral blood mononuclear cells) in the otelixizumab group, in contrast with Received 7 February 2013 and accepted 6 June previously published studies at higher doses of otelixizumab. 2014. CONCLUSIONS Clinical trial reg. no. NCT00678886, clinicaltrials .gov. Otelixizumab was well tolerated in patients with recent-onset type 1 diabetes at a This article contains Supplementary Data online total dose of 3.1 mg, but did not achieve preservation of levels of C-peptide or at http://care.diabetesjournals.org/lookup/ other markers of metabolic control. suppl/doi:10.2337/dc13-0327/-/DC1. *A complete list of investigators can be found in the Supplementary Data. For patients with type 1 diabetes, the risk for the development of serious micro- © 2014 by the American Diabetes Association. vascular and macrovascular complications is proportional to the degree of chronic Readers may use this article as long as the work hyperglycemia, although these complications may remain subclinical during the is properly cited, the use is educational and not pediatric and adolescent years (1). Among intensively treated patients participating for profit, and the work is not altered. care.diabetesjournals.org Aronson and Associates 2747

in the Diabetes Control and Complica- durable remission of diabetes (10). In this 3; and 0.5 mg/day on days 4–8, for a tions Trial, higher C-peptide concentra- study, mice with greater residual b-cell total dose of 3.1 mg. According to tions ($0.20 pmol/mL) at baseline were function at the initiation of treatment each investigator’s clinical judgment, associated with a lower HbA1c concen- were more likely to enter remission. In a prophylaxis or treatment for cytokine tration, and a reduced risk for the de- phase II, double-blind, placebo-controlled release syndrome was administered velopment of diabetes complications study carried out by the Belgian Diabetes each dosing day, 2 h before the study and hypoglycemia (2). Therefore, ther- Registry (BDR), which included 80 pa- drug or placebo infusion, right after the apies that preserve C-peptide con- tients with recent-onset type 1 diabetes, infusion, 6 h after the infusion, and at centrations above these levels may treatment with otelixizumab for 6 con- bedtime. Medications included ibupro- improve the outcome of patients with secutive days, for a total dose of 48–64 fen 400–800 mg to a maximum of 1,800 type 1 diabetes. mg, reduced insulin requirements and mg/24 h or acetaminophen 500–1,000 It is now widely recognized that type 1 preserved b-cell function (11,12). In- mg to a maximum of 4,000 mg/24 h, and diabetes reflects an autoimmune distur- deed, at 36 months, residual b-cell func- diphenhydramine 25–50 mg or other bance in which CD4+ and CD8+ T cells tion was 80% higher in the otelixizumab equivalent antihistamine. destroy insulin-producing b-cells in the group than in the placebo group. These At key study visits (screening, predose pancreas in genetically susceptible indi- beneficial effects were correlated with [#14 days], week 12, month 6, and viduals (3–5). Currently, insulin replace- higher residual b-cell function at baseline month 12), all subjects underwent a ment therapy remains the principal and treatment at a younger age (11). mixed-meal tolerance test (MMTT) treatment for type 1 diabetes (6), but The Durable Response Therapy Eval- with Boost. If the time from screening achieving optimal glycemic control con- uation for Early or New-Onset Type 1 to randomization was .35 days, the tinues to be a persistent challenge (7). Diabetes (DEFEND-1) study was a ran- predose MMTT was used as the baseline Insulin replacement therapy also fails to domized, placebo-controlled, multi- measurement. Fasting C-peptide level address the underlying disorder. With national, phase III trial designed to was measured prior to the meal, and the the discovery of several therapies that evaluate the efficacy and safety of maximum stimulated C-peptide level can change the progressive loss of insulin- otelixizumab after a single course of was the maximum observed C-peptide producing b-cells, this treatment treatment in subjects with new-onset level after the meal. Blood samples paradigm may be challenged. The po- type 1 diabetes. A lower dose than the for the measurement of C-peptide tential of interdiction of the underlying phase II study was chosen in order to secretion were taken at the following autoimmune process to preserve b-cell target a lower rate of Epstein Barr virus intervals in adults: 10 min before time function could facilitate glucose control, (EBV)reactivationthanseeninthe 0(210 min); immediately before the reduce long-term complications, and phase II study (75%) (12). The primary subject started drinking Boost (desig- address the practical challenges of day- outcome analysis of the DEFEND-1 trial nated as the time 0 sample); and 15, to-day disease management, which was to compare the change from the 30, 60, 90, and 120 min after time 0. has a substantial impact on patients’ baseline C-peptide area under the curve For adolescent subjects (,18 years of quality of life. (AUC) at month 12 between the otelix- age) at the time of screening, blood Otelixizumab represents a novel, izumab group and the placebo group samples were taken at the 120 min targeted, T-cell immunomodulator (13,14). The large demographic data time point only; at subsequent visits, designed to induce long-term remis- set from this study afforded an oppor- blood samples were taken at all time sion with a short course of therapy. tunity to investigate the relationship be- points. The C-peptide AUC was calcu- Otelixizumab, a chimeric monoclonal tween fasting and stimulated C-peptide lated. The mean daily insulin dose was antibody that targets the CD3/T-cell re- levels as well as the impact of optimal computed as the mean of the subject’s ceptor, has been genetically modified to glycemic control on C-peptide levels. daily total insulin use recordings (in in- remove the glycosylation site in the Fc ternational units [IU] per kilogram) domain, thus diminishing complement RESEARCH DESIGN AND METHODS over a 7-day period, within 14 days of or Fc receptor binding and reducing DEFEND-1 Design study visits, the predose MMTT, and the risk of inflammatory adverse reac- DEFEND-1 was a randomized, placebo- prior to the study drug. tions secondary to cytokine release (8). controlled, double-blind, multicenter Vials of the study drug were sent to Otelixizumab downregulates pathogenic study conducted in the U.S., Canada, the study centers under blinded condi- T cells and upregulates T regulatory cells, Germany, Denmark, Spain, Finland, the tions. Because the person who prepared thus inhibiting the autoimmune process U.K., Italy, and Sweden. Subjects pro- the study drug solutions could become responsible for type 1 diabetes (9). The vided consent to participate under ap- unblinded as a result of close observa- potential utility of otelixizumab in the proved protocols from local institutional tion of the material, that person was not management of type 1 diabetes has review boards and were randomized involved in any other aspect of the been demonstrated in animal and human in a 2:1 ratio to receive treatment with study. The investigator and other person- studies. In the nonobese diabetic (NOD) otelixizumab or placebo. Subjects nel who administered the study drug mouse model of spontaneous autoim- received a series of eight intravenous and/or performed study assessments mune diabetes, otelixizumab at a total infusionsdone infusion per day over 8 did not watch or participate in the prep- dose of 8 mg yielded a 53% remission of consecutive days. The doses of otelixi- aration of the study drug. Inadvertent diabetes, with as little as 30% CD3/T-cell zumab administered were 0.1 mg on unblinding can occur as a result of the receptor complex modulation, inducing a day 1; 0.2 mg on day 2; 0.3 mg on day cytokine release signs and symptoms 2748 Otelixizumab in Recent-Onset Type 1 Diabetes Diabetes Care Volume 37, October 2014

experienced by some subjects who re- levels, and differences by age group for cytometry experiments included anti- ceive otelixizumab, and the transient de- these and other metabolic variables in human Foxp3 and TcRalpha+TcRbeta creases in lymphocyte counts that are adults and adolescents. Using baseline (An Der Grub Bio Research GmbH). Anti- expected to occur. Investigators, other data, glucose variability was evaluated bodies from BD Bioscience included study center staff, and the study sponsor in relationship to C-peptide level. The anti-human CD2 fluorescein isothiocya- were blinded to laboratory data such as primary efficacy end point was to nate, CD2 phycoerythrin, CD25 phycoer- lymphocyte counts during the infusion compare the change from baseline in ythrin, CD8 peridinin chlorophyll, and period, to reduce the possibility of un- the stimulated serum C-peptide level CD4 allophycocyanin. C-peptide level blinding. An unblinded independent med- (mean AUC over the 2-h period after was measured using the competitive ical monitor did have access to such an MMTT) between the otelixizumab IMMULITE C-Peptide chemiluminescent laboratory data. Although the unblinded group and the placebo group at 12 immunometric assay and an IMMULITE independent medical monitor did not months (16). Secondary efficacy vari- series analyzer. know subject treatment group assign- ables included mean total daily insulin ments, that person could potentially be- use over 7 consecutive days, sum- Statistical Analyses come unblinded by reviewing data. marized at baseline and key visits, Demographic and other baseline char- During the first 12 months after adminis- and HbA1c level, insulin use, and insu- acteristics were summarized using de- tration of the study drug, study clinic visits lin dose–adjusted HbA1c level at 12 scriptive statistics. The intent-to-treat occurred frequently (weekly to monthly). months. The following two measures (ITT) population was used for efficacy of glucose variability were evaluated and pharmacodynamic analyses. A Inclusion Criteria for the relationship to C-peptide: aver- mixed-model repeated-measures meth- Male or female subjects were 12–45 age daily risk range (ADRR) (17), and odology was used with a “missing at ran- years of age with a diagnosis of new- mean amplitude of glycemic excursions dom” assumption in which the missing onset (#90 days between the initial di- (MAGEs), both at 12 months (18). data may depend on the prior observa- agnosis and the first dose of study drug) Investigators encouraged subjects to tions (i.e., the conditional independence type 1a autoimmune diabetes based on get as close as possible to an HbA con- assumption). The model provides unbi- American Diabetes Association and 1c centration of 6%. All adverse events ased least-squares mean estimates for World Health Organization criteria. All were recorded. Hypoglycemic events the treatment groups at the assessment subjects required insulin treatment cur- that were reported by subjects between time points. The per protocol (PP) pop- rently or at some point between the study visits were classified, as defined by ulation was defined as ITT subjects who date of diagnosis and administration of the American Diabetes Association and had completed all scheduled infusions the first dose of the study drug. Subjects Food and Drug Administration guidelines, of the study drug and received the cor- were positive for one or more of the as follows: severedrequiring assistance rect doses, and for whom there were following autoantibodies typically associ- of another person to actively administer recorded MMTT C-peptide data, with ated with type 1 diabetes: glutamic acid carbohydrate, glucagon, or other re- at least one pre-MMTT and at least decarboxylase; tyrosine phosphatase– suscitative actions; documenteddtypical two post-MMTT samples, at both base- like protein; zinc transporter auto- symptoms of hypoglycemia accompa- line and month 12. The PP population ; or insulin, if using insulin nied by a measured plasma glucose also did not use a prohibited concomi- for ,7 days. Subjects demonstrated evi- concentration of #3.9 mmol/L; and tant medication while in the study. The dence of residual functioning b-cells as asymptomaticdnot accompanied by primary efficacy end point was analyzed measured by a stimulated C-peptide typical symptoms of hypoglycemia but using a repeated-measures mixed- level of .0.20 nmol/L during an MMTT with a measured plasma glucose con- effects model with change from baseline when fasting blood glucose levels were centration of #3.9 mmol/L (19,20). Per- C-peptide AUC, age group, continent .70 and #200 mg/dL, with a maximum turbation of EBV immunity was assessed (North America or Europe), treatment stimulated C-peptide level of #3.50 by EBV DNA viral load (an EBV DNA viral group, visit, and treatment group by nmol/L. Subjects $18 years of age at load of .10,000 copies/106 peripheral visit interaction as prespecified inde- screening were required to have a BMI blood mononuclear cells [PBMNCs] was pendent variables. Assumptions made of ,32 kg/m2. For subjects ,18 years of considered to be above normal) each in the power calculation were based age at screening, the BMI was required day from day 1 to day 8. If a subject on a comparable study carried out by to be in the ,95th percentile for age had positive test results, the test was the BDR, in which 80 subjects received and sex, according to U.S. Centers for repeated weekly for 2 weeks or until either otelixizumab or placebo. In the Disease Control and Prevention (15). the count decreased to ,10,000 BDR study, the observed difference be- Subjects needed to be in generally copies/106 PBMNCs (whichever time tween the otelixizumab and placebo good health, with no significant and/or was longer). Serologic tests were also groups in mean change from baseline active disease in any body system and performed for antibodies to hepatitis B tomonth12inC-peptideAUCwas no clinically significant abnormal labora- surface and core . The lympho- 0.22 (nmol/L 3 min)/min. Therefore, tory values. cyte subsets, CD3 saturation, and CD3 this was used as the assumed value for Objectives modulation were measured by flow cy- the clinically meaningful treatment ef- Using DEFEND-1 baseline data, the post- tometry, and data were analyzed using fect for otelixizumab. With these as- intervention results compared mixed- the FCS Express V3 Clinical Edition soft- sumptions, at a two-sided significance meal–stimulated and fasting C-peptide ware. Antibodies that were used in flow level of 0.05, a sample size of 180 care.diabetesjournals.org Aronson and Associates 2749

subjects (120 in the otelixizumab group were reported. Exogenous insulin use eligibility criteria for a variety of rea- and 60 in the placebo group) provided and HbA1c end points were compared sons, including exclusions for BMI, un- 90% power to detect a treatment effect using the same repeated-measures mixed- related laboratory abnormalities, and of 0.22 (nmol/L 3 min)/min on the pri- effects model as the primary end point. lack of proven autoimmunity, with the mary efficacy end point with a two-sided The composite outcomes HbA1c/ most common single reason being low t test. Assuming a drop-out rate of 25%, exogenous insulin use and C-peptide C-peptide response. The 272 subjects the total planned sample size for the AUC/HbA1c/exogenous insulin use were were randomized to otelixizumab (n = study was 240 subjects (160 in the ote- compared between treatments using 181) and placebo (n = 91) (Fig. 1). Data lixizumab group and 80 in the placebo the O’Brien nonparametric ranks test were analyzed for the ITT and PP popu- group).Inadditiontotherepeated- procedure. lations. Since there were no significant measures analysis, ANCOVA of the change Glucose Variability differences between these two types of from baseline in C-peptide AUC to month Two measures of glucose variability analyses, the ITT analysis is presented 12 was performed on the ITT population. were evaluated for the relationship to here. The ANCOVA model contained terms C-peptide: ADRR and MAGE. Spearman for baseline C-peptide AUC, continent, correlations were used to assess associ- Demographics of Enrolled Subjects age, and treatment. In the ITT analysis, Subject demographics are noted in ations among C-peptide AUC, HbA1c missing month 12 data were imputed concentration, MAGE, and ADRR. Sub- Table 1. No major baseline differences using the last observation carried forward. jects were divided into tertiles based were noted among study subjects. Fur- A subject was considered a responder on C-peptide AUC to further explore thermore, there were no major differ- if, at a given visit, the subject had an the association between C-peptide level ences noted among subjects from # HbA1c concentration of 6.5% and and other variables. different countries. There were slight mean total daily insulin use over 7 con- differences in the level of mean stimu- secutive days of ,0.5 IU/kg/day dur- RESULTS lated C-peptide level between the sub- ing the 2 weeks preceding the visit. Screen Study Description group of adolescent subjects compared Responders were compared by two- Three hundred eighty-one subjects with adult subjects that were not clini- sided tests of equality based on the nor- were assessed for eligibility, and 272 cally relevant (0.81 vs. 0.72 nmol/L/min, mal approximation to the binomial dis- subjects met the entry criteria. One hun- respectively). The prevalence of type 1 tribution; Hochberg-adjusted P values dred nine patients did not meet the diabetes–associated antibodies was

Figure 1—Flow chart: enrollment of the DEFEND study. 2750 Otelixizumab in Recent-Onset Type 1 Diabetes Diabetes Care Volume 37, October 2014

Table 1—Baseline characteristics more frequently reported in the placebo Placebo group Otelixizumab group group compared with the otelixizumab Characteristics (n =91) (n =181) group. Although the incidence was low, pharyngitis and viral infection were Age, years 25.3 6 7.2 24.8 6 6.6 more frequent in the placebo group Sex, n (%) compared with the otelixizumab group. Female 31 (34.1) 64 (35.4) Male 60 (65.9) 117 (64.6) Papular rash was reported only in the Ethnicity, n (%) otelixizumab group, but maculopapular Hispanic/Latin American 3 (3.3) 8 (4.4) rash was reported in both groups with Non-Hispanic/Latin American 88 (96.7) 173 (95.6) proportionately more in the placebo Race, n (%) group. American Indian or Alaska Native 0 (0) 1 (0.6) Asian 2 (2.2) 1 (0.6) Mechanistic Studies African American 2 (2.2) 6 (3.3) Multiracial 2 (2.2) 4 (2.2) Transient reductions in lymphocytes Caucasian 85 (93.4) 169 (93.4) and modulation of CD3 T-cell receptor BMI, kg/m2 23.6 6 3.0 23.6 6 3.3 were observed during the dosing period Weight, kg 72.2 6 12.2 71.9 6 14.6 in otelixizumab-treated patients (Fig. 3A and B, respectively), which is consistent HbA1c % 7.26 6 1.55 7.25 6 1.29 with observations in previous studies in- mmol/mol 56 6 16.9 56 6 14.1 vestigating the same regimen (data not Time-normalized stimulated C-peptide shown). However, the maximum reduc- AUC, (nmol/L 3 min)/min* 0.70 6 0.36 0.75 6 0.42 tion in lymphocytes (36.3% relative to Maximum stimulated C-peptide, nmol/L* 1.00 6 0.51 1.03 6 0.55 predose values) was less than that ob- Data are the mean 6 SD, unless otherwise specified. *Placebo n = 90; otelixizumab n =179. served during therapy in the phase II trial using otelixizumab (12) and in previous trials with (21), an- other anti-CD3 monoclonal antibody. similar among all groups. More detailed secondary end points of mean HbA1c information is shown in Table 1. level (6.91 6 0.147% vs. 7.04 6 Anti-CD3 treatment in the NOD mouse 0.116%, respectively; P =0.289)(Fig. has been shown to increase regulatory Two-hour MMTT AUCdPrimary End 2B and Supplementary Table 2) and T-cell populations in addition to its other Point mean daily insulin dose (0.43 6 0.026 effects (22). We examined changes in the The primary end point was the change vs. 0.39 6 0.017 IU/kg, respectively; number of CD4+CD25+FoxP3+ T cells in from baseline in 2-h C-peptide AUC from P = 0.276) (Fig. 2C and Supplementary those individuals treated with otelixizu- an MMTT at month 12. Using an ITT Table 2). There were no significant dif- mab, and while, as expected, there was analysis, there was no difference noted ferences between the otelixizumab and a transient reduction during the dosing between placebo-treated and drug- placebo groups in other secondary end period in otelixizumab-treated patients, treated patients (20.20 6 0.037 vs. points, including ADRR and MAGE (Sup- no treatment differences were observed –0.22 6 0.025 nmol/L, respectively; plementary Table 2). beyond day 14 (Supplementary Fig. 1). P =0.58)(Fig.2A). There was also no difference in the change from baseline Adverse Events CONCLUSIONS in C-peptide level between placebo- Adverse events were more common in treated and otelixizumab-treated pa- the otelixizumab group, and included, The DEFEND-1 trial recruited a popula- tients after adjustment for age, conti- for example, headache, fever, rash, nau- tion that was representative of the type nent, and baseline C-peptide level sea, which were consistent with known 1 diabetes population. Unfortunately, (20.21 6 0.030 vs. 20.21 6 0.021 side effects of anti-CD3 antibodies the cumulative dose chosen for this nmol/L, respectively; P =0.81).Pre- (Supplementary Table 3). No subjects study, 3.1 mg otelixizumab, was not ef- specified subgroup analyses of baseline in the otelixizumab arm had EBV, as de- fective in preserving C-peptide level, as C-peptide AUC, adolescent and adult termined by PCR results of .10,000 had been previously seen in the phase II 6 subjects, North American and European copies/10 PBMNCs, whereas in an ear- study (11). Modulation of the CD3/T-cell lier phase II trial (12), 75% of study sub- receptor complex has been shown to be subjects, baseline insulin use, HbA1c level, BMI, gender, and type and num- jects had symptomatic disease with effective in preserving C-peptide levels ber of positive autoantibodies were per- reactivation of EBV after receiving in previous studies using both otelixizu- formed; all failed to demonstrate a 48–64 mg otelixizumab. Only one sub- mab (11,12) and teplizumab, another significant difference between the two ject (an adult in the placebo group) had chimeric/humanized anti-CD3 monoclo- groups (Supplementary Table 1). EBV, as determined by a PCR result of nal antibody (21,23,24). However, it is .10,000 copies/106 PBMNCs at weeks 6 important to take into account the rela- Secondary End Points and 12. The incidence of adverse events tive dosages used in these individual clin- There was no significant difference from related to the perturbation of EBV im- ical trials when conducting cross-study baseline to month 12 between the pla- munity was similar between the treat- and cross-compound comparisons. The cebo and otelixizumab groups in the ment groups. Oropharyngeal pain was three otelixizumab efficacy studies, care.diabetesjournals.org Aronson and Associates 2751

Figure 2—Comparison of otelixizumab vs. placebo for C-peptide mean AUC (A), HbA1c levels (B), and insulin use over 12 months (C).

BDR (48 mg) (11), DEFEND-1, and subse- effect on C-peptide levels was seen, cytokine release syndrome were signifi- quently Durable-Response Therapy Eval- with a reduction in insulin dose for up cantly milder, and no reactivation of EBV uation For Early or New-Onset Type 1 to 4 years, although this was at the ex- was seen. Teplizumab delivered at a Diabetes (DEFEND-2) (25) (3.1 mg pense of severe symptoms of cytokine dose of 17 mg appears to deliver an in- doses), had very different risk/benefit release syndrome and reactivation of termediate efficacy response (positive profiles. With respect to otelixizumab, EBV. At a dose of 3.1 mg, no efficacy effect on C-peptide levels at ;6 at a dose of 48 mg, a sustained beneficial was seen, although symptoms of months), with an intermediate effect 2752 Otelixizumab in Recent-Onset Type 1 Diabetes Diabetes Care Volume 37, October 2014

Figure 3—Lymphocyte depletion in otelixizumab-treated vs. placebo-treated patients (A) and modulation of CD3 on the cell surface (B). BL, baseline; D, day; DD, dosing day; EOI, end of infusion; H, hour; M, month; W, week; TCR, T-cell receptor.

on cytokine release syndrome symptoms administration causes the modulation of by week 3. The reasons for the lack of (24). Therefore, it appears that if a suffi- CD3/T-cell receptor complexes from the congruence between dosing of anti-CD3 cient dose of antibody is administered, a cell surface of T cells, and their subse- therapy in the NOD mouse and in type 1 meaningful clinical effect can be ob- quent transient redistribution from the diabetes patients remains unclear, but served. In this study, other factors, bloodstream into tissues and lymph may include species differences in pe- such as level of C-peptide at baseline, nodes (26). Relatively low levels of CD3 ripheral blood versus target organ age, and sex, were examined to deter- binding and downmodulation induced marker dynamics and/or differences in mine whether a cohort of subjects could by short treatment of anti-CD3 Fab2 the pharmacologic behavior at the site be found that showed efficacy, but these fragments in the NOD mouse led to life- of action of a Fab2 fragment versus an subgroup analyses were also not statis- long reversal of established diabetes in intact monoclonal antibody. Further pre- tically significant (Supplementary Table this model of type 1 diabetes (10). In clinical and clinical studies will be 2). contrast, in the original clinical proof- needed to understand and improve the One measure of the pharmacological of-concept study (BDR study), the ad- translatability of this model. Likewise, activity of anti-CD3 therapy is the rela- ministration of a 48- to 64-mg cumula- the relationship between the level of tive level of T-cell lymphopenia observed tive dose of otelixizumab resulted in an CD3 modulation or lymphopenia and ef- in the peripheral blood. Although otelix- almost complete loss of circulating CD3+ ficacy has not been established clinically izumab appears not to be a fully deplet- T cells in patients for the duration of with otelixizumab. It appears that the ing antibody, based on animal studies, its treatment, recovering to predose levels dose administered in the BDR study care.diabetesjournals.org Aronson and Associates 2753

may be at the top of the dose response guide dosing of therapy. Therapy with ongoing care of study subjects; participated in from the perspective of CD3 modulation both anti-CD3 and anti-CD20 appears the formatting, editing, writing, and reviewing at least, and data from clinical studies to be most effective when each of the of the manuscript; was fully responsible for all content and editorial decisions; and was in- with teplizumab suggest that subsaturat- drugs is administered at a dose that can volved at all stages of manuscript development. ing doses of an anti-CD3 monoclonal an- markedly reduce the respective cell R.A. is the guarantor of this work and, as such, tibody may be efficacious (21). populations within the bloodstream had full access to all the data in the study and Dose exploration studies had been (28,29). It may be that markers such as takes responsibility for the integrity of the data and the accuracy of the data analysis. conducted prior to the start of the these, or hopefully ones that may be DEFEND-1 study but had a limited scope, more disease specific, can be found to References primarily related to safety and tolerabil- help guide dosing in future trials. In ad- 1. Brink SJ. Complications of pediatric and ad- ity, in patients with established disease dition, the heterogeneity of type 1 dia- olescent type 1 diabetes mellitus. Curr Diab Rep and with no control or placebo group. betes in terms of age of onset, residual 2001;1:47–55 While the regimen chosen was shown b-cell function, and HLA genotype 2. The Diabetes Control and Complications Trial to be tolerable with respect to cytokine should be taken into account in future Research Group. Effect of intensive therapy on release syndrome and EBV reactivation, trial designs (30). In conclusion, while a residual beta-cell function in patients with type 1 diabetes in the diabetes control and complica- clinical evidence for the efficacy of the dose reduction of otelixizumab appears tions trial. A randomized, controlled trial. Ann DEFEND-1 dose regimen was lacking. to effectively improve its safety profile, Intern Med 1998;128:517–523 Further dose finding in new-onset dia- such a reduction sacrifices efficacy. Fur- 3. Gillespie KM. Type 1 diabetes: pathogenesis beticpatientsisrequiredtoestablish ther dose-ranging studies should be un- and prevention. CMAJ 2006;175:165–170 fi 4. Roep BO. The role of T-cells in the pathogen- the ef cacy and therapeutic window of dertaken to effectively determine a esis of Type 1 diabetes: from cause to cure. otelixizumab. viable therapeutic window with proven Diabetologia 2003;46:305–321 Reactivation of latent herpes viruses efficacy and a tolerable safety profile. 5. Mallone R, van Endert P. T cells in the path- is a common event in clinical studies ogenesis of type 1 diabetes. Curr Diab Rep 2008; – with immunomodulating agents. Of par- 8:101 106 6. Mehta SN, Wolfsdorf JI. Contemporary man- ticular concern is the reactivation of Funding and Duality of Interest. This work agement of patients with type 1 diabetes. En- EBV, due to the oncogenic potential of was supported by Tolerx, Inc. Writing and docrinol Metab Clin North Am 2010;39:573–593 this virus (27). It was felt that the level editorial assistance was provided by James 7. Edelman SV, Weyer C. Unresolved challenges of EBV reactivation (identified as a syn- Kesslick of Publication CONNEXION (Newtown, with insulin therapy in type 1 and type 2 diabe- PA), which was contracted by Tolerx, Inc., for fi drome similar to acute mononucleosis, tes: potential bene t of replacing amylin, a sec- these services. ond beta-cell hormone. Diabetes Technol Ther manifesting with sore throat, fever, and R.A. and P.A.G. received honoraria from Tolerx, 2002;4:175–189 cervical adenopathy between day 16 Inc., for participation in Advisory Board meet- 8. Hale G, Rebello P, Al Bakir I, et al. Pharmaco- and day 21 after the first infusion) ob- ings, and have acted as consultants for GSK. E.B. kinetics and antibody responses to the CD3 anti- served with otelixizumab in the original received honoraria from GSK for participation in body otelixizumab used in the treatment of type 1 Advisory Board meetings. Atlanta Diabetes As- – BDR study was unacceptably high, at diabetes. J Clin Pharmacol 2010;50:1238 1248 sociates, the employer of B.W.B., has received 9. Chatenoud L. CD3-specific antibody-induced 75% (12), and that a lower dose needed grant and research support from Tolerx, Inc. P.P. active tolerance: from bench to bedside. Nat to be found that could avoid this com- has received support in the form of grants from Rev Immunol 2003;3:123–132 plication. Of note, no significant reacti- Tolerx, Inc., and GSK. No other potential conflicts 10. Mehta DS, Christmas RA, Waldmann H, vationofEBV(EBVDNAviralload of interest relevant to this article were reported. Rosenzweig M. Partial and transient modulation . 6 GSK has made the placebo and baseline data of the CD3-T-cell receptor complex, elicited by 10,000 copies/10 PBMNCs) was from the DEFEND studies available in a deiden- low-dose regimens of monoclonal anti-CD3, is noted with the dose of otelixizumab tified, patient-level format. They would be sufficient to induce disease remission in non-obese used in the DEFEND-1 trial. The severity happy to receive proposals for secondary data diabetic mice. Immunology 2010;130:103–113 of the adverse events in the present trial analyses from external researchers once 11. Keymeulen B, Walter M, Mathieu C, et al. was, in general, milder than that seen in DEFEND-1 and DEFEND-2 primary publications Four-year metabolic outcome of a randomised are accepted for publication. Please contact controlled CD3-antibody trial in recent-onset the phase II trial and was mostly grade 1 Dr. Jill Donaldson ([email protected]) for type 1 diabetic patients depends on their age or 2. Furthermore, all symptoms con- further information. and baseline residual beta cell mass. Diabetolo- nected with potential cytokine release Author Contributions. R.A., P.A.G., E.B., and gia 2010;53:614–623 were diminished with this dose, along P.P. participated as local Principal Investigators 12. Keymeulen B, Vandemeulebroucke E, with pretreatment of patients with ibu- in the conducting of the trial, reviewed and Ziegler AG, et al. Insulin needs after CD3- edited the manuscript, were fully responsible antibody therapy in new-onset type 1 diabetes. profen or acetaminophen and antihis- for all content and editorial decisions, and were N Engl J Med 2005;352:2598–2608 tamines. Infections and other adverse involved at all stages of manuscript develop- 13. Palmer JP, Fleming GA, Greenbaum CJ, et al. events were more frequent in the ote- ment. J.S.C. contributed to data collection and C-peptide is the appropriate outcome measure lixizumab group, but were not signifi- interpretation of the results, reviewed and for type 1 diabetes clinical trials to preserve cant and were also generally mild edited the manuscript, was fully responsible beta-cell function: report of an ADA work- for all content and editorial decisions, and was shop, 21-22 October 2001. Diabetes 2004;53: (grade 1 and 2). Higher doses of otelix- involved at all stages of manuscript develop- 250–264 izumab could lead to better efficacy with ment. T.W.D. assisted in the design of the study, 14. Food and Drug Administration. Considera- still tolerable levels of adverse events participated as a local Principal Investigator in tions for allogeneic pancreatic islet cell products and should be explored. the conducting of the trial, reviewed and edited [article online], 2009. Available from http:// the manuscript, was fully responsible for all www.fda.gov/downloads/BiologicsBloodVaccines/ To date, trials of immunotherapy content and editorial decisions, and was in- GuidanceComplianceRegulatoryInformation/ have suffered from the lack of a second- volved at all stages of manuscript development. Guidances/CellularandGeneTherapy/UCM182441. ary immunologic end point that could B.W.B. participated in the recruitment and pdf. Accessed 11 June 2014 2754 Otelixizumab in Recent-Onset Type 1 Diabetes Diabetes Care Volume 37, October 2014

15. Centers for Disease Control and Prevention. from treatment and prevention [article online], 2008. adolescent type 1 diabetes: DEFEND-2, a ran- Body mass index [article online], 2011. Avail- Available from http://www.fda.gov/downloads/ domized, placebo-controlled, double-blind, able from http://www.cdc.gov/healthyweight/ Drugs/GuidanceComplianceRegulatoryInformation/ multi-centre study. Diabet Med 2014;31:399– assessing/bmi/index.html. Accessed 11 June Guidances/ucm071624.pdf. Accessed 11 June 402 2014 2014 26. Waldron-Lynch F, Henegariu O, Deng S, 16. Greenbaum CJ, Mandrup-Poulsen T, McGee 21. Herold KC, Gitelman S, Greenbaum C, et al.; et al. Teplizumab induces human gut-tropic reg- PF, et al.; Type 1 Diabetes Trial Net Research Immune Tolerance Network ITN007AI Study ulatory cells in humanized mice and patients. Sci Group; European C-Peptide Trial Study Group. Group. Treatment of patients with new onset Transl Med 2012;4:118ra12 Mixed-meal tolerance test versus glucagon Type 1 diabetes with a single course of anti- 27. Loechelt BJ, Boulware D, Green M, et al.; Type stimulation test for the assessment of beta- CD3 mAb Teplizumab preserves insulin produc- 1 Diabetes TrialNet /Mycophenolic cell function in therapeutic trials in type 1 di- tion for up to 5 years. Clin Immunol 2009;132: Acid Study Group. Epstein-Barr and other her- abetes. Diabetes Care 2008;31:1966–1971 166–173 pesvirus infections in patients with early onset 17. Kovatchev BP, Otto E, Cox D, Gonder- 22. Chatenoud L. The use of CD3-specific anti- type 1 diabetes treated with daclizumab and Frederick L, Clarke W. Evaluation of a new mea- bodies in autoimmune diabetes: a step toward mycophenolate mofetil. Clin Infect Dis 2013;56: sure of blood glucose variability in diabetes. Di- the induction of immune tolerance in the clinic. 248–254 abetes Care 2006;29:2433–2438 Handb Exp Pharmacol 2008;221–236 28. Pescovitz MD, Greenbaum CJ, Krause- 18. Service FJ, Molnar GD, Rosevear JW, 23. Herold KC, Hagopian W, Auger JA, et al. Steinrauf H, et al.; Type 1 Diabetes TrialNet Ackerman E, Gatewood LC, Taylor WF. Mean Anti-CD3 monoclonal antibody in new-onset Anti-CD20 Study Group. , B-lymphocyte amplitude of glycemic excursions, a measure type 1 diabetes mellitus. N Engl J Med 2002; depletion, and preservation of beta-cell function. of diabetic instability. Diabetes 1970;19:644– 346:1692–1698 N Engl J Med 2009;361:2143–2152 655 24. Sherry N, Hagopian W, Ludvigsson J, et al.; 29. Pescovitz MD, Torgerson TR, Ochs HD, 19. Workgroup on Hypoglycemia, American Di- Proteg´ e´ Trial Investigators. Teplizumab for et al.; Type 1 Diabetes TrialNet Study Group. abetes Association. Defining and reporting hy- treatment of type 1 diabetes (Proteg´ e´ study): Effect of rituximab on human in vivo antibody poglycemia in diabetes: a report from the 1-year results from a randomised, placebo- immune responses. J Allergy Clin Immunol American Diabetes Association Workgroup on controlled trial. Lancet 2011;378:487–497 2011;128:1295–1302, e5 Hypoglycemia. Diabetes Care 2005;28:1245– 25. Ambery P, Donner TW, Biswas N, 30. Pozzilli P. Type 1 diabetes mellitus in 1249 Donaldson J, Dayan CM. Efficacy and safety of 2011: heterogeneity of T1DM raises ques- 20. Food and Drug Administration. Diabetes mel- low-dose otelixizumab anti-CD3 monoclonal tions for therapy. Nat Rev Endocrinol 2012; litus: developing drugs and therapeutic biologics antibody in preserving C-peptide secretion in 8:78–80