GAD-treatment of children and adolescents with recent-onset type 1 diabetes preserves residual insulin secretion after 30 months

GAD-treatment of children and adolescents withrecent-onset type 1 diabetes preserves residualinsulin secretion after 30 months

Johnny Ludvigsson1,2*Mikael Chéramy1

Stina Axelsson1

Mikael Pihl1

Linda Åkerman1

Rosaura Casas1

for the clinical GAD-studygroup in Sweden

1Department of Clinical andExperimental Medicine, LinköpingUniversity, Linköping, Sweden2Pediatric clinic, County Council ofÖstergötland, Linköping, Sweden

*Correspondence to:Johnny Ludvigsson, Department ofClinical Experimental Medicine,Faculty of Health Sciences, LinköpingUniversity Hospital, SE-58185Linköping, Sweden.E-mail: [email protected]

Trial registration: Phase II trial;NCT00435981, Phase III trial;NCT00723411.

Abstract

Background This study aimed to analyse data from two different studies(phase II and phase III) regarding the safety and efficacy of treatment withalum formulated glutamic acid decarboxylase GAD65 (GAD-alum) at 30 monthsafter administration to children and adolescents with type 1 diabetes.

Methods The phase II trial was a double-blind, randomised placebo-controlled study, including 70 children and adolescents who were followedfor 30 months. Participants received a subcutaneous injection of either 20 μgof GAD-alum or placebo at baseline and 1 month later. During a subsequentlarger European phase III trial including three treatment arms, participants re-ceived two or four subcutaneous injections of either 20 μg of GAD-alum and/orplacebo at baseline, 1, 3 and 9 months. The phase III trial was prematurelyinterrupted at 15 months, but of the 148 Swedish patients, a majority com-pleted the 21 months follow-up, and 45 patients completed the trial at30 months. Both studies included GAD65 auto-antibodies-positive patientswith fasting C-peptide ≥0.10 nmol/l. We have now combined the results ofthese two trials.

Results There were no treatment related adverse events. In patients treatedwith 2 GAD-alum doses, stimulated C-peptide area under the curve had de-creased significantly less (9 m: p< 0.037; 15 m: p< 0.032; 21 m: p< 0.003and 30 m: p< 0.004), and a larger proportion of these patients were also ableto achieve a peak stimulated C-peptide >0.2 nmol/L (p< 0.05), as comparedwith placebo.

Conclusion Treatment with two doses of GAD-alum in children and adoles-cents with recent-onset type 1 diabetes shows no adverse events and preservesresidual insulin secretion. Copyright © 2013 John Wiley & Sons, Ltd.

Keywords type 1 diabetes; children; GAD-alum treatment; C-peptide; GAD65

Introduction

Type 1 diabetes (T1D) is a very serious disease that causes substantial morbid-ity and mortality in spite of an intensive treatment with multiple daily injec-tions of insulin, controlled diet and self-monitoring of blood glucose [1,2].

RESEARCH ARTICLE

Received: 28 March 2013Revised: 19 November 2013Accepted: 26 November 2013

Copyright © 2013 John Wiley & Sons, Ltd.

DIABETES/METABOLISM RESEARCH AND REVIEWSDiabetes Metab Res Rev 2014; 30: 405–414.Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/dmrr.2503

Residual insulin secretion decreases the risk of bothsevere hypoglycaemia [3] and ketoacidosis [4], and aquite modest β-cell function, with stimulated C-peptideabove 0.2 nmol/L, has been reported to reduce long-termcomplications [5]. Most attempts with immune inter-vention to preserve residual β-cell function have resultedin limited effects and have often been associated withadverse effects [6–21]. Treatment with anti-CD3 mono-clonal antibodies showed encouraging efficacy in phaseII trials [22,23], but later phase III trials failed to reachtheir primary endpoints [24,25]. As T1D is such a seriousdisease, certain risks and adverse events may have to beaccepted, but it is of course desirable to find both safeand tolerable treatment modalities, especially in childrenand adolescents. There are reasons to believe that auto-antigen treatment could be an alternative concept inmodulating the immune response to stop or at least delaythe destructive process leading to T1D [26]. In T1D,insulin is certainly such an auto-antigen [27], but so far,insulin has not stopped the immune process. Since weonce discovered the 64kD antigen [28], later shown tobe Glutamic Acid Decarboxylase 65 (GAD65), there havebeen a number of experimental studies followed by earlysafety-finding and dose-finding studies in humans. Thesestudies finally led to a study including latent autoimmunediabetes in adults patients [29] and later also clinical tri-als involving children and adolescents with T1D [30–32].

In a Swedish phase II trial, we administered two doses(2D) of 20 μg GAD65 in an alum-formulation (GAD-alum)or placebo (alum alone) to 70 young patients with recent-onset T1D at baseline and 1 month later in order to testwhether it would reduce or halt the loss of residual insulinsecretion (NCT00435981). The results showed significantpreservation of both fasting and stimulated C-peptideafter 30 months with no treatment-related adverse events[30], and preservation of C-peptide remained still after4 years in patients who were treated within 6 monthsfrom diagnosis [33]. As the efficacy was most pronouncedin patients with short duration of diabetes when receivingtheir first GAD-alum injection, the following phase III trial(NCT00723411) was designed to include only patientswith diabetes duration less than 3 months. During theEuropean phase III trial, patients were randomised intothree arms, one arm similar to the phase II trial with 2D20 μg GAD-alum, followed by two placebo injections. Inparallel, the second arm received four doses 20 μg ofGAD-alum (4D), whereas the third arm received fourplacebo injections. Inclusion criteria were similar to thephase II trial, although the patients were recruited notonly in Sweden (n=148) but also in nine European coun-tries (n=186). However, exclusion criteria differed as thepatients in the phase III trial were allowed to receive influ-enza vaccinations in parallel to the GAD-alum treatment,something that was not allowed in the phase II trial.

The phase III trial failed to reach primary endpoint[change in C-peptide AUC after a mixed meal tolerancetest (MMTT)] at 15 months [31]. There was however asignificant preservation of C-peptide in several pre-specified subgroups, which were interpreted as supportof concept. In addition, no treatment-related adverseevents were observed. Although the trial was interruptedprematurely, a majority of the Swedish patients hadcompleted the 21-month follow-up, and 45 patients hadalso completed the trial after 30 months, of whom 31individuals had received similar treatment as in the phaseII trial (2D n=15, placebo n=16). We have herecombined clinical data from the phase III trial with thosefrom the earlier phase II trial to test whether thisstrengthens or weakens the concept of GAD-alum treat-ment. Both studies were approved by the Research EthicsCommittee at the Faculty of Health Sciences, LinköpingUniversity, Sweden.

Material and methods

The phase II and phase III study groups

In 2005, a total of 70 patients were randomised to adouble-blind treatment with either 20 μg of recombinanthuman GAD65 formulated in alum (Diamyd®, DiamydMedical, Stockholm, Sweden; 35 patients) or placebo(the same formulation without GAD65; 35 patients). Fora detailed description of inclusion/exclusion criteria andstudy design, see [30]. Briefly, patients received a subcu-taneous primary injection of either GAD-alum or placeboon day 1 followed by a boost dose 1 month later. Partici-pants were 10–18 years of age and had been diagnosedwith T1D within the 18 months preceding screening.Inclusion criteria demanded presence of GAD65 auto-antibodies (GADA) and fasting C-peptide levels above0.1 nmol/L. All patients were treated with multiple dailyinjections of insulin or insulin pump with a target HbA1c

level of less than 6.5% (<58 mmol/mol; correspondingto <7.5% Diabetes Control and Complications Trial value).The pre-specified primary efficacy end point was the changebetween baseline and month 15 in the fasting C-peptidelevel, and the pre-specified secondary efficacy end pointswere changes between baseline and various pre-specifiedtime points, up to month 30, in fasting and stimulatedC-peptide levels and glycated haemoglobin values.

In 2008, we started the recruitment of patients to aEuropean phase III trial. For a detailed description ofinclusion/exclusion criteria and study design, see [31].Briefly, 334 patients aged 10–20 years and with a diseaseduration of less than 3 months were randomly assigned toone of the following three treatments arms: (1) 4D of

406 J. Ludvigsson et al.

Copyright © 2013 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2014; 30: 405–414.DOI: 10.1002/dmrr

20 μg GAD-alum at baseline, 1, 3 and 9 months, (2) 2D of20 μg GAD-alum at baseline and 1 month followed byplacebo at 3 and 9 months, (3) 4D of placebo at baseline,1, 3 and 9 months. As during the phase II trial, inclusioncriteria required presence of GADA and fasting C-peptidelevels above 0.1 nmol/L, and all patients were treated withmultiple daily injections of insulin or insulin pump with atarget HbA1c level of less than 6.5%. The primary outcomewas the change in the stimulated serum C-peptide level[mean area under the curve (AUC) over the 2-h periodafter a mixed-meal tolerance test] between the baselinevisit and the 15-month visit. In contrast to the phase II trial,influenza vaccinations were accepted in parallel during thephase III trial.

The intervals for patient follow-up from baseline up to30 months were identical in the two trials (i.e. baseline,1, 3, 9, 15, 21 and 30 months). Thus, an MMTT wasperformed after 3, 9 15, 21 and 30 months in accordancewith the European study on estimation of β-cell function[34]. This test includes ingestion of 6 ml Sustacal/Boost®

per kilogramme of body weight, with a maximum of360 mL, to be ingested within 5 min. The meal test wasperformed in the morning (between 7:00 and 10:00 h)after an overnight fast, in which no eating, smoking ordrinking (with the exception of water) occurred after22:00 h the preceding day. The patients used no short-acting insulin for at least 6 h before the test. Patients on

continuous subcutaneous insulin infusion continued theirnormal basal rate but received no added boluses for atleast 6 h prior to the test. Blood samples for C-peptideanalysis were collected before, 30, 60, 90 and 120 minafter the beginning of the MMTT. Other evaluationsincluded neurological assessments, clinical examination,haematology, biochemistry, blood glucose, insulin re-quirement (units per kilogramme body weight and 24 h)and a careful registration of adverse events.

Baseline characteristics of patients included in thephase II and III trials are described in Table 1. The numberof individuals participating in each treatment arm aftercombining the study cohorts from the phase II and III tri-als is described in Table 2.

Table 1. Baseline characteristics, according to study group

Phase III Phase II

Characteristic 4D (n=49) 2D (n=49) Placebo (n=50) 2D (n=35) Placebo (n=34)

Age 13.3±2.1 13.3±2.2 13.4±2.5 13.8±2.3 12.8±1.9Days since diagnosis 74.6±20.3 73.3±19.9 71.0±20.6 9.9±5.3 8.8±5.5Gender distribution, n (%)Female 25 (51) 26 (53) 20 (40) 23 (66) 18 (53)Male 24 (49) 23 (47) 30 (60) 12 (34) 16 (47)HLA classification, n (%)Very high risk 14 (29) 18 (37) 16 (32) NA NAHigh risk 21 (44) 23 (47) 23 (46) 18 (51) 16 (47)Moderate risk 8 (17) 4 (8) 8 (16) 9 (26) 7 (21)Low risk 5 (10) 4 (8) 3 (6) 8 (23) 11 (32)Tanner, n (%)1 6 (12) 10 (20) 5 (10) 4 (11) 7 (21)2+3 17 (35) 12 (25) 16 (32) 8 (23) 10 (29)4+5 26 (53) 27 (55) 29 (58) 23 (66) 17 (50)C-peptide (nmol/L)Fasting C-peptide 0.30±0.17 0.25±0.11 0.26±0.14 0.33±0.19 0.35±0.23Stimulated C-peptide AUC 0.71±0.35 0.66±0.26 0.64±0.24 0.62±0.28 0.71±0.43Glycated haemoglobin (%) 6.84±0.65 6.81±0.91 6.87±1.04 7.22±1.27 7.14±0.95Insulin dose (IU/day/kg) 0.59±0.29 0.62±0.31 0.55±0.23 0.66±0.30 0.66±0.28Fasting plasma glucose (mmol/L) 6.28±2.16 5.95±1.74 5.69±1.19 9.40±4.00 8.80±3.30Median GADA (units/mL) 204 312 190 601 861Median IA-2A (units/mL) 472 620 736 125 552

HLA, human leukocyte antigen; 4D, four doses of GAD-alum; 2D, two doses of GAD-alum; AUC, area under the curve; GADA, glutamic aciddecarboxylase antibodies; IA-2A, insulinoma antigen 2 antibodies.Values are mean±SD unless stated otherwise. HLA risk classification is based on HLA class II type. The Tanner puberty stage ranges from 1to 5, with higher stages indicating more developed genitalia.

Table 2. The number (n) of participating individuals at eachtime-point and for all treatment regimens, after combiningthe study cohorts from the phase II and III trials

3 m 9 m 15 m 21 m 30 m

4D phase III 48 48 48 39 142D phase II+ III 84 84 84 72 50Placebo phase II+ III 84 84 83 73 50

4D, four doses of GAD-alum; 2D, two doses of GAD-alum.

The GAD-Alum Phase II and III Trials 407


C-peptide analysis

In the phase II trial, C-peptide levels were measured inserum samples with a time-resolved fluoroimmunoassay(AutoDELFIA™ C-peptide kit, Wallac). Results were vali-dated with inclusion of a C-peptide control modulecontaining a high, a medium and a low-level control ineach assay (Immulite, DPC). A 1224 MULTICALC® prog-ramme (Wallac) was used for automatic measurementand result calculation, and measurements were expressedin pmol/mL. During the phase III trial, serum C-peptideanalysis was performed centrally by BARC Laboratories(Ghent, Belgium) using an Immulite 2000C-peptide kiton an Immulite 2000 analyzer with calibration standardson the basis of the World Health Organization’s NationalInstitute for Biological Standards and Controls (ReferenceStandard 84/510). Preserved β-cell function was deter-mined by the change in fasting and stimulated C-peptide(AUC) from baseline. Comparison between the twomethods showed an acceptable correlation (r=0.914,p< 0.001) when duplicate samples were analysed withthe two methods, but as different standard curves wereused, we do not translate them into one series.

HbA1c analysis

In the phase II trial, HbA1c was analysed by an immuno-logical method, calibrated against the Swedish nationalstandard Mono-S and continuously controlled againstthe External Quality Assurance in Laboratory Medicinein Sweden reference method, whereas HbA1c in the phaseIII trial was determined centrally by BARC laboratories.

HbA1c values from the phase II trial have been convertedinto phase III values by adding 0.9%.

Statistical analyses

As data sets were determined to be significantly differentfrom a Gaussian distribution using Shapiro–Wilk’s test;non-parametric tests corrected for ties were used.Unpaired analyses were performed using the Kruskal–Wallis test followed by Mann–Whitney U-test, and correla-tions were analysed with Spearman’s rank correlationcoefficient test. Differences within groups were calculatedby Friedman’s test followed by Wilcoxon signed-rank test.A probability level of <0.05 was considered statisticallysignificant. Calculations were performed using IBM SPSS

Statistics version 20 (SPSS Inc.).

Results

Adverse events

During the phase II trial, seven patients were reported tosuffer serious adverse events, whereas 15 patients reportedserious adverse events during the phase III trial (Table 3).None of the serious adverse events were considered to betreatment-related, and there was no difference in self-reported severe hypoglycaemias in the GAD-alum-treatedgroups compared with the placebo groups.

There were no treatment-related adverse eventsreported in patients treated with GAD-alum except forslight transient reaction at the injection site, similar in

Table 3. Serious adverse events, according to study group (number of events/number of patients)

Type of serious adverse event

Phase III Phase II

4D (n=49) 2D (n=49) Placebo (n=50) 2D (n=35) Placebo (n=34)

Anxiety — 1/1 — — —

Appendicitis — 1/1 — — —

Cessation of insulin administration — — — — 1/1Concussion 1/1 1/1 — — —

Diarrhoea — — — 1/1 —

Fractures 1/1 — — 1/1 1/1Gastroenteritis — 3/3 1/1 — —

Hyperglycaemia 2/1 — 2/1 — —

Hypoglycaemic seizure — 1/1* — — 2/1Ketoacidosis 1/1 4/1 1/1 2/2 —

Knee trauma — — — 1/1 —

Mononucleosis — — — — 1/1 †

Pyelonephritis/Sepsis — — 1/1 — —

Streptococcal tonsillitis — — — — 1/1 ‡

4D, four doses of GAD-alum; 2D, two doses of GAD-alum.*The patient also contributed to the reported hypoglycaemic seizure incidents in this group.†The patient did not receive a second injection and was withdrawn from the study.‡The patient also displayed high glycated haemoglobin level and contributed to one of the reported ketoacidosis incidents.



the active and placebo groups. The results from the neuro-logical part of the physical examination did not show anyabnormal findings, which could be related to the treat-ment, and no other parts of the physical examinationrevealed anything abnormal.

C-peptide

Treatment with GAD-alum did not significantly reduce theloss of stimulated C-peptide or improve clinical outcomesover a 15-month period in the phase III European trial[31]. In a previous phase II study including Swedishpatient’s significant preservation of both fasting andstimulated C-peptide was shown after 30 months(Figure 1(A–B)). Although the phase III study was closedafter 15 months, a majority of the Swedish patients com-pleted their 21 months visit (n=115), and a subgroupcompleted the 30 months follow-up (n=45), but analysisof C-peptide data for the 21-month and 30-month visitsrevealed that both fasting and stimulated C-peptide levelsdeclined from baseline in all treatment arms, with no sig-nificant differences between the groups (Figure 1(C–D)).

When combining the treatment regimens from bothphase I and phase II trials, 2D patients had significantlyless decline of stimulated C-peptide AUC at 9, 15, 21 and30 months compared with placebo (Figure 2(A)). There

was a difference in fasting C-peptide between the 2Dand placebo groups at 21 months (p=0.048), whereas atrend was observed at 30 months (p< 0.1) (Figure 2(B)).A larger proportion of 2D patients were also able toachieve a peak stimulated C-peptide level >0.2 nmol/Lfrom baseline to 30 months compared with placebo(p< 0.05) (Figure 2(C)). In addition, a larger proportionof 2D patients also preserved >25% of their baseline C-peptide than patients treated with placebo (p< 0.01) (datanot shown).

When we next looked at the small subgroup that com-pleted all the visits, a reduced loss in fasting (p=0.04)and stimulated C-peptide AUC (p=0.012) at 30 monthswas observed in patients who received 2D of GAD-alumcompared with placebo (Figure 3(A–B)), although themagnitude of C-peptide preservation was somewhat lessin the phase III subgroup than in the previous phase IItrial. In addition, significantly more 2D patients preservedmore than 25% of their initial stimulated C-peptide ascompared with patients in the placebo arm after30 months (p=0.012; Figure 3(C)), and the proportionof patients with a C-peptide peak above 0.2 nmol/Ltended to be larger in the 2D regimen than in the placebogroup (p=0.061; Figure 3(D)).

To confirm that the subgroup of 2D patients followedfor 30 months was representative of the entire study co-hort, fasting and stimulated C-peptide in the 2D group

Figure 1. Mean changes in stimulated and fasting C-peptide according to treatment regimen in the Phase II and III trials. Changes in(A) stimulated and (B) fasting C-peptide (mean±SEM), reported as a change from baseline, in patients from the Phase II trial receiv-ing two doses of GAD-alum (2D, grey circles) or placebo (white circles). Changes in (C) stimulated and (D) fasting C-peptide (mean±SEM), reported as a change from baseline, in Swedish patients from the phase III trial receiving four doses of GAD-alum (4D, blackcircles), two doses of GAD-alum (2D, grey circles) or placebo (white circles). Significant differences are indicated as p-values and con-sidered when p<0.05, trend are considered as p<0.1



were compared between the 30 months subgroup(n=15), the Swedish cohort (n=49) and the entirestudy cohort (n=108). There were no statistically signif-icant differences between the 2D groups at any time point(data not shown).

Clinical efficacy

During the phase II trial, no differences in HbA1c orinsulin doses were observed between the actively treatedpatients and the placebo group. As the phase III trialwas terminated at primary endpoint (15 month), no dataon insulin treatment and HbA1c were safely collected andsaved thereafter.

Discussion

The phase II trial was aimed to study the safety and effi-cacy of GAD-alum treatment during the 30-month trialperiod, with a primary endpoint of preserved fastingserum C-peptide at 15 months. Even though no significantdifference in fasting serum C-peptide was found at15 months between the actively treated and placebo

group, a significant preservation of stimulated C-peptidemeasured was observed. At the end of the extensionperiod (i.e. 30 months), a statistically significant differ-ence was found in fasting C-peptide as well [30]. Thetreatment was well tolerated and showed no treatment-related adverse events. We have shown in several reportsthat GAD-alum treatment had a GAD-specific influence onthe immune system, with increased secretion of moststudied cytokines in response to GAD-stimulation but witha tendency towards Th2-deviation in both cytokinesecretion and in change of GADA IgG subclass proportions[35–39]. However, no specific biomarker could be identi-fied as these GAD65-specific immunological effects did notcorrelate to the clinical efficacy of the treatment.

On the basis of the encouraging results of the phase IIstudy, we initiated a European phase III trial in theautumn 2008, including three treatment arms. The pri-mary endpoint of the phase III trial was changed in stim-ulated C-peptide after 15 months, measured as AUC afteran MMTT, but the study unfortunately failed to reach thisendpoint [31]. This led to a rapid decision by the involvedcompanies to stop the trial, which otherwise was designedto continue for 30 months. The reason for this decision

Figure 2. (A–C) Mean changes in stimulated and fasting C-peptide according to treatment regimen from the combined Phase II and IIItrials. Changes (A) stimulated and (B) fasting C-peptide (mean±SEM), reported as a change from baseline, in patients receiving fourdoses of GAD-alum (4D, black circles), two doses of GAD-alum (2D, grey circles) or placebo (white circles). (C) The proportion ofpatients (%) achieving a peak stimulated C-peptide level >0.2 nmol/L from baseline to 30 months in patients receiving four dosesof GAD-alum (4D, black circles/line), two doses of GAD-alum (2D, grey circles/line) or placebo (white circles/dotted line). Significantdifferences are indicated as p-values and considered when p<0.05, trends are considered as p<0.1



was pessimistic view on possible efficacy after longer du-ration, and political/economical, as failure of the primaryendpoint would not make the study useful for marketingthe product even if there would have been efficacy at 30months. As discussed in the original article [31], therewere several possible explanations to the different clinicalresults between phase ii and phase III trials. However,before the phase III trial was ended, a small subgroup of45 Swedish patients had already reached the last30 months follow-up visit. Out of them, 15 patients (2Darm) were actively treated in a similar way as the patientsin the phase II trial, and 15 had received placebo.Statistical assessment of these patients showed that thosein the 2D group had significantly less decline of bothfasting and stimulated C-peptide than the placebo groupat 30 months. This was a surprising finding, as previoussubgroup analyses of the European study cohort showeda clinical efficacy at 15 months when excluding Swedishpatients [31]. The difference seems to be explainedby the fact that while C-peptide in placebo and 4Dpatients continued to decline between the 21-monthand 30-month visits, 2D patients did not lose furtherC-peptide. Our observations are based on a small numberof patients and have to be interpreted cautiously, but theysuggest that 2D of GAD-alum did have an effect onpreservation of residual insulin secretion, although theeffect was not seen until after 30 months. This might

be due to the slow loss of C-peptide in diabetic teen-agers during the first 12–15 months after diagnosis.It has been shown that the decline in C-peptide dur-ing the first year after diagnosis of T1D in teenagersmay be very slow, as seen in a large European cohortof patients [40]. Thus, even an effective decrease ofthe autoimmune process might not be reflected in clin-ical parameters within 15 months. Indeed, the previousphase II trial did not meet its primary endpoint of animprovement in fasting C-peptide at month 15, butfasting C-peptide was significantly higher in GAD-alum-treated patients after 30 months [30].

When combining 2D and placebo patients from the twotrials, we observed a significant preservation of sti-mulated C-peptide and a trend towards preservation offasting C-peptide in the 2D group, compared withplacebo. A higher proportion of 2D patients were also ableto achieve a peak stimulated C-peptide value >0.2 nmol/L (p< 0.05) from baseline to 30 months, which have beenregarded as clinically relevant for the prevention of long-term complications [5]. Once again, we can confirm thatGAD-alum treatment is very easy to perform, tolerablefor the patients and without any treatment-relatedadverse events. The preservation of C-peptide may havewell clinically meaningful long-term effects, with adecrease of both acute and late complications, eventhough the secretion of endogenous insulin is not

Figure 3. Mean changes in stimulated and fasting C-peptide according to treatment regimen for the Swedish subgroup of patientswho completed the 30 months visit. Changes in (A) stimulated and (B) fasting C-peptide, reported as a change (%) from baseline,in patients receiving four doses of GAD-alum (4D, black circles), two doses of GAD-alum (2D, grey circles) or placebo (white circles).(C) The proportion of patients (%) preserving >25% of their initials C-peptide AUC from baseline to 30 months and (D) the propor-tion of patients achieving a peak stimulated C-peptide level >0.2 nmol/L from baseline to 30 months after treatment with four dosesof GAD-alum (4D, black circles), two doses of GAD-alum (2D, grey circles) or placebo (white circles). Significant differences are indi-cated as p-values and considered when p<0.05, trends are considered as p<0.1



sufficient to cause reduction of insulin dose or lowerHbA1c in otherwise strictly treated patients.

Altogether, these results may support the concept ofGAD-alum treatment; however, it is obvious that theclinical efficacy needs to be further improved, not leastas another phase II trial with GAD treatment in theUnited States failed [32]. One possibility is to aim thetreatment to more selected subgroups of patients onthe basis of experience from previous trials while con-tinuing the search of biomarkers and thereby identifyingpatients more suitable for the treatment. According toour experience so far, the treatment, given within afew months after diagnosis, seems to have best chancesto succeed when selecting patients with reasonablygood residual β-cell function at diagnosis and not toosevere metabolic state. The patients should have classi-cal autoimmune T1D with typical human leukocyteantigen-risk, whereas other patients regarded as T1Dbut perhaps with other genetic background or traitsrelated to type 2 diabetes may not be suitable forauto-antigen treatment, at least not as the sole therapy.Even after careful consideration of inclusion criteria, itmight be wise to try strengthening the treatment bycombining this autoantigen treatment with othertherapies. Anti-CD3 monoclonal antibodies still seemto be the most efficacious treatment even though phaseIII trials have failed [24,25]. However, because of theadverse events and heavy treatment, a more carefulselection of suitable patients and further studies usinganti-CD3 may be required before combination therapiesincluding anti-CD3 are initiated. Further treatmentstrategies may also include the combination of GAD-alum with additional autoantigens, as there are sugges-tions that both oral insulin [41] and Diapep277 [42]may have some efficacy in mitigating the autoimmuneprocess causing β-cell destruction. In addition, GAD-alum might also be administered in combination withVitamin D, which seems to have several desirable ef-fects on dendritic-cell and T-cell function [43,44] inaddition to positive effects on the β-cells as well as in-sulin sensitivity. As persistent inflammation may be aproblem, even when the autoimmune process is miti-gated, it might be wise to consider anti-inflammatoryagents [45] during combination therapies. Finally, allinterventions should as a basis have as good traditionaldiabetes treatment as possible as increase β-cell loadper se may contribute to deterioration of the β-cell de-structive process [46].

In conclusion, although the results of phase III GADtrials were disappointing, the positive indications ofefficacy in the few patients who completed the trial,together with the positive results when combining thephase II and III study populations, seem to support theconcept of GAD-alum treatment. In addition, GAD-alum

is simple to administer and considered safe, and ratherthan giving up this concept, further efforts are neededto improve the efficacy through combination therapy inselected patient groups.

Acknowledgements

Co-authors phase II trial

Gun Forsander, Nils-Östen Nilsson, Bengt-Olof Samuels-son, Jan Åman, Eva Örtqvist, Helena Larsson and AustePundziute-Lyckå.

Co-authors phase III trial

Ulf Samuelsson Linköping SWE, Helena Elding LarssonMalmö SWE, Jan Åman Örebro SWE, Gunilla Kördel GävleSWE, Jan Neiderud Helsingborg SWE, Göran LundströmKalmar SWE, Eva Albinsson Karlstad SWE, AnnelieCarlsson Lund SWE, Maria Nordvall Norrköping SWE,Hans Fors Trollhättan SWE, Carl-Göran Arvidsson VästeråsSWE, Stig Edvardson Växjö SWE, Ragnar Hanås UddevallaSWE, Karin Larsson Kristianstad SWE, Björn RathsmanStockholm SWE, Henrik Forsgren Hudiksvall SWE, HelenaDesaix Borås SWE, Gun Forsander Göteborg SWE, Nils-Östen Nilsson Halmstad SWE, Carl-Göran ÅkessonJönköping SWE, Päivi Keskinen Tampere FIN, Riitta VeijolaOulu FIN, Timo Talvitie Seinäjoki FIN, Klemens RaileBerlin GER, Thomas Kapellen Leipzig GER, Walter BurgerBerlin GER, Andreas Neu Tübingen GER, Ilse EngelsbergerBochum GER, Bettina Heidtmann Hamburg GER, SuzanneBechtold München GER, David Leslie London UK,Francesco Chiarelli Chieti ITA, Alessandro Cicognani Bolo-gna ITA, Giuseppe Chiumello Milan ITA, Franco CeruttiTorino ITA, Gian Vincenzo Zuccotti Milan ITA, AndreaScaramuzza Milan ITA, Luis Castaño Bilbao ESP, ItxasoRica Baracaldo ESP, Raquel Barrio Madrid ESP, MariaClemente Barcelona ESP, Maria José López Garcia ValenciaESP, Mercedes Rodriguez Zaragoza ESP, Isabel GonzalezMadrid ESP, Juan Pedro Lopez Siguero Malaga ESP,Mirentxu Oyarzabal Pamplona ESP, H.M. Reeser TheHague NL, Roos Nuboer Amersfoort NL, Dr. PaulineStouthart Sittard-Geleen NL, Natasa Bratina LjubljanaSLO, Nina Bratanic MD Ljubljana SLO, Marc de KerdanetRennes FRA, Jacques Weill Lille Sud FRA, Nicole SerToulouse FRA, Pascal Barat Bordeaux FRA, Anne MarieBertrand Besançon FRA, Jean-Claude Carel Paris FRA,Rachel Reynaud Marseilles FRA, Regis Coutant AngersFRA and Sabine Baron Nantes FRA.

The GAD-vaccination studies have been possible,thanks to several collaborators at the Division of Pediat-rics and Diabetes Research Centre, Linköping University,and we want to thank Lena Berglert, Ingela Johanssonand Gosia Smolinska Konefal for the laboratory



assistance; the research nurses Eva Isacsson, Ann-MarieSandström and all the physicians and nurses involved atdifferent Swedish clinics. The studies have beensupported by The Swedish Research Council (K2008-55X-20652-01-3), the Juvenile Diabetes Research Foun-dation (JDRF grant 17-2011-249), Barndiabetesfonden(The Swedish Child Diabetes Foundation) and theResearch Council of Southeast Sweden.

Conflict of interest

Diamyd Medical has been/is sponsor for the phase II/IIItrials and has also given financial support for investiga-tor-initiated mechanistic studies. Diamyd Medical AB assponsor was involved in the planning, monitoring andquality assurance of the phase II and phase III studiesaccording to ICH Good Clinical Practice.

References

1. The Diabetes Control and ComplicationsTrial Research Group. The effect ofintensive treatment of diabetes on thedevelopment and progression of long-term complications in insulin-dependentdiabetes mellitus. N Engl J Med 1993;329: 977–986.

2. Bojestig M, Arnqvist HJ, Hermansson G,Karlberg BE, Ludvigsson J. Declining in-cidence of nephropathy in insulin-dependent diabetes mellitus. N Engl JMed 1994; 330: 15–18.

3. The Diabetes Control and Compli-cations Trial Research Group. Effect ofintensive therapy on residual beta-cell function in patients with type1 diabetes in the diabetes control andcomplications trial. A randomized,controlled trial. Ann Intern Med 1998;128: 517–523.

4. Madsbad S, Alberti KG, Binder C, et al.Role of residual insulin secretion inprotecting against ketoacidosis ininsulin-dependent diabetes. Br Med J1979; 2: 1257–1259.

5. Steffes MW, Sibley S, Jackson M,Thomas W. Beta-cell function and thedevelopment of diabetes-related compli-cations in the diabetes control and com-plications trial. Diabetes Care 2003; 26:832–836.

6. Pescovitz MD, Greenbaum CJ, Krause-Steinrauf H, et al. Rituximab, B-lymphocyte depletion, and preservationof beta-cell function. N Engl J Med2009; 361: 2143–2152.

7. Ludvigsson J, Heding L, Lieden G,Marner B, Lernmark A. Plasmapheresisin the initial treatment of insulin-dependent diabetes mellitus in chil-dren. Br Med J (Clin Res Ed) 1983;286: 176–178.

8. Feutren G, Papoz L, Assan R, et al.Cyclosporin increases the rate andlength of remissions in insulin-dependent diabetes of recent onset. Re-sults of a multicentre double-blind trial.Lancet 1986; 2: 119–124.

9. The Canadian-European RandomizedControl Trial Group. Cyclosporin-induced remission of IDDM after earlyintervention. Association of 1 yr ofcyclosporin treatment with enhancedinsulin secretion. Diabetes 1988; 37:1574–1582.

10. Eisenbarth GS, Srikanta S, Jackson R,et al. Anti-thymocyte globulin and pred-nisone immunotherapy of recent onsettype 1 diabetes mellitus. Diabetes Res1985; 2: 271–276.

11. Chase HP, Butler-Simon N, Garg S,McDuffie M, Hoops SL, O’Brien D. A trialof nicotinamide in newly diagnosed pa-tients with type 1 (insulin-dependent)diabetes mellitus. Diabetologia 1990;33: 444–446.

12. Pozzilli P, Visalli N, Signore A, et al.Double blind trial of nicotinamide inrecent-onset IDDM (the IMDIAB IIIstudy). Diabetologia 1995; 38: 848–852.

13. Coutant R, Landais P, Rosilio M, et al.Low dose linomide in type I juvenile dia-betes of recent onset: a randomisedplacebo-controlled double blind trial.Diabetologia 1998; 41: 1040–1046.

14. Ludvigsson J, Samuelsson U, JohanssonC, Stenhammar L. Treatment with anti-oxidants at onset of type 1 diabetes inchildren: a randomized, double-blindplacebo-controlled study. Diabetes MetabRes Rev 2001; 17: 131–136.

15. Ludvigsson J, Samuelsson U, ErnerudhJ, Johansson C, Stenhammar L, BerlinG. Photopheresis at onset of type 1diabetes: a randomised, double blind,placebo controlled trial. Arch Dis Child2001; 85: 149–154.

16. Mastrandrea L, Yu J, Behrens T, et al.Etanercept treatment in children withnew-onset type 1 diabetes: pilot ran-domized, placebo-controlled, double-blind study. Diabetes Care 2009; 32:1244–1249.

17. Couri CEB, Oliveira MCB, Stracieri ABPL,et al. C-peptide levels and insulin in-dependence following autologous nonm-yeloablative hematopoietic stem celltransplantation in newly diagnosed type1 diabetes mellitus. J Am Med Assoc2009; 301: 1573–1579.

18. Raz I, Elias D, Avron A, Tamir M,Metzger M, Cohen IR. Beta-cell functionin new-onset type 1 diabetes andimmunomodulation with a heat-shockprotein peptide (DiaPep277): a rando-mised, double-blind, phase II trial.Lancet 2001; 358: 1749–1753.

19. Schloot NC, Meierhoff G, Lengyel C,et al. Effect of heat shock protein peptideDiaPep277 on beta-cell function in

paediatric and adult patients withrecent-onset diabetes mellitus type 1:two prospective, randomized, double-blind phase II trials. Diabetes Metab ResRev 2007; 23: 276–285.

20. Lazar L, Ofan R, Weintrob N, et al. Heat-shock protein peptide DiaPep277 treat-ment in children with newly diagnosedtype 1 diabetes: a randomised, double-blind phase II study. Diabetes Metab ResRev 2007; 23: 286–291.

21. Rother KI, Brown RJ, Morales MM, et al.Effect of ingested interferon-alpha onbeta-cell function in children with new-onset type 1 diabetes. Diabetes Care2009; 32: 1250–1255.

22. Herold KC, Gitelman SE, Masharani U,et al. A single course of anti-CD3 mono-clonal antibody hOKT3gamma1(Ala-Ala) results in improvement in C-peptide responses and clinical parame-ters for at least 2 years after onset oftype 1 diabetes. Diabetes 2005; 54:1763–1769.

23. Keymeulen B, Vandemeulebroucke E,Ziegler AG, et al. Insulin needs afterCD3-antibody therapy in new-onset type1 diabetes. N Engl J Med 2005; 352:2598–2608.

24. Sherry N, Hagopian W, Ludvigsson J,et al. Teplizumab for treatment of type1 diabetes (protege study): 1-year re-sults from a randomised, placebo-controlled trial. Lancet 2011; 378:487–497.

25. Gottlieb P, Pozzilli P. DEFEND presentedduring session: treatment of type 1diabetes-update on clinical trialsADA71st scientific sessions. San DiegoCalifornia, 2011.

26. Ludvigsson J. Adequate doses ofautoantigen administered using the ap-propriate route may create toleranceand stop autoimmunity. Diabetologia2009; 52: 175–176.

27. Jasinski JM, Eisenbarth GS. Insulin as aprimary autoantigen for type 1Adiabetes. Clin Dev Immunol 2005; 12:181–186.

28. Baekkeskov S, Nielsen JH, Marner B,Bilde T, Ludvigsson J, Lernmark A.Autoantibodies in newly diagnoseddiabetic children immunoprecipitatehuman pancreatic islet cell proteins.Nature 1982; 298: 167–169.



29. Agardh CD, Lynch KF, Palmer M, Link K,Lernmark A. GAD65 vaccination: 5 yearsof follow-up in a randomised dose-escalating study in adult-onset autoim-mune diabetes. Diabetologia 2009; 52:1363–1368.

30. Ludvigsson J, Faresjo M, Hjorth M, et al.GAD treatment and insulin secretion inrecent-onset type 1 diabetes. N Engl JMed 2008; 359: 1909–1920.

31. Ludvigsson J, Krisky D, Casas R, et al.GAD65 antigen therapy in recently diag-nosed type 1 diabetes mellitus. N Engl JMed 2012; 366: 433–442.

32. Wherrett DK, Bundy B, Becker DJ, et al.Antigen-based therapy with glutamicacid decarboxylase (GAD) vaccine in pa-tients with recent-onset type 1 diabetes:a randomised double-blind trial. Lancet2011; 378: 319–327.

33. Ludvigsson J, Hjorth M, Cheramy M,et al. Extended evaluation of the safetyand efficacy of GAD treatment of childrenand adolescents with recent-onset type 1diabetes: a randomised controlled trial.Diabetologia 2011; 54: 634–640.

34. Greenbaum CJ, Mandrup-Poulsen T,McGee PF, et al. Mixed-meal tolerancetest versus glucagon stimulation testfor the assessment of beta-cell functionin therapeutic trials in type 1 diabetes.Diabetes Care 2008; 31: 1966–1971.

35. Axelsson S, Hjorth M, Akerman L,Ludvigsson J, Casas R. Early inductionof GAD(65)-reactive Th2 response intype 1 diabetic children treated withalum-formulated GAD(65). DiabetesMetab Res Rev 2010; 26: 559–568.

36. Hjorth M, Axelsson S, Ryden A,Faresjo M, Ludvigsson J, Casas R.GAD-alum treatment induces GAD65-specific CD4+CD25highFOXP3+ cellsin type 1 diabetic patients. ClinImmunol 2011; 138: 117–126.

37. Skoglund C, Cheramy M, Casas R,Ludvigsson J, Hampe CS. GAD autoanti-body epitope pattern after GAD-alumtreatment in children and adolescentswith type 1 diabetes. Pediatr Diabetes2012; 13: 244–250.

38. Axelsson S, Cheramy M, Hjorth M, et al.Long-lasting immune responses 4 yearsafter GAD-alum treatment in childrenwith type 1 diabetes. PLoS One 2011;6: e29008.

39. Cheramy M, Skoglund C, Johansson I,Ludvigsson J, Hampe CS, Casas R.GAD-alum treatment in patients withtype 1 diabetes and the subsequent ef-fect on GADA IgG subclass distribution,GAD65 enzyme activity and humoral re-sponse. Clin Immunol 2010; 137: 31–40.

40. Pozzilli P, Schloot NC, Hosszúfalusi N,et al. Time dependent C-peptide decline

in 4411 patients with recent onset type1 diabetes followed for up to 10 years:a meta-analysis from 8 Europeancentres. EASD Presentation Abstract161 2011.

41. Skyler JS, Krischer JP, Wolfsdorf J, et al.Effects of oral insulin in relatives ofpatients with type 1 diabetes: the dia-betes prevention trial–type 1. DiabetesCare 2005; 28: 1068–1076.

42. Raz I, Pozzilli P. Current Ongoing Trialsand the Issue of Multiple InterventionStrategies in Newly Diagnosed patientsWorld Diabetes Congress. IDF: Dubai,2011.

43. Mathieu C, Gysemans C, Giulietti A,Bouillon R. Vitamin D and diabetes.Diabetologia 2005; 48: 1247–1257.

44. Boonstra A, Barrat FJ, Crain C, Heath VL,Savelkoul HF, O’Garra A. 1alpha,25-Dihydroxyvitamin d3 has a direct effecton naive CD4(+) T cells to enhance thedevelopment of Th2 cells. J Immunol2001; 167: 4974–4980.

45. Mandrup-Poulsen T, Pickersgill L,Donath MY. Blockade of interleukin 1in type 1 diabetes mellitus. Nat RevEndocrinol 2010; 6: 158–166.

46. Ludvigsson J. Why diabetes incidenceincreases – a unifying theory. Ann N YAcad Sci 2006; 1079: 374–382.



Documents

GAD-treatment of children and adolescents with recent-onset type 1 diabetes preserves residual insulin secretion after 30 months