Article Text

Original article
Anti-NKG2D monoclonal antibody (NNC0142-0002) in active Crohn's disease: a randomised controlled trial
  1. Matthieu Allez1,
  2. Brett E Skolnick2,
  3. Maria Wisniewska-Jarosinska3,
  4. Robert Petryka4,
  5. Rune Viig Overgaard5
    1. 1Department of Gastroenterology, APHP, Hôpital Saint Louis, INSERM UMRS 1160, Paris Diderot, Sorbonne Paris-Cité University, Paris, France
    2. 2Novo Nordisk, Inc., Princeton, New Jersey, USA
    3. 3Department of Gastroenterology, Medical University of Lodz, Lodz, Poland
    4. 4NZOZ Vivamed, Warsaw, Poland
    5. 5Novo Nordisk A/S, Søborg, Denmark
    1. Correspondence to Professor Matthieu Allez, Department of Gastroenterology, Hôpital Saint-Louis, 1, Avenue Claude Vellefaux, Paris 75010, France; matthieu.allez{at}


    Objective Anti-NKG2D (NNC0142-0002) is an antagonising human immunoglobulin G4 monoclonal antibody that binds to natural killer group 2 member D (NKG2D) receptors, which are expressed by T cells and innate lymphoid cells, and may be linked to mucosal damage in Crohn's disease (CD).

    Design Seventy-eight patients (aged ≥18 and ≤75 years) with CD for ≥3 months, Crohn's Disease Activity Index (CDAI) ≥220 and ≤450 and either C-reactive protein ≥10 mg/L or endoscopic evidence of inflammation, were randomised 1:1 to a single subcutaneous (SC) dose of 2 mg/kg anti-NKG2D or placebo. Primary endpoint was change in CDAI (ΔCDAI) from baseline to week 4. Prespecified significance level was 10% for CDAI endpoints. A futility analysis was instituted due to slow recruitment.

    Results Primary endpoint was not significantly different between anti-NKG2D and placebo (week 4 ΔCDAI=–16); however, there was a significant difference by week 12 (ΔCDAI=–55; p≤0.10). Significant improvements were noted in the non-failure to biologics subgroup (treated with anti-NKG2D (n=28)) from week 1 onward. Greater effects of anti-NKG2D were also observed in patients with baseline CDAI ≥330. Frequencies of adverse events (AEs) were comparable between anti-NKG2D and placebo. Most AEs were mild (49%) or moderate (43%). No antidrug antibodies were observed.

    Conclusions A single SC dose of 2 mg/kg anti-NKG2D did not reduce disease activity at week 4 versus placebo, but the difference was significant at week 12, and effects were evident in key subgroups. These data support further development of anti-NKG2D in IBD.

    Trial registration number NCT01203631.


    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Video 1

    Significance of this study

    What is already known on this subject?

    • The activating receptor natural killer group 2 member D (NKG2D) is increased on the surface of several mucosal T-cell subsets, which exhibit proinflammatory properties.

    • A number of NKG2D ligands are expressed on epithelial cells and are upregulated in the inflamed mucosa in IBD.

    • Administration of an NKG2D-blocking antibody attenuated the development of a murine transfer-induced colitis.

    • Anti-NKG2D (NNC0142-0002) is an antagonising human immunoglobulin G4 monoclonal antibody that binds specifically to NKG2D receptors.

    What are the new findings?

    • This was a randomised, double-blind, parallel group trial of a single subcutaneous dose of 2 mg/kg anti-NKG2D or placebo in 78 patients with Crohn's disease.

    • The primary endpoint (change in Crohn's Disease Activity Index (CDAI) from baseline to week 4) was not significantly different between anti-NKG2D and placebo; however, there was a significant difference by week 12.

    • When patients with high baseline CDAI (≥330) were partitioned into tertiles based on exposure, larger magnitude CDAI changes were observed with higher concentrations.

    How might it impact on clinical practice in the foreseeable future?

    • This clinical proof of principle trial provides appropriate support for further investigation of this novel anti-NKG2D antibody in Crohn's disease. Future trials should perform appropriate dose ranging and dose regimen evaluations.


    Crohn's disease (CD) is a chronic IBD characterised by uncontrolled immune responses.1 ,2 Therapy for CD is based on suppression of the immune system by blockade of inflammatory processes with immune suppressants or biologic therapies. Progress has been considerable over the last decade, mainly due to the development and extensive usage of anti-tumour necrosis factor (TNF) monoclonal antibodies. Despite initial efficacy, long-term benefit is observed in less than half of patients with CD treated with anti-TNF antibodies.3 Biologics with new targets have been developed, including monoclonal antibodies targeting the trafficking of immune cells, yet therapies with novel mechanisms of action are still required.

    The persistence of intestinal inflammatory lesions in CD is mediated by an active crosstalk between immune and non-immune cells, and T cells are key players in this pathogenic process.4 The inflamed mucosa is heavily infiltrated with activated T lymphocytes, which produce inflammatory cytokines, exhibit cytotoxic properties and contribute to mucosal damage.5 ,6 The accumulation of these immune cells relies on an active recruitment from the bloodstream, a sustained cell cycling and diminished susceptibility of cells to undergo apoptosis. T-cell activation relies on the recognition of specific antigens by the T-cell receptor and the concomitant delivery of a costimulatory signal. Interestingly, mucosal T cells may express innate receptors that provide this costimulatory signal. Natural killer group 2 member D (NKG2D) is an activating receptor present on the surface of natural killer (NK) cells, some NK T cells, CD8+ cytotoxic T cells, gamma–delta T cells and CD4+ T cells, under certain conditions.7 ,8

    The ligands that bind to human NKG2D are major histocompatibility complex class I-related molecules A and B and UL-16-binding proteins, all of which have increased expression with cellular stress.9 A number of these NKG2D ligands are expressed on epithelial cells and are upregulated in the inflamed mucosa in IBD.10–12 Thus, the intestinal epithelium may modulate a variety of T-cell responses through direct interactions via the NKG2D pathway.4 ,9

    An increased expression of NKG2D on CD4+ T cells is observed in CD.10 CD4+NKG2D+ T cells exhibit specific cytotoxic activity and are able ex vivo to kill target cells expressing NKG2D ligands and are also an important source of inflammatory cytokines (eg, TNFα, interferon (IFN) γ and interleukin-17 (IL-17)).10 ,13 The production of these cytokines is strongly enhanced ex vivo by costimulation of the T-cell receptor and the NKG2D receptor.10 ,13 Interestingly, most of the T-cell oligoclonal expansions found in the inflamed mucosa of patients with CD correspond to CD4+ T cells expressing NKG2D.14 The implication of CD4+NKG2D+ T cells in gut inflammation has been further demonstrated in a murine model of transfer-induced colitis.15 ,16 Administration of a specific NKG2D-blocking antibody decreased NKG2D expression on CD4+ T cells and attenuated the development of colitis. NKG2D may also modulate the function of other T-cell subsets including CD8+ T cells and NK cells, particularly cytotoxicity, as shown in coeliac disease.17 ,18 These data support the potential role of the NKG2D pathway in the overactivation of effector T cells in CD.

    We present here the results of a randomised, double-blind, placebo-controlled trial in patients with moderately to severely active CD that assessed the efficacy and safety of anti-NKG2D (NNC0142-0002). Anti-NKG2D is an antagonising human immunoglobulin G4 monoclonal antibody that binds specifically to NKG2D receptors. The compound is demonstrated to be blocking, non-depleting and non-activating, and it does not fix complement or induce antibody-dependent cell-mediated cytotoxicity (NCT01203631).


    Trial design

    This was a multicentre, multinational, randomised, double-blind, parallel-group, placebo-controlled, single-dose trial in patients with moderately to severely active CD conducted from February 2011 to May 2013. Eligible patients were randomised in a 1:1 ratio to one of two parallel treatment arms to receive either a single subcutaneous (SC) dose of 2 mg/kg anti-NKG2D (NNC0142-0002) or placebo (figure 1).

    Figure 1

    Consort diagram showing trial design. CDAI, Crohn's Disease Activity Index; CRP, C-reactive protein; NKG2D, natural killer group 2 member D.

    The dose selected for evaluation was based on pharmacokinetic/pharmacodynamic (PK/PD) modelling of data from a trial (NN8555-3618, NCT00927927) in 24 patients with rheumatoid arthritis, which primarily assessed safety and tolerability following single-dose administration of anti-NKG2D in the range of 0.0002 to 7.5 mg/kg. No safety concerns were identified in this trial. The administration of anti-NKG2D as a single dose up to 7.5 mg/kg was further supported by findings from the multiple dose part of the NN8555-3618 trial; these data were considered adequate to justify initiation of a trial in patients with CD with a dose of 2 mg/kg having been selected to achieve a median duration of full receptor occupancy of 12 weeks.

    The protocol and its amendments were approved according to local regulations by appropriate health authorities and by independent ethics committees/institutional review boards. The trial was performed in accordance with Declaration of Helsinki and International Conference on Harmonisation Good Clinical Practice.


    The trial was conducted at 52 centres in Belgium, Canada, France, Hungary, Israel, Poland, the Russian Federation and the USA. Eligible patients included men and women aged between 18 and 75 years with CD for ≥3 months with a Crohn's Disease Activity Index (CDAI) score19 of ≥220 and ≤450, with active disease confirmed during screening by either the presence of an elevated C-reactive protein (≥10 mg/L) or endoscopic verification of active ulceration confirmed by a central reader. Patients with or without failure to prior biologic therapy for CD were included. Non-failure to biologic therapy included biologic-naïve patients or patients who discontinued biologic therapy for reasons other than lack of efficacy or intolerance. Failure to biologic therapy included primary or secondary failure to a single biologic therapy (eg, after an appropriate therapeutic trial at the approved dose) or intolerance to therapy (eg, unable to achieve doses or treatment duration because of drug-related adverse reactions). All patients provided written informed consent to participate.

    Patients were excluded if they had known or suspected hypersensitivity or previous exposure to anti-NKG2D, body mass index ≥38.0 kg/m2, indeterminate colitis, UC, coeliac disease or irritable bowel syndrome. Patients with any of the following were also excluded: symptomatic bowel obstruction; short bowel syndrome; current ileostomy or colostomy; surgical bowel resection within 6 months prior to randomisation; clinically relevant undrained abscess; past or current malignancy, with the exception of treated and cured basal or squamous cell carcinoma or cervical carcinoma occurring >12 months prior to randomisation; history of dysplasia in the colon; chronic or ongoing active infectious disease requiring hospitalisation or intravenous anti-infectious treatment within 4 weeks of randomisation; evidence of herpes zoster or cytomegalovirus infection that resolved <2 months prior to randomisation; a positive QuantiFERON-TB Gold test, or a history of active tuberculosis within the past 3 years; or signs of severe, progressive or uncontrolled renal, hepatic, haematological, cardiac, pulmonary, neurological, ophthalmological or cerebral disease. Women of childbearing potential who were pregnant, breast feeding or intended to become pregnant were also excluded.

    Patients also were excluded if they had received investigational products or vaccinations within 4 weeks prior to randomisation or were receiving total parenteral nutrition or opiates for abdominal pain. Permitted medications included stable dosages of 5-aminosalicylates, prednisone (≤20 mg/day), budesonide (≤9 mg/day), azathioprine, 6-mercaptopurine and methotrexate. Prednisone and budesonide dosages must have been stable for 2 weeks before randomisation. Patients who had changed dosages or discontinued 5-aminosalicylates, mesalamine or sulfasalazine treatment within 4 weeks of screening also were excluded.

    Randomisation and interventions

    Patients were randomised to receive 2 mg/kg of anti-NKG2D or placebo (both supplied by Novo Nordisk A/S, Søborg, Denmark) administered SC in the abdominal region at week 0 and were evaluated through week 24 for safety. To maintain blinding of trial personnel, an unblinded and trial-independent pharmacist prepared the trial products and dilutions, and filled the syringes. Patients, investigators, trial site personnel and sponsor were unaware of treatment assignments. Randomisation was completed through a central computer-generated scheme stratified by biologic failure status (failure/non-failure to previous biologic therapy) and CDAI score (CDAI <330/≥330). Access to blinded patient treatment information was possible for medical emergencies only. Patients were evaluated at the time of randomisation and drug administration (week 0) and again at weeks 1, 2, 4, 8, 12 and at weeks 16, 20 and 24 (for safety).


    The primary efficacy endpoint was the change in disease activity assessed by CDAI from baseline (week 0) to week 4. Patient-reported assessments were performed daily using an interactive voice response system. Selected investigator assessments and laboratory values were also recorded using interactive voice/Web response system.

    Secondary endpoints included the change in CDAI from baseline to weeks 1, 2, 8 and 12, the proportion of patients in clinical remission (defined as achieving a CDAI score <150) at weeks 1, 2, 4, 8 and 12, and the proportion of patients who achieved clinical response (reduction in CDAI score ≥100) from baseline to weeks 1, 2, 4, 8 and 12. Disease activity was also assessed by the Harvey–Bradshaw Index (HBI) at weeks 4 and 12, and number of draining fistulas at week 12. The Short Form-36 (SF-36) and IBD Questionnaire (IBDQ) were self-administered by patients at weeks 4 and 12.

    PK parameters were assessed to describe the concentration–time profile, including maximum observed serum/plasma concentration (Cmax) and time to maximum concentration (Tmax). PD parameters included anti-NKG2D occupancy of the NKG2D receptor on lymphocyte subsets in blood to assess the durations of full occupancy and of detectable occupancy in blood. The expression of NKG2D on the surface of peripheral NK and CD8+ T cells was based on two different assays using flow cytometry, measuring the number of receptors on the cells' surface in molecules of equivalent fluorochrome (MEF) units.

    Adverse events (AEs) (including local tolerability at injection site) were recorded at each visit by trial personnel, laboratory evaluation and spontaneous patient reports, and information on death, serious AEs and AE withdrawals were recorded in case report forms. Determinations were made as to date of onset, event description, severity, time course, outcome and causal relationship to trial drug. Serious AEs were recorded throughout the trial in a similar manner as for AEs through week 24. AEs were coded using Medical Dictionary for Regulatory Activities V.15.1.

    Physical examinations, vital signs, 12-lead electrocardiogram and laboratory assessments—which included haematology and biochemistry panels, urinalysis and determination of C-reactive protein concentrations, calprotectin levels and antibodies to anti-NKG2D—were measured throughout the trial. If rescue therapy was initiated, the type of rescue medication was recorded in the case report form. Concomitant medications were recorded at every visit. In patients who had a baseline endoscopy, an optional second endoscopy with biopsies at week 12 required a separate informed consent.

    Statistical analysis

    A sample size of 100 randomised patients (90 completers) was planned to ensure 80% power to detect a difference between active treatment and placebo at week 4. This was based on a 1:1 randomisation ratio, a difference in mean change from baseline to week 4 on the CDAI of 45 points, with an estimated SD of 85 on the CDAI, at a prespecified 10% significance level (two-sided t-test). All patients randomised, dosed and contributing with post-dosing data were included in the full analysis set, used for analysis of disease activity. All patients randomised and dosed were included in the safety analysis set.

    The primary CDAI endpoint was analysed using a mixed-effects model for repeated measures (MMRM) with fixed factors for treatment, visit, treatment-by-visit, stratum-by-visit interaction and stratification (failure/non-failure to previous biologic therapy, CDAI score at baseline <330 or ≥330 and in patients with a baseline calprotectin level >250 µg/g), with patient as a random factor. Baseline CDAI and baseline CDAI-by-visit interaction were included as covariates. All post-dosing measurements from baseline up to week 4 or time of rescue medication administration, whichever came first, were included in the primary analysis of the primary endpoint. The prespecified significance level for CDAI endpoints was 10% (p≤0.10; two-sided t-test). Sensitivity analyses were conducted on change in CDAI scores from baseline to week 4 using an analysis of covariance (ANCOVA) model for patients who had not received rescue medication (using treatment and stratification as factors, and baseline as covariate). This was repeated using the full analysis set week 4 data, or a last observation carried forward approach, before the use of rescue medication or dropout at week 4.

    Mean changes in CDAI from baseline to weeks 1 and 2 were also based on the MMRM model (including postdosing assessments through week 12 or time of first use of rescue medication or dropout, whichever came first), and mean change from baseline to weeks 8 and 12 was based on a mixed-effects model analogous to the sensitivity analyses model. CDAI clinical remission and response were analysed using Fisher's exact test and logistic regression analysis (weeks 1, 2, 4, 8 and 12). The logistic regression model contained treatment, strata and CDAI at baseline as a covariate. HBI was analysed using ANCOVA at weeks 4 and 12. IBDQ and SF-36 were analysed using MMRM at weeks 4 and 12. The fixed and random factors applied were similar to those described for the primary endpoint. The prespecified significance level was 5% (p≤0.05; two-sided t-test) for analyses that did not include CDAI. All other disease activities, safety endpoints and PK/PD parameters are presented as descriptive statistics. Statistical analyses were performed using SAS V.9.3 (SAS Institute, Cary, North Carolina, USA).


    Randomisation and baseline characteristics

    A total of 205 patients were screened from 32 sites in eight countries from February 2011 to May 2013; 127 patients were not eligible based on exclusion criteria (n=73), inclusion criteria not being met (n=50) and other reasons (n=4). Due to slow recruitment, a futility analysis, prespecified in the protocol, based on change in CDAI at week 4 in 74 patients, resulted in discontinuation of further recruitment. Seventy-eight patients were included and randomised to receive anti-NKG2D (n=40) or placebo (n=38).

    The demographic and baseline characteristics of the randomised patients (n=78) are summarised in table 1. Patients were well matched in the two groups, but the geometric mean for C-reactive protein and calprotectin were numerically higher in the anti-NKG2D group. Twenty patients discontinued (10 in each group) due to ineffective therapy (n=8), withdrawal criteria (n=7), AEs (n=4) and non-compliance (n=1) (figure 2).

    Table 1

    Demographic and baseline characteristics (n=78)

    Figure 2

    Consort diagram showing patient flow. NKG2D, natural killer group 2 member D.

    Efficacy in the full analysis set

    Anti-NKG2D was not significantly better than placebo at week 4 in the primary endpoint analysis (ΔCDAI mean difference, −16 (90% CI −49 to +16); p=0.40). A significant difference was achieved at week 12 (ΔCDAI mean difference, −55 (90% CI −103 to −8); p≤0.10) (figure 3A). The mean change in HBI score was also significantly different between treatment groups at week 12 (ΔHBI score mean difference: −2.7 (95% CI −4.6 to −0.8); p=0.01) (see online supplementary figure S1).

    Figure 3

    Disease activity in patients who were anti-NKG2D-treated and placebo-treated expressed as mean (SEM) change (Δ) in CDAI score from baseline to week 12 (week 4, primary outcome) for (A) all patients and (B) in the ‘non-failure to biologic therapy’ subgroup. CDAI, Crohn's Disease Activity Index; NKG2D, natural killer group 2 member D. *p Value versus placebo analysed using (A) mixed-effects model for repeated measures and (B) analysis of covariance.

    Rates of clinical response and remission were assessed. There was a significantly higher rate of clinical response (reduction in CDAI score ≥100 from baseline) at week 1 (p≤0.10), and a significantly higher rate of clinical remission (CDAI score <150) at week 2 in patients who received anti-NKG2D compared with placebo (see online supplementary figure S2, left panel and online supplementary figure S3, left panel), although these were secondary exploratory endpoints.

    There was an improvement in the change from baseline in the physical component of the SF-36 in patients receiving anti-NKG2D compared with placebo at week 12 (3.46 (95% CI 0.09 to 6.84); p≤0.05), but there was no significant difference in the global IBDQ score (table 2). The number of draining fistulas was too low for a meaningful comparison. A second optional endoscopy was performed in nine patients to obtain biopsy material, without assessment of endoscopic severity.

    Table 2

    Mean change in SF-36 and IBDQ scores in patients who were anti-NKG2D-treated and placebo-treated (n=78)

    Efficacy in subgroups

    Fifty-five randomised patients had no previous failure to biologic therapy, which included 46 patients who were biologic naïve. In this prespecified subgroup (‘non-failure to biologic therapy’), there were significant differences in the CDAI score between treatment groups from week 1 through week 12 (p≤0.10) (figure 3B). There were also significant improvements (p≤0.10) in clinical response (weeks 1, 4 and 12) and remission (weeks 2 and 8) with anti-NKG2D compared with placebo in the ‘non-failure to biologic therapy’ subgroup (see online supplementary figure S2, right panel and online supplementary figure S3, right panel). In the subgroup of 46 patients who were biologic naïve, there were also differences in the CDAI score among patients who received anti-NKG2D as compared with placebo from week 1 through week 12 (see online supplementary figure S4).

    In the subgroup of patients with a baseline calprotectin level >250 µg/g (n=50/78, 64%), there was a significant difference for anti-NKG2D versus placebo at week 8 (ΔCDAI mean difference, −66 (90% CI −114 to −18); p=0.026) and week 12 (ΔCDAI mean difference, −108 (90% CI −170 to −46); p=0.006) (figure 4). Among patients who have a baseline C-reactive protein above 10 mg/L, we found a decrease in C-reactive protein at week 12 in patients who received anti-NKG2D as compared with placebo (see online supplementary figure S5).

    Figure 4

    Disease activity in patients who were anti-NKG2D-treated and placebo-treated expressed as mean (SEM) change (Δ) in CDAI score from baseline to week 12 (week 4, primary outcome) in the subgroup of patients with a baseline calprotectin level >250 µg/g. CDAI, Crohn's Disease Activity Index; NKG2D, natural killer group 2 member D. *p-value versus placebo analysed using mixed-effects model for repeated measures.


    The median concentration–time profile of anti-NKG2D reflects a Cmax of approximately 13 μg/mL (figure 5A). The profile indicates linear PKs for high concentrations (approximately >1 µg/mL) and non-linear PKs for lower concentrations, as observed after week 8. Due to the non-linear PKs, a single half-life could not be determined. A median NKG2D receptor occupancy of 80% was observed up to 8 weeks for patients who were anti-NKG2D-treated. At week 12, median occupancy was <20% (figure 5B). To further explore the relation between disease burden and anti-NKG2D concentration, patients with high CDAI scores (≥330) were partitioned into tertiles based on exposure (mean concentration of anti-NKG2D); larger magnitude changes in CDAI were observed with higher concentrations (figure 5C). No signs of exposure–response at week 4 were observed for patients with baseline CDAI <330.

    Figure 5

    Pharmacokinetic and pharmacodynamic outcomes in patients who were anti-NKG2D-treated and placebo-treated. Data shown include (A) median and IQR for serum anti-NKG2D concentration with a horizontal line at the LLOQ, (B) NKG2D receptor occupancy, (C) CDAI exposure–response according to baseline CDAI score (≥330) and (D) NKG2D receptors on the surface of NK and CD8+ T cells. CDAI, Crohn's Disease Activity Index; LLOQ, lower limit of quantification; NK, natural killer; NKG2D, natural killer group 2 member D.

    Pharmacodynamics: biomarkers

    Treatment with anti-NKG2D significantly decreased the expression of NKG2D on NK cells and CD8+ T cells (free and bound to anti-NKG2D) until week 8 and started to normalise when receptor occupancy was <20% at week 12 (figure 5D) compared with placebo. There were no statistically significant differences between anti-NKG2D and placebo in mean cytokine levels (IL-6, IL-15, TNFα and IL-1Rα) at any evaluated time point. No clinically relevant changes in histological score (Geboes criteria) were observed in either treatment groups, although only a small subset (n=9) of patients who were anti-NKG2D-treated had a post-treatment biopsy.

    Safety and immunogenicity

    A total of 180 AEs were reported for 56/78 patients (72%) during the trial (table 3). The proportions of AEs were comparable between treatment groups, 29 (73%) of patients who were anti-NKG2D-treated reported 93 AEs and 27 (71%) of patients who were placebo-treated reported 87 AEs. The majority of AEs were mainly mild (49% of events) or moderate (43% of events) in severity. A total of 14 (8%) events were evaluated as severe and occurred with comparable frequencies in the anti-NKG2D and placebo groups (8 vs 6, respectively). None of the severe events were judged as trial-drug-related. Five serious AEs were reported in the anti-NKG2D group (exacerbation of CD (n=4) and Clostridium difficile infection (n=1)), and two serious AEs were reported in the placebo group (exacerbation of CD (n=1) and nephrolithiasis (n=1)). None were judged to be trial-drug-related. There were no deaths during the trial. Four patients withdrew from the trial due to exacerbation of CD; none were considered trial-drug-related.

    Table 3

    Summary of overall AEs and by frequency (≥5%) in the safety analysis set (n=78)

    The most frequent AEs (≥5%) were GI disorders, fever, anaemia, arthralgia and nasopharyngitis (table 3). Overall, 29% of patients who were placebo-treated and 15% of patients who were anti-NKG2D-treated had possible or probable treatment-related AEs, the majority of which were general disorders/administration-site conditions, GI disorders and infections/infestations. No safety concerns were reported throughout the trial, with respect to urinalysis parameters, viral screens, physical examinations and electrocardiogram results. Binding antidrug antibodies were found in three samples collected prior to dosing from two patients who were randomly assigned to placebo treatment. No antidrug antibodies or neutralising antibodies were detected in the anti-NKG2D group.


    Both innate and adaptive immune responses in the intestinal mucosa are highly regulated, involving multiple cell types and pathways. Several trials have demonstrated an increased proportion of effector T cells and an abnormal expression of NK receptors on T cells in patients with CD.10 ,13 ,20 Activation of the NKG2D pathway in mucosal T cells in CD could contribute to the increased production of proinflammatory cytokines and enhance cytotoxicity. Also, neutralising anti-NKG2D antibodies have been shown to reduce inflammatory-induced colitis in murine models.15 ,16 The aim of the present trial was to determine whether blocking the NKG2D-activating receptor on these cells would result in beneficial effects in CD.

    Treatment with a single anti-NKG2D injection had an effect on clinical measures that included a significant effect on change in CDAI at week 12 mirrored by a similar response in the HBI at week 12 as well as a change in the physical component summary score of the SF-36. Although the magnitude of improvement appears moderate (55-point difference from placebo), it consistently improved in magnitude from week 1. In the prespecified subgroup of patients with non-failure to biologics, representing ∼70% of the sample, the effects of anti-NKG2D were seen as early as week 1 with consistently greater differences between anti-NKG2D and placebo observed through week 12.

    Although the trial was discontinued due to a futility analysis, resulting in only 78 of the planned 100 patients being randomised, it is interesting to note a few observations from post hoc analyses based on drug exposure levels. For patients with the worse disease at baseline (CDAI ≥330), there appear to be preliminary indications that higher exposure levels, based on the area under the curve at week 4, were consistent with the largest magnitude change in CDAI. In aggregate, these data suggest that higher anti-NKG2D doses may further improve clinical outcomes.

    This was the first trial with a monoclonal antibody targeting the activating receptor NKG2D in CD. A single 2 mg/kg dose was selected for evaluation. The PK data demonstrate linear kinetics for high concentrations (approximately >1 μg/mL), achieving a Cmax of ∼13 μg/mL resulting in >80% receptor occupancy being maintained through week 8, with occupancy dropping to <20% by week 12. This single dose level also significantly decreased the expression of NKG2D on NK and CD8+ T cells through week 8, with normalisation of expression occurring coincidentally with the reduced receptor occupancy at week 12 continuing through week 24, the last follow-up visit.

    There are important limitations to the interpretation of the current data, which include the reduced sample size from that initially planned, limiting the potential generalisability of the above results. In addition, as this is the first trial of anti-NKG2D in CD, the absence of any initial dose findings in CD resulted in the decision to evaluate only 2 mg of anti-NKG2D. The initial trial (NCT00927927) was performed only in patients with rheumatoid arthritis using doses up to 4 mg, but a decision was taken to only evaluate 2 mg in this trial. This was done even though in the first human dose trial no relevant toxicities were demonstrated, nor did the trial inform as to a maximum tolerated dose. This conservative approach was taken to ensure the safe conduct of this initial trial in CD. Also, despite relevant lines of evidence that multiple doses over a treatment interval are typically required for optimising effect, only a single dose was explored in this trial.

    Anti-NKG2D is a novel mechanism as compared with current therapies. A decline in NKG2D expression was observed on peripheral CD8+ T cells and NK cells in patients exposed to anti-NKG2D. We hypothesise that this is probably due to an intracellular internalisation of the receptor. The expression of NKG2D at the surface of immune cells is probably mandatory for their activation through this pathway. Activation of NKG2D, particularly on CD8+ T cells and NK cells, triggers cellular proliferation, cytokine production and cell killing.8 An aberrant expression of NKG2D has been shown in CD, specifically on CD4+ T cells.14 CD4+ T cells expressing NKG2D are expanded in the inflamed tissue of patients with CD. These CD4+ T cells are terminally differentiated and a significant fraction of these cells corresponds to expanded clones.14 The production of proinflammatory and cytotoxic cytokines is strongly enhanced ex vivo by costimulation of the T-cell receptor and the NKG2D receptor.10 ,13 Blocking NKG2D could limit the activation and proliferation of these CD4+ effector T cells. Taken together, these data suggest that several subsets of immune cells expressing NKG2D, including CD4+ and CD8+ T cells and NK cells, could be targeted by the anti-NKG2D antibody.

    Intestinal epithelial cells can regulate T-cell responses in the intestinal mucosa.21 Several NKG2D ligands, upregulated on epithelial cells in the inflamed tissue of patients with IBD, may activate several subsets of effector T cells to proliferate, to produce proinflammatory cytokines and exert cytotoxic functions. For these reasons, we hypothesise that blocking of NKG2D receptors may inhibit interactions between immune and non-immune cells (including lymphoepithelial interactions) and decrease the activation and function of effector cells.

    There were no relevant safety signals in this trial that would preclude further investigation. NKG2D is also involved in responses to various pathogens and tumours. Thus, blocking NKG2D could potentially favour infections and tumours. However, functional assays demonstrated an effect on the fraction of terminally differentiated lymphocytes, which expand in CD. Thus, blocking NKG2D should not have a major effect on immune responses maintaining homeostasis in the intestinal mucosa.

    In conclusion, this clinical proof of principle trial provides appropriate support for further investigation of this novel anti-NKG2D antibody in CD. Future trials should carefully select an appropriate dose range and should incorporate multiple doses over the first 4–16 weeks. It would also be relevant to consider how to evaluate the requirements for maintenance therapy beyond the initial induction period.


    Editorial assistance was provided by PAREXEL, which was funded by Novo Nordisk.


    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


    • Collaborators Belgium: Gert Van Assche, Marc Ferrante; Canada: Denis Petrunia; France: Laurent Peyrin-Biroulet, Guillaume Cadiot, Jean-Louis Dupas; Hungary: Gábor Horváth, Tünde Kristóf, Ágnes Salamon, Róbert Schnabel, Patricia Varga; Israel: Yona Avni, Fred Konikoff, Alexandra Lavy, Ori Segol, Chaim Shirin; Poland: Tomasz Mach, Jerzy Rozciecha, Grazyna Rydzewska; The Russian Federation: Olga Polikarpovna Alexeeva, Lyudmila Gennadievna Lenskaya, Marina Fedorovna Osipenko, Vladimir Ilich Simanenkov, Alexander Vasilievich Tkachev, Alexey A. Yakovlev; USA: Charles F. Barish, Michael S. Epstein, Robert A. Hardi, Daniel J. Pambianco, John E. Poulos, Ziad Younes.

    • Contributors MA and BES were involved in the trial design, implementation, analyses and interpretation of data, and manuscript preparation. MW-J and RP were involved in trial execution and manuscript preparation. RVO was involved in the trial design, analysis and manuscript preparation.

    • Funding Supported by Novo Nordisk A/S.

    • Competing interests MA received honoraria from Novo Nordisk, MSD, Abbvie, Ferring, Genentech, TxCell, Janssen, Pfizer, GSK, Hospira and UCB. BES was an employee and stockholder of Novo Nordisk at the time the trial was conducted. RVO is an employee and stockholder of Novo Nordisk.

    • Patient consent Obtained.

    • Ethics approval Independent ethics committees/institutional review boards.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Further information can be provided by Novo Nordisk A/S.