Research ArticleRheumatoid Arthritis

Citrullinated peptide dendritic cell immunotherapy in HLA risk genotype–positive rheumatoid arthritis patients

See allHide authors and affiliations

Science Translational Medicine  03 Jun 2015:
Vol. 7, Issue 290, pp. 290ra87
DOI: 10.1126/scitranslmed.aaa9301


In animals, immunomodulatory dendritic cells (DCs) exposed to autoantigen can suppress experimental arthritis in an antigen-specific manner. In rheumatoid arthritis (RA), disease-specific anti–citrullinated peptide autoantibodies (ACPA or anti-CCP) are found in the serum of about 70% of RA patients and are strongly associated with HLA-DRB1 risk alleles. This study aimed to explore the safety and biological and clinical effects of autologous DCs modified with a nuclear factor κB (NF-κB) inhibitor exposed to four citrullinated peptide antigens, designated “Rheumavax,” in a single-center, open-labeled, first-in-human phase 1 trial. Rheumavax was administered once intradermally at two progressive dose levels to 18 human leukocyte antigen (HLA) risk genotype–positive RA patients with citrullinated peptide–specific autoimmunity. Sixteen RA patients served as controls. Rheumavax was well tolerated: adverse events were grade 1 (of 4) severity. At 1 month after treatment, we observed a reduction in effector T cells and an increased ratio of regulatory to effector T cells; a reduction in serum interleukin-15 (IL-15), IL-29, CX3CL1, and CXCL11; and reduced T cell IL-6 responses to vimentin447–455–Cit450 relative to controls. Rheumavax did not induce disease flares in patients recruited with minimal disease activity, and DAS28 decreased within 1 month in Rheumavax-treated patients with active disease. This exploratory study demonstrates safety and biological activity of a single intradermal injection of autologous modified DCs exposed to citrullinated peptides, and provides rationale for further studies to assess clinical efficacy and antigen-specific effects of autoantigen immunomodulatory therapy in RA.


Rheumatoid arthritis (RA) is a common chronic, systemic inflammatory disease, leading to significant pain, disability, premature mortality, and individual and societal economic burden (1). It is currently incurable and is managed with drugs that nonspecifically modulate the immune inflammatory response [disease-modifying anti-rheumatic drugs (DMARDs) and newer biologic (bDMARDs)]. Early intervention and combination therapy are established standards of care in RA. In the most intensively managed early arthritis clinics, reported remission rates range from 30 to 65% (2). Adverse events limit current treatments, and incomplete response and loss of efficacy are common. For example, nonspecific suppression of inflammation or immunity is associated with an increased risk of infection. Individualization of therapy to the autoimmune pathogenesis, genetic background, and relevant autoantigens of RA patients may improve outcomes and reduce adverse events with greater specificity.

A number of immunotherapeutic approaches have been developed on the basis of sub-immunogenic delivery of autoantigens, including collagen, methylated bovine serum albumin (mBSA), and citrullinated fibrinogen in animal models, resulting in antigen-specific deletion of effector T (Teff) cells and/or induction or expansion of regulatory T (Treg) cells (35). In RA, oral delivery of various autoantigens, including collagen, human cartilage gp39, and dnaJp1, has been trialed, with the aim of sub-immunogenic presentation by mucosal dendritic cells (DCs) (68). DCs are professional antigen-presenting cells that migrate from peripheral tissues to draining lymph nodes, where they present antigens in a human leukocyte antigen (HLA)–restricted manner to T cells (9). Animal experiments demonstrate that delivery of immunomodulatory DCs exposed to autoantigen can suppress experimental arthritis in an antigen-specific manner (1012). However, translation to clinical trials is challenging because multiple methods have been shown to modulate DC function and T cell responses, and multiple autoantigens have been described in RA, to which T cell responses are difficult to measure in vitro. Furthermore, the scope of what can be achieved in a clinical trial of DC immunotherapy is limited by cost and safety considerations.

The strongest RA genetic association maps to a shared conserved region of HLA-DRβ 70–74 termed the “shared susceptibility epitope (SE),” which includes HLA-DRB1*04:01, *04:04, *01:01, and *04:05 molecules (13, 14). Disease-specific autoantibodies specific for citrullinated peptide antigens (ACPA or anti-CCP) are found in the serum of about 70% of RA patients and are strongly associated with HLA-DRB1 SE alleles (15). Consistent with the proposed autoimmune mechanism, synovial biopsies of ACPA-positive patients demonstrate high levels of lymphocytic infiltration (16). Furthermore, circulating autoreactive T cells recognizing citrullinated autoantigens were identified in the peripheral blood (PB) of HLA-DRB1 SE+ RA patients with in vitro assays or peptide–HLA-DR tetramers (1719). Structural studies demonstrated that citrulline is accommodated for presentation within the P4 pocket of HLA-DRB1*04:01 and 04:04 (20, 21). These data suggest that ACPA+ HLA-DRB1 SE+ RA patients are an appropriate target population for initial investigation of immunomodulatory therapy consisting of modified autologous DCs exposed to putative citrullinated autoantigenic epitopes.

Activation of the transcription factor nuclear factor κB (NF-κB) drives the coordinate up-regulation of molecules associated with DC antigen presentation (22). Multiple genes driving NF-κB pathway activation are associated with RA susceptibility (23). Furthermore, NF-κB subunits are overexpressed in RA synovial tissue (24, 25). DCs deficient in the NF-κB subunit, RelB, suppressed preexisting immune responses in vivo in an antigen-specific manner through induction of a suppressive CD4+ Treg cell population (26, 27). Suppression of NF-κB, including RelB, in DCs can be achieved pharmacologically in vitro with Bay11-7082, a specific, irreversible inhibitor of NF-κB, in mice and humans (26, 28). A single subcutaneous dose of DCs modified with Bay11-7082 suppressed established arthritis in an antigen-specific manner in mice (11).

To test this strategy in humans, we carried out an exploratory phase 1 trial of DCs modified with Bay11-7082 and exposed to citrullinated peptides in ACPA+ RA patients with HLA-SE risk alleles (Fig. 1). Because efficacy and tolerability were unknown and immune mechanism was untested in patients, our aims were to assess safety and to describe the immunological and clinical effects of a single dose of autologous modified DCs and citrullinated peptides in DMARD-treated RA patients with any level of disease activity and any disease duration in a first-in-human study.

Fig. 1. Concept and schema of immunomodulatory therapy with DCs and citrullinated peptides in HLA risk genotype–positive patients with RA.

(A and B) HLA-DRB1 SE+ RA patients (A) are the target population for initial investigation of immunomodulatory therapy of modified autologous DCs (B) exposed to putative citrullinated autoantigenic epitopes (A). HLA-DR risk allotypes bind citrullinated self-peptides due to a critical lysine residue at position 71 of the peptide binding groove (A). DCs generated ex vivo from PB monocytes in the presence of the NF-κB inhibitor Bay11-7082 are exposed to citrullinated peptides and then injected intradermally (I.D.) (B). Immune response is expected in draining lymph nodes (LN) on Teff and Treg cells, with expected impact on joint symptoms through regulation of inflammation (A). (C) Schema for ex vivo production, injection, and follow-up. Each follow-up assessment included clinical and laboratory evaluation of toxicity (table S2); phone follow-up evaluated clinical evidence of toxicity only. Assessments included clinical and laboratory evaluation of RA disease activity, using DAS scores. At each assessment, PB mononuclear cells (PBMCs) and serum were collected for analysis of cell populations by flow cytometry (table S3), in vitro responses to antigens (table S4), serum analytes (table S5), anti-CCP titer, and peptide-specific ACPA (fig. S5 and table S4). Fasting insulin and glucose were measured at baseline and 1 month, and patients recorded fasting glucose measurements for the first 2 days after Rheumavax.



We recruited 34 ACPA+ RA patients carrying HLA-DRB1 SE alleles, 9 of whom were treated with low-dose and 9 with high-dose Rheumavax; 16 controls were untreated as outlined in the flowchart (fig. S1). Baseline characteristics and outcomes for each group are shown in Table 1 and for individual participants in table S1. Among those treated, 55% carried at least one HLA-DRB1*04:01 allele and 45% carried other SE alleles. The median [interquartile range (IQR)] disease duration was 3 years (1.5 to 4 years), and the baseline disease activity score (DAS28CRP) was 2.43 (1.54 to 3.81) for the low-dose group. Median disease duration was 2 years (1 to 4 years), and baseline DAS28 was 2.2 (1.56 to 3.26) for the high-dose group. Low baseline DAS28 are as expected for a cohort of RA patients with short disease duration treated with multiple DMARDs (29). Baseline characteristics did not differ significantly between treated groups and controls (Table 1).

Table 1. Demographic details of treated and control patients.

DAS, disease activity score; RF, rheumatoid factor; MTX, methotrexate; SSZ, sulfasalazine; HCQ, hydroxychloroquine; LEF, leflunomide; TNFi, tumor necrosis factor inhibitor; N/A, not applicable.

View this table:


Patients received a single intradermal administration of Rheumavax to the upper thigh (doses indicated in Table 1 and table S1). Predicted adverse events included injection site and draining lymph node reactions, allergy or anaphylaxis, hypoglycemia (preclinical toxicity studies), and disease flare. Observed adverse events included transient new leukopenia or lymphopenia in six patients, transient anemia in two, and transient new elevation of liver transaminases in two. One patient had a self-limited headache and one had a single fasting glucose of 2.9 μM measured by a glucometer on the day after Rheumavax. All adverse events were grade 1 out of a maximum 4, and there was a comparable frequency in low- and high-dose groups (Tables 1 and 2 and tables S1 and S2).

Table 2. Clinically relevant adverse events (treatment and possibly treatment-related).

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase.

View this table:

Effects on Teff and Treg cells, antigen-specific responses, and systemic inflammation

To assess T cell effects, we followed changes in PB CD4+ Teff and Treg cells by flow cytometry after Rheumavax (table S3 and fig. S2). Initial assessment was 6 days after Rheumavax, when effects of DCs on peripheral T cell immunity were anticipated. We focused on changes at day 6 and 1 month, which were most likely attributable to Rheumavax. Figure 2 compares changes in %CD4+CD25+CD127+ Teff, CD4+CD25hiCD127 Treg, and the ratio of Treg/Teff in patients receiving low-dose, high-dose, or no Rheumavax over 1 month. We observed a reduction in % Teff by at least 25% in 11 of 15 patients assessed within the first month after Rheumavax. In controls, the % Teff decreased by ≥25% in 0 of 5 (P < 0.01 relative to treated patients, Fisher’s exact test; Fig. 2A). Whereas the % Treg increased by ≥25% in only 5 of 15 treated patients and in 0 of 5 controls (Fig. 2B), the Treg/Teff ratio increased by ≥25% in 11 of 15 treated patients and in 1 of 5 controls assessed over the same time frame, consistent with an altered balance of regulatory to effector CD4+ T cells (Fig. 2C). Changes in Teff, Treg/Teff, and disease activity in each patient over 6 months after Rheumavax are detailed in fig. S4. Interindividual variability was evident. Because the timing of the peak change in Teff or Treg/Teff ratio varied, we calculated the peak % increase (Emax) or decrease (Emin) of each change above baseline after Rheumavax. We observed at least one of these T cell effects in all but two treated patients (patients 1 and 16, fig. S4 and Fig. 2D). In controls, the median Teff Emin was 105 (88 to 127) and Treg/Teff Emax was 96 (89 to 102).

Fig. 2. Changes in Teff and Treg cells in treated and control individuals.

(A to C) PB % CD4+CD25+CD127+ Teff (A), % CD25hiCD127CD4+ Treg cells (B), and Treg/Teff ratio (C) for individuals at baseline (month 0), 6 days, and 1 month after low- or high-dose Rheumavax, and controls at baseline and 1 month. (D) The maximum decrease in % CD4+CD25+CD127+ Teff and the maximum increase in % CD25hiCD127/CD25+CD127+CD4+ T cells (Treg/Teff) were calculated in each patient after Rheumavax relative to baseline (that is, Emax = peak value/baseline value × 100%; Emin = trough value/baseline value × 100%). Emin Teff and Emax Treg/Teff values are plotted with median and IQR. n = 9 each for low-dose (black symbols) and high-dose (green symbols) Rheumavax; n = 5 for controls.

Although PB autoantigen-specific proliferative responses were weak, we previously demonstrated interleukin-6 (IL-6) production by CD4+ T cells in response to citrullinated autoantigenic peptides in RA patients (18). We assessed ex vivo antigen-specific T cell proliferative and IL-6 responses to citrullinated peptides delivered with Rheumavax and to aggrecan84–103–Cit93, which was not delivered with Rheumavax, in treated patients and controls. Stimulation indices for peptide-specific proliferative responses were generally <2, as described previously (18). We therefore compared IL-6 responses by group over 1 month before and after Rheumavax by linear mixed-effect (LME) modeling. Individual responses varied considerably and the group × time interaction was significant only for IL-6 responses to vimentin447–455–Cit450, with reduction in the Rheumavax group relative to controls (P < 0.05, with correction for baseline) (Fig. 3A). We compared proliferative responses to tetanus toxoid antigen in treated patients and controls as an indicator of immune responsiveness toward nominal foreign antigen. As demonstrated previously, RA patients demonstrate relative T cell anergy toward tetanus toxoid (18). There was no significant difference in the response of the Rheumavax-treated and control groups over the first month after Rheumavax; however, the tetanus toxoid response increased in many subjects after Rheumavax (Fig. 3B).

Fig. 3. Changes in citrullinated antigen-specific IL-6 responses and tetanus toxoid–specific proliferative responses in treated and control individuals.

PBMCs (2 × 105 per well) were incubated for 5 days with aggrecan84–103–Cit93, vimentin447–455–Cit450, fibrinogen α chain717–725–Cit720, and collagen type II1237–1249–Cit1240 (0 or 30 μg/ml) or with tetanus toxoid [4 Limes flocculation units (Lf)/ml]. IL-6 production and T cell proliferation were measured by cytokine bead array. The stimulation indices for peptide-stimulated/unstimulated IL-6 and T cell proliferation were calculated and are plotted for each individual at baseline (0) and 1 month. Black symbols, low dose; green symbols, high dose. (A) Citrullinated peptide–specific IL-6. (B) Tetanus toxoid–specific proliferation. n = 17 for Rheumavax; n = 10 for controls. Differences in Rheumavax-treated and control group responses over time were compared by LME for group ×time interaction. *P = 0.02.

Initial disease activity assessment was 1 month after Rheumavax. We fitted linear regression models to identify immunologic features associated with change in DAS28 at 1 month (Materials and Methods and Table 3). After FDR correction, day 6 reduction in Teff, day 6 and 1 month reduction in anti-CCP immunoglobulin A (IgA)/IgG level, and peak increase in Treg/Teff were associated with reduced DAS28 at 1 month (Fig. 4, A to C, fig. S4, and Table 3). P values for Emin Teff and Emax Treg/Teff were 0.01 and 0.07, respectively (R2 = 0.41 and 0.24, respectively), after removing a potential outlier data point with a change in DAS28 of −2.79. Changes in anti-CCP IgA/IgG levels were not correlated with changes in DAS28 in controls (P = 0.64). In contrast to the changes observed in anti-CCP level, when assessing citrullinated peptide–specific ACPA IgG responses to peptides not included in Rheumavax (table S4), we found no difference in the proportion of treated and control patients with positive assays (fig. S5).

Table 3. Linear regression analysis of change in DAS at day 30 on laboratory features.

Features with P ≤ 0.05 (t test) are shown in the table. The intercepts are not shown. % Foxp3+ Treg is the % of the CD4+CD25hiCD127lo Treg cells expressing Foxp3. Emax = peak value/baseline value × 100%; Emin = trough value/baseline value × 100%. FDR, false discovery rate (calculated using Benjamini-Hochberg correction).

View this table:
Fig. 4. Changes in T cells, proinflammatory cytokines, and anti-CCP levels in patients treated with Rheumavax and controls.

(A and B) The relationship between the change in DAS4v at 1 month after Rheumavax with the changes in Teff at day 6 (A) and Emax Treg/Teff (B) are plotted, showing regression lines. (C) Serum levels of anti-CCP antibodies were determined using the CCP3.1 enzyme-linked immunosorbent assay (ELISA) assay. The relationship between change in DAS4v at 1 month after Rheumavax with the change in anti-CCP level at day 6 after Rheumavax is plotted, showing regression line. P and R2 values are shown in Table 3. (D) Change in C-reactive protein (CRP) at 1 month plotted relative to DC dose/kg. The regression line is shown (r2 = 0.37). Black symbols, low dose; green symbols, high dose. (E) Serum levels of IL-15, CXCL11, CX3CL1, IL-29, and PYY were determined by luminex assay at baseline, day 6 (d6), and 1 month after Rheumavax (closed circles) and at a 1-month interval in controls (open circles). Log-transformed values are plotted as median and IQR. n = 9 for controls; n = 18 for Rheumavax. Differences at 1 month were estimated with the Mann-Whitney test (*P = 0.046, IL-15; P = 0.028, CXCL11; P = 0.024, CX3CR1; P = 0.045, IL-29; P = 0.027, PYY).

These features suggested systemic and immunological effects of Rheumavax. We observed that CRP was more likely to fall in patients receiving higher doses of DCs per kilogram (P = 0.016, Fig. 4D). To further explore anti-inflammatory effects, we fitted an LME model for each of 106 serum protein analytes (table S5) collected at baseline, day 6, and 1 month. Reductions in 11 proteins had significant group-time effects discriminating Rheumavax-treated patients from controls at 1 month, correcting for baseline (table S6). The concentrations of IL-15, CX3CL1 (fractalkine), CXCL11 (ITAC), IL-29, and peptide YY were significantly lower in Rheumavax-treated patients at 1 month but not at baseline relative to controls (Fig. 4E).

Clinical effects

To determine immediate clinical effect or flares, we focused on the first month after Rheumavax, by dose group (Fig. 5). Among the 18 treated patients at baseline, 9 had a swollen joint count (SJC) ≥1 (DAS28 >2.4) and thus could be evaluated for clinical improvement (“active disease”). In the low-dose group, five of nine were active. DAS28 and SJC decreased in four of four patients evaluated at 1 month. Patient 1 was evaluated at 2 months: both DAS28 and SJC were reduced. None of the remaining four patients flared: maximum DAS28 was 2.64 at 1 month. In the high-dose group, four of nine were active. DAS28 decreased in all four, and SJC decreased in three and did not change in one. None of the remaining five patients flared: maximum DAS28 was 2.49. DAS28 and SJC followed no trend among controls (Fig. 5, A and B). When evaluated longitudinally, the active disease group had a median change in DAS28 of −0.84 (−1.52 to 0.60) at 1 month and −0.45 (−1.58 to 0.02) at 6 months, whereas the inactive disease group had a median change of 0.48 (−0.05 to 0.61) at 1 month and 0.58 (0.03 to 1.07) at 6 months after Rheumavax. The control group had a median change of 0.12 (−1.64 to 1) at 1 month and −0.58 (−2.94 to 0.65) at 6 months (Fig. 5C). All groups including the control group were actively managed for RA during this 6-month period. Neither the ratio of modified DC/DC HLA-DR nor CD40 mean fluorescence intensity (MFI) (fig. S6) was associated with 1-month change in DAS28 (P = 0.55 and 0.91, respectively). DMARDs were sometimes reduced or stopped during the follow-up period for clinical reasons (fig. S4). Clinical improvements that had occurred after Rheumavax were destabilized in several cases with changes in DMARDs (for example, patients 8 and 15; fig. S4). In patient 7, a knee injury precipitated a general flare (fig. S4).

Fig. 5. Clinical effect of Rheumavax.

(A) Disease activity score (DAS4v) at baseline and 1 month after treatment in RA patients treated with low and high doses of DCs and in controls. Patients with active disease (SJC ≥ 1) and inactive disease (SJC = 0) are denoted by the dashed and solid lines, respectively. (B) SJC at baseline and 1 month after treatment in RA patients in low- and high-dose DC groups and in controls. (C) Change in DAS relative to baseline for 6 months plotted in active and inactive groups and controls. Grouped data are displayed as median and IQR. n = 9 for Rheumavax groups; n = 11 for control group. Groups were compared at each time point using the Kruskal-Wallis test with Dunn’s multiple comparison test: **P = 0.005, Rheumavax active versus inactive; P = 0.033, Rheumavax active versus control; *P = 0.013, active versus inactive; P = 0.040, inactive versus control. Control data not available at 2 months.


Individualization of therapy to the autoimmune pathogenesis, genetic background, and relevant autoantigens of RA patients may improve patient outcomes and reduce adverse events with greater specificity. Such individualized approaches apply the principles of immune tolerance (30). However, translation to clinical trials is challenging. Effective long-term disease suppression by restoration of immune tolerance will likely require intervention in a multidose regimen very early after or even before disease onset (31). Furthermore, the safety of autoantigenic peptide immunomodulatory therapy in RA and the best vehicles to achieve this are as yet unknown (32). As a first step, in this first-in-human phase 1 study, we wished to explore safety and immunological and clinical effects of a single dose of immunodulatory DCs and citrullinated antigen in a small trial of DMARD-treated HLA SE+ RA patients within the first years after diagnosis. We found that a single administration of 0.6 × 106 to 4.5 × 106 Rheumavax in DMARD-treated HLA SE+ RA patients was well tolerated. We tested two dose levels: the high dose was chosen on the basis of the maximum predicted yield after venesection of 250 ml of blood and our experience in melanoma patients (33). Human studies suggest that intradermal delivery of ≤5 × 106 mature DCs optimizes migration to draining lymph nodes (34). Although we did not carry out labeling studies of Rheumavax, we previously demonstrated migration of labeled immature DCs to draining lymph nodes after intradermal delivery (35). Preparation of Rheumavax from venesected blood limited DC dose delivery: two of nine patients received <1 × 106 DCs, and nine of nine received <5 × 106 DCs. Furthermore, two patients had self-limiting anemia during follow-up, potentially due to previous venesection.

The relationship of change in CRP to dose suggests a systemic anti-inflammatory effect of Rheumavax. Consistent with this notion, we observed significant reductions in activated Teff cells and proinflammatory cytokines and chemokines in Rheumavax-treated patients relative to untreated controls. Given that DCs modified with Bay11-7082 did not secrete higher levels of anti-inflammatory cytokines before administration (fig. S8) and that DCs were administered in low numbers to a noninflamed intradermal site, we hypothesize that this anti-inflammatory effect relates to the rapid change in activated Teff after treatment. In mice, we showed that Teff proliferation in response to antigen is reduced in lymph nodes draining the site of antigen-exposed modified DCs (26). The association of reduced DAS28 at 1 month, with day 6 and peak reduction in Teff supports the conclusion that reduction in joint inflammation is related to the Teff response. Teff may have decreased after Rheumavax by several mechanisms, including deletion or anergy in response to antigen recognition, regulation, or apoptosis due to reduction in homeostatic cytokines (36). In this regard, we investigated changes in a large number of serum analytes without FDR correction to identify possible immune response proteins for validation in future trials. Although it is impossible to draw firm conclusions in this small trial, the analytes differing between Rheumavax-treated subjects and controls are of interest. IL-15 is a key inflammatory cytokine in RA (37), which can sensitize autoreactive T cells to citrullinated peptides (38). Thus, IL-15 reduction could arrest autoreactive Teff, potentially with more widespread effects than just peptide-specific Teff. The Treg/Teff ratio increased in treated patients associated with the reduction in Teff. The relative increase in Foxp3+ Treg cells could have resulted from decreased inflammatory cytokine effects on Foxp3 expression and enhanced Treg stability in the face of fewer Teff (39, 40). Improvements in T cell function, including recall responses to antigen and Treg cell function, have been described with TNF inhibition (41, 42). Although serum TNF was not measured, we observed a reduction in several proinflammatory cytokines and chemokines relative to controls. Reduction in CX3CL1 has been shown to correlate with response to TNF inhibitors in RA patients (43).

Here, we delivered immunomodulatory DCs and citrullinated antigen in a selected HLA at-risk population of RA patients. Immune assays suggested multiple immunoregulatory and anti-inflammatory effects of Rheumavax on T cells, serum cytokines, and anti-CCP and that some of these were associated with clinical effects. However, the exploratory studies carried out are limited by small sample size and interindividual variability. The results of antigen-specific assays were not conclusive. Although IL-6 responses to vimentin447–455–Cit450 decreased after Rheumavax relative to responses in control subjects, citrullinated peptide–specific responses varied between individuals. Assays of citrullinated self-peptide responses in vitro are limited by low sensitivity. Moreover, sensitivity to detect a consistent change in antigen-specific T cell responses among treated patients would have been limited in a single-dose trial (44). Therefore, to adequately assess antigen-specific responses, more sensitive assays, such as pHLAII tetramers or intracellular cytokine responses after in vitro restimulation, would need to be carried out in trials of larger numbers of subjects given multiple doses of citrullinated peptide immunomodulatory therapy (45). Nevertheless, in the current trial, tetanus toxoid–specific T cell proliferative responses were not suppressed after Rheumavax, suggesting that Rheumavax was not broadly immunosuppressive to a foreign antigen response.

Our study has limitations. The primary outcomes of the trial were safety and effects on immune function. For this, and logistic reasons, the trial was open-labeled and assessed only a single Rheumavax administration. The sample size and trial design were limited by cost of production of autologous DCs and potential for disease flare; therefore, we did not deliver irrelevant control antigens or unmodified DCs. Because open-labeled trials are subject to placebo effects, larger, placebo-controlled, double-blind studies are needed to assess clinical efficacy of antigen-specific immunotherapeutic approaches. It is important that future trials of immunomodulation develop and include mechanistic studies as well as clinical outcomes to advance the field for further mechanistic understanding to underpin the development of antigen-specific approaches in RA. Inclusion in future trials of a control arm receiving immunomodulatory DCs without peptide or with irrelevant peptide would aid interpretation of antigen specificity. Furthermore, two control groups were recruited to compare serial clinical and immune markers in patients who were not treated with Rheumavax. These comparisons may be confounded by the small numbers of patients per group with variability in disease activity and DMARD treatments, as well as fewer post-baseline visits in the controls. For these reasons and because immunological and clinical effects become more difficult to attribute to a single intervention with increasing time, we limited almost all analyses to the first month after Rheumavax treatment and to a 1-month follow-up in controls.

Self-peptide response assays could be optimized with advances in technology for analysis of antigen-specific T cells. The peptides delivered in Rheumavax would be predicted to bind the different HLA-DR SE molecules with variable affinity based on core sequence binding motifs (20), and indeed, fibrinogen β chain433–441 may not have bound all the HLA-DR molecules (fig. S7). Using anchor residue frequencies, citrullinated peptides could be further optimized for binding to RA-associated HLA-DR molecules.

In this preliminary study, a single intradermal injection of autologous modified DCs exposed to citrullinated peptides was safe, and immunoregulatory and anti-inflammatory effects were observed in HLA risk genotype–positive RA patients. Further studies are needed to assess clinical efficacy and associated immune effects of antigen-specific immunotherapy in RA.


Study patients

Inclusion criteria were a diagnosis of RA with symptoms for at least 3 months and treated with at least one DMARD or bDMARD, positive anti-CCP, and carriage of HLA-DRB1 SE. Exclusion criteria were serious infection or major surgery within 28 days; history or family history of atopy or positive radioallergosorbent (RAST) test; malignancy; significant cardiovascular, renal, liver, neurological, or skin disease; positive hepatitis or HIV serology; pregnancy; or drug abuse. Eleven control RA patients were recruited and studied according to the same schedule as the 18 treated patients, except that flow cytometric analyses and T cell responses to antigen could not be evaluated for technical reasons in the controls. To compare changes in cell populations longitudinally in RA patients not receiving Rheumavax, an additional five anti-CCP+, HLA-DRB1 SE+ control RA patients were recruited: flow cytometry and in vitro assays were carried out twice, at visits 1 month apart (control group 2, Table 1 and Fig. 2).

Study design

Between 10 November 2009 and 31 March 2011, we conducted a single-center, open-labeled, first-in-human, controlled, prospective phase 1 trial at the Princess Alexandra Hospital, Brisbane, in which autologous modified DCs pulsed with four citrullinated peptide antigens, designated “Rheumavax,” were administered once intradermally in two groups: a low dose of 1 × 106 DCs and a high dose of 5 × 106 DCs (Fig. 1B, Table 1, table S1, and full protocol). RA patients were recruited and studied according to the protocol and statistical analysis plan, outlined in the flowchart (fig. S1). Written informed consent was obtained from all patients, and the study was approved by the human ethics committees of the Metro South District and University of Queensland, Brisbane, Australia, and by the Therapeutic Goods Administration of Australia under the Clinical Trial Notification (CTN) scheme. Patients’ safety was monitored and assessed against specifically designed toxicity criteria (table S2) at continuous time points. Safety data 1 month after administration of the low dose were evaluated by the Ethics Committee before commencing the high-dose group (table S1).

Preparation of Rheumavax

In preliminary experiments, we determined that addition of Bay11-7082 to monocyte-derived DC cultures of RA patients was immunomodulatory: It dose dependently reduced their capacity to stimulate allogeneic T cells in mixed lymphocyte cultures (fig. S8, A and B), similar to healthy control Bay11-7082 DCs (28). In these cultures, levels of HLA-DR, CD40, CD80, and CD86 varied between DCs generated from different RA donors. Cell surface expression of HLA-DR and CD80 consistently decreased as the concentration of Bay11-7082 increased, whereas CD40 and CD86 expression varied (fig. S8C). CCR7 was expressed by 100% of Bay11-7082 DCs with an intensity equivalent to DCs (fig. S8D). Cell death increased as Bay11-7082 concentrations increased above 3 μM, consistent with the requirement of NF-κB for cell survival. Cytokine secretion, including IL-10 by DCs and Bay11-7082 DCs, was similar, except for significantly lower secretion of TNF by Bay11-7082 DCs relative to DCs (fig. S8D).

At the preentry visit, the concentration of Bay11-7082 was titrated in monocyte-derived DC cultures to where HLA-DR expression was reduced while maintaining DC viability relative to control DCs. In all patients, the concentration of Bay11-7082 ranged from 2 to 2.5 μM. Before injection, modified DCs were exposed to the following peptides to permit presentation by HLA-DR molecules: collagen type II1237–1249–Cit1240, fibrinogen α chain717–725–Cit720, fibrinogen β chain433–441–Cit436, and vimentin447–455–Cit450 (table S4). All cultures met release criteria for viability and sterility. HLA-DR and CD40 MFI was used as an internal control as evidence that the Bay11-7082 had modified the DCs relative to untreated DCs. Preset release criteria of 40% reduction in MFI of HLA-DR and CD40 relative to untreated DCs were found to be too stringent in most patients [median (IQR) actual reduction in HLA-DR, 17% (10 to 33), and CD40, +5% (11 to +24)]. Because the trial was exploratory and the absolute value of this internal control was not expected to affect safety, the trial investigators agreed to record the values and release the DCs for administration. The yield of modified DCs determined the final dose delivered to the patient, and varied between 0.5 × 106 to 1 × 106 DCs (7.2 × 103 to 1.7 × 104 DCs/kg) in the low-dose group and 2 × 106 to 4.5 × 106 DCs (2.7 × 104 to 6.2 × 104 DCs/kg) in the high-dose group (Table 1 and table S1). DCs were resuspended in 0.5 ml of sterile phosphate-buffered saline for intradermal injection.


Citrullinated aggrecan, vimentin, collagen type II, and fibrinogen α and β chain peptides predicted to bind the HLA-DRB1 SE were synthesized by Auspep Pty. Ltd. to 95% purity and tested for endotoxin and sterility, and their identities were confirmed by N-terminal sequencing. Peptides were reconstituted in water and stored at −20°C. Final dilutions were made with medium.

HLA-DR–binding assay

The method was adapted from that previously described (46). Flat-bottom 96-well plates were coated with 100 μl per well of HLA-DR–specific monoclonal antibody L243 at 20 μg/ml in 50 mM sodium carbonate buffer for 2 hours at 37°C. Wells were washed and blocked with 0.2% BSA overnight at 4°C. Lysates of the homozygous Epstein-Barr virus–transformed B lymphoblastoid cell lines expressing appropriate DR haplotypes were prepared, and cell debris was removed by centrifugation. Lysate (100 μl per well) was added overnight at 4°C. Wells were washed, and test peptides were added at 3- to 100-fold molar excess with biotinylated CLIP peptide and then incubated for 48 hours at 37°C. Plates were washed and incubated with neutravidin horseradish peroxidase in binding buffer for 1 hour, developed with o-phenylenediamine dihydrochloride, and then detected in a Varioskan plate reader.

Flow cytometric analysis of PBMCs

Thawed PBMCs (2 × 105) were incubated for 1 hour at 37°C and then stained with three panels of surface and intracellular markers to assess leukocyte populations (table S3, fig. S3). Antibodies were sourced from BD Pharmingen, eBioscience, BioLegend, and Invitrogen. Tetramers for CD1d were obtained from Dale Godfrey (Monash University, Melbourne, Victoria, Australia). Data were collected on the Gallios flow cytometer (Beckman Coulter) and analyzed using Kaluza software (Beckman Coulter). Live cells were identified as Aqua-negative, and gates were set on the basis of isotype and FMO controls.

Luminex analysis of serum

Serum samples were collected and frozen within 16 hours at −80°C. A high-throughput multiplex assay analyzed a select panel of 106 analytes (table S5) detectable after thaw in human serum samples with Beadlyte technology (MPXHCTYO-60K; Millipore) using the Bio-Plex Luminex-100 Station (Bio-Rad), as described (47). Human bead mates determined the levels of several cytokines with the Human Multi-Cytokine Flex Kit using detection protocol B (Upstate) and duplicate standards. Markers were analyzed in triplicate from a 600-μl sample. Bio-Plex Manager 4.1 software with a five-parameter logistic curve-fitting algorithm was applied for standard curve calculations.

Anti-CCP and ACPA fine specificity assays

Initial screening of the patients was carried out by Queensland Health Pathology Laboratories using the anti-CCP2 ELISA (Axis-Shield). Antibody reactivity against the citrullinated (Cit) and the uncitrullinated form of linear peptides (vimentin1–16–Cit3, fibrinogen α chain27–43–Cit36, vimentin59–74–Cit64, fibrinogen β chain36–52–Cit44, and enolase5–20–Cit9) was determined by ELISA (48) (table S4). Cutoff values for the Cit-specific responses were calculated as previously described (49). Sera collected at each time point from all patients were tested using anti-CCP3.1 IgG/IgA ELISA (Quanta Lite), with titration of serum and quantification based on the standard curve to a maximum of 1000 semiquantitative units/ml.

T cell proliferative and cytokine responses in vitro

PBMCs were thawed from cryopreserved samples, and then 2 × 105 PBMCs per well were incubated with tetanus toxoid (4 Lf/ml) (Chiron Vaccines) or aggrecan84–103–Cit93, vimentin447–455–Cit450, fibrinogen β chain433–441–Cit436, fibrinogen α chain717–725–Cit720, and collagen type II1237–1249–Cit1240 (0 or 30 μg/ml) in complete medium (RPMI, 1× PSG, 1× sodium pyruvate, 10% human serum) for 5 days. T cell proliferation was assessed by [3H]thymidine incorporation (18). IL-2, IL-4, IFN-γ (interferon-γ), IL-10, IL-6, IL-17, and TNF were measured in day 5 supernatants using BD Cytometric Bead Array Human Th1/Th2/Th17 kits (BD Pharmingen). Samples were read on the BD FACSArray bioanalyzer system.


The patients were followed longitudinally according to the flowchart (fig. S1). Not all treated subjects attended for all visits; the patterns of missingness were random for all study parameters. Analysis of safety, efficacy, and immunological parameters was descriptive, including interparticipant comparisons using nonparametric methods. Inter- and intraparticipant comparisons over time used longitudinal methods, including linear regression and linear mixed models. Variables collected longitudinally were expressed as change (Δ) with respect to the nontreated baseline. We determined the maximum increase or decrease in % T cell populations (Emax, Emin), as previously described (that is, Emax = peak value/baseline value × 100%; Emin = trough value/baseline value × 100%) (50). Differences between groups were compared using two-tailed unpaired Student’s t tests, Mann-Whitney test for nonnormally distributed data, and Kruskal-Wallis tests with Dunn’s correction for multiple testing, where three or more groups were compared with a significance level α of 0.05. Variance for nonnormally distributed data is displayed as median ± IQR, otherwise as mean ± SD.

The immune response data set includes 39 features, and the proteomics data set includes 116 features. The statistical analyses used the statistical software R, including the stats and MASS packages. Proteomics data were log-transformed, and missing values were imputed using the NIPALS algorithm (51).

Univariate linear regression model. The immune response data were analyzed with linear regression models. Each predictor (feature) was expressed in terms of change of expression after Rheumavax treatment (day 6 or 30) with respect to the nontreated baseline. The response of interest in the model is the DAS between day 30 (after treatment) and the baseline, denoted by ∆DASD30 = DASD30 − DASbaseline for each individual.

To identify features in which changes of expression were associated with ∆DASD30, we performed a linear regression model on each feature separately. Those with a regression coefficient significantly different from 0 are indicated in Table 3 (P < 0.05).The normality of data was assessed using quantile-quantile plots before the application of linear regression models.

LME model. To investigate the treatment effect in serum analytes and in vitro IL-6 responses to peptides at 1 month relative to baseline, we performed an LME model with time (t) and Rheumavax treatment (trt) as fixed effects and random intercepts. We report features with an interaction effect (t*trt) and F test P < 0.05.


Fig. S1. Flowchart of the study.

Fig. S2. Gating strategy for CD4+ Teff and Treg cells.

Fig. S3. Gating strategy for B cell/T cell/Tfh and NKT/DC/monocyte/NK panels shown in table S3.

Fig. S4. Effects of Rheumavax on disease activity, Teff and Treg cells, tetanus toxoid immunity, and anti-CCP titer in each treated patient.

Fig. S5. Proportion of each group with citrullinated peptide–specific ACPA for the duration of the study.

Fig. S6. Ratio of HLA-DR and CD40 expression by modified DC relative to DC before administration.

Fig. S7. Frequency of anchor residues from naturally eluted peptides from HLA-DRB1*01:01, DRB1*04:01, and DRB1*04:04.

Fig. S8. Generation of monocyte-derived DCs from RA PB in the presence of increasing concentrations of Bay11-7082 reduces HLA-DR expression and the capacity to stimulate allogeneic T cells.

Table S1. Characteristics, clinical outcomes, and treatment side effects for the 18 study subjects.

Table S2. Rheumavax safety toxicity criteria.

Table S3. Flow cytometry panels used in the analysis of PBMCs.

Table S4. Peptides used in this study.

Table S5. Serum analytes measured in this study.

Table S6. LME model of serum analytes.

Source data (excel file)


  1. Acknowledgments: We thank A. Casgrain who produced Fig. 1; A. Tran who carried out clinical assessments; D. White and P. Vecchio who assisted with trial design and management; D. Godfrey who provided CD1d tetramer; and K. Lau, L. van Toorn, N. Bobby, and E. Duggan who provided technical assistance. Funding: Supported by National Health and Medical Research Council (NHMRC) grants 301244, 351439, and 569938; an AusIndustry Biotechnology Innovation Fund grant; a Queensland Government Innovation Start-Up Scheme grant; and the Princess Alexandra Hospital Foundation. R.T. was supported by Arthritis Queensland and an Australian Research Council (ARC) Future Fellowship, S.S. by an NHMRC Peter Doherty Fellowship, and B.J.O.S. by Queensland Government Smart State Fellowship and Arthritis Queensland. A.W.P. is an NHMRC Senior Research Fellow. L.A.T. is a recipient of a De Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) ZonMw Vidi grant. S.K.P. received infrastructure funding support from Therapeutic Innovation Australia. K.-A.L.C. was supported in part by the Australian Cancer Research Foundation for the Diamantina Individualised Oncology Care Centre at The University of Queensland Diamantina Institute. Author contributions: R.T. had full access to all of the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: R.T., B.J.O.S., J.E.C., L.A.T., S.K.P., and K.-A.L.C. Acquisition, analysis, or interpretation of data: all authors. Drafting of the manuscript: H.B. and R.T. Critical revision of the manuscript for important intellectual content: all authors. Obtained funding: R.T. Technical support: H.J.N., S.C.L., S.S., N.R., H.P., B.T.L., J.N., M.E.G.B., C.H., L.A.T., A.W.P., and N.L.D. Statistical analysis and support: A.M.M., S.K.P., and K.-A.L.C. Competing interests: R.T. has filed provisional patents surrounding technology for targeting DCs for antigen-specific tolerance and is a director of the spin-off company, Dendright, which is commercializing vaccines that target DCs to suppress autoimmune diseases in collaboration with Janssen Biotech Inc. A.M.M. was supported from this commercial source. L.A.T. is a recipient of a fellowship from Janssen Biologics. S.K.P. has acted as a consultant and speaker for Novartis and Amylin Pharmaceuticals Inc. and has received grants for clinical studies from Merck, Bristol-Myers Squibb, Novo Nordisk, and Pfizer. The study funders had no role in the design or conduct of the study; collection, management, analysis, and interpretation of the data; preparation of the manuscript; or decision to submit the manuscript for publication. The other authors declare no competing interests. Data and materials availability: Trial registration:; identifier: ACTRN12610000373077.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article