Article Text

Download PDFPDF

Extended report
Antimodified protein antibody response pattern influences the risk for disease relapse in patients with rheumatoid arthritis tapering disease modifying antirheumatic drugs
  1. Camille P Figueiredo1,2,
  2. Holger Bang3,
  3. Jayme Fogagnolo Cobra4,
  4. Matthias Englbrecht1,
  5. Axel J Hueber1,
  6. Judith Haschka5,
  7. Bernhard Manger1,
  8. Arnd Kleyer1,
  9. Michaela Reiser1,
  10. Stephanie Finzel1,
  11. Hans-Peter Tony6,
  12. Stefan Kleinert7,
  13. Joerg Wendler7,
  14. Florian Schuch7,
  15. Monika Ronneberger7,
  16. Martin Feuchtenberger8,
  17. Martin Fleck9,
  18. Karin Manger10,
  19. Wolfgang Ochs11,
  20. Matthias Schmitt-Haendle11,
  21. Hanns-Martin Lorenz12,13,
  22. Hubert Nuesslein14,
  23. Rieke Alten15,
  24. Joerg Henes16,
  25. Klaus Krueger17,
  26. Jürgen Rech1,
  27. Georg Schett1
  1. 1Department of Internal Medicine 3, University of Erlangen-Nuremberg, Erlangen, Germany
  2. 2Division of Rheumatology, Universidade des Sao Paulo, Sao Paulo, Brazil
  3. 3Orgentec Diagnostika GmbH, Mainz, Germany
  4. 4Instituto de Reumatologia de Sao Paulo, Sao Paulo, Brazil
  5. 5Department of Internal Medicine 2, The Vinforce Study Group, Saint Vincent Hospital, Vienna, Austria
  6. 6Department of Internal Medicine 2, University of Wurzburg, Wurzburg, Germany
  7. 7Rheumatology Practice, Erlangen, Germany
  8. 8Rheumatology Practice and Department of Internal Medicine 2, Clinic Burghausen, Burghausen, Germany
  9. 9Department of Rheumatology and Clinical Immunology, Asklepios Medical Center Bad Abbach, Germany
  10. 10Rheumatology Practice, Bamberg, Germany
  11. 11Rheumatology Practice, Bayreuth, Germany
  12. 12Division of Rheumatology, Department of Internal Medicine V, University of Heidelberg, Heidelberg, Germany
  13. 13ACURA Center for Rheumatic Diseases Baden-Baden, Baden-Baden, Germany
  14. 14Rheumatology Practice, Nuremberg, Germany
  15. 15Schlosspark Clinic, Berlin, Germany
  16. 16Department of Internal Medicine 2, University of Tubingen, Tubingen, Germany
  17. 17Rheumatology Practice, Munich, Germany
  1. Correspondence to Professor Georg Schett, Department of Internal Medicine 3, Rheumatology and Immunology, University Clinic of Erlangen-Nuremberg, Erlangen; Ulmenweg 18, Erlangen 91054, Germany; georg.schett{at}


Objective To perform a detailed analysis of the autoantibody response against post-translationally modified proteins in patients with rheumatoid arthritis (RA) in sustained remission and to explore whether its composition influences the risk for disease relapse when tapering disease modifying antirheumatic drug (DMARD) therapy.

Methods Immune responses against 10 citrullinated, homocitrullinated/carbamylated and acetylated peptides, as well as unmodified vimentin (control) and cyclic citrullinated peptide 2 (CCP2) were tested in baseline serum samples from 94 patients of the RETRO study. Patients were classified according to the number of autoantibody reactivities (0–1/10, 2–5/10 and >5/10) or specificity groups (citrullination, carbamylation and acetylation; 0–3) and tested for their risk to develop relapses after DMARD tapering. Demographic and disease-specific parameters were included in multivariate logistic regression analysis for defining the role of autoantibodies in predicting relapse.

Results Patients varied in their antimodified protein antibody response with the extremes from recognition of no (0/10) to all antigens (10/10). Antibodies against citrullinated vimentin (51%), acetylated ornithine (46%) and acetylated lysine (37%) were the most frequently observed subspecificities. Relapse risk significantly (p=0.011) increased from 18% (0–1/10 reactivities) to 34% (2–5/10) and 55% (>5/10). With respect to specificity groups (0–3), relapse risk significantly (p=0.021) increased from 18% (no reactivity) to 28%, 36% and finally to 52% with one, two or three antibody specificity groups, respectively.

Conclusions The data suggest that the pattern of antimodified protein antibody response determines the risk of disease relapse in patients with RA tapering DMARD therapy.

Trial registration number 2009-015740-42; Results.

  • Ant-CCP
  • Rheumatoid Arthritis
  • Treatment

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.


Sustained control of inflammation is the key treatment goal in patients with rheumatoid arthritis (RA).1 ,2 Earlier initiation of disease modifying antirheumatic drug (DMARD) therapy, more effective use of these drugs and defined treatment goals have led to substantially improved outcomes in RA. Hence, more patients than ever before reach a state of remission characterised by the virtual absence of signs and symptoms of disease. This development became particularly evident, when large cohort studies such as the Norvegian (NOR)-DMARD were analysed, showing doubling of remission rates in the last decade.3

Due to the steady increase of patients in remission the possibility of tapering and stopping DMARD treatment becomes more important.4–6 Several studies have addressed this topic when prospectively tapering or even stopping DMARD treatment in patients in disease remission.7–10 Data from these studies show that only a fraction of patients with RA has the privilege to successfully de-escalate antirheumatic treatment. Therefore, risk prediction of disease relapse after tapering or stopping treatment becomes important, not lastly because it will facilitate the decision to either continue or taper DMARD treatment.11

We have previously shown that presence of anticitrullinated protein autoantibodies (ACPAs) determines the risk for disease relapse in patients tapering DMARD treatment.12 ACPA-positive patients with RA relapse significantly more often than ACPA-negative patients, suggesting that autoimmunity against citrullinated proteins influences the risk of reoccurrence of disease. In the last years it has become clear that autoimmunity in RA targets citrullinated proteins and extends to other protein modifications such as protein homocitrullination, also known as carbamylation,13 and acetylation.14 Hence, the individual patient with RA is characterised by a spectrum of antibodies against post-translationally modified proteins, which may determine her/his risk of recurrence of disease. In contrast to rheumatoid factor, antimodified protein antibody (AMPA) response is much less affected by the inflammatory disease activity of RA and hence is likely to be also maintained in patients with RA in sustained remission.

In this exploratory study our first aim was to examine the pattern of AMPA response in patients with RA including 10 different modifications and antigenic targets including citrullination, homocitrullination/carbamylation and acetylation of proteins. This comprehensive autoantibody analysis was performed in patients with RA included to the RETRO (Reducing therapy in rheumatoid arthritis patients in ongoing remission) study, who were all in sustained remission at entry into the study. Our second aim was to test whether the AMPA profile has a biological significance, that is, whether it influences the risk for disease relapse when DMARD treatment is tapered or stopped.


Patients and inclusion criteria

RETRO is a multicentre, randomised, open, prospective, controlled, parallel-group study (EudraCT number 2009-015740-42). Details of the study are described elsewhere.10 ,12 The primary objective of this study is to evaluate the possibility of tapering or stopping DMARDs in patients fulfilling the American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 classification criteria for RA.15 To be enrolled patients had to have RA for at least 12 months and sustained clinical remission with a disease activity score (DAS) 28 based on erythrocyte sedimentation rate (ESR) of less than 2.6 for at least 6 months.16 ,17 In addition, patients had to receive stable treatment with synthetic and/or biological DMARDs without alteration in dose for at least 6 months. The study was approved by the ethics committee of the Friedrich-Alexander-University of Erlangen-Nuremberg, all local ethics committees of the external centres as well as the Paul Ehrlich Institute and was conducted according to the ethical principles of the Declaration of Helsinki.

Treatment and follow-up

Patients were randomised into three different trial arms: Arm 1 (continuation) kept existing conventional and/or biologic DMARD regimen at full-dose for 12 months. Arm 2 (tapering) reduced the dose of all conventional and/or biologic DMARDs by 50%. Arm 3 (stop) reduced the dose of all conventional and/or biologic DMARDs by 50% for the first 6 months before entirely stopping all DMARDs. Details on the mode of tapering DMARDs are outlined elsewhere.10 The primary efficacy parameter was DAS28-ESR disease activity assessed at baseline and after 3 months, 6 months, 9 months and 12 months. Relapse of disease was defined as leaving DAS28 remission corresponding to a DAS28-ESR score of >2.6.

Peptide synthesis

The mutated vimentin peptide sequence (GRVYAT-Cit-SSAVR) with arginine at position 7 substituted by citrulline (Cit) derived from the vimentin protein used in the mutated citrullinated vimentin (MCV) assay18 was used to introduce the different peptide modifications (figure 1 and online supplementary table S1). For further references also the corresponding native citrullinated vimentin (GGVYAT-Cit-SSAVR) and vimentin (GRVYATRSSAVR) peptides were used in the analyses. Modifications were homocitrullination (=carbamylation; GRVYAT-HomoCit-SSAVR), acetylation of ornithine (GRVYAT-Orn(ac)-SSAVR) and acetylation of lysine (GRVYAT-Lys(ac)-SSAVR). All these modifications were done at the arginine in position 7. In addition, in order to analyse the influence of neighbouring amino acid residues peptides we also alternatively introduced citrulline (G-Cit-VYATRSSAVR), homocitrulline (G-HomoCit-VYATRSSAVR) or acetylated lysine (G-Lys(ac)-VYATRSSAVR) at the arginine in position 2 instead of position 7 (‘inverse peptides’) into the same peptide sequence. In addition, acetylated histone (GGKGGG-LysAc-AARKKA) and non-histone (SRSSGG-LysAc-GSKEAS) peptides were synthesised. The sequences were selected according to heat maps and sequence logos for acetylation sites identified on proteins residing in the nucleus, cytosol, mitochondria or endoplasmic reticulum-Golgi, published recently.19 Crude fractions after peptide synthesis were purified using high-performance liquid chromatography. Quality control of purity of peptides was done by mass spectrometry and analytical high-performance liquid chromatography (see online supplementary figure S1).

Figure 1

Protein modifications serving as antigenic targets. The scheme illustrates the main protein modification used for autoantibody detection. Top: Backbone peptide sequence unmodified vimentin showing arginine residues. Below: peptides identical in length with modifications (symbolised by ‘X’) of respective arginine residues leading to standard and inversely modified peptides. Bottom: nature of modifications showing citrullinated residues (centre), homocitrullinated/carbamylated residues (right), acetylated lysine residues (left) and acetylated ornithine residues (bottom). Cit, citrulline; acetyl-Orn, acetylated ornithine; homoCit, homocitrulline (carbamylated lysine); acetyl-Lys, acetylated lysine.

Antibody testing

Baseline serum samples were available from 94 of the first 101 patients of the RETRO study. AMPA reactivity was determined by ELISA: citrulline, mutated citrulline, inversely mutated citrulline, acetylated ornithine, homocitrulline, inverse homocitrulline, acetylated lysine, inverse acetylated lysine, acetylated histone, acetylated non-histone and arginine (as negative control). In addition antibodies against cyclic citrullinated peptide 2 (CCP2) were measured.

For ELISA, streptavidin-coated microtitre plates (Nunc) were coated with 0.5 µg/mL biotinylated peptide in 100 µL phosphate-buffered saline (PBS). Then, serum diluted 1:100 in PBS supplemented with 1% bovine serum albumin was added. After washing horseradish peroxidase conjugated affinity purified goat antibody against the human IgG-Fcγ fragment (Dianova) was added and the reaction visualised by 3,3′,5,5′-tetra-methyl benzidine. Absorbance at 450 nm was determined using an ELISA reader (Rainbow Reader, Tecan). Each serum sample was tested in duplicates.

Cut-offs were defined by the mean+3*SD of ELISA optical densities (ODs) obtained from a population of healthy controls (n=112). Results are summarised in online supplementary table S2 showing OD values between 0.24 and 0.34. We therefore used a conservative cut-off of 0.4 throughout all peptide reactivities. Measurements of interassay variability were assayed in six replicates with three samples (low, medium and high value) in a single run. Interassay precisions were determined with four samples in five independent tests on 1 day. The results for each peptide are summarised in online supplementary table S3.

Statistical analysis

In descriptive analyses patients with different autoantibody reactivities were analysed for demographic and disease-related parameters. Descriptive results are stated in medians and IQRs due to deviation from normal distribution. Corresponding inferential comparisons of subgroups were calculated using Kruskal–Wallis or Mann–Whitney U tests for numerical variables and Fisher's exact χ2 tests for nominal characteristics. Kaplan–Meier plots were used to illustrate relapse rates over the 12 months with respect to autoantibody reactivity to (1) 0–1, (2) 2–5 or (3) more than 5 protein modifications. Additionally, Kaplan–Meier plots were set up for autoantibody categories, resembling (1) 0, (2) 1, (3) 2 or (4) 3 positive categories (citrullination, homocitrullination/carbamylation and acetylation). SPSS software V.21 was used for calculations. p Values ≤0.05 were considered statistically significant.


Prevalence of antimodified protein antibodies in the RETRO patients

We first analysed reactivity to 10 different modified peptides in the 94 RETRO patients, where serum was available at baseline. No response to negative control (unmodified arginine) peptide was found (0/94; 0%; table 1). With respect to citrullination, citrullinated peptide responses were observed in 44.6% (42/94), mutated citrullinated peptide responses in 55.3% (52/94) and inverse mutated citrullinated peptide responses only in 13.8% (13/94) of the patients. With respect to acetylation, acetylated ornithine responses (48/94; 51.1%) as well as acetylated lysine responses were frequent (29/94; 30.8%), while responses to inversely acetylated lysine peptides were rarer (14/94; 14.9%). Responses to homocitrullinated/carbamylated and inversely homocitrullinated peptides were observed in 30.8% (29/94) and 29.8% (28/94) of the patients, respectively. Finally 27.6% (26/94) of the patients showed responses to acetylated histones and only 5 (5.3%) showed reactivity against acetylated non-histone proteins (figure 2A). Prevalence of CCP2 reactivity was already determined earlier10 attributing to 56.3% of the patients. When CCP2-positive and CCP2-negative patients were separately analysed, a wider AMPA response was observed in CCP2-positive patients (figure 2B). CCP2-positive patients recognised a mean (±SD) number of 4.6 (±2.7) peptides (median: 5; IQR: 3,7). While responses in CCP2-negative patients were lower recognising 1.1 (±1.8) peptides, they were not completely negative with respect to AMPA response (figure 2C).

Table 1

Baseline characteristics of the patients

Figure 2

Antimodified protein autoantibody reactivity pattern in the RETRO patients. (A) Entire cohort; (B) cyclic citrullinated peptide 2 (CCP2)-positive patients; (C) CCP2-negative patients. Each dot resembles one patient. Y axes show optical densities (ODs) measuring the extinction of light at 410 nanometres. Cut-off between positive and negative values at 0.4 is indicated by dashed line. X axes indicate the respective autoantibody reactivity: Cit, citrulline; mCit, mutated citrulline; mCit.inv, mutated citrulline inverse; AcOrn, acetylated ornithine; HC, homocitrulline (carbamylation); HC.inv, homocitrulline invers; AcLys, acetylated lysine; AcLys.inv, acetylated lysine invers; AcHist, acetylated histone; AcNonHist, acetylated non-histone; Arg, arginine.

Reactivity patterns to antimodified proteins in the RETRO patients

Analysis of individual autoantibody reactivity does not completely depict the pattern of response in the individual patient. Therefore, we searched for specific autoantibody profiles, hypothesising that patients can react to one, several or multiple modified proteins. Different patient profiles are shown in figure 3: Apart from those, who did not show any reactivity to modified proteins (example shown in figure 3A) we identified patients with specific reactivity to either citrullinated (figure 3B), homocitrullinated/carbamylated (figure 3C) or acetylated (figure 3D) peptides. Rarely patients with mere antiacetylated histone responses were observed (figure 3E). In addition, we identified those, who showed combined citrullination/homocitrullination responses (figure 3F) or those additionally including acetylated proteins (figure 3G). Finally, we observed patients with broad-spectrum immune responses, which also included the inversely modified antigens either excluding (figure 3H) or including antihistone responses (figure 3I).

Figure 3

Examples for antimodified protein autoantibody profiles in the RETRO patients. Each graph resembles one patient example: (A) negative, (B) anticitrulline reactivity, (C) antihomocitrulline (carbamylation) reactivity, (D) antiacetylation reactivity, (E) antiacetylated histone reactivity, (F) anticitrulline/homocitrulline reactivity, (G) anticitrulline/acetylation reactivity, (H) broad and (I) very broad antimodified protein reactivity. Y axes show optical densities (ODs) measuring the extinction of light at 410 nanometres. Cut-off between positive and negative values at 0.4 is indicated by dashed line. X axes indicate the respective autoantibody reactivity: Cit, citrulline; mCit, mutated citrulline; mCit.inv, mutated citrulline inverse; AcOrn, acetylated ornithine; HC, homocitrulline (carbamylation); HC.inv, homocitrulline invers; AcLys, acetylated lysine; AcLys.inv, acetylated lysine invers; AcHist, acetylated histone; AcNonHist, acetylated non-histone; Arg, arginine.

Heat map of antimodified protein autoantibody reactivity

To further define the interactions of autoantibody reactivity in the RETRO patients we established a heat map, which summarised the coexistence of different autoantibody reactivities. This heat map is based on the per cent overlap between two specific antibody reactivities. In general substantial overlap between certain but not all antibody reactivities was observed (figure 4A). For instance, overlaps between citrulline, mutated citrulline and acetylated ornithine responses were high and found in about two-thirds of the patients. Also, their respective concordance with anti-CCP2 responses was high. In contrast, antiacetylated lysine antibodies showed a much lower overlap with anticitrullinated protein reactivity, while they typically coappeared with acetylated ornithine and also homocitrulline reactivity. Others, such as reactivity against inverse citrulline or acetylated non-histone proteins showed only little overlap with numbers consistently below a third of the patients. Venn diagrams were used to describe the spectrum of overlaps in more detail (figure 4B). These diagrams illustrated that many patients show autoantibody responses to several different rather than individual modified proteins as well as certain preferences such as a central core with broad-spectrum reactivity, the citrullination/homocitrullination overlap and the triple cluster with citrullination/homocitrullination/acetylated ornithine.

Figure 4

Heat map and Venn diagrams for antimodified protein autoantibody interactions in the RETRO patients. (A) Heat map: The following autoantibody reactivities are plotted: Cit, citrulline; mCit, mutated citrulline; mCit.inv, mutated citrulline inverse; AcOrn, acetylated ornithine; HC, homocitrulline (carbamylation); HC.inv, homocitrulline invers; AcLys, acetylated lysine; AcLys.inv, acetylated lysine invers; AcHist, acetylated histone; AcNonHist, acetylated non-histone; Arg, arginine. Numbers indicate the percentage of overlap between two specific autoantibody reactivities. Colours indicate the intensity of the link: >70% (dark red), >60% (light red), >50% (dark orange), >40% (light orange), >30% (dark yellow), >20% (light red), <20% (green). (B) Venn diagrams: Overlaps between mutated citrulline, homocitrulline (carbamylation), acetylated ornithine and acetylated lysine responses are shown. Numbers indicate patients.

Pattern of AMPA reactivity determines the risk of relapse of RA

We next questioned whether the pattern of AMPA reactivity influences the risk for relapse during DMARD tapering. We therefore linked AMPA profiles to outcomes in the RETRO study (relapse or sustained remission). When patients were categorised according to the number of AMPA reactivities, we found that those with no or only one reactivity had a low risk for relapse (18%), while the risk increased in those with 2–5 reactivities (38%) and more than 5 reactivities (55%; figure 5A). Similar observations were made when categorising according to number of specificity groups (citrullination, homocitrullination and acetylation): Hence, patients, who did not show AMPA reactivity had the lowest risk (18%), while relapse rates steadily increased with the recognition of higher numbers of specificity groups, 28% with one, 36% with two and finally 52% with three specificity groups (figure 5B). In this respect it did not matter whether citrullinated peptides (37.2%), acetylated ornithine (39.1%) or acetylated lysine (39.4%) participated in the overall reactivity, with the only exception of homocitrullinated peptides (51.4%), showing high overlap with other reactivities (figure 4B) and therefore favouring multireactivity patterns. Kaplan–Meier curves showed that patients with more widespread autoantibody reactivity faced faster and more pronounced loss of remission. In exploratory statistics, differences were significant with both stratifications according to autoantibody reactivity (figure 5C) and specificity groups (figure 5D). When separately assessing the three treatment groups we found that relapse rates were consistently low (<30%) in patients continuing their DMARDs independent of AMPA profile and in patients who had no or low AMPA reactivity, while increasing to over 80% in patients with wide AMPA responses stopping the DMARDs.

Figure 5

Antimodified protein autoantibody (AAB) reactivity and relapse risk in the RETRO patients. (A and B) Bar graphs showing the risk (in %) to experience disease relapses during tapering and stopping of disease modifying antirheumatic drugs over 1 year. X axes show the pattern of antimodified protein AAB reactivity: (A) groups according to the number of positive AAB reactivities (minimum=0, maximum=10); (B) groups according to the number of positive specificity groups (citrullinated, homocitrullinated/carbamylated or acetylated) in the respective patient (minimum=0, maximum=3). (C and D) Kaplan-Meier plots showing loss of remission status according to AAB reactivities (0–1, 2–5 and >5; C) or specificity groups (0–3; D). Exploratory statistics for group comparisons were performed by applying exact χ2 test. (E) Risk charts for relapses based on antimodified protein autoantibody (AMPA) reactivities (left), AMPA specificity groups (middle) and anticyclic citrullinated peptide 2 (CCP2) AAB status (right) for patients either continuing, tapering or stopping disease modifying antirheumatic drugs.


Our data reveal a substantial, so far unrecognised heterogeneity in the pattern of AMPA response in patients with RA. By simultaneously assessing the responses against citrullinated, homocitrullinated/carbamylated and acetylated proteins including their inverse modifications we were able to draw a comprehensive picture of AMPA autoimmunity in RA. In addition we linked this heterogeneity of AMPA responses to biological differences by analysing the risk for recurrence of disease when tapering DMARD therapy. Our findings support the biological relevance of autoimmunity in RA by showing that patients with a more widespread AMPA response face a higher risk for disease relapse. These findings extend previous observations that CCP2 antibody positivity determines the risk for disease relapse when tapering DMARDs,10 but now allows a more differentiated prediction of relapse risk based on characterisation of autoimmunity against post-translationally modified proteins.

The majority of the RETRO patients revealed positive ACPA results with substantial but not complete overlap with CCP, citrullinated vimentin or mutated citrullinated vimentin. Furthermore, the entire AMPA response was more pronounced in CCP+ than CCP− patients, supporting the central role for anticitrullinated protein antibody testing in detection of autoimmunity of RA. Nonetheless, significant AMPA reactivity can be found in a subset of CCP2-negative patients, highlighting previous observations20 and indicating that more comprehensive antibody testing may be required in thoroughly characterising patients with RA. Overlaps of ACPA were particularly high with antiacetylated ornithine antibodies, while they were lower for antihomocitrullinated/carbamylated peptide or antiacetylated lysine responses.

AMPA responses differed considerably in individual patients with RA including monospecificity to citrullinated, homocitrullinated/carbamylated or acetylated peptides, respectively, as oligospecificity or even polyspecificity against several of these peptide modifications. Citrullination is an enzymatic process, catalysed by peptidyl-arginine deiminases that deiminate arginine to citrulline. ACPA response is meanwhile the best-studied autoimmune feature of RA.21–23 Importantly, these antibodies are linked to human leucocyte antigen (HLA) DR4 epitopes,24 precede the onset of RA25 ,26 and are associated to smoking.27 Moreover, osteoclasts containing vast amounts of citrullinated proteins are considered as an important cellular target for these antibodies.28 Further support for their biological relevance comes from observations that more pronounced ACPA responses are linked to the immediate onset of RA29 and faster disease progression.30 ,31

Homocitrullination (carbamylation) is a non-enzymatic process when cyanate reacts with amino groups of lysine and arginine. Cyanate carbamylates lysine residues to form ε-carbamyl-lysine (homocitrulline) altering protein structure and function. Cyanate formation and homocitrullination occur at sites of inflammation, particularly when the enzyme myeloperoxidase is released from neutrophils converting thiocyanate to cyanate.32 Antihomocitrullinated/carbamylated antibodies are found in about 45% of patients with RA and precede the onset of RA.13 ,33 Notably, homocitrulline is structurally very similar to citrulline, differing only in one carbon atom. Hence, it is not surprising that cross-reactive antibodies exist, which were first demonstrated, when rabbits immunised with homocitrullinated human albumin developed ACPA.34 In accordance, we and others have found substantial overlaps between ACPA and homocitrullinated/carbamylated peptide antibodies.35 Nonetheless, divergent responses exist in a subset of patients with RA reflecting previously observed differences in HLA-DR association and smoking between homocitrullinated/carbamylated peptide antibodies and ACPA.

The third major group of AMPA recognises acetylated peptide residues. Acetylation is a reversible enzymatic process, where acetyl groups are added to free amines of lysine. Lysine acetylation is highly tissue specific and considered as a dynamic process that requires tight balance between acetylases and deacetylases.19 ,36 Acetylation of histones and transcription factors plays a key role in nuclear transcription regulation.37–39 Acetylation of cytoplasmic proteins regulates cytoskeleton dynamics, energy metabolism and autophagy40 ,41 and is influenced by the host microbiome and certain dietary components.42–44 Hence, microbiome changes associated with the development of RA, such as colonisation with Prevotella copri, could be linked to the development of antiacetylated protein immune responses, providing an alternative pathway for the formation of an AMPA response.44 Recently, antibodies against acetylated lysine have been found in patients with early RA at similar frequency as homocitrullinated/carbamylated antibodies.14 In fact, acetylated lysine is identical to homocitrulline except for one terminal amine, which is replaced by a methyl moiety. In accordance, our study showed that one of the most pronounced overlaps of antiacetylated lysine antibodies is with homocitrullinated/carbamylated antibodies.

As a limitation of this study, AMPA were analysed in a post hoc setting. Hence, future studies, which a priori include AMPA as a prognostic factor in their protocol are warranted to confirm the predictive role of AMPA for disease relapse. On the other hand the RETRO study represents an ideal population for such analysis because of their protocol-defined inclusion criteria and prospective tapering strategy. A further limitation is that easy-to-use comprehensive AMPA testing yet needs to be developed to implement this strategy in daily clinical practice, which is so far confined to testing for anticitrullinated protein antibodies.

In summary, this exploratory study shows that a broader AMPA response is associated with higher rates of relapse of RA. Hence, the composition of the AMPA response influences the persistence of RA and the likelihood to successfully taper or withdraw DMARDs. It can be speculated that different mechanisms lead to either citrullinated, homocitrullinated/carbamylated or acetylated protein immune responses. If so, it may be less likely to erase more rather than one precipitating factor, making patients with a more widespread immune response more resistant to successful DMARD tapering. On the other hand, our observations will allow a more individualised decision making for DMARD tapering based on the profile of the patient's AMPA response.



  • Handling editor Tore K Kvien

  • Contributors JFC, AJH, BM, AK, MR, H-PT, SK, JW, FS, MR, MFe, MFl, KM, WO, MS-H, H-ML, HN, RA, JH and KK collected the data. CPF, HB and JH analysed the data. ME performed statistical analyses. SF, AJH, JR and GS designed the study. CPF, HB and GS wrote the manuscript.

  • Funding This study was supported by the Deutsche Forschungsgemeinschaft (SPP1468-IMMUNOBONE; CRC1181), the Bundesministerium für Bildung und Forschung (BMBF; project METARTHROS), the Marie Curie project OSTEOIMMUNE, the TEAM project of the European Union and the IMI funded project BTCure. CPF was supported by Ciencias sem Frontieras from the Conselho Nacional de Desenvolvimento Cientifico e Teconologico do Brasil (CNPq 200175/2014-9).

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval The study was approved by the ethical committee of the University Clinic of Erlangen.

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