A prospective cohort study of 14-3-3η in ACPA and/or RF-positive patients with arthralgia
© van Beers-Tas et al. 2016
Received: 16 December 2015
Accepted: 15 March 2016
Published: 1 April 2016
14-3-3η (eta) is a novel serum/plasma protein biomarker involved in the upregulation of inflammatory and joint damage factors. We analysed the association of 14-3-3η with the development of clinically apparent arthritis in a cohort of subjects with arthralgia and positivity for at least one serologic marker: rheumatoid factor (RF) or anti-citrullinated protein antibody (ACPA).
Measurement of 14-3-3η in plasma collected on entry into the cohort. For this study, 144 subjects with a minimum of 2.5 (median and maximum 5) years of follow-up were available. The relationship between presence and levels of 14-3-3η and development of arthritis was investigated.
Arthritis occurred in 43 (30 %) of the 144 subjects after a median of 15 months. 14-3-3η was detectable up to 5 years before onset of clinical arthritis and was present significantly more often (36 % versus 14 %; relative risk 2.5, 95 % confidence interval 1.2–5.6; p = 0.02) and at significantly higher levels (median 0.95 versus 0.28 ng/ml; p = 0.02) in subjects developing arthritis compared with those who did not. 14-3-3η levels/positivity and ACPA, but not RF, were univariately associated with the development of arthritis while generalized linear model analysis with RF and ACPA as obligatory factors could not return an incremental benefit with 14-3-3η.
14-3-3η was detectable prior to the onset of arthritis and was associated with arthritis development in arthralgia subjects pre-selected for positivity of RF or ACPA. Its power to predict onset of arthritis independent of ACPA and RF requires a new study in which patients are not pre-selected based on ACPA and/or RF seropositivity.
Keywords14-3-3ŋ protein Rheumatoid arthritis Prediction Anti-citrullinated protein antibodies Arthralgia
The focus on the management of rheumatoid arthritis (RA) is increasingly towards early detection and treatment. Better prediction of the development of RA will potentially allow preventive interventions in these at-risk individuals. Recently, we published a prediction rule for the development of arthritis in rheumatoid factor (RF) and/or anti-citrullinated protein antibodies (ACPA) positive (seropositive) arthralgia patients . Patients could be divided into high, intermediate or low risk categories quite accurately with an area under the receiver operator characteristic (ROC) curve (AUC) of 0.82 at 5 years. However, this is still inadequate for individual patient care – assuming the availability of a preventive intervention; therefore additional biomarkers to improve the predictive arthritis risk algorithm is required.
Several such potential biomarkers were recently described. Examples are anti-carbamylated protein antibodies (anti-Carp) , peptidyl arginine deiminase type 4 (anti-PAD-4) , and a high interferon gene score . Two of these biomarkers were discovered within the same patient group as we will describe here [2, 4]. Such biomarkers may help to improve prediction, but also offer new insights into the course of events leading to clinical arthritis.
Serum 14-3-3η (eta) is a novel protein biomarker showing potential in predicting radiographic deterioration in early and advanced RA [5, 6]. 14-3-3 proteins belong to a family of seven isoforms known to bind to and regulate the biologic activity of various intracellular proteins . Overexpression of 14-3-3 proteins is associated with worse outcomes in various diseases, such as cancers, neurodegenerative diseases and Creutzfeldt-Jakob's disease. The 14-3-3η isoform is expressed at higher levels in patients with arthritis compared with healthy individuals, which is thought to be related to 14-3-3η's direct ability to induce factors linked to inflammation and radiographic damage . 14-3-3η has been shown to induce inflammatory factors such as interleukin (IL)-1 and -6, and is linked to the process of joint damage as it also induces factors such as receptor activator of nuclear factor-kB ligand (RANKL) and matrix metalloproteinase (MMP) 1.
In this study, we analysed the association of baseline 14-3-3η with the development of clinically apparent arthritis in a cohort of subjects with arthralgia who were pre-selected based on being positive for at least one serologic marker: rheumatoid factor (RF) or anti-citrullinated protein antibody (ACPA).
From the Reade seropositive arthralgia cohort, the first 144 participants (with ≥30 months of follow-up or development of arthritis, included between 2004 and 2008) were used. This cohort was set up to determine clinical and serological risk factors for development of arthritis, and comprises subjects at risk of arthritis, as defined by arthralgia (no history and no presence of clinically diagnosed arthritis at the time of their first physical examination and no erosions on X-rays of hands and feet) and positivity for at least one serologic marker: ACPA or RF .
At baseline, all participants had clinical and demographic data collected (including visual analogue scale pain, morning stiffness, total of painful and swollen joints) and provided a plasma sample through standard phlebotomy procedures. Enrolment was based on being positive for ACPA and/or RF. Plasma was stored at −20 °C until blinded batch analyses were performed. Following baseline assessments, all participants were re-assessed at regular 12-month intervals over 5 years with emphasis on the development of clinical arthritis. An extra visit could be scheduled in case of arthritis development. Arthritis was defined based on the presence of at least one swollen joint on physical examination of 44 joints by a trained medical doctor (WB or LAS), who was aware of the status of ACPA and RF in the patient. In case of (uncertain) arthritis according to the first observer, the final judgment on presence or absence of arthritis was determined by a senior rheumatologist, who was unaware of the serostatus in the patient (DS). The study was approved by the Ethics Committee of Slotervaart Hospital and Reade, Amsterdam, The Netherlands, and written informed consent was obtained from all study participants.
Detection of biochemical markers
Baseline plasma was assessed for 14-3-3η levels using the quantitative 14-3-3η enzyme-linked immunosorbent assay (ELISA, Augurex Life Sciences Corp, Vancouver, Canada). Positivity for 14-3-3η was defined as ≥0.19 ng/ml based on the manufacturer's recommended cut-off, and at 2 times and 4 times this cut-off. The development, validation and calibration of the assay are detailed in a recent publication . ACPA was measured by an anti-CCP2 ELISA (Axis Shield, Dundee, UK) and immunoglobulin M rheumatoid factor (IgM-RF) by an in-house ELISA as described previously . The cut-off level for ACPA positivity was set at ≥5 arbitrary units/ml (AU/ml), according to the manufacturer's instructions. The cut-off level for IgM-RF positivity was set at ≥30 international units/ml (IU/ml).
The primary outcome chosen was arthritis, not rheumatoid arthritis to prevent circularity, as ACPA and RF are present in the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) RA criteria and 14-3-3η is not. Continuous, normally distributed data were presented as mean (standard deviation) and two-tailed t tests were used to establish whether significant differences existed between groups. Non-normally distributed data were presented as median (interquartile range) and analysed by Mann-Whitney U tests. Fisher's exact test was used to identify if positivity for any of the serologic variables investigated (ACPA, RF, and 14-3-3η) was significantly associated with arthritis development over 5 years. Spearman’s rank correlation coefficients expressed the relationship between 14-3-3η and the other serological markers ACPA and RF. Cox-proportional hazards survival analysis tested whether 14-3-3η can predict time to arthritis development.
Generalized linear models (GLM) assessed whether 14-3-3η was independently associated with the development of arthritis within 5 years. We used GLM with binomial outcome and log-link function, rather than standard logistic regression, because of the opportunity to describe relative risks (RR) instead of odds ratios, as this is a more proper association measure for describing results from prospective cohort studies. Since enrolment in the study implied that a subject was either ACPA or RF positive (or both) and no data was obtained in a group negative on both ACPA and RF, we jointly corrected for ACPA and RF status using a categorical variable distinguishing the three groups: (1) only RF positive, (2) only ACPA positive, (3) both RF and ACPA positive. Thereafter we created a variable containing 14-3-3η at different cut-off points (as mentioned above). In the GLM we first put in the categorical variable, after which we added 14-3-3η. The generated p values for 14-3-3η can then be interpreted as follows; if significance is found then the 14-3-3η test adds predictive value to the ACPA and RF test in the case one or both of these tests are positive. Note that this significance will imply that the additive value is the same for all three categories. To test whether predictive performance of 14-3-3η depends on the outcome of the ACPA and RF test, we also performed interaction analysis (by adding the interaction between the categorical variable and 14-3-3η in multivariable analyses). This interaction analysis will reveal whether 14-3-3η has more predictive capacity in one of the three groups. All analyses were performed with SPSS version 21 (IBM Corp, Armonk, NY, USA).
Baseline characteristics of study participants
(n = 144)
(n = 43)
(n = 101)
Time until end of follow-up (censoring or arthritis; months)
Tender joint count 53
Visual analogue scale pain
Use of NSAIDs (%)
Fulfilment of 2010 ACR/EULAR classification criteria for RA (%)
≥0.19 ng/ml (%)
≥0.40 ng/ml (%)
≥0.80 ng/ml (%)
Serological biomarkers 14-3-3η, ACPA and RF
Univariate and multivariable association of 14-3-3η, ACPA and RF with arthritis development
Univariate logistic regression
RR (95 % CI)
Multivariable logistic regression
RR (95 % CI)
RF and ACPA
Adding 14-3-3η to the above categorical variable
This study presents data that 14-3-3η is present in the pre-clinical phase of arthritis development, since there was a greater proportion of positivity of this marker at study entry together with higher expression in a pre-selected cohort of ACPA- and/or RF-positive arthralgia subjects who developed arthritis, compared with those who did not. This may be related to 14-3-3η’s ability to induce various inflammatory and joint degradative factors [6, 8].
Since 14-3-3η is an inflammatory mediator, the mechanism through which it is related to the development of arthritis may be different from that of autoantibodies such as ACPA and RF, whose levels tend to remain static or unchanged over the course of one’s disease. In this regard, a possible link of 14-3-3η with non-specific measures of inflammation, such as ESR and CRP, might be revealing. However, measurements of ESR and CRP at baseline were related neither to development of arthritis nor to 14-3-3η positivity in this cohort (data not shown). Another difference with autoantibodies might be the dynamic nature of serum 14-3-3η, which was supported by a study in first-degree relatives of indigenous North Americans with RA . This study population was not suitable for serial measurements since half of the patients developed arthritis shortly after inclusion and therefore missed a secondary measurement of 14-3-3η.
In this study, although 14-3-3η was associated with the development of arthritis, baseline 14-3-3η levels and positivity at three cut-off points did not add predictive value to the combination of ACPA and RF. This is most likely influenced by both the pre-selection method for this cohort, the ascertainment of arthritis, as well as the dynamic nature of 14-3-3η. In particular, the blinded confirmatory rheumatologist reviewed only those suspected of developing arthritis, and not all 144 subjects. Since the unblinded physician was making the initial assessment of arthritis development, if a bias did exist from their knowledge of ACPA and RF status, the blinded confirmatory physician would have reviewed a predominance of, say, ACPA-positive subjects and therefore identified more arthritis among those who were ACPA positive.
For clinical practice it would be very useful if 14-3-3η positivity could enhance the prediction of (rheumatoid) arthritis when combined with ACPA and RF. One such study that would enable such an analysis comes from a cohort of subjects based on clinically suspect arthralgia at risk for RA rather than on the basis of positive serology results. In addition to this, a prospective cohort recruited based on the presence of either of the three markers ACPA, RF or 14-3-3η may avoid any underestimation of the predictive capacity of 14-3-3η . Another suggestion would be to use a design which includes serial measurements of 14-3-3η. The OMERACT working group has recommended this design in guidelines to study soluble biomarkers, aimed at clinical validation of their predictive capacity, particularly for prognostic end points . The major limitation of baseline assessment alone has been repeatedly emphasized, particularly for responsive biomarkers such as 14-3-3η, which could vary considerably over the course of disease, and also with therapeutic intervention. This is highlighted in a recent publication describing clinical validation of IL-6 as a predictor of an event where longitudinal, but not baseline assessment alone, was predictive of structural damage in RA . The publication is about progression of radiographic damage, but it applies to other end points as well. However, a single assessment does conform with the clinical situation where a decision is often made to follow the patient or not.
Another limitation of this study was that the primary outcome could not be rheumatoid arthritis, as ACPA and RF are part of the 2010 ACR/EULAR classification criteria and 14-3-3η is not. We explored alternative outcomes such as subjects fulfilling the ACR/EULAR criteria regardless of serostatus and the development of erosions, but not enough subjects were positive for either to allow meaningful analysis.
In conclusion, we have shown that 14-3-3η is often present in arthralgia subjects positive for ACPA and/or RF prior to the development of arthritis, and was associated with the development of arthritis. In this cohort of subjects pre-selected for ACPA and/or RF positivity the added predictive value of 14-3-3η, both levels and different cut-off points, could not be established. Further studies are warranted to assess the combined utility of these three markers in predicting the development of arthritis.
anti-citrullinated protein antibodies
anti-carbamylated protein antibodies
peptidyl arginine deiminase type 4
Dirkjan van Schaardenburg
enzyme-linked immunosorbent assays
erythrocyte sedimentation rate
generalized linear model
immunoglobulin M rheumatoid factor
Lotte van de Stadt
non-steroidal anti-inflammatory drugs
receptor activator of nuclear factor-kB ligand
We thank Wouter Bos and Lotte van de Stadt for collecting the cohort patient data.
We also thank Charlie Goldsmith and Peter van de Ven for their expert statistical advice.
This project was funded by the Netherlands Organisation for Health Research and Development (61000010) and Augurex Life Sciences Corp.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- van de Stadt LA, Witte BI, Bos WH, van Schaardenburg D. A prediction rule for the development of arthritis in seropositive arthralgia patients. Ann Rheum Dis. 2013;72(12):1920–6.View ArticlePubMedGoogle Scholar
- Shi J, van de Stadt LA, Levarht EW, Huizinga TW, Toes RE, Trouw LA, et al. Anti-carbamylated protein antibodies are present in arthralgia patients and predict the development of rheumatoid arthritis. Arthritis Rheum. 2013;65(4):911–5.View ArticlePubMedGoogle Scholar
- Kolfenbach JR, Deane KD, Derber LA, O'Donnell CI, Gilliland WR, Edison JD, et al. Autoimmunity to peptidyl arginine deiminase type 4 precedes clinical onset of rheumatoid arthritis. Arthritis Rheum. 2010;62(9):2633–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Lubbers J, Brink M, van de Stadt LA, Vosslamber S, Wesseling JG, van Schaardenburg D, et al. The type I IFN signature as a biomarker of preclinical rheumatoid arthritis. Ann Rheum Dis. 2013;72(5):776–80.View ArticlePubMedGoogle Scholar
- Maksymowych WP, Landewe R, van der Heijde D. Serum 14-3-3η: a rheumatoid arthritis biomarker. Arthritis Rheum. 2011;63(10):S358.Google Scholar
- Maksymowych WP, Naides SJ, Bykerk V, Siminovitch KA, van Schaardenburg D, Boers M, et al. Serum 14-3-3eta is a novel marker that complements current serological measurements to enhance detection of patients with rheumatoid arthritis. J Rheumatol. 2014;41(11):2104–13.View ArticlePubMedGoogle Scholar
- van Heusden GP. 14-3-3 proteins: regulators of numerous eukaryotic proteins. IUBMB Life. 2005;57(9):623–9.View ArticlePubMedGoogle Scholar
- Maksymowych WP, van der Heijde D, Allaart CF, Landewe R, Boire G, Tak PP, et al. 14-3-3eta is a novel mediator associated with the pathogenesis of rheumatoid arthritis and joint damage. Arthritis Res Ther. 2014;16(2):R99.View ArticlePubMedPubMed CentralGoogle Scholar
- Bos WH, Dijkmans BA, Boers M, van de Stadt RJ, van Schaardenburg D. Effect of dexamethasone on autoantibody levels and arthritis development in patients with arthralgia: a randomised trial. Ann Rheum Dis. 2010;69(3):571–4.View ArticlePubMedGoogle Scholar
- Nielen MM, van Schaardenburg D, Reesink HW, van de Stadt RJ, van der Horst-Bruinsma IE, de Koning MH, et al. Specific autoantibodies precede the symptoms of rheumatoid arthritis: a study of serial measurements in blood donors. Arthritis Rheum. 2004;50(2):380–6.View ArticlePubMedGoogle Scholar
- Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT. Bingham 3rd CO, et al. 2010 rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Ann Rheum Dis. 2010;69(9):1580–8.View ArticlePubMedGoogle Scholar
- Hitchon CA, Smolik I, Meng X, Robinson D, El-Gabalawy HS. Serum 14-3-3eta are elevated in indigenous North Americans with rheumatoid arthritis and may predict imminent synovitis in their in at-risk first degree relatives. Arthritis Rheum. 2015;67(10). http://acrabstracts.org/abstract/serum-14-3-3eta-are-elevated-in-indigenous-north-americans-with-rheumatoid-arthritis-and-may-predict-imminent-synovitis-in-their-at-risk-first-degree-relatives/. Accessed 1th Dec 2015.
- Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD Statement. BMC Med. 2015;13:1.View ArticlePubMedPubMed CentralGoogle Scholar
- Maksymowych WP, Landewe R, Tak PP, Ritchlin CJ, Ostergaard M, Mease PJ, et al. Reappraisal of OMERACT 8 draft validation criteria for a soluble biomarker reflecting structural damage endpoints in rheumatoid arthritis, psoriatic arthritis, and spondyloarthritis: the OMERACT 9 v2 criteria. J Rheumatol. 2009;36(8):1785–91.View ArticlePubMedGoogle Scholar
- Baillet A, Gossec L, Paternotte S, Etcheto A, Combe B, Meyer O, et al. Evaluation of serum interleukin-6 level as a surrogate marker of synovial inflammation and as a factor of structural progression in early rheumatoid arthritis: results from a French national multicenter cohort. Arthritis Care Res. 2015;67(7):905–12.View ArticleGoogle Scholar