Recent eLetters

Dear Editor,

Radner and colleagues report a systematic literature search analysing the numerous articles and conference proceedings which examined the performance of the 2010 ACR/EULAR classification criteria for rheumatoid arthritis (RA) (1). The comprehensive evaluation identified if the 2010 criteria were correctly applied as suggested in the original publication, explored the performance of the criteria according to different methods to assess the criteria components and with use of different reference standards, and finally, directly compared the results obtained upon classification when using the 2010 or the 1987 criteria.
The overall sensitivity of the criteria was 82% and overall specificity was 61%, when applied to the intended target population. Eight studies and five meeting abstracts directly compared 1987 and 2010 criteria using different reference standards within different target populations. When excluding patients with other diagnosis, the 2010 ACR/ EULAR criteria demonstrated almost 21% higher sensitivity compared with 1987 ACR criteria, whereas specificity was 16% lower. Therefore, Radner et al. conclude that the 2010 criteria are more sensitive than the 1987 criteria at the cost of a slight decrement in specificity, which might increase the possibility that a few non-RA patients are classified as RA patients and, for example, entered into clinical trials (1). Another recent systematic literature review and a meta-analysis including 6 full papers and 4 abstracts reported identical performance stating that the new classification criteria have good sensitivity, lower specificity and an overall moderate diagnostic accuracy (2). Sakellariou et al. interpret the results as confirmation of the value of the criteria for classification but not for diagnosis (2).
Indeed, the first prospective study of consecutive patients seen in routine care of an outpatient clinic of a university rheumatology centre using the doctor's diagnosis as the gold standard, different from using medication start, found that the sensitivity and specificity of the 2010 criteria in classifying RA were 97% and 55%, respectively, compared with the 1987 RA criteria which were 93% and 76%, respectively (3). More specifically, 66.7% of systemic lupus erythematosus patients, 50% of osteoarthritis, 37.5% of psoriatic arthritis and 27.2% of others fulfilled the new criteria and could have been incorrectly classified as RA. Thus, testing the criteria for the first time in a broad spectrum of rheumatological diseases seen in routine rheumatology care confirm the concern that the poor specificity may lead to over- and misdiagnosis of patients with RA, leading to inappropriate medication use (4, 5). Vunkemann and van de Laar reviewing the performance of the criteria very recently concluded:"Especially, when the classification criteria are used as diagnostic criteria this carries the risk of overtreatment. It remains to be determined whether or not the new criteria when used to diagnose and treat patients provide an acceptable balance between efficacy and safety and in these days also of major importance, cost-effectiveness"(6). In keeping with overclassification, several evaluation studies revealed that patients classified as RA according to the 2010 criteria are more likely to achieve drug-free remission than those who meet the 1987 criteria (7, 8, 9). Moreover, an ongoing prospective study of early arthritis demonstrated that 51% of 2010 criteria "non-RA" patients compared to 86% of patients with RA were treated with methotrexat (MTX) in the first year, suggesting that the rheumatologists in their clinic had a more aggressive approach to early arthritis during the same period than the rheumatologists treating the cohorts that were used to derive the criteria (10). Obviously, MTX is neither a "gold standard" for RA nor a static feature, as rheumatologists have a tendency to treat earlier and more aggressively. In addition, MTX and DMARD medications are also prescribed for other chronic inflammatory diseases, such as psoriatic arthritis and peripheral spondyloarthritis. Most importantly, classification criteria serve as a tool to arrive at homogeneous groups of patients with comparable features to make data obtained by different researchers at different places comparable why they should have a high specificity (preferably close to 100%), in order to prevent misclassification and inclusion of patients who do not have the disease (11). Also a balance of sensitivity and specificity is required in validation of criteria sets for use in clinical trials and epidemiologic studies (12).
Obviously, these requirements are hardly met if in the appropriate intended population the area under curve for receiver operating characteristic curves are rather weak between 0.72 and 0.78 indicating misclassification in 22% to 28% and limited accuracy to separate RA from other early arthritides (10, 13, 14). Overall, the question arises whether the loss of specificity is a price worth paying to use the criteria for classification in clinical trials.

In conclusion, the good news of better sensitivity is considerably limited by the loss of specificity and related risk of overtreatment of patients with self-limiting disease as RA with potentially toxic agents, even considering that poor recognition and inadequate intervention in the earliest phases of inflammatory arthritis may occur more often. The improvement of the accuracy of diagnosis and classification of early and very early RA remain a continuous challenge. Recently we reviewed proposals and suggestions how to overcome the problems and limitations of the 2010 criteria by clinical practice and future research (4):
1) the rheumatologist as the expert strikes a balance between possible or probable RA, depending on the level of confidence (15),
2) the rheumatologist uses a diagnostic certainty scale at baseline (0 to 100 visual analog scale) (16),
3) use of the prediction rule developed by van der Helm-van Mil et al. (17, 18) to estimate the chance of progression to RA in individual patients presenting with undifferentiated arthritis,
4) use of imaging techniques (sonography, MRI) to identify erosions earlier (19, 20, 21),
5) testing the discriminative value of HLA-B27 and diagnostic programs for reactive arthritis (22, 23),
6) testing likelihood ratios for diagnostic decision-making based on the Bayesian approach (24),
7) use of automated, multiplex biomarker assay testing for autoantibodies, cytokines, and bone-turnover products (25). Finally, future research should focus on validating the 2010 criteria in terms of important long-term outcomes in RA such as radiological damage, disability and mortality (5). In this regard, most recently a study of early arthritis patients evaluated the ability of the 2010 ACR/ EULAR and the 1987 ACR classification criteria to predict radiographic progression after 10 years of follow-up (27). The data suggests that both classification criteria predict poorly erosive disease in patients with early RA. The discriminative power of the 2010 criteria is only slightly better than that of the 1987 criteria with area under the curve of 0.72 and 0.65, respectively. Another most recent study reported that the 2010 criteria appear to be as efficient as the 1987 criteria in identifying increased risk of mortality but the 2010 criteria identify a greater proportion of at-risk patients soon after their first presentation to health care with a hazard ratio of 1.35 compared to 1.24 (28).

References

1 Radner H, Radner H, Neogi T, Smolen JS, et al. Ann Rheum Dis Published Online First: [April 16, 2013] doi:10.1136/ annrheumdis-2013-203284

2 Sakellariou G, Scire` CA, Zambon A et al. Performance of the 2010 Classification Criteria for Rheumatoid Arthritis: A Systematic Literature Review and a Meta-Analysis. PLoS ONE 2013;8:e56528.

3 Kennish L, Labitigan M, Budoff S et al. Utility of the new rheumatoid arthritis 2010 ACR/EULAR classification criteria in routine clinical care. BMJ Open 2012;2:e001117.

4 Zeidler H. The need to better classify and diagnose early and very early rheumatoid arthritis. J Rheumatol 2012;39:212-7.

5 Humphreys JH, Symmons DP. Postpublication validation of the 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for rheumatoid arthritis: where do we stand? Curr Opin Rheumatol 2013;25:157-63.

6 Vonkeman HE, van de Laar MA. The new European League Against Rheumatism/American College of Rheumatology diagnostic criteria for rheumatoid arthritis: how are they performing? Curr Opin Rheumatol 2013;25:354-9.

7 Cader MZ, Filer A, Hazlehurst J, et al. Performance of the 2010 ACR/EULAR criteria for rheumatoid arthritis: comparison with 1987 ACR criteria in a very early synovitis cohort. Ann Rheum Dis 2011;70:949-55.

8 De Hair MJ, Lehmann KA, van de Sande MG, et al. The clinical picture of rheumatoid arthritis according to the 2010 American College of Rheumatology/European League Against Rheumatism criteria: is this still the same disease? Arthritis Rheum 2012;64:389-93.

9 Krabben A, Huizinga TW, van der Helm-van Mil AH. Undifferentiated arthritis characteristics and outcomes when applying the 2010 and 1987 criteria for rheumatoid arthritis. Ann Rheum Dis 2012;71:238-241.

10 Britsemmer K, Ursum J, Gerritsen M, van Tuyl LH et al. Validation of the 2010 ACR/EULAR classification criteria for rheumatoid arthritis: slight improvement over the 1987 ACR criteria. Ann Rheum Dis 2011;70:1468- 70.

11 van der Helm-van Mil AH, Huizinga TWJ. The 2010 ACR/EULAR criteria for rheumatoid arthritis: do they affect the classification or diagnosis of rheumatoid arthritis? Ann Rheum Dis 2012;71:1596-98.

12 Johnson SR, Goek ON, Singh-Grewal D et al. Classification criteria in rheumatic diseases: A review of methodologic properties. Arthritis Rheum 2007;57:1119-33.

13 van der Linden MP, Knevel R, Huizinga TW et al. Classification of rheumatoid arthritis: Comparison of the 1987 American College of Rheumatology criteria and the 2010 American College of Rheumatology/European League Against Rheumatism criteria. Arthritis Rheum 2011;63:37-42.

14 Varache S, Cornec D, Morvan J et al. Diagnostic accuracy of ACR/EULAR 2010 criteria for rheumatoid arthritis in a 2-year cohort. J Rheumatol 2011;38:1250-7.

15 Benhamou M, Rincheval N, Roy C et al. The gap between practice and guidelines in the choice of first-line disease modifying antirheumatic drug in early rheumatoid arthritis: Results from the ESPOIR cohort. J Rheumatol 2009;36:934-42.

16 Morvan J, Berthelot JM, Devauchelle-Pensec V et al. Changes over time in the diagnosis of rheumatoid arthritis in a 10-year cohort. J Rheumatol 2009;36:2428-34

17 van der Helm-van Mil AH, Detert J, le Cessie S et al. Validation of a prediction rule for disease outcome in patients with recent-onset undifferentiated arthritis: Moving toward individualized treatment decision-making. Arthritis Rheum 2008;58:2241-7.

18 Huizinga TW, van der Helm-van Mil A. Quantitative approach to early rheumatoid arthritis. Bull NYU Hosp Joint Dis 2011;69:116-21.

19 Freeston JE, Wakefield RJ, Conaghan PG et al. A diagnostic algorithm for persistence of very early inflammatory arthritis: The utility of power Doppler ultrasound when added to conventional assessment tools. Ann Rheum Dis 2010;69:417-9.

20 Boutry N, do Carmo CC, Flipo RM et al. Early rheumatoid arthritis and its differentiation from other joint abnormalities. Eur J Radiol 2009;71:217-24.

21 Filer A, de Pablo P, Allen G et al. Utility of ultrasound joint counts in the prediction of rheumatoid arthritis in patients with very early synovitis. Ann Rheum Dis 2011;70:500-7.

22 Hülsemann JL, Zeidler H. Diagnostic evaluation of classification criteria for rheumatoid arthritis and reactive arthritis in an early synovitis outpatient clinic. Ann Rheum Dis 1999;58:278-80.

23 Soderlin MK, Borjesson O, Kautiainen H et al. Annual incidence of inflammatory joint diseases in a population based study in southern Sweden. Ann Rheum Dis 2002;61:911-5.

24 Corrao S, Calvo L, Licata G. The new criteria for classification of rheumatoid arthritis: What we need to know for clinical practice. Eur J Intern Med 2011;22:217-9.

25 Chandra PE, Sokolove J, Hipp BG et al. Novel multiplex technology for diagnostic characterization of rheumatoid arthritis. Arthritis Res Ther 2011;13:R102.

26 M?kinen H, Kaarela K, Huhtala H et al. Do the 2010 ACR/EULAR or ACR 1987 classification criteria predict erosive disease in early arthritis? Ann Rheum Dis 2013;72:745-7.

27 Humphreys JH, Verstappen SM, Hyrich KL et al. 2010 ACR/EULAR classification criteria for rheumatoid arthritis predict increased mortality in patients with early arthritis: results from the Norfolk Arthritis Register. Rheumatology (Oxford) 2013 Mar 5. [Epub ahead of print] doi:10.1093/rheumatology/ket113.

Conflict of Interest:

None declared

Dear Editor,

We read with interest the manuscript by Aggarwal et al. entitled “Patients with non-Jo-1 anti-RNA-synthetase autoantibodies have worse survival than Jo-1 positive patients” 1. This large cohort study provides important information on outcomes for patients with antisynthetase syndrome (ASS), based on the specificity of the anti-RNA-synthetase autoantibody subtypes. Interestingly, the authors decided to include patients with anti-EJ, OJ and KS -tRNA-synthetase autoantibodies, something which had not yet been done in the previous studies2,3, due to the rarity of these autoantibodies. The conclusion by Aggarwal et al. confirmed our previous data showing a worse prognosis for non-Jo-1 patients as compared with Jo-1 patients2. However, this study prompts questions on the two following points:

1. The authors showed that a longer delay in diagnosing non-Jo1 patients was a major predictor of poor survival. For this, they used a multivariate Cox model analysis, adjusted for diagnosis delay, as well as for the following parameters: gender, ethnicity, age at initial and final diagnosis. However, the model was not tested with any of the variables that have clearly been shown to correlate with either poor prognosis (i.e. interstitial lung disease1 and pulmonary hypertension1,4) or with better survival (i.e. the presence of a myositis at ASS diagnosis2). Although the main objective is of course to decrease the diagnosis delay in all patients, it would be quite valuable to know whether this delay is an independent predictor of survival after adjusting for these variables.

2. The results shown in Table 3 and in Figure 1 are difficult to interpret since no information is provided about the censored data (overall median patient follow-up <3 years vs survival evaluations >5 years). In Table 3, the absolute number of patients evaluated at 5 and 10 years is not given, which leads to some confusion: do the percentages of patients correspond to the ratio of living patients to total number of patients at diagnosis, or to the probability of survival, as estimated with the Kaplan-Meyer method? Similarly, there are no marks to help identify the censored data in the Kaplan-Meyer curve of Figure 1, making it difficult to identify the number of censored patients during the follow-up period.

We thank the authors for addressing these issues and for providing additional data that will be useful for understanding the factors associated with survival of patients with ASS.

References

1. Aggarwal R, Cassidy E, Fertig N, et al. Patients with non-Jo-1 anti-tRNA-synthetase autoantibodies have worse survival than Jo-1 positive patients. Ann Rheum Dis 2013 Feb 26. [Epub ahead of print] PubMed PMID: 23422076.

2. Hervier B, Devilliers H, Stanciu R, et al. Hierarchical cluster and survival analyses of antisynthetase syndrome: phenotype and outcome are correlated with anti-tRNA synthetase antibody specificity. Autoimmun Rev 2012;12:210-7 .

3. Marie I, Josse S, Decaux O, et al. Comparison of long-term outcome between anti-Jo1- and anti-PL7/PL12 positive patients with antisynthetase syndrome. Autoimmun Rev 2012;11:739-45.

4. Hervier B, Meyer A, Dieval C, et al. Pulmonary hypertension in antisynthetase syndrome: prevalence, etiology and survival. Eur Respir J 2013 Feb 8. [Epub ahead of print] PubMed PMID: 23397301.

Conflict of Interest:

None declared

Dear Editor,

The severity of rheumatoid arthritis (RA) is highly variable between patients and currently known risk factors explain only part of this variance.(1) Much research is dedicated to identify additional new risk factors. Such factors may shed light on the processes underlying progression of RA and many risk factors together may enable risk stratification and individualized treatment of RA.

With interest we read the study by Moller et al, showing that RA- patients with anaemia have more severe radiological progression.(2) Although anaemia in RA is generally considered to be a consequence of chronic inflammation, this recent study -based on RA patients included in the Swiss SCQM-database- observed that the association between anaemia and joint damage was independent of the association between disease activity (measured with the DAS28ESR and cDAI) and joint damage. This led to the presumptions that anaemia in RA captures disease processes that are unmeasured by established disease activity markers (e.g. subclinical inflammation) and that evaluation of the haemoglobin level may help to identify patients with rapid radiological progression.

Since in science replication of findings is relevant to ascertain the validity, we evaluated the association between anaemia at first presentation and disease severity over 7 years in 676 early RA patients included in the Leiden Early Arthritis Clinic.(1) Two outcome measures were studied. First, radiological progression; 3502 sets of hand and feet radiographs were made with yearly intervals and scored according to the Sharp-van-der-Heijde method by one reader (ICC 0.91).(3) Second, disease persistency was assessed by evaluating its counterpart, achieving DMARD- free sustained remission.(4) Analyses were done using multivariate normal regression analysis and cox regression.(5) All analyses were adjusted for age, gender and treatment strategy.(1) The haemoglobin level was determined at first presentation. The WHO definitions for the presence and severity of anaemia were used.(6)

24.1% of the RA-patients had anaemia at first presentation. These patients were older and had a higher swollen joint count (SJC), ESR and CRP (Table 1). Patients with anaemia had more severe joint damage progression (beta 1.03, p 0.012, indicating a 1.03 higher rate of joint destruction per year, which equals 1.037= 23% more joint damage after 7 years). Similar to M?ller et al, we also adjusted for ESR, SJC and RF; this did not affect the association (beta 1.03, p 0.040) (Figure 1A). When evaluating the three haemoglobin categories a "dose-dependent" effect was observed (beta 1.03, p 0.002). Also this association was significant after adjusting for ESR, SJC and RF (beta 1.02, p 0.032) (Figure 1B). All analyses remained significant when the CRP was included as covariate instead of the ESR (data not shown).(7) Patients with anaemia tended to achieve DMARD-free remission less often than patients without anaemia (HR 0.57, 95% CI 0.34-0.95, p 0.031), also after adjusting for ESR, SJC and RF (HR 0.59, 95% CI 0.34-1.02, p 0.056) (Figure 1C).

Analysis on this population-based inception cohort revealed that within RA anaemia is independently associated with radiographic progression. As clinical inflammation in RA may predominantly affect the feet,(8) we evaluated a 66-SJC that -in contrast tot the DAS28ESR and cDAI used by Moller et al- included the MTP-joints. Nonetheless, also we observed that anaemia independently predicted disease severity. This may indicate that anaemia indeed reflects subclinical inflammation. Future studies are required to unravel this association.

Table 1 Characteristics of RA patients per haemoglobin category. Table available on the monthly Epage.

Figure 1. Joint destruction (Sharp-van der Heijde scores) and DMARD- free sustained remission curves over 7 years follow-up in RA patients, categorized according to the presence (A, C) or severity (B) of anaemia. Figure available on the monthly Epage.

References

(1) de Rooy DP, van der Linden MP, Knevel R, et al. Predicting arthritis outcomes--what can be learned from the Leiden Early Arthritis Clinic? Rheumatology (Oxford) 2011;50:93-100.

(2) Möller B, Scherer A, Forger F, et al. Anaemia may add information to standardised disease activity assessment to predict radiographic damage in rheumatoid arthritis: a prospective cohort study. Ann Rheum Dis Published Online First: 16 March 2013. doi:10.1136/annrheumdis-2012-202709

(3) Knevel R, Krabben A, Brouwer E, et al. Genetic variants in IL15 associate with progression of joint destruction in rheumatoid arthritis: a multicohort study. Ann Rheum Dis 2012;71:1651-7.

(4) van der Woude D, Young A, Jayakumar K, et al. Prevalence of and predictive factors for sustained disease-modifying antirheumatic drug-free remission in rheumatoid arthritis: results from two large early arthritis cohorts. Arthritis Rheum 2009;60:2262-71.

(5) Knevel R, Tsonaka R, Cessie SL, et al. Comparison of methodologies for analysing the progression of joint destruction in rheumatoid arthritis. Scand J Rheumatol Published Online First: 20 February 2013. doi:10.3109/03009742.2012.7286182013

(6) WHO. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. Vitamin and Mineral Nutrition Information System. Geneva, World Health Organization, 2011 (WHO/NMH/NHD/MNM/11.1) (http://www/who.int/vmnis/indicators/haemoglobin.pdf, accessed [March 2013]).

(7) Wells G, Becker JC, Teng J, et al. Validation of the 28-joint Disease Activity Score (DAS28) and European League Against Rheumatism response criteria based on C-reactive protein against disease progression in patients with rheumatoid arthritis, and comparison with the DAS28 based on erythrocyte sedimentation rate. Ann Rheum Dis 2009;68:954-60.

(8) Knevel R, Kwok KY, de Rooy DP, et al. Evaluating joint destruction in rheumatoid arthritis: is it necessary to radiograph both hands and feet? Ann Rheum Dis 2013;72:345-9.

Conflict of Interest:

None declared

Dear Editor,

We thank Clockaerts and colleagues for their thoughtful commentary on our paper and will address a number of their comments. They expressed concern regarding lack of verification of our data related to statin use, duration of use, and dosage, and indicated these data were based on self- report. While we acknowledged in our paper[1] that we did not have dosage data, the verification was that participants brought their medications with them during yearly follow-up visits and we were able to confirm that statins were prescribed for these patients. This approach was not as valid as a pill count but, in our view, is more rigorous than self-report.

Clockaerts and colleagues accounted for duration of statin use by examining effects for those with 1-119 days, 120-364 days and 365 days or greater of statin use at 50% or more of the recommended daily dosage. Our approach was to use the yearly follow-up visit prescription data indicating 0, 1, 2, 3 or 4 years of statin use to examine the effects on changes in OA structural progression, pain and function. While we did not have dosage data, we believe that our approach reasonably accounted for differing durations of statin usage in our sample. In the end, an assumption of compliance was made in both studies[1,2] that medications were taken as prescribed.

In our opinion, neither our study[1] nor the work of Clockaerts et al[2] provides convincing evidence of a subgroup(s) who might be most responsive to OA-related benefits, if any, of statins. Clockaerts and colleagues suggested it is more plausible that statins could be useful in the early as compared to later stages of OA. They went on to recommend studies of statin use in persons with Kellgren and Lawrence (KL) scores of 0 at baseline. While we did not conduct subgroup analyses, we had 684 persons in our sample who had a baseline KL grade of 0 on one knee and 1,286 who had a baseline grade of 0 or 1 in at least 1 knee. If an association between statin use or duration of use and OA disease progression existed for patients with either no or very mild disease of at least one knee, the random intercept analyses in the structural equation models would have detected these relationships. No associations were found. When describing their analyses for persons with and without KL scores of 0, Clockaerts and colleagues reported that "when we used those separate definitions of incidence and progression, we found similar results."p645 It is not clear to us why Clockaerts and colleagues suggest statins may be more effective DMOADS for persons with either no or mild OA.

Clockaerts et al correctly indicated that we did not report the occurrence of muscle pain in our sample. They also suggested that muscle pain is an important side effect of statins and should be taken into account when considering pain scores for OA; the implication would be that, if statins caused muscle pain, then statin-related muscle pain might interfere with knee OA pain and function measures. On further reflection, this appears unlikely. First, our pain measure asked the participant to "rate the pain that you've had in your right/left knee during the past 7 days," requesting a focal pain report, not a more generalized pain assessment. Second, Clockaerts et al cite the work of Armitage who summarized potential adverse musculoskeletal effects of statin use including myopathy (defined as muscle symptoms accompanied by muscle enzyme elevation 10-fold-or-higher above the upper limit of normal) or myalgia (muscle symptoms with enzyme elevation up to 10-fold) [3]. Armitage indicated myopathy is a rare event particularly for standard doses, about 11 per 100,000 person-years of follow-up. Given that we had approximately 3,000 person-years of statin use in our study, we suspect myopathy influences were minimal (less than one represented in our panel). Regarding myalgia risk, Armitage reviewed several randomized trials that failed to support the assertion that statins cause myalgia or muscle cramps. For example, in one trial, the frequency of unexplained muscle pain or weakness was 32.9% in simvastatin-treated subjects and 33.2% in placebo-treated subjects[4]. Symptoms of myalgia are frequent among patients in the age ranges commonly needing cholesterol-lowering therapy, but the causal relationship of statins to myalgia is unsupported.

Thus, given the focused pain questions, the very low anticipated frequency of true myopathy and absence of evidence that statins are causally associated with myalgia, we suspect that any misreporting of muscle symptoms as knee OA pain and reduced function would likely have occurred in both statin-treated and statin-untreated groups about equally.

We agree that additional study is needed and we fully endorse the recommendation by Clockaerts and colleagues that future studies of the potentially protective effects of statins on OA incidence and progression should be targeted toward sub-groups who are most likely to benefit. Unfortunately, our study does not assist in identifying a subgroup whose osteoarthritis may be more likely to respond favorably to statins.

References

1. Riddle DL, Moxley G, Dumenci L. Associations between statin use and changes in pain, function and structural progression: a longitudinal study of persons with knee osteoarthritis. Ann Rheum Dis. 2013;72:196-203.

2. Clockaerts S, Van Osch GJ, Bastiaansen-Jenniskens YM, et al. Statin use is associated with reduced incidence and progression of knee osteoarthritis in the Rotterdam study. Ann Rheum Dis 2012;71:642-647.

3. Armitage J. The safety of statins in clinical practice. Lancet 2007;370:1781-1790.

4. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet 2002;360:7-22.

Conflict of Interest:

We have no competing interests. This "response" is submitted in response to the comments by Clockaerts and colleagues who were responding to our paper published in ARD (Riddle et al. Associations Between Statin Use and Changes in Pain, Function and Structural Progression: A Longitudinal Study of Persons with Knee Osteoarthritis. ARD. 2013 Feb;72(2):196-203.

Dear Editor,

Vincent et al present a thorough review of available clinical research on antidrug antibodies to tumour necrosis factor blocking agents (1). We agree to a large extent with their interpretation of the data with regard to pathofysiology, assay characteristics and variation in drug and antidrug antibody serum levels. However, the conclusion that measurement of (anti)drug levels should therefore be used for clinical decision making in non-responding biological patients with inflammatory diseases to save costs, prevent adverse events and improve disease activity, including their proposed algorithm, seems flawed and is thusfar insufficiently supported by evidence. In our comment we will focus on RA, but for other inflammatory disease the same comments can be made.

Firstly, we would argue that the most promising application for therapeutic drug monitoring (TDM) to save costs or adverse events is not to predict response to the next treatment option in non responding patients, but rather to predict successful dose reduction and stopping in patients who are doing well. All treatment alternatives in non responding patients employ either another biologic or a higher dose of the same biologic, so even optimal channelling of patients can only lead to better disease control, not to saved costs or prevented adverse events. In patients doing well however, it could be possible to use TDM to optimise dose reduction or stopping compared to a trial and error strategy. This way, unnecessary flares and unneeded drug exposition (cost and adverse events) could be prevented.

Secondly, when TDM is done in non responding patients and a low serum drug level is found it is not sensible, at least in RA, to increase the dose as suggested. The better choice would be to change to another biologic, either TNF blocker or with another mechanism of action. Increasing the dose has a much lower chance of response in rheumatic diseases (2,3) and is associated with more adverse events and costs (4) than switching to another biological, and is therefore probably not cost effective. In addition to these arguments, there is also a lack of specific empirical data to support the proposed decision tree. The theoretical framework for TDM indicates that any test that is used for TDM in clinical practice should meet the following requirements: 1/ uncertainty exist on an important clinical outcome, 2/ a reliable and valid test is available that is strongly associated with this outcome and gives additional information about this outcome above simple test, 3/ the use of the test has treatment consequences and 4/ the use of the test is cost effective (adapted from Aarnoutse et al, 5). The proposed algorithm of Vincent et al does not fulfill these criteria. Firstly, increasing the dose in arm 1 is arguably not safe and (cost)effective. Also, arm 2 and 3 converge to the same treatment option, negating the possible value of testing. For arm 4, switching to another class of biological, a clearly higher post test chance of response has not been consistently demonstrated.

Finally and most importantly, the possible advantages of TDM in this context - better disease control - has never been demonstrated compared to usual tight control care.
The appropriate clinical study design to assess the value of TDM is a prediction cohort study in patients with either active disease and starting or switching a biological, or patients with low disease and withdrawing medication. After the clinical outcome has been measured prospectively, the sensitivity and specificity of baseline (anti)drug levels for (non)response measured using a validated test should be estimated using ROC analyses. Thereafter, prediction modelling including all other known predictors (eg clinical, CRP) should be done to determine the additive value of TDM above regular clinical tight control. A clear change from pretest to posttest chances should be demonstrated, expressed preferably in numbers needed to diagnose, and cost effectiveness measures should be provided. Finally, confirmation of the cost effectiveness in a so called diagnostic study - a randomised trial comparing test based treatment with usual care - would be considered to be the golden standard in proving the value of any diagnostic or prognostic test, including TDM. The current review describes 76 studies done in the last 13 years of which 36 in RA assessing (anti)drug levels for the five anti-TNF agents. These studies include basic labstudies, cross-sectional studies and longitudinal non interventional studies. Unfortunately, however, due to the specific design limitations, not one of these studies was able to provide test characteristics for an important clinical outcome, including sensitivity and specificity, and pretest and posttest chances, let alone cost effectiveness analyses. Of note, the cost effectiveness analysis of TDM in adalimumab done in RA patients by Krieckaert et al (6) is a Markov modeling study showing that TDM using adalimumab (anti)drug levels could be cost-effective, based on the presumption that (anti)adalimumab serum levels are predictive for successful dose reduction, and this has indeed yet to be established. Also, modeled TDM guided dose reduction was compared to no change in adalimumab treatment instead of the more valid comparison with clinically guided doses reduction.

We are currently aware of three studies in rheumatic diseases specifically designed to assess the value of TDM, two of which have not been published yet (7-9). These studies assessed respectively the predictive value of infliximab (anti) drug levels in RA patients to predict response after initiation, prediction of successful dose reduction, and in ankylosing spondylitis patients prediction of response after dose increase. Interestingly, all three studies failed to demonstrate any relevant contribution of TDM for clinical decision making in the respective contexts. Other studies are currently underway to further explore this for other biologicals in other diseases (e.g. DRESS study in RA, TAXIT study in IBD).
In conclusion, TDM has no proven value yet in biological treatment, and should not be advocated at this moment. The most promising context seems TDM guided dose reduction in patients doing well. We strongly support research in this field, as it might enable us to treat our patients better, resulting in optimal disease control, better safety, and lower costs. However, a proposed TDM algorithm should be logically sound and empirically tested with scrutiny. We can therefore wholeheartedly join Vincent et al in their plea for significant research investment in this topic, both for test development as well as execution of appropriate clinical studies.

References

1. Vincent FB, Morand EF, Murphy K, et al. Antidrug antibodies (ADAb) to tumour necrosis factor (TNF)-specific neutralizing agents in chronic inflammatory diseases: a real issue, a clinical perspective. Ann Rheum Dis 2013;72:165-78.

2. Pavelka K, Jarosov K, Such D, et al. Increasing the infliximab dose in rheumatoid arthritis patients: a randomised, double blind study failed to confirm its efficacy. Ann Rheum Dis 2009;68:1285-9.

3. van den Bemt BJF, den Broeder AA, Snijders GF, et al. Sustained effect after lowering high dose infliximab in patients with rheumatoid arthritis: a prospective dose titration study. Ann Rheum Dis 2008;67:1697-701.

4. Bongartz T, Sutton AJ, Sweeting MJ, et al. Anti-TNF antibody therapy in rheumatoid arthritis and the risk of serious infections and malignancies: systematic review and meta-analysis of rare harmful effects in randomized controlled trials. JAMA 2006;295:2275-85

5. Aarnoutse RE, Schapiro JM, Boucher CA, et al. Therapeutic drug monitoring: an aid to optimising response to antiretroviral drugs? Drugs 2003;63:741-53.

6. Krieckaert C, Nair SC, Nurmohamed MT, et al. Evaluating the cost-effectiveness of personalized treatment with adalimumab using serum drug level and anti-adalimumab antibodies in rheumatoid arthritis patients. Ann Rheum Dis 2012;71(suppl 3):104.

7. Inman RD, Davis JC Jr, Heijde D vd, et al. Efficacy and safety of golimumab in patients with ankylosing spondylitis: results of a randomized, double-blind, placebo-controlled, phase III trial. Arthritis Rheum 2008;58:3402-12.

8. van der Maas A, van den Bemt BJF, van den Hoogen FHJ, et al. Baseline (anti-) infliximab serum trough levels do not predict successful down-titration or cessation of infliximab in Rheumatoid Arthritis patients with long term low disease activity. Submitted 2013.

9. van den Bemt BJF, den Broeder AA, Wolbink GJ, et al. The combination of disease activity and infliximab serum trough levels for early prediction of (non-) response to infliximab-treatment in patients with rheumatoid arthritis. Submitted 2013.

Conflict of Interest:

None declared