The diagnosis and prognosis of early arthritis: Comment on the editorial by Scott

To the Editor:

We read with interest the recent editorial by Scott regarding the diagnosis and prognosis of early arthritis. The editorial contributes to the discussion about how scientists should deal with the diagnostic and prognostic aspects of early arthritis in the future. We are concerned, however, about the discussion of certain important issues and the possible confusion that may arise.

In Table 1, Scott compares our diagnostic study, performed in unselected patients with early arthritis, with 2 prognostic studies in patients with rheumatoid arthritis (RA). We believe that this comparison is not warranted and may be misleading. The fact that the diagnostic criteria set used in our study was developed in the form of a clinical prediction model for persistent/erosive arthritis may have confused Scott. Yet, as long as the etiology of RA remains unknown, the 2 “gnoses” needed to direct management strategies in early arthritis—diagnosis (Gk, knowledge of discrimination) and prognosis (Gk, knowledge of outcome)—are fused. In other words, for both the patient and the clinician confronted with recent-onset arthritis, the relevant issue is discrimination (at an early stage) between different prognoses, not the question of whether the patient will fulfill a set of classification criteria.

Scott also states that our report suggests that the anti–cyclic citrullinated peptide antibody (anti-CCP) enzyme-linked immunosorbent assay (ELISA) has limited value as a diagnostic criterion for early RA. The limited diagnostic value of the anti-CCP ELISA is, however, a misinterpretation by Scott rather than a result of the study, which demonstrated that the anti-CCP ELISA independently and significantly contributes to the performance of the predictive model. The odds ratios of the anti-CCP test are higher than those of the other 6 criteria of the model. In his conclusion, Scott states that special tests such as the anti-CCP ELISA should not be part of the prognostic criteria, because they are not widely available. An attitude like this would preclude any further developments in rheumatology. When proper diagnostic research and validation in different patient populations prove that a new test is of additional value in the diagnosis of RA, it should be made part of predictive models for RA. In the same way, the rheumatoid factor has become a generally accepted tool in RA diagnosis.

Scott states that the cases in our study are highly selected, which is untrue. The patients with early arthritis were referred directly by local general practitioners, which makes our study population based. He also suggests that it is not necessary to introduce new diagnostic criteria for RA, but that it is sensible to produce new criteria to define the prognosis of early arthritis. This appears to be a contradiction, because differentiating patients with early arthritis by means of prognostic criteria is diagnosis. Scott probably is referring to the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 classification criteria for RA, which are generally used as diagnostic criteria for RA and as the gold standard.

Contrary to Scott, we think that the time has come to discontinue use of these criteria, for both the diagnosis and classification of RA. The ACR 1987 classification criteria for RA were formed by consensus and lack a clear etiologic, pathophysiologic, or epidemiologic basis. The criteria are unsuitable both as diagnostic criteria for early arthritis and as a gold standard for RA. In our study, 15% of the patients with persistent or persistent erosive arthritis could not be classified according to any of the international classification criteria during 2 years of followup. These cases were classified in the study as undifferentiated arthritis. What should the clinician tell such unclassifiable patients? How should he or she treat them? So far, therapeutic trials have been performed only with patients fulfilling international classification criteria. This has resulted in a remarkable situation: for patients with persistent/erosive arthritis who do not fulfill any classification criteria (undifferentiated arthritis), who represent ∼30% of patients with early arthritis, there is no evidence at all from therapeutic studies. Some of these patients probably have a disease that is not immunologically different from RA according to the classification criteria. Nevertheless, these patients are often excluded from etiologic or pathophysiologic studies on RA.

We think that the time has come to develop and validate prediction models for persistent/erosive arthritis. International collaboration is necessary to develop an efficient model that is generally applicable. Next, clinical therapeutic trials are needed in populations of patients with recent-onset arthritis in order to solve the therapeutic dilemma with which clinicians are confronted in daily practice. For these trials, a predictive model that discriminates between patients with different risks for persistent/erosive arthritis is indispensable. Also, for researchers who study the etiopathogenesis of rheumatic diseases, such a model is important because it gives them the opportunity to study immunologic phenomena in an early phase of the disease.