Authors reply: Clinical presentation, course, and prognosis of patients with mixed connective tissue disease

Dear Editor,

We thank Dechosal et al. for their comments on our article [1]. The authors raise questions about the nosology of mixed connective tissue diseases (MCTDs), systemic sclerosis (SSc), and its classification criteria [2] and, thus, the status of MCTD as a distinct connective tissue disease (CTD) or as a subgroup of SSc. They propose that, in a population of anti-U1 RNP patients, a lower point threshold (e.g., 6 points) based on non-immunologic parameters from the ACR/EULAR 2013 criteria might be sufficient to classify a patient as SSc. They recommend comparing the clinical course of MCTD patients with ≥6 points in the ACR/EULAR 2013 SSc classification to definite SSc patients with different antibody subtypes, suggesting a potential reclassification of MCTD as SSc if outcomes are comparable.

First, we would like to emphasize that anti-ribonucleoprotein U1 (anti-U1RNP) antibodies are not specific to SSc because they can be found in all differentiated CTD. Thus, their prevalence is high in systemic lupus erythematosus (SLE) (25%–35%) [3, 4] or myositis with nonspecific/associated antibodies (15%) [5] compared to 6%–8% in SSc (6%–8%) [6, 7], 4.7% in Sjögren's syndrome (SS) [8], and below 1% in rheumatoid arthritis patients [9]. Therefore, anti-U1RNP antibodies cannot and should not be considered SSc-specific. Furthermore, the authors contend that “because autoantibodies are mutually exclusive in SSc, the positivity for SSc-related antibody is largely unlikely—although not impossible, as suggested by the two patients with anti-centromere antibodies—in case of U1-RNP positivity.” This assertion does not hold true for anti-U1RNP in SSc. In 2022, our colleagues from the Royal Free Hospital reported on 119 SSc patients with anti-RNP antibodies. Among these, 43 patients (36.1%) had anti-U1RNP antibodies in association with another SSc-specific or related autoantibody, including anti-topoisomerase 1 in 21 patients (48.8%), anti-centromere in 8 (18.6%), anti-RNA polymerase III in 4 (9.3%), anti-U3RNP in 6 (14%), and anti-PmScl in 4 (9.3%). In this study, anti-U1RNP was the most common antibody associated with an SSc-specific antibody (anti-topoisomerase, anti-centromere, and anti-RNA polymerase III), accounting overall for 73% of the 63 patients with more than one autoantibody [10].

Second, as Dechosal et al. will recall, the ACR/EULAR 2013 criteria were established with the aim of assembling homogeneous patient groups for SSc studies. To achieve this, van den Hoogen et al. evaluated candidate items for classification using both data and expert clinical judgment. They subsequently reduced the number of these items and assigned weights. The classification system underwent repeated testing and adaptation using prospectively collected SSc cases and “scleroderma-like disorder” derivation cohorts, which included MCTD patients. The selection of a threshold of 9 points aimed to achieve a sensitivity and specificity of 0.97 and 0.88, respectively, despite only about two thirds of SSc patients having specific SSc antibodies. Notably, antinuclear antibodies (ANA) or anti-U1RNP were included neither in the classification criteria nor in the items tested. The system underwent testing for specificity and sensitivity in SSc cases (with only 51% of patients having SSc-specific antibodies) and scleroderma-like disorder controls, resulting in a sensitivity of 0.92 and a specificity of 0.91. The classification system was also assessed against expert opinion (n = 16 experts). Considering cases scoring ≥9 as SSc, those scoring <9 were deemed not to be SSc or controversial. Therefore, 9 was established as the threshold and not 8, 7, 6, or less, regardless of the antibody specificity. If this holds true for U1-RNP antibodies, it is important to note that it is also applicable to patients with anti-U3 RNP antibodies or anti-Pm-ScL antibodies. As a result, dedicated studies for these antibodies have utilized 9 as a threshold, not 6 [11, 12]. Thus, there is no reason to use a threshold of 6 in patients with U1-RNP as for any other specific or nonspecific SSc antibodies (e.g., anti-SSA). Additionally, the authors of the classification criteria explicitly state that “[these criteria] are not to be applied to patients who have a scleroderma-like disorder.” The use of a score ≥6 in patients without SSc-specific antibodies—especially in a “scleroderma-like disorder” such as MCTD—is entirely inappropriate. Overall, it is not valid to consider less than 9 points from the ACR/EULAR classification criteria to exclude patients with SSc.

Third, Dechosal et al. suggested comparing the clinical course of these “SSc-like” U1-RNP patients with patients with definite SSc with other antibody subtypes (from historical cohorts). We thank the authors for this suggestion, but we would argue that such a comparison should be made between SSc-U1-RNP⁺ and SSc-U1-RNP⁻ patients and those with MCTD in order to obtain a definitive answer. Along this line, Dechosal et al. proposed that “if the clinical presentation is similar and the only difference is the antibody subtype, should we not lump these disorders within the same nosological entity, with a common name—i.e. systemic sclerosis?” We believe that our study shows that this suggestion may be overly restrictive and invalid. Indeed, not every MCTD patient evolves into SSc, as highlighted in our work [1]. Thus, only 16% of MCTD patients evolve into SSc-RNP⁺, which is even lower than in the cohort of patients with very early diagnosis of SSc reported in Bellando-Randone et al.’s study. In that study, patients with Raynaud's phenomenon, ANA (which may include U1 RNP), and puffy fingers had a 79% chance of evolving into SSc over 5 years [13]. This percentage is much higher than in our study, indicating that these two populations are very different. Additionally, 10.6% and 4.5% of our patients evolve into SLE and SS, respectively, which is quite unusual in an SSc population, also highlighting differences in outcome.

To conclude, we express our gratitude to Dechosal et al. for their comments; however, we strongly disagree with the proposal to lump MCTD and SSc under the same name. Such a decision would be a simplistic mistake, whereas striving to enhance both the SSc and MCTD classifications could be a common objective.

The authors declare no conflicts of interest.