The diagnosis of idiopathic interstitial pneumonia requires the exclusion of other known causes of interstitial lung disease (ILD), among them connective tissue diseases (CTD). The most conservative recommendations suggest the determination of rheumatoid factor (RF) and antinuclear antibodies (ANA) together with a questionnaire for the purpose of ruling out a CTD, although it is considered that screening for the presence of other antibodies (anti-Ro/SS-A, anti-La/SS-B, anti-Jo-1, etc.) could be useful in selected cases.1,2 As a consequence of these recommendations, it is common to encounter positive findings in the physical examination, questionnaires and/or immunology laboratory tests in patients that do not meet the classification criteria for a particular CTD, but are diagnosed as having a syndrome that has had many names, one of which is undifferentiated connective tissue disease (UCTD).3 This disorder is characterized by a considerable heterogeneity in terms of the demographic, clinical and morphological (histological and tomographic) characteristics as well as the outcome.4–8 The existence of highly sensitive classification criteria proposed by a number of authors has contributed to the heterogeneity.4,5,9,10 An intersociety consensus was recently published that attempts to clarify the definition of this disorder referred to as “interstitial lung disease (or interstitial pneumonia) with autoimmune features”, although the proposed criteria have not been validated.11 Beneath the hypothesis there is an underlying idea that this condition affects at least 2 different groups of patients. First, there would be a group of patients with inflammatory ILD, associated with an incomplete CTD that might or might not be detected throughout the course of the disease. Secondly, there would be another group of patients with fibrotic ILD, with a clinical profile similar to that of idiopathic pulmonary fibrosis (IPF), with false positive results according to the immunology laboratory, accompanied by nonspecific extrapulmonary clinical manifestations that are prevalent in the general population.12 In this context, the main objective of our study was to group the patients according to the presence or absence of certain clinical manifestations or immunological features, hoping to observe that these groups had different tomographic and/or functional expressions.

This is an observational study involving retrospective cohorts. We reviewed the medical records of all the patients hospitalized in the ILD clinic of the Instituto Nacional de Enfermedades Respiratorias Ismael Cosio Villegas (INER) (Mexico) between January 2012 and December 2014, and of all of the outpatients attended to in the ILD unit of Hospital María Ferrer (HMF) in Buenos Aires, Argentina, between January 2012 and June 2015. The patients provided their written consent. We included those patients with ILD who met the Kinder criteria for UCTD (Table 1).3 We excluded patients who met the criteria for a defined CTD in accordance with the criteria of the American College of Rheumatology or for some other cause of known interstitial disease.

We recorded demographic data (age, sex, smoking behavior), clinical findings (respiratory signs/symptoms and those associated with CTD), data provided by the immunology laboratory (ANA, RF and anti-Ro/SS-A, anti-Jo-1, anti-Scl-70, anticentromere and anti-U1 ribonucleoprotein [RNP] antibodies), data on respiratory function (percentage of predicted forced vital capacity [FVC%] and percentage of predicted diffusing capacity of the lung for carbon monoxide [DLCO%] at baseline, as well as changes in FVC% during follow-up), high-resolution computed tomography (HRCT) and surgical lung biopsy. Follow-up included respiratory function test every 4–6meses. With respect to the clinical manifestations, this being a retrospective study, we analyzed those that are usually dealt with in visits to the pulmonologist specialized in ILD in the participating centers (those included in the Kinder criteria). Other more specific manifestations that require the presence of another specialist for their evaluation (for example, mechanic's hands or Gottron's papules) were not analyzed. Data related to treatment were also recorded. In every case, this was indicated by the treating physician according to his or her criteria.

In this report, we found that the presence of at least 1 manifestation that was “highly specific for CTD” (Raynaud's phenomenon, xerophthalmia or arthritis) was associated with a higher prevalence of women, lower age at onset and a more favorable course in terms of changes in FVC% over time, which was confirmed after adjusting for potential confounders. This profile differs from that classically observed in patients with IPF, a disease in which there is a predominance of men, advanced age and rapid functional deterioration.12 One datum to point out with respect to our results is the fact that there were significant clinical and radiological differences between the cohorts. Although we do not have a definitive explanation for this finding, it could be that the differences are related to the moment at which the patients are evaluated (during a hospital stay in the INER and in the outpatient clinic in HMF).

During the baseline evaluation of a patient with ILD, the presence of clinical manifestations related to a CTD or a positive test for autoantibodies represents a dilemma for a physician, who should decide whether to consider the patient a carrier of idiopathic ILD (IPF) and utilize antifibrotic agents, or consider the case of a ILD associated with a CTD and administer immunosuppressive therapy.19,20 The detection of clinical manifestations that can contribute to the differentiation of these phenotypes appears to be a relevant finding.

Our results support the proposals of other authors in terms of the impact of clinical manifestations on the course of this group of patients.5 The intersociety consensus recently published includes proposals for new classification criteria (not yet validated) in which the clinical manifestations occupy a specific domain (together with a serological and a morphological domain).11 This consensus eliminates certain clinical manifestations included in previous criteria (for example, symptoms of gastroesophageal reflux, weight loss) and includes others with greater specificity, such as Raynaud's phenomenon and arthritis. The patients who meet the proposed criteria and do not satisfy those of a defined CTD have an “interstitial lung disease with autoimmune features”. We believe that the objective of the results of our study are along the lines of that proposed in this consensus, in the attempt to exclude highly sensitive manifestations that can mean the erroneous inclusion of individuals with IPF in the group of patients with ILD associated with CTD.

With respect to the serological data, neither the presence of high ANA titers nor the specific patterns according to IIF were helpful in differentiating between the groups presented in the hypothesis of this study. This contrasts with findings reported by other groups, which stress the role of the detection of autoantibodies in the evaluation of these patients.4,21 We consider that the information provided by the immunology laboratory may be useful, although we know that it can lead to false positives and negatives (even more so in the use of techniques like IIF that depend on the observer) and should be interpreted in accordance with the clinical setting in which it is obtained.

In a recent article that involved patients with a diagnosis of “interstitial lung disease with autoimmune features” according to the abovementioned consensus,11 it was observed that survival was better in those with a pattern not specific for UIP. The authors stressed the importance of the tomographic and histological pattern.22 In our study, the HRCT pattern did not differ among the subgroups evaluated. We found no association with a better functional outcome when a “pattern inconsistent with UIP” was included as a covariate in the multivariate analysis of the change in FVC over time. We point out the low prevalence in our cohort of the typical pattern of UIP according to HRCT findings in comparison with other reports.4,22,23 It could be that there has been a decrease in the strength of the study for detecting differences in the HRCT pattern when compared to the clinical manifestations, immunology laboratory and functional outcome. Our cohort came from 2 centers specialized in respiratory diseases, with physicians who have extensive experience in the evaluation of patients with ILD and numbers of patients similar to those of other cohorts previously reported. Thus, we consider that our findings are representative and comparable to those of other articles.

Our study has certain limitations. The patients were included on the basis of the medical records documented during hospital stays in one center (INER) and from an outpatient clinic in the other (HMF). This may constitute an important difference regarding the severity of the cases in the 2 centers, which implies possible selection bias. On the other hand, as this is a retrospective study with data taken directly from the medical records, the examination concerning the symptoms linked to CTD was not standardized, leading to possible information bias. The study also has limitations with respect to the evaluation of the features of autoimmune disease. The clinical and serological variables taken into account were those proposed in the Kinder criteria. Thus, they exclude manifestations and serological findings specific for CTD, such as mechanic's hands, palmar telangiectasias, Gottron's papules, antisynthetase antibodies other than Jo-1 or anti-cyclic citrullinated peptide antibodies. Many of these variables are now incorporated into the latest intersociety consensus and are valuable in the identification of patients with UCTD.11 Again, we did not analyze the role of capillaroscopy or minor salivary gland biopsy in the evaluation of patients with ILD associated with UCTD, as they are not included in the diagnostic criteria utilized.24,25 Finally, to evaluate the functional outcome, we chose to study the change in FVC% over time, in the knowledge that it is not the best method in patients who have different follow-up periods. The analysis of the time to event would have been a more accurate approach, but we did not have access to the data necessary to carry it out. While we consider that our study provides useful information, and that it is in keeping with the latest intersociety consensus, we believe that there is a need for prospective studies that evaluate survival and involve a larger number of patients to respond more conclusively to the question being proposed.

The present report has not received any type of funding.

The authors declare they have no conflicts of interest.