Background The interstitial lung disease (ILD) is the most frequent form of lung damage in Sjögren's syndrome (SjS). The diagnosis is still nowadays challenging: there are not specific test -autoantibodies, pulmonary function test- or symptoms. Chest CT is the gold standard. Semiquantitative (SQCT) assessment (such Taouli or Goh visual scores) could estimate ILD severity or burdened by relevant inter-rater variability. Quantitative chest CT (QCT) is a promising method to assess primary or secondary ILD, but there are not studies focused on SjS-ILD.
Objectives Herein we designed a study to evaluate if QCT indexes might assess differences in SjS-ILD.
Methods We conducted a multicenter, cross-sectional, and retrospective study, identifying subjects affected of SjS (per modified American-European consensus criteria) and with a thorax CT ordered by local physician by any reason. We enrolled 52 cases, splitted in two cohorts according to the presence/absence of ILD on chest CT. A centralized SQCT assessment was carried out to calculate both Goh and Taouli scores. A DICOM-viewer open-source software (Horos) was used to analyse, blindly and anonymously, all CT images and to calculate QCT indexes (pulmonary (p) and total (t) kurtosis, standard deviation, skewness and mean lung attenuation). QCT indexes and SQCT scores were compared using the Spearman rank and Mann-Whitney tests. ROC curves were calculated to assess the sensibility and specificity of the QCT indexes to detect ILD.
Results Median age was 70.5 (95% CI 65.47–73.0); 96.2% were female. 9 cases have ILD, and 41 were not affected - two patients not assessable. The disease duration (till CT thorax) have a median of 21 (95% CI -6 – 42.41) months. 42.3% have secondary SjS and 57.7% a primary SjS. All QCT scores, except tSDev, were statistically different in ILD comparing those without ILD; the most relevant were: pKurt 2.05 (IQR 0.389 – 3.78) vs 4.34 (IQR 2.74- 6.0), p=0.008; pSdev (105.6 (IQR 101.4 - 122.95) vs 96.4 (IQR 92.38 to 104.94), p=0.006); pSkew (1.49 (IQR 0,49 -1,94) vs 2.01 (IQR 1.59 – 2.38), p=0.011); tKurt 7.62 (IQR 3.35 – 9.31) vs 10.00 (IQR 8.60 – 11.42),p=0,012 and tSkew 2.52 (1.63 – 2.85) vs 3.04 (2.81- 3.29), p=0.005). ROC analysis showed that tSkew could best discriminate ILD and those without: 0.8 (95% CI 0.66–0.9), cut off point ≤2.89; sensitivity 88.89% (95% CI 51.8–99.7), specificity 65.85% (95% CI 49.4–79.9), p<0.001. The correlation coefficients for QCT and Goh and Taouli's score ranges around ±0.4.
Conclusions QCT indexes identified some differences between SjS-ILD and those not. These parameters could be useful both to improve imaging diagnosis and also the decision to treat those cases. This innovative tool migh open up a potential research area to be developed in SjS. Moreover, the operator independence of QCT makes it a time-saving method extremely suitable for multi-centre trials. Our pilot study have some handicaps: small sample, low number of events, different reasons for CT performing, no smoking identification, and biological differences between primary and secondary SjS. However, it could be point out that QTC could be a useful tool to identify and quantify ILD in SjS, and further data are need to ascertain this hypothesis.
Disclosure of Interest None declared