DISCUSSION
In this study the change in lung density and volume due to pulmonary fibrosis in patients with SSc evaluated by a soft-ware aided image segmentation method was investigated. According to the study hypothesis, fibrosis of lung tissue should lead to increased lung density and reduced lung volume due to decreased ventilation [15]. Consistent with this hypothesis, it was shown that right and left lung densities were higher in SSc patients with signs of fibrosis than SSc patients without pulmonary fibrosis and healthy controls.
Although parenchymal changes related to fibrosis were detected in approximately 80-100% of SSc patients in early autopsy studies [2,16], it is known that lung involvement in SSc varies between 40-60% when assessed by HRCT studies [17-19]. In line with the literature, it was expected that lung densities would be decreased in the patients without fibrosis-related morphological changes on HRCT compared to healthy subjects indicating sub-clinical and sub-radiological lung involvement. However, lung densities showed no difference between the patients without signs of fibrosis and controls. It hs previously been shown that PF is closely related to anti-topoisomerase antibody positivity and diffuse cutaneous involvement, and thus patients without fibrosis-related morphological changes on HRCT were analyzed after subgrouping them into limited and diffuse cutaneous sub-types [15,18,20]. Our results showed that in dcSSc patients, even if there were no signs of interstitial lung disease (ILD) in HRCT, lung densities were higher than those of lcSSc and control patients. This result suggested that early changes in lung density may appear before the architectural distortion on HRCT in dcSSc compared to lcSSc patients. Numerous studies have confirmed that patients with lcSSc or dcSSc have distinct autoantibody profiles, patterns of organ involvements, progression, and outcomes [21-22]. There is still controversy whether the two subtypes of SSc represent truly distinct diseases or merely different extremes of one disease spectrum [23]. We suggest that the two subtypes differ in terms of lung density in radiologically non-apparent disease, although the clinical significance of this is unknown.
The annual follow-up HRCT scans of the 56 patients without any morphological signs of fibrosis on HRCT were evaluated and it was found that 16 (28.6%) of the patients developed abnormal imaging after a mean of ten years from baseline scans. It was shown that the baseline right and left lung densities were significantly lower in the future PF developers than the non-developers. However, there is little evidence to support the predictive value of decreased lung density for PF development in the future. Therefore its role as a parameter in assessing and monitoring newly diagnosed patients should be investigated in larger cohorts. Considering the heterogeneity of systemic sclerosis-associated interstitial lung disease (SSc-ILD), the potential absence of symptoms in early or mild disease, and the potential impact of ILD on outcomes in patients with SSc, our results may contribute to prediction models in SSc-ILD [24].
There was a weak correlation between lung volumes and left lung density and both FVC and DLCO in patients without PF. However, the association became stronger in PF between the densities and left lung volume and both FVC and DLCO. Based on our results, the gradual development of a relationship with the effect of fibrosis-related changes in the lung parenchyma was evident.
Interestingly, the expected decrease in lung volumes due to fibrosis were inconsistent in some subgroup analyses. We believe that this variability may be due to a change in compensatory capacity of the lungs in response to any volume-decreasing pathology. Thus, due to this possible confounder, the main focus of the study was lung density and we believe that use of lung volume as a measurement parameter may be misleading.
One of the most important limitations of our study was that follow-up HRCT scans were not performed in all patients at regular intervals due to the retrospective design. Thus, we refrain from making precise comments about the time to the onset of PF, especially in patients with dcSSc. The small sample size was another limitation, leading to unreliable statistical analysis in some of the subgroup comparisons.