The present study measured the normal lung volume in patients with iPPFE using 3D-CT. The normal UL volume was less than half of the normal LL volume. Interestingly, longitudinal analyses illustrated that the extent of the decline in the normal UL volume was severe in patients with a lower normal UL volume, but this was not the case for the normal LL volume. Patients with a lower normal UL volume exhibited significantly shorter survival than those with a higher volume and a lower normal UL volume was associated with an increased mortality risk independently of age, gender and FVC. A composite model consisting age, sex and the normal UL volume successfully classified mortality risk into three groups. Our study illustrated that the normal UL volume is helpful for evaluating disease severity and mortality risk in patients with iPPFE.
The present study measured the volume with attenuation between − 950 and − 750 HU as the normal attenuated lung volume using 3D-CT data. Threshold-based measurements cannot distinguish morphological changes or spatial relationships, such as honeycombing, traction bronchiectasis and emphysema [10]. Further, the optimal threshold was not determined and it might differ among different ILDs [10]. However, this method can theoretically exclude high-attenuation fibrotic areas and low-attenuation destructive areas. Indeed, several reports that studied patients with ILDs employed similar thresholds [11–13]. The benefit of 3D-CT analyses in assessing physiology is its safe and patient-friendly method that is applicable to patients with advanced disease or respiratory failure [14]. In particular, because the distribution of ILD lesions is heterogeneous, 3D-CT permits evaluation of each lobe separately, making it ideal for patients with iPPFE. Indeed, the present study revealed a distinct feature of fibrous predominantly in the upper lobes. An obviously decreased normal UL volume was well correlated with FVC (r = 0.542). Thus, the normal UL volume is believed to quantitatively represent morphological features of iPPFE.
The present study consisted of two distinct cohorts of patients with iPPFE, which had several differences. The rate of lower-lobe ILD was higher in the Discovery cohort, which also featured a lower FVC and DLCO than the Validation cohort. Notably, in addition to decreased normal LL volume, patients with iPPFE and lower-lobe ILD exhibited significantly decreased normal UL volume. In clinical practice, certain patients with iPPFE develop lower-lobe ILD during the course of the disease and the presence of lower-lobe ILD was reported to be associated with the risk of mortality [6–9, 15]. Thus, the rate of advanced disease might have been higher in patients in the Discovery cohort.
The present study revealed that patients with iPPFE and decreased normal UL volume had shorter survival. Importantly, the results of Cox proportional analyses illustrated that the hazard ratio for lower normal UL volume was lower than that for normal LL volume. These results indicated that normal UL volume preservation is prognostically more important than normal LL volume preservation and that the normal UL volume is more closely related to the pathogenesis and disease progression of iPPFE than the normal LL volume. Consistently, an annual volume decline was more frequent in the upper lobes than in the lower lobes. Further, the extent of the decline was severe in patients with decreased normal UL volume, suggesting accelerated disease progression in patients with advanced iPPFE.
Using age, sex and the normal UL volume, we developed a composite model that classified patients with iPPFE into three groups according to their mortality risk in the Combined cohort as well as the two separate cohorts. To date, several reports have proposed composite models [6–9]. The sensitivity and specificity of this model for predicting mortality risk were higher than those of the GAP index. Kinoshita et al. also reported a composite scoring model consisting of seven factors, including FVC, pneumothorax history, presence of lower-lobe ILD and Krebs von den Lungen-6 [9]. Thus, simplicity is an advantage of this model. However, its statistical superiority over previously reported models has not been demonstrated, nor has its optimal cut-off been determined. Thus, the issues of overfitting should be acknowledged in this study. Additionally, there were several limitations. First, this was a retrospective study. In addition, quantitative analyses of the normal lung volume by 3D-CT required commercial software and a complicated procedure.
In conclusion, the present study measured the normal UL volume in patients with iPPFE by 3D-CT. The normal lung volume decreased each year predominantly in the upper lobes and the normal UL volume was associated with mortality risk in patients with iPPFE. A composite scoring model including age, sex and the normal UL volume was more predictive of the risk of mortality than the GAP index. Collectively, our results suggested that assessing the normal UL volume is useful for monitoring disease severity and mortality risk in patients with iPPFE.