Distinct profile and prognostic impact of body composition changes in idiopathic pulmonary fibrosis and idiopathic pleuroparenchymal fibroelastosis

Change in body composition with skeletal muscle wasting, a major component of pulmonary cachexia, is associated with mortality in chronic obstructive pulmonary disease and cancer. However, its relevance in interstitial lung diseases (ILDs) remains unclear. We hypothesized changes in body composition would be associated with mortality in ILDs. We measured the cross-sectional-area (ESMCSA) and muscle attenuation (ESMMA) of erector-spinae muscles, as determined by CT-imaging, in patients with idiopathic pulmonary fibrosis (IPF; n = 131) and idiopathic pleuroparenchymal fibroelastosis (iPPFE; n = 43) and controls. Subsequently, implications with prognosis were evaluated. The ESMCSA of ILD patients, but not ESMMA, was significantly smaller than that in controls. Lower ESMCSA with decreased BMI were recorded in iPPFE patients versus IPF patients, whilst IPF patients had decreased ESMCSA without BMI decline. Lower ESMCSA in IPF patients were associated with poorer prognoses. Conversely, decreased ESMMA were associated with worse survival in iPPFE patients. Multivariate analyses showed that ESMCSA in IPF and ESMMA in iPPFE were independent risk factors for mortality. Distinct changes in body composition had prognostic significance among patients with IPF and iPPFE. Lower ESMCSA and ESMMA were independently associated with poor prognosis in IPF and iPPFE, respectively. These results suggest values to measure body composition changes in managing patients with IPF and iPPFE.

Society (ATS) guideline 15 . Compared to the lower-lobe dominant IPF, iPPFE is characterized by upper lobe predominant fibrosis involving the pleura and subpleural lung parenchyma 15 . These distinct types of ILDs cause progressive dyspnea and lung function decline, resulting in pulmonary dysfunction and cachexia syndrome. Body composition changes are typically found in iPPFE; patients often complain of weight loss associated with a slender stature, with a "flattened chest" being a unique peculiar physical finding 18,19 . However, despite their occurrence, the clinical implications of body composition changes in ILDs are currently unclear.
Thus, we hypothesized that changes in body composition would be associated with mortality in ILDs. In this study, we aimed to compare the cross-sectional area (ESM CSA ) and muscle attenuation (ESM MA ) of erector spine muscle, as determined by CT-imaging, in patients with IPF and iPPFE. We also aimed to explore the impact of body composition changes on prognoses in patients with IPF and iPPFE.

Results
Clinical characteristics. The clinical characteristics of 131 patients with IPF and 43 patients with iPPFE are summarized in Table 1 and Supplement Table 1. Among the 131 patients with IPF, 81 patients (61.8%) were diagnosed with IPF by the presence of usual interstitial pneumonia pattern on high-resolution CT images, and 50 patients (38.2%) were histologically diagnosed with IPF via surgical lung biopsies. As expected, there were several differences between the patients with IPF and iPPFE; there were more male smokers and levels of serum Krebs von den Lungen-6 (KL-6) were increased in the IPF group, whilst patients with iPPFE had lower BMIs (median 17.2 [14.7-18.5] kg/m 2 ), occasional "flattened chest" as previously described 18,20,21 , lower % forced vital capacity (FVC)-predicted, and higher partial pressures of carbon dioxide (PaCO 2 ). The levels of serum KL-6 were increased in patients with IPF compared to those in iPPFE patients. Patients with IPF and iPPFE did not differ in terms of predicted values of diffusing capacity of the lung for carbon monoxide (DLCO; n = 52 and n = 24, respectively) or serum levels of albumin and surfactant protein-D (SP-D).  Fig. 2A-C). In iPPFE, conversely, only ESM MA , but not ESM CSA or BMI, effectively determined prognoses ( Fig. 2D-F).   Univariate and multivariate analyses of ESM CSA and ESM MA in idiopathic pulmonary fibrosis and idiopathic pleuroparenchymal fibroelastosis. To determine the impact of the measurements associated with body composition changes on prognosis, we preformed Cox proportion-hazard regression analyses. Univariate analyses in patients with IPF revealed that factors related with muscle wasting, such as ESM CSA , ESM MA , and BMI were significantly associated with mortality. Given that age and sex often deviate in accordance with body mass, we performed adjusted multivariate analyses with age and sex. We excluded DLCO, because it was examined in only 52 patients with IPF. In the multivariate analyses, both ESM CSA and %FVC-predicted values were independently associated with mortality in patients with IPF (Table 4). In the iPPFE cohort, univariate analyses indicated only sex and ESM MA as significant factors for mortality. Following adjustment for age and sex, lower ESM MA values were associated with higher mortality rates in patients with iPPFE (Table 5).

Discussion
To our knowledge, this is the first study to evaluate body composition changes in patients with ILDs by means of quantitative measurement of ESM CSA , ESM MA , and BMI and to explore their clinical implications. First, we found lower ESM CSA , but not ESM MA , in patients with the ILDs versus controls. Among the ILDs, skeletal muscle loss in patients with IPF was not accompanied by weight loss, while patients with iPPFE showed more decline in skeletal muscle mass together with marked lean BMI values compared with those with IPF. Second, regarding the prognostic significance, a decline in ESM CSA was associated with poor prognoses in patients with IPF, and this association was independent of established prognostic factors of IPF, such as age, sex, and pulmonary physiology. In contrast, lower ESM MA values were independently related to poor survival in patients with iPPFE. These results suggest that distinct profiles of body composition changes occurred among IPF and iPPFE. More importantly, these changes were of prognostic significance, which highlights the importance of monitoring ESM CSA and ESM MA in the ILDs. Loss of skeletal muscle, a hallmark of cachexia, is common in advanced lung disease. In addition, increased lipid deposition often occurs with declining muscle mass, and growing evidence supports an important role of fatty acid and intermediates in the regulation of muscle function 22 . Exposure to free fatty acids and excess dietary lipid intake are associated with pathogenesis of muscle wasting. Indeed saturated fatty acids convey detrimental effects upon muscle functions by causing insulin resistance, reactive oxygen species, inflammatory signals, and activate proteolysis, and also impairing protein synthesis and mitochondrial function 22 . Muscle mass loss can be assessed by measuring skeletal muscle cross-sectional area on CT, such as ESM CSA , whereas lipid deposition is evaluated by measures of skeletal muscle attenuation on CT, such as ESM MA 23,24 . When assessing cachexia, BMI does not accurately account for body composition changes and may underestimate the frequency of cachexia in obese patients; termed as sarcopenic obesity, or those who have gained weight due to edema 25,26 . Thus, quantitative analysis of body composition by ESM CSA and ESM MA is superior to BMI for assessing cachexia in COPD and cancer patients [12][13][14] . To date, however, no data are available on the assessment of body composition by ESM CSA and ESM MA in ILDs, such as IPF and iPPFE. Thus, in the present study, we attempted to investigate body composition changes determined by ESM CSA and ESM MA in patients with IPF and iPPFE. Compared with controls, ESM CSA was significantly lower in patients with IPF and iPPFE, but ESM MA was comparable. Among the measurements associated with body composition, ESM CSA correlated fairly with BMI, and weakly with ESM MA . No significant correlation was found between ESM MA and BMI. In addition, ESM CSA correlated weakly with age and pulmonary physiology, whereas no correlation was observed between ESM MA and pulmonary physiology. Collectively, these data suggest that each measurement may, in part, reflect different body composition changes and pathophysiology in ILDs. In cancer patients receiving chemotherapy, a significant decrease in both muscle CSA and attenuation on CT was reported, suggesting that these patients exhibited sarcopenic phenotype with decreased skeletal muscle mass and increased lipid deposition 27 . In COPD, however, although CSA of quadriceps was smaller in patients than in controls, muscle attenuation on CT did not differ significantly between them, which is similar to our observations in IPF and iPPFE 28 . These results suggest that patients with IPF and iPPFE may have decreased skeletal muscle mass without changes in lipid deposition.    Interestingly, a significant difference in body composition changes was found between IPF and iPPFE; lower ESM CSA and smaller BMI values were noted in patients with iPPFE versus IPF, while ESM MA values were similar in both groups. These data indicate that patients with iPPFE have greater loss of skeletal muscle mass, as well as a leaner body, compared to those with IPF, while patients with IPF show only moderate skeletal muscle loss without a change of BMI, suggesting different profiles of body composition changes between IPF and iPPFE. In our cohorts, patients with iPPFE were characterized by upper lobe predominant fibrosis together with more severe restrictive defects on spirometry than those with IPF. These differences might lead enhancements of impaired energy and protein balance, resulted in depletion of both fat and protein stores as reflected in weight loss and muscle wasting in iPPFE patients.
Most importantly, the present study indicates that the measurements associated with body composition were of prognostic significance, independent of age, sex, and pulmonary physiology. In IPF, patients with lower values of ESM CSA , ESM MA , and BMI had significantly worse survival than those with higher values. In iPPFE, patients with lower values of ESM MA showed significantly poorer outcomes. Moreover, multivariate analyses with Cox proportional hazards regression model demonstrated that ESM CSA and ESM MA were independent prognostic factors for IPF and iPPFE, respectively. In patients with cancer, several studies reported that both skeletal muscle mass and its attenuation assessed by CT images were powerful prognostic factors, independent of BMI and clinical staging 12,29 . In COPD, ESM CSA or CT-derived pectoralis muscle fat-free index correlated significantly with prognosis, independent of BMI and pulmonary physiology 13,14 . Thus, as observed in patients with cancer and COPD, the assessment of skeletal muscle by ESM CSA and ESM MA provides additional prognostic information in patients with IPF and iPPFE. Recently, several therapeutic molecules targeting for cachexia and muscle wasting, such as a ghrelin-receptor agonist (anamorelin) and a selective androgen receptor modulator (enobosarm), have been proven to be effective in patients with cancer 30,31 . Thus, it is possible that cachexia and muscle wasting may be a novel therapeutic target in ILDs.
Interestingly, as described above, we found a difference in prognostic factors associated with body composition changes between IPF and iPPFE. The reason for this difference is not clear. Generally, it is conceivable that ESM CSA was a prognostic factor in IPF, because ESM CSA has been reported to be of prognostic significance in other diseases, such as cancer and COPD, and patients with IPF had significantly lower ESM CSA than controls. However, it is unclear why only ESM MA had prognostic significance in iPPFE, despite no difference in ESM MA between iPPFE and controls. In patients with advanced non-small cell lung cancer, Sjoblom and colleagues recently reported that lower skeletal muscle attenuation on CT was independently prognostic for poor survival, whereas cross-sectional muscle area was not 29 . This suggests that skeletal muscle adiposity may be more important than skeletal muscle loss in those patients for predicting prognosis. In addition, Maddocsks and colleagues showed that skeletal muscle attenuation in patients with COPD was not significantly different from that in controls, but the attenuation was associated with physical activity levels and exercise capacity in the patients 28 . In our patients with iPPFE, most of whom already had extremely low skeletal muscle mass, it is suggested that fat deposition assessed by ESM MA may have been more closely associated with prognosis than skeletal muscle mass determined by ESM CSA . Future studies including larger patients with iPPFE will be required to confirm this.
The present study has several limitations. First, although a relatively large number of patients with IPF and iPPFE were enrolled, the data collection method was retrospective. Thus, the impact of longitudinal changes in weight loss, %FVC-predicted decline, ESM CSA , and ESM MA were not evaluated. Additionally, we did not confirm our results among patients with pathologically diagnosed iPPFE. Although definite diagnosis of iPPFE requires pathological evaluation, surgical lung biopsies are not performed in a substantial number of cases in clinical practice due to the lack of curative treatment, limited ventilator reserve, risk of prolonged postoperative pneumothorax, and acute exacerbation 18 . Therefore, we have proposed clinical criteria for iPPFE, which enables us to recruit patients with characteristics similar to those of iPPFE 20 . Second, we measured ESM area at Th12, but did not evaluate the lumbar muscles or the pectoralis, which were explored in patients with COPD and cancer. Third, although D-XA is used to evaluate body composition changes, the associations between axial CT and D-XA measurements are unclear. Thus, future prospective studies are required to overcome these limitations.
In conclusion, the present study investigated body composition changes and their association with prognoses in patients with ILDs, including IPF and iPPFE. Compared with controls, distinct patterns of body composition changes occurred among patients with IPF and iPPFE, and the different measurements associated with body composition had prognostic significance for the two diseases. These results highlight the importance of assessing body composition changes in patients with ILDs. Additionally, management of muscle wasting may improve prognoses and provide a novel therapeutic target among these patients.

Methods
Subjects. This retrospective study was conducted on a cohort of 137 consecutive patients with IPF admitted to Hamamatsu University of School of Medicine between, and a cohort of 44 patients with clinically diagnosed PPFE admitted to the Hamamatsu University of School of Medicine and its nine associated hospitals between 2000 and 2015. Diagnosis of IPF was based on the ATS/ERS/Japanese Respiratory Society (JRS)/Latin American Thoracic Association (ALAT) criteria 17 . Clinical diagnosis of iPPFE was based on the following criteria 20 : (1) radiograph iPPFE pattern on chest CT (defined as bilateral subpleural dense consolidation with or without pleural thickening in the upper lobes, less marked according to Reddy's radiological criteria 21 ); (2) radiological confirmation of disease progression (defined as an increase in upper lobe consolidation with or without pleural thickening and/or a decrease in upper lobe volume on serial radiological assessment); (3) exclusion of other lung diseases with identifiable etiologies (such as connective tissue disease-related ILDs, chronic hypersensitivity pneumonitis, pulmonary sarcoidosis, pneumoconiosis, and active pulmonary infection). Chest CT at diagnosis was unavailable in 6 patients with IPF, and CT-imaging was insufficient for evaluation in one patient with iPPFE (due to spinal implants). Thus, this study enrolled 131 and 43 patients with IPF and iPPFE, respectively.
SCIentIfIC REpoRtS | (2018) 8:14074 | DOI:10.1038/s41598-018-32478-z As characteristics between IPF and iPPFE, such as gender, smoking habitant and body statures were different, this study also enrolled age matched 78 consecutive non-COPD subjects who visited our institute for medical check-up as controls; evaluations were performed using anthropometry, spirometry, and chest CT. None of the control subjects had ILDs, resected lung, active infections, malignancies, and neuromuscular diseases.
The study protocol was approved by the Ethical Committee of Hamamatsu University School of Medicine , and carried out in accordance with approved guideline. The need for patient approval and/or informed consent was waived due to the retrospective nature of the study.
Computed-tomography image analysis. Electronically stored CT-images were used to assess muscle mass. All CT-images were obtained for diagnostic purposes during routine clinical practice. Chest CT was performed in the supine position at full inspiration breath-hold at 120 kVp and approximately 200 mA. Using a modified method, as described in a previous article 12,13,32 , chest CT images were reconstructed in a mediastinal setting (reconstruction kernel FC13). Single slice axial CT-images (contrast unenhanced condition; 5-mm thickness and 5-mm interval) taken at the lower margin of the 12th thoracic vertebra (Th12) were selected to measure ESM CSA . After imaging, ESMs were identified and manually shaded; ESM CSA quantification was based on Hounsfield unit (HU) thresholds (−29 to + 150) and mean ESM MA (HU) levels were assessed as previously described 12 . All CT analyses were independently performed by trained individuals (YS and KY) blinded to the patients' survival statuses and then averaged. Images were analyzed using SYNAPSE VINCENT version 3 (FUJIFILM Medical Systems, Tokyo, Japan).

Data collection.
Clinical data were obtained from the patients' medical records. Laboratory findings, pulmonary, and function test results, obtained at the time of diagnoses, were recorded. Statistical analysis. Discrete variables are expressed as totals (percentages), and continuous variables are expressed as median [interquartile range]. Mann-Whitney test was used to compare continuous variables, and the Kruskal-Wallis test and post hoc analyses were used for multiple comparisons. Fisher's exact tests for independence were used to compare categorical variables. Correlations were analyzed using the Spearman's rank correlation technique. Overall survival time was measured from date of IPF and iPPFE diagnosis. To determine the impact of body composition changes on prognoses, univariate and multivariate analyses were performed using the Cox proportional hazards regression model. Cumulative survival probabilities were estimated above and below the median ESM CSA , ESM MA , and BMI values using the Kaplan-Meier method and logrank test. Statistical analyses were performed using GraphPad Prism Version 6 (GraphPad Software, San Diego, CA, USA) and SPSS Statistics (Ver23, IBM Corporation, Armonk, NY, USA) software. All analyses were two-tailed and P values less than 0.05 were considered significant.