Early decrease in erector spinae muscle area and future risk of mortality in idiopathic pulmonary fibrosis

Computed tomography (CT) assessment of the cross-sectional area of the erector spinae muscles (ESMCSA) can be used to evaluate sarcopenia and cachexia in patients with lung diseases. This study aimed to confirm whether serial changes in ESMCSA are associated with survival in patients with idiopathic pulmonary fibrosis (IPF). Data from consecutive patients with IPF who were referred to a single centre were retrospectively reviewed. We measured the ESMCSA at the level of the 12th thoracic vertebra on CT images at referral and 6 months later (n = 119). The follow-up time was from 817–1633 days (median, 1335 days) and 59 patients (49.6%) died. A univariate Cox regression analysis showed that the decline in % predicted forced vital capacity (FVC) (Hazard ratios [HR] 1.041, 95% confidence interval [CI] 1.013–1.069, P = 0.004), the decline in body mass index (BMI) (HR 1.084, 95% CI 1.037–1.128; P < 0.001) and that in ESMCSA (HR 1.057, 95% CI 1.027–1.086; P < 0.001) were prognostic factors. For multivariate analyses, the decline in ESMCSA (HR 1.039, 95% CI 1.007–1.071, P = 0.015) was a significant prognostic factor, while those in % FVC and BMI were discarded. Early decrease in ESMCSA may be a useful predictor of prognosis in patients with IPF.

an independent prognostic factor for patients with COPD 21 . Miller et al. reported that height-adjusted ESM CSA was significantly associated with 30-day mortality and length of hospital stay after lobectomy in patients with lung cancer 23 . Suzuki et al. also revealed that a smaller ESM CSA in patients with IPF was associated with a poor prognosis 22 .
We queried whether the survival of patients with IPF decreases with a decrease in muscle mass. Resultantly, we hypothesised that a relative decline in ESM CSA over the course of 6 months is associated with decreased survival in patients with IPF. We measured the ESM CSA at the level of the spinous process of 12th thoracic vertebra on CT images at baseline (n = 144) and 6 months later (n = 119) in patients with IPF. The correlation of ESM CSA values in the 119 patients between trained individuals were as follows; r = 0.951(95% confidence intervals [CI] 0.913-0.965), P < 0.001. The Bland-Altman analysis revealed that the agreement between two individuals was excellent (0.934).

Results
The clinical characteristics and ESM CSA of 144 patients at baseline, 119 patients who underwent HRCT after 6 months, and 25 patients who did not undergo HRCT after 6 months are shown in Supplemental Table 1. We could not analyse CT in 25 patients after 6 months. Eleven of the 25 patients died; 5 were transferred; 3 were unchanged in pulmonary function test; 2 were out of timing; 1 was rejected; and 1 experienced acute exacerbation. There were significant differences in age (P = 0.001), body mass index (BMI) (P = 0.039), %FVC (P < 0.001), % deffuse capacity of the lung for carbon monoxide (DL CO ) (P = 0.008), distance walked during the 6 minute walk test (6MWT) (P < 0.001), lowest SpO 2 during the 6MWT (P = 0.046) and ESM CSA (P = 0.009) between patients who underwent HRCT after 6 months and those who did not.
The distributions of ESM cSA declines. The distribution of the relative decline in ESM CSA is shown in Fig. 1. We compared the relative decline in %FVC with that in ESM CSA to determine which was greater. We observed that the relative decline in ESM CSA (6.2 [0.5-11.8] %) was significantly greater than that in %FVC (1.4 [−3.1%-5.0] %) (Student's t-test, P < 0.001).
Correlations between the decline in ESM cSA and other clinical parameters. The correlations between the relative decline in ESM CSA and other clinical parameters were as follows ( Table 2) Figure 2 shows the correlations of relative decline in ESM CSA with the relative decline in %FVC, the relative decline in %DL CO , and the relative decline in BMI.
HRs and 95% CI for each parameter on uni-and multivariate Cox regression analyses in the baseline cohort (n = 144) are shown in Supplemental Table 3.

Kaplan-Meier curves, log-rank tests and number at risk.
Kaplan-Meier curves of the cohort available for 6 months of follow-up (n =119) are shown in Fig. 3. To determine an optimal cutoff value for predicting 3-year mortality, we constructed receiver operator characteristic (ROC) analysis. A cutoff value of 10.5% (AUC = 0.734, specificity: 0.752, sensitivity: 0.553) was identified. The median survival times were as follows: the relative decline in ESM CSA ≥ 10.5%, 602 [480-1269] days; the relative decline in ESM CSA < 10.5%, 1431 [1130-1743] days. The relative decline in ESM CSA ≥ 10.5% had a significantly poorer prognosis (P < 0.001, log-rank test). characteristics with ipf patients with or without eSM cSA decline. The clinical characteristics of patients with IPF with (≥10.5%) or without (<10.5%) ESM CSA decline are shown in Table 4. There were significant difference in baseline BMI (P = 0.049), baseline %FVC (P = 0.002), baseline forced expiratory volume in 1 second (FEV 1 )/FVC (P = 0.002), baseline lowest SpO 2 during 6MWT (P = 0.004), ESM CSA after 6 months (P < 0.001), survival time (P < 0.001), the relative decline in ESM CSA (P < 0.001), the relative decline in %FVC (P = 0.003), the relative decline in %DL CO (P = 0.004) and the relative decline in BMI (P < 0.001). During the 6 months, there were 7 unexpected hospitalisations (3 acute exacerbations, 2 infections and 2 instances of worsening IPF). Pirfenidone was prescribed to 28 patients during the 6 months whereas nintedanib was prescribed to none ( Table 4). The number of cancer and chronic heart failure comorbidities, as well as, the Charlson comorbidity index of the study population, are also shown in Table 4.  Table 3. Prediction of mortality by uni-and multivariate Cox-proportion analyses in the patients available for CT after 6 months (n = 119). Abbreviations: HR, hazard ratio; CI, confidence interval; FVC, forced vital capacity; DLCO, diffuse capacity of the lung for carbon monoxide; ESMCSA, cross-sectional area of erector spinae muscles. *We analysed using n = 114, because 5 cases were missing. www.nature.com/scientificreports www.nature.com/scientificreports/ Decline in ESM cSA and effects of treatment with corticosteroids. No patients had been treated with corticosteroids at baseline, but 11 patients were treated with corticosteroids 6 months later. We compared the relative decline in ESM CSA between the patients treated with corticosteroids and those who were not. In the patients treated with corticosteroids, the relative decline in ESM CSA was 9.6 [3.5-16.8] % after 6 months. In the patients not treated with corticosteroids, the relative decline in ESM CSA was 6.0 [−0.2-10.6] % after 6 months. There was no statistical difference (P = 0.096) between the two groups (Wilcoxon rank-sum test).

Discussion
Baseline ESM CSA is a prognostic factor in patients with IPF 22,24 . Here, we demonstrated that the serial change in ESM CSA is also a prognostic factor in patients with IPF. On multivariate analysis, the decline in ESM CSA was a significant prognostic factor, while those of predicted %FVC and BMI were not.
It is widely known that the decline in FVC is a strong prognostic factor 15,16 . Measurements of serial changes in FVC are used as a gold standard in clinical trials [25][26][27][28] . Interestingly, in the present study, the decline in ESM CSA (HR 1.039, 95% CI 1.007-1.071, P = 0.015) was a significant prognostic factor in the multivariate analyses, while the decline in %FVC was not. This result may indicate that the decline in ESM CSA is a stronger prognostic factor than the decline in %FVC. The decline in ESM CSA was correlated with the decline in %FVC. However, the decline in ESM CSA (6.2 [0.5-11.8] %) was significantly greater than that in %FVC (1.4 [−3.1-5.0] %). It is possible that this result influences the data of multivariate Cox-proportion analysis. We did not evaluate the results of 6MWT because the missing data of 6 months later were not small. Further studies are needed to verify whether the decline in ESM CSA is useful for predicting mortality as compared with serial changes in the lowest SpO 2 and the distance walked during the 6MWT.
Cachexia and sarcopenia have been extensively studied in patients with lung cancer and COPD. Evan's diagnostic criteria for cachexia include evaluation items such as muscle mass, fatigue and weight loss. In an attempt to include a wider evaluation of the burden of cachexia, diagnostic criteria based on an assessment of laboratory data and symptoms in addition to weight have been proposed 17 . The criteria included weight loss ≥5% in 12 months or low BMI (<20 kg/m 2 ) with 3 of 5 of the following features: decrease muscle strength, fatigue, anorexia, low fat-free mass index, of abnormal biochemistry (increased inflammatory markers, anaemia and low serum albumin). We reason that cachexia would be associated with the survival in IPF.  www.nature.com/scientificreports www.nature.com/scientificreports/ Kinsey et al. reported that a smaller pectoralis muscle area, as measured on CT images obtained at the time of diagnosis of non-small cell lung cancer, was associated with poorer overall survival 29 . McDonald et al. reported that patients with a smaller pectoralis muscle area on CT scan tended to have a more severe expiratory air flow obstruction, lower quality of life scores and diminished exercise capacity, as compared with people with a lower BMI 30 . The measurement of the psoas muscle at the L3 or L4 level on CT images is frequently used for evaluating sarcopenia 31 . Canvasser et al. reported a strong correlation between the paraspinous muscle area at Th12 and the total psoas area at L4 (r = 0.72, P < 0.001), and both were associated with lower mortality rates after surgery 32 . Those authors suggest that measuring the area of the erector spinae muscles at the Th12 level might be useful for evaluating skeletal muscles in patients with lung disease who have not undergone an abdominal CT. However, no data on the pectoralis muscles exist in patients with IPF presently. In patients with IPF, chest CT scans are regularly taken in clinical practice, along with chest X-rays and pulmonary function tests. Based on these and our results, measuring erector spinae muscles (ESMs) by chest CT images would be useful in evaluating sarcopenia and cachexia in patients with IPF.
Loss of skeletal muscle mass is considered to be induced by systemic inflammation, inactivity, malnutrition and enhanced energy expenditure [17][18][19][20] . In the present study, the decline in ESM CSA was weakly correlated with the decline in BMI In a previous report, ESM CSA was weakly correlated with percentage of predicted FEV 1 (r = 0.31, P < 0.0004) in patients with COPD 21 . Impaired pulmonary function, exercise-induced hypoxemia and nutrition status might influence the decline in ESM CSA . However, further studies are needed to understand the mechanisms underlying muscle loss in IPF.
Only 11 patients were treated with corticosteroids in this study, and there was no significant difference in the decline in ESM CSA . However, we cannot rule out the possibility that muscle atrophy due to the side effects of corticosteroids affected the ESM CSA . Further studies are needed to explore this possibility. Moreover, the influence of appetite loss side effects of anti-fibrotic drugs on the ESM CSA should be investigated in future studies.
Recently, several studies have reported the association between lung disease and ESM CSA . The ESM CSA was significantly associated with health-related quality of life and prognostic physiological parameters in patients with Mycobacterium avium complex lung disease 33 . A smaller ESM CSA was significantly associated with a lower level of activities of daily living at the end of treatment for pneumonia 34 .
The present study has the following limitations. First, in the study, the results were obtained by a retrospective analysis of all Japanese patients from a single centre. The sample size was small and there was no replication cohort. Further studies are required to confirm our observations in other external validation cohorts. Second, the diagnoses of IPF were according to the 2011 international guideline. The diagnosis of patients who initially presented before 2011 was confirmed by MDD according to the 2011 guidelines. In these patients, the diagnosis may be more accurate due to a consideration of the MDD time course. The limitation is that the MDD diagnosis before and after 2011 may not be the same. Third, 25 of 144 patients (17.4%) were not available for CT images to compare the changes at 6 months. Eleven patients died within 6 months, and CT images of the rest of the patients were not available for several reasons. Fourth, in the present study, approximately 16.7% (30 of 180 patients) of the CT of patients with IPF excluded the 12th thoracic spinous process level. This might be because the lungs of patients with IPF are smaller than those with COPD. This is another limitation of the present study. Fifth, we did not adjust for treatment in the survival analysis. The reason was that there was bias because a little evidence in anti-fibrotic drugs for IPF existed before 2014.
In conclusion, we have demonstrated that a smaller ESM CSA is a prognostic predictor in patients with IPF. A reduced ESM CSA after 6 months was an independent prognostic predictor in patients with IPF.

Methods
This single centre retrospective study was performed in accordance with the amended Declaration of Helsinki. The ethical review board of Tosei General Hospital, which contributed cases to this study, provided approval for the study (approval number 745). The study was also approved by the ethical review board of the Nagoya City University Hospital (approval number 60-18-0210), where the CT analyses were performed. Given that the data were analysed anonymously, the ethical review boards did not ask for the patients' approval or informed consent. The opt-out document can be found on the website of Tosei General Hospital.
Patients. During the period of June 2008 to July 2013, the clinical records of 180 consecutive patients with IPF, referred to the Tosei General Hospital without corticosteroid or anti-fibrotic drug pre-treatments, were retrospectively reviewed. IPF was diagnosed by multidisciplinary discussion (MDD) according to 2011 international guidelines 1 . The diagnosis of patients who initially presented before 2011 was confirmed by MDD according to the 2011 guidelines before May 2015.
The following patients were excluded: one patient, who experienced acute exacerbations at the initial visit; patients who had other comorbidities, such as lung cancer (n = 1), infectious diseases or congested cardiac failure at the initial presentation; thirty patients, whose chest CT lacked the level of the spinous process of the 12th thoracic vertebrae and four patients, who underwent lung transplantation because due to the small number of patients and the consideration of death and lung transplantation as different outcomes.
Ultimately a total of 144 patients were enrolled. After approximately 6 months, Chest CT images were available for 119 patients. The interval between performing the CT at baseline and the second time was 190 [169-217] days.
Besides, the duration between the day of performing the chest CT on the first visit to the last visit or death was recorded. Other cases were contacted to confirm their life-or-death status by telephone.
Pulmonary function tests and six-minute walk test. All patients underwent pulmonary function tests by spirometry (CHESTAC-55 V; Chest, Tokyo, Japan), according to the American Thoracic Society and European Respiratory Society (ATS/ERS) criteria 35 . The DL CO was also measured (CHESTAC-55 V). The values of FVC, FEV 1 and DL CO were measured according to ATS/ERS recommendations 36 . We calculated %FEV 1 , %FVC and %DL CO based on the patients' height, age and sex per the Japanese guidelines 37 . We then conducted 6MWTs without supplemental oxygen in accordance with ATS guidelines 38 .
The comorbidities and the Charlson comorbidity index. The Charlson comorbidity index was calculated according to a previously reported method 39 .
Statistical analyses. Continuous variables were presented as medians and interquartile ranges. Categorical variables were presented as numbers and percentages. The differences between patients who underwent HRCT after 6 months and those who did not were analysed using the Student's t-test or the Wilcoxon rank-sum test. Spearman's rank correlation coefficients were used to test for correlations between ESM CSA and clinical parameters and those between the relative decline in ESM CSA and clinical parameters. Univariate and multivariate Cox regression analyses were performed to evaluate the relationship between each variable and mortality. We included factors with P-values < 0.05 in the univariate analysis for the multivariate analyses. Survival times were estimated using the Kaplan-Meier method and compared with the log-rank test. To determine the optimal cutoff value for predicting 3-year mortality, we constructed ROC curves. P-values less than 0.05 were considered significant. The statistical analyses were conducted using JMP statistical software (version 14; SAS Institution Japan Ltd, Japan). Since we could not analyse the r (95%CI) with JMP statistical software (version 14), we analysed r (95%CI) using SPSS (version 26; IBM, Japan).