AST/ALT ratio as a predictor of mortality and future exacerbations of PM/DM-ILD——a retrospective cohort study with 522 cases CURRENT

Objective To assess the associations between aspartate transaminase/alanine transaminase ratio (DRR) and mortality in patients with Polymyositis/dermatomyositis associated interstitial lung disease (PM/DM-ILD). This was a retrospective cohort study, which included 522 patients with PM/DM-ILD whose DRR on admission were tested at West China Hospital of Sichuan University during the period from January 1, 2008 to December 31, 2018. Cox regression models were used to estimate hazard ratios for mortality in four predefined DRR strata (≤ 0.91, 0.91–1.26, 1.26–1.73 and > 1.73), after adjusting for age, sex, DRR stratum, diagnosis, overlap syndrome, hemoglobin, platelet count, white blood cell count, the percentage of neutrophils, neutrophil/lymphocyte ratio, albumin, creatine kinase, uric acid/creatinine ratio, triglycerides or low density lipoprotein. Higher DRR (> 1.73) was an independent predictor of 1-year mortality in multivariate Cox regression analysis (hazard ratio 3.423, 95% CI 1.481–7.911, p = .004). Patients with higher DRR more often required use of mechanical ventilation and readmission for acute exacerbation of PM/DM-ILD at 1-year follow-up. Higher DRR on admission for PM/DM-ILD patients are associated with increased mortality, risk of mechanical ventilation and hospitalization in 1-year follow-up. This low-cost, easy-to-obtain, rapidly measured biomarker may be useful in the identification of high-risk PM/DM-ILD patients that could benefit from intensive management. diagnosis,


Introduction
Polymyositis (PM) and dermatomyositis (DM) are common branches of connective tissue disease characterized by muscle weakness and skeletal muscle inflammation, which gradually involves other organs, especially lungs. [1] Interstitial lung disease (ILD) is a major complication of PM/DM with a 4 diagnosis of clinically amyopathic dermatomyositis (CADM) met the criteria developed by Sontheimer and colleagues [22]. Patients with acute hepatic failure or if they had suffered acute myocardial infarction which needed heart bypass surgery or percutaneous coronary intervention in 3 months were excluded. In addition, we excluded patients with chronic renal failure, especially dialysis patients, that significantly influence serum homeostasis, and patients with strenuous exercise within a week.

Study design
Patient demographics, laboratory and clinical parameters were collected from the medical record. Patients were evaluated on admission and were followed up for 1 year by monthly outpatient visits in the first 6 months and quarterly in the last 6 months. Routine clinical and laboratory examination were included in those visits. Complete blood count, liver enzymes and muscle enzymes were done every visit and high-resolution computed tomography was done every 3 months. The recording of mechanical ventilation (MV), noninvasive or invasive, was limited to hospital records of West China Hospital and recording of readmission for acute exacerbation of PM/DM-ILD (AEPM/DM-ILD) was limited to records of our hospital and confederate hospitals. Acute exacerbation was defined as worsening of dyspnea within 30 days, new radiographic opacities, the presence of new rash, recurrent muscle weakness or elevated serum levels of muscle enzymes. [23][24][25][26] Survival data were retrieved from the electronic records of any accessible hospital in the southwest of China.

5
The primary end-points were all-cause mortality at 1 year. Secondary outcomes included need for MV and re-hospitalization for AEPM/DM-ILD in 1 year.

Statistical analysis
Summary statistics for normally distributed quantitative variables were expressed as means and standard deviations. For non-normally distributed variables, we use median and interquartile range.
Categorical data were summarized by ratios and percentages. Comparisons between quantitative variables were analyzed by the student's t, Mann-Whitney U-test and Kruskal-Wallis tests based on variable distribution, Bonferroni's multiple comparison tests were used for multiple group comparisons. Categorical variables were analyzed by Chi-squared test or Fisher's Exact test.
Multicollinearity analysis using Spearman's correlation test (r > 0.5) was performed to identify the collinearity between the variables before further analysis.
For the analysis of the objectives, survival analyses and Cox regression analyses were implemented.
In detail, we grouped the patients based on quartile of DRR at admission and used the first quartile as the reference group for all subsequent analyses. The times to death according to the DRR strata was evaluated with Kaplan-Meier survival curves and log-rank tests. We conducted univariate analysis by a univariate Cox regression model to assess the relationship between variables and outcomes. Then, we did a multivariable Cox regression to adjust potential confounders and all variables with P<0.1 in the univariate analysis were subsequently entered into the model as potential predictors. All statistical analyses were performed with the IBM SPSS Statistical version 23.0 (IBM, Armonk, NY, USA SPSS) and graphs were drawn by Graphpad Prism 5.0 (Graphpad Software, La Jolla, CA, USA). A twosided p-value less than 0.05 was considered statistically significant.

AST/ALT ratio (DRR) and associations with clinically relevant outcomes and mortality rates
We further divided patients by DRR to explore the relative mortality risk across the DRR strata (Table 2). Crude all-cause mortality rates within one year across the DRR strata were 6.8%, 8.6%, 10.7% and 23.1% (p < .001), which showed a J-shaped association. Moreover, patients with higher DRR required MV within one year more often (p = .006). Figure 1 shows the cumulative survival of patients with different DRR strata. Apparently, higher DRR was associated with a poorer prognosis than lower DRR strata (stratum 1 VS stratum 4, p < .001; stratum 2 VS stratum 4, p = .001; stratum 3 VS stratum 4, p = .006), while the prognoses of patients with lower DRR strata did not differ. Table 3 shows the unadjusted hazard ratio (HR) for mortality and clinically relevant outcomes across the DRR strata, and Table 3-5 shows the HRs adjusted by different variables across the DRR strata. In general, the unadjusted HRs for readmission, MV and all-cause mortality showed a J-shaped association across the DRR strata, which means high DRR stratum was associated with greater risks for readmission, MV and mortality than low DRR strata. Using the DRR stratum 1 as a reference, the unadjusted HRs (95% CIs) for readmission within one year were 1.00, 1 As shown in Table 4, the adjusted HRs (95% CI) for readmission, MV and mortality also exhibited a Jshaped association across the DRR strata. The adjusted HRs (95% CI) for readmission were 1.00,

Discussion
The present study investigated the association between DRR and mortality. Our analysis demonstrated a significant J-shaped association between DRR and all-cause mortality, with the lowest mortality occurring in individuals with DRR less than 0.91. Additionally, patients with increased DRR more often required use of MV and readmission for AEPM/DM-ILD at 1-year follow-up. To our knowledge, this is the first study that has assessed DRR, a widely available and rapidly measured biomarker, as a predictor of clinically important outcomes in a retrospective cohort of patients with PM/DM-ILD.
Several mechanisms may be involved in the presence of high DRR in PM/DM-ILD. First, prolonged hypoxaemia, caused by exacerbation of diffusing capacity for carbon monoxide (DLco), that is further increased with the development of ILD may result in increased pulmonary artery pressures, leading to increased right ventricle afterload, which promotes hepatocyte injury and AST level increased more.
[27] Secondly, PM/DM is characterized by autoimmune conditions that target muscles to some degrees [28], and elevated serum CK, AST, ALT levels have been associated with increased levels of injury to skeletal muscle. [27,29] The positive association of CK with the DRR in our cohort provides support for this mechanism. What's more, pulmonary infection, mainly bacterial, is the most common cause of acute exacerbation in patients with ILD. In the present study, almost all patients who complained of respiratory symptoms, such as cough and dyspnea, were confirmed suffering varying degree of lung infection in high-resolution computed tomography. Theoretically, serum infectious indicators in these patients, like WBC and neutrophil/lymphocyte ratio (NLR), would increase [30].
However, we found that the difference of WBC between DRR strata was statistically significant: WBC decreased with the increase of DRR, while that of NLR was not, which was a more sensitive marker in infectious diseases. A same trend with WBC was seen in HB as well. There is a possible explanation that patients in higher DRR strata present higher activity of autoimmunity, which attacks blood cells, , succeeded to predict short-term or long-term prognosis in patients with PM/DM-ILD. The fact that high DDR acted as an independent predictor of 1-year mortality in our study suggest that this easy-to-obtain biomarker may be used to identify high-risk patients that require more intensive treatment.
Previous studies have demonstrated higher DDR was associated with mortality in some malignant diseases, such as renal cell carcinoma, primary hepatic carcinoma and upper tract urothelial cancer.
A retrospective study reported significant associations between serum DRR and renal vein invasion, renal capsule infiltration and renal pelvis involvement.
[45] Another retrospective cohort study in 698 patients receiving nephrectomy has suggested that DRR is related to the prognosis of such patients, but the retrospective design may reduce the generalizability of the application of DRR. [17] And a study of 414 patients with primary hepatic carcinoma in China observed significant associations between DRR and mortality rate.
[46] Although all the previously mentioned studies provide evidence that DRR is increased in more severe disease, no study had reported associations between DRR and PM/DM-ILD. In the present study, high DRR was an independent predictor of 1-year mortality in the multivariate Cox regression analysis, which suggests a possible role for this biomarker as a predictor of long-term mortality which needs to be evaluated in larger studies. In the evaluation of AST and ALT levels we need to take into account that these enzymes increase in a very sensitive but nonspecific way in several forms of tissue damage and inflammation, especially in muscle and liver tissues, all of which are very dynamic processes in patients with PM/DM-ILD. [10] AST and ALT are influenced by several factors including cardiovascular disease, food intake, exercise, renal dysfunction and liver diseases. In the present study we have excluded patients with chronic renal failure and acute hepatic failure, and we excluded patients with strenuous exercise within a week, but we cannot exclude other possible confounders that may have influenced our results.
However, despite these possible limitations, we believe that the data from our cohort provide evidence for a possible role of DDR as a biomarker that is associated with disease severity and may identify patients with worse prognosis in hospitalized patients with PM/DM-ILD. A recent study showed that female is associated with poor outcomes in Chinese patients with PM/DM-ILD.
[50] The predominance of female patients in stratum 4 of the present study also suggests that the elevated proportion of female may be associated with increased disease related outcomes in our cohort.
However, the fact that higher DDR continued to be an independent predictor of 1-year mortality, even after adjustment for the presence of female, further supports the possible role of DDR as a clinically relevant biomarker in PM/DM-ILD.
The present study presents some limitations. First, as in all retrospective cohort studies, the 10 retrospective nature of data collection could not be avoided. And approximately a third of patients from the analysis were lost to follow-up within one year, of which 54 patients lost immediately after their discharge from the hospital. We decided to include them since this was an exploratory noninterventional study evaluating the possible role of DRR as a predictor of clinically relevant outcomes in patients with PM/DM-ILD which we wanted to maintain integrity of these data and come to the conclusion as accurate as possible. Secondly, for the same reason, we were not able to collect high-quality data on cause-specific mortality.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.     Collinearity results