The C-reactive protein to albumin ratio is a novel prognostic factor in patients with stable coronary artery disease following percutaneous coronary intervention

These results further support the stable independent association between CAR and poor prognosis of SCAD.

prevent the risk factors contributing to the poor prognosis of SCAD. The root cause of SCAD was identi ed to be atherosclerosis. The pathological mechanism underlying the initiation and progression of atherosclerosis and induction of plaque instability is considered to be caused by in ammation produced by immune cells [2][3]. In previous studies, in ammatory markers such as CRP, hypersensitive CRP, and interleukin-6 have been shown to be reliable long-term prognostic factors for SCAD [4][5][6]. And we noticed that when the body is in an in ammatory state, the rate of Alb synthesis is reduced, which results in a series of malignant clinical events [7]. It is widely acknowledged that Alb is a critical protein throughout the body, involved in the transport of things such as growth hormones or therapeutic drugs to their targets, and the presentation of surface antigens during viral or bacterial infections. Normal Alb levels are therefore essential for the body to maintain health. Recently, others have highlighted that there was also strong correlation between Alb and cardiovascular disease [8,9]. Therefore, both in ammatory cytokines and Alb can be used to predict the prognosis of cardiovascular disease, but is there a better index to replace both CRP and Alb? A new in ammatory indicator: the CRP to Alb ratio (CAR), was found to aid in the assessment of acute coronary syndrome severity, coronary ow status, and could predict the prognostic outcome in ST-elevation myocardial infarction patients, which can better re ect the body's in ammatory status [10][11][12]. However, the prognostic value of CAR in patients with SCAD is unclear.
Therefore, we conducted this secondary data analysis based on previously published data to explore the association between the CAR and poor prognosis of SCAD patients [9]. The ndings from such a study may indicate a convenient and useful marker for the further prognosis risk appraisal of SCAD patients.

Study design and population
We download raw data uploaded by Suzuki and his colleagues [9] from the "DATADRYAD" database (www.datadrya-d.org). According to the Dryad Terms of Service, we could use these data for secondary analysis based on different research hypotheses and quote Dryad packets(Dryad data package: Suzuki et al. (2019), https://doi.org/10.5061/dryad.fn6730j). The study design used here has been described in a previous study [9]. In brief, this was a retrospective cohort study that continuously enrolled patients with newly diagnosed SCAD (except old myocardial infarction) and hospitalized in Shinonoi General Hospital between October 2014 and October 2017. All patients received elective percutaneous coronary intervention (PCI) treatment, excluding patients with malignant tumors. Since ethical approval and informed consent were obtained in previous studies, this study no longer requires ethical research approval. Based on the original data provided by Suzuki et al., we carried out further data collation. First, replace the missing data with NA: aspartate transaminase (AST, n = 1), alanine aminotransferase (ALT, n = 1), hemoglobin A1c (HBA1c, n = 14), triglyceride (TG, n = 4), total cholesterol (TC-Chol, n = 33), highdensity lipoprotein cholesterol (HDL, n = 6), and CRP (n = 8). The items containing CRP or Alb missing data were deleted (n = 8). Then, CAR was calculated as the concentration of CRP divided by the concentration of Alb. The clinical follow-up data was provided by hospital. To explore the prognostic impact of CAR on SCAD patients, we de ned the study endpoint using any MACE that occurred during the follow-up, including all-cause death, non-fatal myocardial infarction, and non-fatal stroke.

Data Collection
Routine blood chemistry tests were conducted after admission, including measurements for hemoglobin (Hb), Alb, estimated glomerular ltration Ra (eGFR), CRP, AST, ALT, HbA1c, TC-Chol, TG, HDL-Chol and lowdensity lipoprotein cholesterol (LDL-Chol). The concentrations of Alb and CRP were determined using a LABOSPECT008 automatic analyzer (Hitachi Ltd, Tokyo, Japan). The measurement range of Alb analysis was 1.0-8.0 g/dl, the coe cient of variation was 5%, the minimum detection limit was 0.1 g/dl, and the lowest detection value of CRP was 0.01 mg/dl. Clinical baseline data and general conditions of patients after admission were recorded by trained doctors; this data included age, sex, body weight, height, systolic blood pressure (SBP), diastolic blood pressure (DBP), left ventricular ejection fraction (LVEF), diabetes mellitus (DM), atrial brillation (AF), hypertension (HT), peripheral artery disease (PAD), old cerebral infarction (OCI), past smoker, drug applications, and types of coronary artery lesions and stents.

De nitions
Coronary angiography and PCI procedures were performed following standard guidelines, and all patients were given aspirin and thienopyridines at least two days before PCI. During the process of PCI, the intervention strategy and stent selection were decided by the operator and the patient after consultation.
The diagnosis of SCAD needed to meet one of the following criteria: (1) coronary angiography con rmed ≥ 90% epicardial coronary artery stenosis. (2) evidence of stress-induced ischemia via any clinical stresstesting modality and (3) exercise-induced chest pain with coronary angiography con rmed coronary artery stenosis ≥ 75%.

Statistical analysis
To better understand the association between CAR and the poor prognosis of patients with SCAD, the population was divided into three equal groups according to the percentage; the continuous variable distributions were expressed as the means and standard deviation, while the categorical variable was expressed as % or n. The Kolmogorov-Smirnov test was used to measure the normal distribution of values. Differences between the mean and proportion of each group were compared, one-way ANOVA or Kruskal-Wallis H test (nonparametric variables) were used for continuous variables, chi-square test was used for categorical variables. Cumulative endpoint event survival curves were determined by the Kaplan-Meier method and event curves of disparate outcomes were compared by using the log-rank test. The Cox proportional hazard regression model analyzed the prognostic value of CAR in SCAD, and the adjusted hazard ratio (HR) to estimate the 95% con dence interval (CI) was used to evaluate the prognostic risk. Before the variables were included in the multivariate regression equation, collinearity between variables was tested using multiple linear regression based on the variance in ation factor (VIF) [13]. Variables with VIF > 5 were considered as showing severe multicollinearity. According to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement, we also display the results of the unadjusted (crude model), minimum adjustment analysis (model I), and full adjustment analysis (model II) [14]. The goodness of t for each model was evaluated by the − 2log likelihood. Model discrimination was evaluated utilizing the concordance index (C index). In order to avoid bias caused by differences in CAR groups, we carried out a sensitivity analysis by treat CAR as a categorical variable and calculated the P for trend. Furthermore, the GAM was used to identify the nonlinear association between CAR and the poor prognosis of patients with SCAD. If a nonlinear correlation was observed, a two-piecewise linear regression model was used to calculate the threshold effect of CAR on the poor prognosis of SCAD according to the smoothing curve, and the in ection point was calculated automatically by the recursive method and the maximum model likelihood was used at the in ection point. Also, we used the Cox proportional hazard model to analyze the consequences of each subgroup (age, BMI, Hb, LDL-Chol, and PAD). For continuous variables, we converted them into categorical variables according to the clinical entry point, and each strati cation was adjusted according to all factors except the strati cation factor itself. Likelihood ratio tests inspected the interactions of subgroups. R package V.3.4.3 (http://www.r-project.org; 2020-02-03 The R Foundation), Empower (R) (www.empowerstats.com; X&Y Solutions Inc.) and SPSS (V.23.0) were used to perform all statistical analyses. P < 0.05 was the criterion for signi cance.

Baseline features and population studied
Selected patients (204 patients) were diagnosed with SCAD for the rst time (mean age, 72 years; male, 69%). The CRP and Alb concentrations of all patients were assessed at admission between October 2014 and October 2017. We divided patients into three equal groups according to the CAR. Eight patients without CAR noted were excluded from the study. Of the 196 remaining patients enrolled in the study, 66 patients were placed into the high-level CAR group, 65 patients were placed in the low-level CAR group, and 65 were placed in the middle-level CAR group. Table 1 summarizes the baseline clinical characteristics of the cohort. We found that in the group with the highest CAR, participants were generally older and had higher CRP concentrations but lower levels of eGFR, LVEF, HbAlc, and HDL-Chol. Moreover, there were no signi cant differences in medication history, coronary artery disease types, or concomitant diseases among these diverse groups. All participants were followed up for an average of 793.75 ± 430.26 days. During the follow-up period, 18 people died. Among them, 6 died of cardiovascular causes, 3 had a non-fatal myocardial infarction, and 11 had stroke events. Additionally, in order to assess the power of our study, we conducted post-hoc power analysis. Using GPower version 3.1.9.7, we showed that a sample size of 196 respondents yielded a power of 100%, an effect size (f) of 44.295, and a signi cance level of 0.05 (two-sided).   Table 2. Univariate analysis showed that age, BMI, Hb, LDL-Chol, CAR, PAD, and warfarin use were positively correlated with the poor prognosis of SCAD. Then the signi cant variables (P < 0.05) in univariate analysis and non-collinear variables (except collinear variables body weight and TC-Chol) were incorporated into the multivariate regression model. In this study, the Cox proportional hazard regression model was used to evaluate the association between CAR and the poor prognosis of SCAD. We evaluated the adjusted and unadjusted models separately (Table 3). In the crude model, there was a positive correlation between CAR and the poor prognosis of SCAD (HR = 1.03, 95% CI:1.01-1.06, P < 0.001), while the result of the minimum adjusting model (model I, adjusted for age, BMI, PAD) showed no signi cant changes compared to crude model (HR:1.03, 95% CI:1.01-1.05, P = 0.018). And after adjusting the full model (model II, adjusted for age, BMI, Hb, LDL-Chol, PAD, and warfarin), we could still see a positive correlation between them (HR = 1.03, 95% CI:1.01-1.06, P = 0.02). For the sensitivity analysis, we disposed of CAR as a categorical variable. In model II, we found that patients with a high CAR level increased the MACE risk by 2.76 times compared with a low CAR level (P for trend = 0.024). And in Kaplan-Meier analysis (Fig. 1), the long-term MACE-free survival rate of patients with high CAR levels decreased signi cantly (P = 0.015). Furthermore, the goodness of t statistics of the two models were 235.158 and 216.411; respectively, the model II had the best model t. And the C index for the two models were 0.751 and 0.79 (Table 4), which means that model I and model II predict each outcome fairly well.

Analysis of nonlinear relationships between CAR and the poor prognosis of SCAD
In this present research, CAR was a continuous variable; we also used the GAM to identify the nonlinear relationship between CAR and the poor prognosis of SCAD. Figure 2 represents a curvilinear relationship (adjusted for age, BMI, Hb, LDL-Chol, PAD, and warfarin) between the MACE risk of CAR and SCAD. The in ection point of CAR calculated by the two-piecewise linear regression model was 3.933 (log-likelihood ratio test P = 0.02). As we anticipated, there was a positive correlation between CAR on the left side of the in ection point (CAR ≤ 3.933) and poor prognosis of SCAD (HR:1.48, 95% CI:1.10-1.99, P = 0.009). On the right hand of the in ection point (CAR > 3.933), the positive correlation between the two tends to be saturated (HR:1.01, 95% CI:0.97-1.05, P = 0.64; Table 5).

Subgroup Analyses
We further explored other risks that could contribute to the association between CAR and MACE in patients with SCAD, and estimated the factors that might affect the results by using subgroup analysis.
Here, we used age, BMI, Hb, LDL-Chol, and PAD as strati ed variables; each strati cation was adjusted according to all the strati cation factors, except for the strati cation factor itself. The Cox proportional hazard model was used to explore the robustness of the results of each subgroup, and the effects of these variables were observed (Table 6). Through the likelihood ratio test of the interaction between subgroups, we noted no statistical signi cance for any strati ed variables (P for interaction > 0.05). These results further support the stable independent association between CAR and poor prognosis of SCAD.

Discussion
In this work, we retrospectively analyzed the prognostic ability of CAR in 196 eligible SCAD patients. As far as we know, this was the rst study to investigate the association between CAR and MACE risk in patients with SCAD. Our results show that there was an independent positive correlation between CAR and the poor prognosis of patients with SCAD (HR = 1.03,95%CI:1.01-1.06, P = 0.02, P for trend = 0.024). This association was independent of other risk factors, and there was a nonlinear relationship. We used a two-piecewise linear regression model to calculate that the in ection point of CAR was 3.933 (loglikelihood ratio test P = 0.02).

Page 13/19
CRP is a clinically used in ammatory marker to assess the in ammatory state of disease. In cardiovascular disease, CRP stimulates in ammatory cells to promote atherosclerotic plaque formation. This is achieved by increasing the expression of endothelial adhesion molecules and inhibiting production of endothelial nitric oxide, which can result in the instability of atherosclerotic plaques [15].
CRP also has a similar effect in cerebrovascular diseases and high CRP levels often indicate poor prognosis [16]. Alb is often used to evaluate the nutritional status of the body [17]. Previous studies found that Alb concentrations were linked to the poor prognosis of patients with SCAD [9]. In these instances, low Alb levels could be related to poor syntax score and lower prognostic nutritional index [18,19].
In recent years, CAR has attracted much attention as a novel predictor of coronary artery disease severity and prognosis [10][11][12]18]. We hypothesized that CAR might have prognostic value for predicting the prognosis of patients with SCAD. What is exciting is that our research con rmed that there was an independent positive correlation between CAR and the poor prognosis of patients with SCAD (HR = 1.03, 95%CI:1.01-1.06, P = 0.02, P for trend = 0.024). Also, CAR was more objectively determined than traditional prognostic factors such as sex, age, BMI, etc., and is an easily quanti able and inexpensive prognostic factor for patients with SCAD. Effectively controlling the occurrence and development of in ammation is essential in the prevention and treatment of cardiovascular disease, but a deeper understanding of the pathogenesis of the cardiovascular disease is necessary. In the article "In ammation and Atherosclerosis: The End of a Controversy" written by Dr. Hansson GK [20], it was also mentioned that more attention should be paid to the preventive effect of anti-in ammatory chemicals on cardiovascular diseases and the clinical research of in ammation and atherosclerosis. Thankfully, several targeted anti-in ammatory drugs are currently being tested [21,22]. The results have suggested that reducing in ammation could reduce the incidence of recurrent cardiovascular events, but their longterm e cacy needs further observation. In addition, many randomized controlled trials have shown that statins not only reduce lipid concentrations but they also have anti-in ammatory effects. This reduction in in ammation can effectively improve the clinical outcome of patients with coronary artery disease (CAD) [23][24][25]. Combined with our research and practice, we suggest that patients diagnosed with SCAD should strictly carry out secondary prevention of CAD. In particular, the use of statins should be adjusted to an appropriate dose according to CAR, blood lipid levels, and other biochemical indicators. In order to reduce the risk of residual MACE, CAR indicators should be regularly monitored. We will continue to expand the scope of our research and carry out multi-center research to con rm our ndings further to identify potential mechanisms and associations.

Limitations
Although we thought our ndings were valid, our research still had some limitations. First, our cases were mainly from a single medical center, which will limit the external applicability of the results. Second, we only measured the CRP and Alb at admission and doesn't consider dynamic changes in these levels over time. Third, although transaminase levels were normal in some people in our cohort, the study did not rule outpatients with liver failure. Finally, this study was a secondary analysis based on a previous study and the data had been screened by Suzuki et al. [9], due to the raw data limitation, we were unable to obtain details about the patients treated with PCI. Moreover, the differences in the mode of operation, the operator's pro ciency, X-ray exposure time, and other factors cannot be evaluated, which may lead to some information bias. We will further explore the association of CAR with the poor prognosis of patients with SCAD by collecting our data in the future.

Conclusion
Our study showed that there was an independent positive correlation between CAR and the poor prognosis of patients with SCAD. The measurement of CAR at admission might be bene cial for early strati cation and intervention to prevent MACE.  Figure 1 Kaplan-Meier analysis for cumulative MACE-free survival with low CAR level, meddle CAR level and high CAR level.