Abstract
Background. To evaluate the prognostic value of preoperative activated partial thromboplastin time (APTT) in patients who underwent coronary artery bypass grafting (CABG). Methods. All data were extracted from the Medical Information Mart for Intensive Care III (MIMIC-III) database. The study population was divided to two groups according to the optimal cut-off value of APTT calculated by X-tile software, and Cox proportional hazard model was used to define independent effect of APTT on 4-year mortality. Survival curves were estimated by the Kaplan-Meier method, and the area under the receiver-operating characteristic curve (AUC) was calculated to compare APTT with other severity scores. Propensity score matching (PSM) analysis were applied to ensure the robustness of this study. Results. A total of 2,706 patients were included. The optimal cut-off value of APTT for 4-year mortality was 44 seconds. The Cox proportional hazard model showed that patients with had a significantly higher risk of all-cause death than those with both before (HR (95% CI), 1.42 (1.16-1.74), ) and after PSM (HR (95% CI), 1.47 (1.14-1.89), ). The survival curves showed that patients with longer APTT had a significantly lower 1-year and 4-year cumulative survival probability. The ROC of APTT combined with other severity scores significantly increased predictive ability for 1-year and 4-year mortality. Conclusions. A longer APTT (≥44) was associated with a higher risk of mortality and can serve as a prognostic predictor in CABG patients.
1. Introduction
Coronary artery bypass grafting (CABG) is a common procedure in cardiac surgery and a gold standard intervention in cases of severe multivessel coronary artery disease [1, 2]. For cardiac surgery, especially CABG with cardiopulmonary bypass (CPB), postoperative bleeding remains a significant source of morbidity and mortality for patients [3]. Anticoagulation therapy is involved before, during and after CABG surgery [4]. Therefore, coagulation function needs to be focused on the CABG procedure. As we known, the coagulation process is complex as a process involving multiple factors and multiple pathways [5]. Although the majority of cardiac surgical patients have no clinical evidence of bleeding diathesis, a substantial proportion may have subtle bleeding tendencies that manifest only after exposure to these hemostatically damaging effects of CPB [6]. A preoperative blood test that could accurately predict those patients who will bleed excessively after CPB would be of great practical value. Tests used for routine evaluation of the coagulation system are activated partial thromboplastin time (APTT) and international normalized ratio (INR) [7]. Previous studies have shown that early APTT is a predictor of 30-day and 1-year mortality in ST-elevation myocardial infarction patients treated with percutaneous coronary intervention and unfractionated heparin [8]. For trauma patients, APTT at admission was also a predictor of 1-year mortality [9, 10]. Therefore, we wanted to know whether the preoperative APTT as an appropriate indicator would be of prognostic value for CABG. So, using the open-source Multiparameter Intelligent Monitoring in Intensive Care III (MIMIC-III) database, we performed a retrospective study aiming to explore the impact of preoperative APTT on the prognosis of CABG-related surgery.
2. Methods
2.1. Database
The study data was extracted from a publicly available database, the Medical Information Mart for Intensive Care III (MIMIC-III) [11], comprising comprehensive and anonymous data of patients admitted to ICU of the Beth Israel Deaconess Medical Center from 2001 and 2012. Thus, the informed consent was waived by the institutional review boards (IRB). One of our authors was approved and authorized to utilize this database (Record ID: 37650993) by IRB of the Massachusetts Institute of Technology (MIT) and Beth Israel Deaconess Medical Center (BIDMC). The present study complied with the corresponding guidelines.
2.2. Patient Selection and Outcome
Patients in MIMIC-III who underwent CABG during this admission were collected according to ICD-9 code. Those who were younger than 18 years old or older than 89 years old, and those who were followed up for less than four years were excluded. The primary endpoint of this study was 4-year mortality, and the secondary endpoint was 1-year mortality.
2.3. Data Extraction
We applied pgAdmin4 based on PostgreSQL 9.6 for data management and the Structured Query Language (SQL) for data extraction. We collected the following data: baseline demographic information such as age, gender, and ethnicity; severity scores including Sequential Organ Failure Assessment (SOFA) and Simplified Acute Physical Score II (SAPS II); comorbidities including hypertension, diabetes mellitus (DM), peripheral vascular disease, myocardial infarction (MI), congestive heart failure (CHF), chronic pulmonary disease, renal failure, liver disease, and obesity. Vital signs within 24 h after ICU admission including mean systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR), respiratory rate, temperature, and percutaneous oxygen saturation (SpO2) were used. The initial values of laboratory tests after ICU admission including white blood cell (WBC), hemoglobin, platelet, sodium, potassium, creatinine, glucose, and APTT were extracted. Treatments including mechanical ventilation, continuous renal replacement treatment (CRRT), and vasopressor use were also selected for analysis. As the proportion of missing values of the collected variables was less than 1%, samples with missing values were discarded in further analysis.
2.4. Statistical Analysis
The study population was divided into two groups according to the optimal cut-off value of APTT for 4-year mortality calculated by X-tile software. Data were summarized as medians [interquartile ranges (IQRs)] for continuous variables and number with percentages for categorical variables. Data were compared using the Mann–Whitney test for continuous variables and Pearson’s test or Fisher’s exact test for categorical variables appropriately.
Propensity score matching (PSM) analysis was applied to ensure the robustness of the present study. The logistic regression model was used to calculate the propensity score, in which the predefined variables included demographic information (age, gender, and ethnicity), and all variables that were statistically different at baseline (hypertension, DM, peripheral vascular disease, MI, CHF, chronic pulmonary disease, renal failure, obesity, SBP, HR, temperature, WBC, platelet, potassium, creatinine, and CRRT). Meanwhile, 1 : 1 nearest neighbor matching method was used, and the caliper width value was set as 0.02 in this study. The distribution of propensity scores for the two groups before and after matching were depicted to show common support domains, and histograms for absolute standardized differences for baseline variables before and after matching were depicted to indicate a balance.
The Kaplan-Meier curves were depicted to determine whether APTT could affect 1-year and 4-year mortality and compared by log-rank tests. Univariate and multivariable Cox proportional hazard models were used to define independent effect of higher APTT on 4-year mortality in CABG patients. Model I was adjusted for gender and age, while model II was adjusted for the variables with in univariate Cox regression analysis. Receiver-operating characteristic (ROC) curves were depicted, and the area under the curve (AUC) was calculated to compare APTT with other severity scores. Subgroup analysis were also applied to ensure the stability of our findings in diverse subgroups, and interaction analysis were performed. All above analysis were performed using R version 4.0.3 and a two-side was considered significant.
3. Results
3.1. Baseline Characteristics before and after PSM
After the application of selection criteria, 2706 eligible patients were included in our study cohort. The study sample was divided into two groups according to the result calculated by X-tile software: group I (, ) and group II (, ) (Supplement File 1). After propensity-score matching, a total of 640 patients with lower APTT were matched with 640 patients with higher APTT. The distribution of propensity scores for the two groups and the histograms for absolute standardized differences for baseline variables before and after matching indicated a good balance (Supplement Files 2 and 3). Before PSM, the baseline characteristics and significant differences of two groups were summarized in Table 1. Overall, the median age of the study patients was 68.8 (60.0-76.4) years, and approximately 26.5% of them were female. Patients with high APTT tended to be older and female ( values < 0.05). They had the higher prevalence of hypertension, DM, peripheral vascular disease, MI, CHF, chronic pulmonary disease, renal failure, and obesity (all values < 0.05). They may have the higher values of SBP, HR, temperature, WBC, platelet, potassium, creatinine, SOFA, and SAPS II (all values < 0.05). Furthermore, patients with high APTT were more likely to receive CRRT (). After PSM, the differences of variables mentioned above were balanced (Table 2).
3.2. Outcome of Patients before and after PSM
Before PSM, compared with , patients with higher APTT had longer ICU LOS and longer in-hospital mortality, 30-day mortality, 90-day mortality, 1-year mortality, and 4-year mortality (all values < 0.001) (Table 3). After PSM, the similar significant differences were found in matched patients (all values < 0.05) (Table 3). The results before and after matching showed that the patients with higher APTT had longer ICU LOS and higher mortality.
Kaplan-Meier analysis indicated higher mortality risk in high APTT group. Kaplan-Meier curves were used to evaluate the association between APTT level and long-term all-cause mortality. As shown in Figures 1(a) and 1(b), the survival curves showed that patients with had a significantly lower 1-year (log-rank test: ) and 4-year (log-rank test: ) cumulative survival probability compared to patients with . Additionally, after PSM, the survival curves (Figures 2(a) and 2(b)) indicated that the higher APTT values were still significantly associated with lower cumulative survival probability of 1 year (log-rank test: ) and 4 years (log-rank test: ).
(a)
(b)
(a)
(b)
3.3. Cox Regression Analysis Indicated Higher Mortality Risk in High APTT Group
We fitted two Cox regression models to demonstrate the independent effects of APTT on 4-year outcome. Model 1 was adjusted for gender and age, while model 2 was adjusted for the variables with in univariate Cox regression analysis. As shown in Table 4, compared to patients with , patients with had higher risk of 4-year all-cause death (model 1: HR (95% CI), 1.81 (1.49-2.21); ; model 2: HR (95% CI), 1.42 (1.16-1.74); ). After further analysis of matched cohort, a longer APTT was still regarded as an independent risk factor for 4-year all-cause mortality (model 1: HR (95% CI), 1.52 (1.19-1.95); ; model 2: HR (95% CI), 1.47 (1.14-1.89); ).
3.4. Ability of APTT to Predict 1-Year and 4-Year Mortality
Receiver-operating characteristic (ROC) curves were depicted, and the area under the curve (AUC) was calculated to compare APTT with other severity scores. The AUCs of APTT for 1-year and 4-year mortality were only 0.673, and 0.628, respectively (Figures 3(a) and 3(b)). For 1-year mortality, the predictive ability of APTT combined with SAPS II (AUC: 0.736) was superior to SAPS II (AUC: 0.704) alone, and the significant difference was found between two groups (DeLong’s test: ). Meanwhile, the predictive power of APTT combined with SOFA (AUC: 0.701) was superior to SOFA (AUC: 0.643) alone with significant difference (DeLong’s test: ). The similar promotion of predictive ability was found for 4-year mortality, the AUC of APTT combined with SAPS II increased to 0.701 compared to SAPS II alone (ACU: 0.688; DeLong’s test: ), and the AUC of APTT combined with SOFA elevated to 0.653 compared to SOFA alone (ACU: 0.620; DeLong’s test: ). The results indicated that the combination of APTT and traditional severity score had a good predictive value for 1-year and 4-year mortality, respectively.
(a)
(b)
3.5. Subgroup Analysis
Subgroup analysis was applied to ensure the stability of our findings in diverse subgroups. The patients with had a higher risk of 4-year death compared to those with in most subgroups except for the patients who had liver disease (HR (95% CI), 3.35 (0.98-11.5); ), underwent CRRT (HR (95% CI), 1.27 (0.66-2.44); ), never used mechanical ventilation (HR (95% CI), 2.29 (0.96-5.44); ) and vasopressor (HR (95% CI), 1.65 (0.95-2.87); ) (Table 5). There was only significant interaction in CHF subgroup.
4. Discussion
This study explored the association between preoperative APTT and mortality among patients who underwent CABG with a 4-year follow-up. The results showed that a preoperative APTT longer than 44 seconds was a reliable predictor of 1-year and 4-year mortality. We observed for the first time the value of APTT in predicting mortality of cardiac surgery patients.
APTT, with the highest sensitivity, reflects the integrity of the intrinsic pathway of coagulation and is an index to demonstrate coagulation factor deficiency, especially in preoperative routine coagulation screening [12]. APTT is widely used to monitor the anticoagulant effect of intravenous heparin applications [13]. Therefore, for patients with CABG who use heparin for coronary heart disease before surgery, APTT has its natural advantages in evaluating preoperative coagulation function. In the present study, it was obviously observed that prolonged APTT increased the short-term and long-term mortality. Moreover, Cox-regression analysis demonstrated that APTT still showed good independent predictive value. Therefore, we believed that preoperative APTT had an important reference value for the prognosis of CABG patients.
There is no clear contraindication standard for the clinical coagulation index at present, which needs to be determined by comprehensively considering the severity of the disease, surgical model, and prognostic judgment [7]. For CABG surgery, the situation is much more complicated due to the application of preoperative anticoagulant drugs. Previous study showed that preoperative APTT was greater than 40 seconds in the group of severe bleeding after CABG [14]. Our study showed that APTT greater than 44 seconds was associated with 4-year mortality in CABG patients. Therefore, further studies are needed to determine the proper APTT value for predicting the prognosis of CABG patients. Although this APTT value could vary in different database, it could be used as a reference for prognostic analysis.
For subgroup analysis, APTT maintained its predictive capability regardless of age, gender, and most of the comorbidities. There were some results that drew our attention. Patients with obesity had a 6.17-fold higher risk of 4-year mortality with an , while patients without obesity had only a 2.2-fold higher risk of 4-year mortality. Obesity is a recognized risk factor for thrombosis. Obesity makes the body in a high coagulant state affect the number of platelets and coagulation factor activity and damage the primary and secondary hemostasis ability [15–17]. If combined with prolonged APTT, the risk of bleeding increased, which could increase the risk of mortality. Though the value was 0.064, which was not significant, it might be caused by the small number of people after grouping.
And we observed that a longer APTT was more associated with a bad prognosis in patients without CHF compared with patients with CHF. As such a result we thought it was owing to CHF which is an important risk factor for the prognosis of CABG [1]. CHF was one of the major adverse cardiovascular events (MACE), so in the presence of CHF the prognostic efficacy of other factors was weakened. But in the absence of CHF, a longer APTT still showed a better predictive value, for which it was reasonable to include APTT in the prediction models. Moreover, the results of ROC curve analysis showed that APTT significantly increased the AUCs when it was added to the SOFA score and SAPS II. This once again demonstrated the important value of APTT for the prognostic evaluation of CABG.
Some limitation to this study included the following: (1) the follow-up outcome could be affected by some cofounders, regarding the severity of the disease and the operative procedure; however, this database analysis was retrospective cohort study, and these situations were not recorded; (2) the study population was US adults based on MIMIC III database; thus, our results might be not applicable to different race; and (3) because this database analysis was a single center research and the sample size was relatively small, the multicenter prospective research is necessary to verify our conclusions.
5. Conclusions
In conclusion, we put forward that a longer APTT (≥44) was associated with a higher risk of mortality and can serve as a prognostic predictor in CABG patients. Studies of large multicenter populations are necessary for further validation.
Data Availability
The data used in the study was extracted from MIMIC III database.
Ethical Approval
The data involving participants was approved and authorized by the institutional review boards (IRB) of the Massachusetts Institute of Technology (MIT) and Beth Israel Deaconess Medical Center (BIDMC). And the informed consent was waived by IRB according with the corresponding institutional requirements.
Conflicts of Interest
There are no conflicts of interests of any of the authors.
Authors’ Contributions
Zhang H has full access to the data in the study and takes responsibility for the accuracy of the data analysis. All authors are responsible for the study design; Zhang H for the statistical analysis of the data; Zhang H and Wei X for the drafting of the manuscript; Wei X for the revision of the manuscript; all authors for the material and method support; and Wei X for the study supervision.
Acknowledgments
This work was supported by the Liaoning Provincial Department of Education Annual Scientific Research Funding Project in 2019 (no. ZF2019014).
Supplementary Materials
Supplementary 1. Supplement file 1: the visual output of X-tile software for the optimal cut-off value of APTT (44 seconds) for 4-year mortality.
Supplementary 2. Supplement file 2: the distribution of propensity scores for the two groups before and after matching.
Supplementary 3. Supplement file 3: the histograms of propensity scores for the two groups before and after matching.