The liver has dual roles in governing the physiological coagulation cascade. Patients with liver cirrhosis have impaired synthesis of coagulation factors and reduced platelet function that can secondarily affect hemostasis, anticoagulation, and fibrinolysis. It is commonly observed that patients with cirrhosis have prolonged coagulation time and some considered these patients as “naturally-anticoagulated.” However, the decreased levels of procoagulants accompany the diminished levels of natural anticoagulants, and therefore patients with cirrhosis are also at risk for thrombotic complications, including portal vein thrombosis and venous thromboembolism.[1,2] In addition, the altered hemostasis, coagulation, and fibrinolysis in cirrhotic patients resulted in “shifted hemostasis” which also leads to increased bleeding. The use of anticoagulants is thus controversial in patients with liver cirrhosis for whom ischemic heart disease is encountered.
Patients with acute myocardial infarction (AMI) have high mortality and require timely coronary intervention and meticulous medical treatment. Although medical therapy for AMI has been advancing at fast pace in recent years, in patients with liver cirrhosis with concomitant AMI, the management can be complex. In a recent nationwide study in the United States, ST elevation myocardial infarction mortality in patients with cirrhosis is higher compared with patients without cirrhosis. However, this mortality difference declined from 1999 to 2009, likely because of higher coronary artery stent utilization for patients with cirrhosis. In the studies of medication use in cirrhotic patients with AMI in Asia, mortality were related to liver complications in cirrhotic patients, and the combined cardiac and liver complications in cirrhotic patients with AMI.[6,7] Therefore in this study, we aimed to investigate the clinical outcome of AMI in patients with liver cirrhosis in Asia.
2.1 Data source
Taiwan's National Health Institute (NHI) Program started in 1995 and provides 99.5% coverage for the 23 million residents in Taiwan. The NHI Research Database (NHIRD) provides all dates of inpatient and outpatient services, diagnosis, prescriptions, examinations, operations, and expenditures, and data are updated biannually. With over 95% of Taiwan's population consists of Han Chinese, our study is considered to be uniform in ethnic background. The NHI system offers detailed follow-up information on medication, intervention, admission, outpatient clinic, and emergency visit of patients. In addition, accurate records of health reimbursement are ensured by prescription of medications and arrangement of interventions being followed by appropriate examinations and indications. On the other hand, false reimbursement claims can result in magnified penalty. The Institutional Review Board of Chang Gung Memorial Hospital Linkou Branch approved this study (IRB No. 201800177B1).
2.2 Study patients
By searching electronic medical records from the NHIRD between January 1, 2001 and December 31, 2013, we retrieved patients with a principal diagnosis of AMI admission. The date of discharge from the index admission was defined as the index date. We excluded patients who had missing information, patients <20 years old, had previous AMI, had previous coronary intervention, and had liver transplant. We further separated patients into cirrhotic patients and non-cirrhotic patients, using the diagnosis of cirrhosis being 2 consecutive outpatient diagnoses or 1 inpatient diagnosis before the index date. Both the diagnoses of AMI and liver cirrhosis in NHIRD have been validated against hospital electronic medical records in previous studies.[8,9]
2.3 Covariate and study outcomes
Diseases were detected using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes. Covariates included age, sex, clinical medical history of hypertension, diabetes mellitus, hyperlipidemia, heart failure, peripheral arterial disease, atrial fibrillation, history of stroke, chronic kidney disease, end-stage renal disease, and malignancy at baseline. The comorbidity was defined as having 2 outpatient diagnoses or 1 inpatient diagnosis in the previous year. Most diagnostic codes of these comorbidities have been validated in previous NHIRD studies.[8,10,11] Liver cirrhosis or AMI related complications were also defined according to the diagnosis of ICD-9-CM (Supplement Table 1, http://links.lww.com/MD/D963).
In-hospital outcomes were analyzed and detected by ICD-9-CM code or Taiwan NHI order code. Outcomes of primary interest included all-cause mortality, recurrent myocardial infarction (MI), major cardiac and cerebrovascular events (MACCE, including recurrent myocardial infarction, revascularization such as percutaneous coronary intervention [PCI] and coronary artery bypass grafting [CABG], ischemic stroke and heart failure, and cardiovascular death), and liver outcomes (any liver outcome, hepatic encephalopathy, ascites tapping, spontaneous peritonitis, esophageal varices bleeding, and hypoalbuminemia), major bleeding. All-cause mortality was defined by withdrawal from the national health insurance. The diagnostic codes of heart failure and stroke have also been validated.[11,13,14] The liver-related outcomes were also reported in our previous work. Each patient was followed until the day of outcome occurrence, death or December 31, 2013, whichever came first.
2.4 Statistical analysis
To reduce bias due to confounding, we performed a propensity score matching (PSM) with a 1:2 ratio which each patient in the cirrhosis group was matched with 2 counterparts in the non-cirrhosis group. The covariates to calculate propensity score (the predicted probability to be cirrhosis group derived from logistic regression) included demographics (age and sex), monthly income, urbanization level, 10 comorbidities, hospital level, coronary intervention at the index admission, stent type in patients who received stent and the index date (listed in Table 1). The matching was processed using a greedy nearest neighbor algorithm with a caliper of 0.2. The quality of matching was checked using the absolute standardized mean difference (ASMD) between the groups after matching which a value <0.1 was considered to have negligible difference between groups.
As to in-hospital outcome, we compared the risk of binary outcome (i.e., in-hospital mortality) and continuous outcome (i.e., ICU duration days) by using generalized estimating equation (GEE) in which the correlation among patients within the same matching pair was considered. We compared the risk of all-cause mortality between groups using a Cox proportional hazard model. The risk of other time to event outcomes between groups was compared using a subdistribution hazard model which considered death during the follow up as a competing risk. Matching pairs were stratified in both Cox and subdistribution hazard models. We generated the plot of cumulative incidence rate using subdistribution cumulative incidence function for time to event outcomes except all-cause mortality (i.e., major composite liver outcome). Unadjusted cumulative event rate of all-cause mortality was calculated and plotted. The study group (cirrhosis vs non-cirrhosis) was the only explanatory variable in GEE, Cox, and subdistribution hazard models. A P value <.05 was considered to be statistically significant and no adjustment of multiple testing (multiplicity) was made in this study. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC), including procedure of “psmatch” for propensity score matching, “genmod” for generalized estimating equation, “phreg” for survival analysis, and the macro of “%cif” for cumulative incidence function.
3.1 Study population
A total of 150,887 patients admitted to hospital with a principal diagnosis of AMI during 2001 and 2013 were identified in the NHIRD. After exclusion criteria, a total of 132,610 patients with AMI were eligible for analysis. Among these, 3217 had history of cirrhosis and 129,393 had no history of cirrhosis. Using 1:2 propensity score matching, there were 3217 patients in the liver cirrhosis group and 6434 patients in non-cirrhosis group analyzed (Fig. 1), with a mean follow up of 2.8 ± 3.3 years. Baseline clinical characteristics of study patients before and after matching were given in Table 1. After matching, there were no significant differences between cirrhosis group and non-cirrhosis group in all clinical characteristics of age, sex, income level, urbanization level, comorbidity, hospital level, coronary intervention at index admission, and patients who received stent. In patients with cirrhosis, baseline clinical variable are shown in Table 2, using previously defined terms on complication of cirrhosis.
3.2 Epidemiology of AMI in cirrhosis
During 2001 to 2013, there was an increasing number of patients with AMI (P trend <.001). During the same period of time, there were also slightly increased cirrhotic patients with AMI (P trend = .004). For both cirrhotic patients with AMI and non-cirrhotic patients with AMI, in-hospital mortality decreased during this period of time (P trend <.001 and P trend <.001, respectively) (Fig. 2).
3.3 In-hospital outcome
There were significantly lower number of ventricular tachycardia (VT)/ventricular fibrillation (VF) and the use of intra-aortic balloon pump in cirrhotic patients with AMI compared with non-cirrhotic patients with AMI. On the other hand, there was significantly higher number of patients with GI bleeding requiring intravenous proton pump inhibitor in cirrhotic patients with AMI compared with non-cirrhotic patients with AMI (Table 3).
3.4 One-year clinical outcome
In cirrhotic patients with AMI, subsequent coronary and cerebrovascular events were lower in comparison to non-cirrhotic patients, with higher all-cause mortality observed from adverse liver related outcomes and bleeding (Fig. 3A). In terms of cardiovascular events, there were significantly lower cumulative incidence of both recurrent MI and MACCE in cirrhotic patients with AMI compared with non-cirrhotic patients with AMI (hazard ratio [HR] = 0.82, confidence interval [CI] = 0.71–0.94, P = .006 and HR = 0.86, 95% CI = 0.79–0.92, P < .001, respectively) (Fig. 3B and C). However, there was significantly higher cumulative incidence of liver related outcome in cirrhotic patients with AMI compared with non-cirrhotic patients with AMI (HR = 2.27, 95% CI = 2.06–2.51, P < .001) (Fig. 3D). Therefore, there was significantly higher all-cause mortality in cirrhotic patients with AMI compared with non-cirrhotic patients with AMI (HR = 1.30, 95% CI = 1.23–1.38, P < .001).
Our study has the following major findings. The mortality rate of cirrhotic patients with AMI is improving and is comparable non-cirrhotic patients with AMI between 2001 and 2013. This is the first and largest study to directly compare the clinical outcome of cirrhotic versus non-cirrhotic patients with AMI using propensity score matching. Compared with non-cirrhotic patients with AMI, cirrhotic patients with AMI had lower rates of VT/VF and intra-aortic balloon pump (IABP) use during in-hospital course, and lower rates of recurrent MI and MACCE during 1-year follow up.
In patients with liver cirrhosis, the frequently encountered clinical scenarios are bleeding and thrombosis. The “shifted balance” of coagulation system in cirrhotic patients however, could not provide us with a definitive direction how this altered hemostasis would behave during the event of AMI. Early autopsy reports have noted lower prevalence of MI in cirrhotic patients compared with the general population.[21,22] In clinical studies, the prevalence of MI was also significantly lower in cirrhotic patients compared with non-cirrhotic patients (1.7% vs 6.4%). In a recent study using registry-based matched case–control study, prevalence of subclinical but obstructive coronary artery disease was neither higher nor lower in cirrhotic patients although the prevalence of the more benign non-obstructive coronary artery disease was higher. Therefore in cirrhotic patients with AMI, we are concerned if the coagulopathy secondary to cirrhosis would modify the clinical course of atherosclerotic and atherothrombotic complications during and post-AMI, which are recurrent MI, and a composite of recurrent MI, revascularization (PCI/CABG), ischemic stroke, and heart failure that may be summarized as MACCE.
In previous study of ST elevation myocardial infarction (STEMI) in acute care hospitals across United States, STEMI mortality in patients with cirrhosis was higher compared with patients without cirrhosis. The age-adjusted mortality rates were 31% for cirrhotic and 11% for non-cirrhotic patients with AMI in 1999 that improved over years to 17% for cirrhotic and 9% with non-cirrhotic patients with AMI in 2009. The adjusted mortality odds ratio 2.54 in 1999 and 1.45 in 2009 probably related to higher coronary artery stent utilization for patients with cirrhosis. In our study, the propensity score matched subjects had mortality rates of 17.2% in cirrhotic and 16.6% rates in non-cirrhotic patients with AMI over entire study period 2001 to 2013. This may be related to our study population being elder with mean age 68.6 ± 13.2 in the cirrhotic and 69.0 ± 14.0 in the non-cirrhotic patients with AMI, compared with the US study population with median age 66.7 in cirrhotic and 67.1 in non-cirrhotic patients with AMI in 1999, and 62.3 in cirrhotic and 62.0 in non-cirrhotic patients with AMI in 2009.
In this study, the 13-year annual rate of AMI has been increasing at a steady pace from 2001 to 2013. During the same period of time, number of cirrhotic patients with AMI seemed only very slightly increased, translating into that liver cirrhosis may have certain inherent protection against ischemic heart disease. With improved medical treatment options and coronary intervention strategies, cirrhotic patients with AMI have improved in-hospital mortality over time and is comparable to non-cirrhotic patients with AMI. A closer look at in-hospital events showed a significantly lower rates of VT/VF and IABP use in cirrhotic patients with AMI compared with non-cirrhotic patients with AMI, reflecting possibly decreased severity of coronary artery obstruction in these patients. Understandably, the “shifted hemostasis” as mentioned earlier, also predisposed cirrhotic patients with AMI with significantly higher rate of GI bleeding leading to intravenous use of proton pump inhibitor, compared with non-cirrhotic patients with AMI.
During the 1-year follow up, our results showed higher rates of all-cause mortality in cirrhotic patients compared with non-cirrhotic patients with AMI, possibly relating to the concurrent cirrhotic complications of hepatic encephalopathy, ascites, esophageal varices, coagulopathy, and hypoalbuminemia. However, in terms of cardiovascular events of recurrent MI and MACCE that included stroke, cirrhotic patients with AMI were associated with lower events compared with non-cirrhotic patients.
Although previously it was debated exactly how cirrhotic patients are protected against ischemic events, our study suggested that naturally-anticoagulated system may affect the atherosclerosis in the coronary and cerebrovascular arteries. As seen in this cohort in Table 1, cirrhotic patients with AMI had significantly less percentage of patients requiring coronary intervention (both percutaneous coronary intervention and coronary artery bypass grafting), and stent deployment (both bare-metal stent and drug-eluting stent) before propensity score matching. Therefore it is possible that cirrhotic patients had lower levels of coronary plaque burden. On the other hand, cirrhotic patients with AMI had significantly higher event rates in terms of liver related outcomes including but not limited to hepatic encephalopathy, ascites tapping, spontaneous peritonitis, esophageal varices bleeding, compared with non-cirrhotic patients without AMI.
In summary, this is the first and largest study to directly compare clinical outcome of cirrhotic patients with AMI and non-cirrhotic patients with AMI. Our results suggested that cirrhotic patients were associated with decreased coronary and cerebrovascular events.
There are several limitations in epidemiologic data from NHIRD. First, due to the limitation of claim-based NHIRD where clinical evaluation and laboratory data were unavailable for the patients, the traditional risk stratification using Child-Pugh criteria in patients with liver cirrhosis could not be performed. However, we used modified Child-Pugh score classification to stratify the degree and severity of liver cirrhosis as previously published. Second, using ICD-9-CM codes for patient screening may miss some cases for conditions not coded correctly. However, ICD-9-CM codes against hospital electronic medical records (EMR) have been performed in the validation studies for NHIRD, the ICD codes have as sensitivity up to 99% for positive predictive value against the gold standard EMR. Third, the all-cause mortality was defined by withdrawal from the NHI, and the cause of death was defined as the discharge diagnosis within the last 3 months of NHI withdrawal. Therefore in certain patients the cause of deaths could only be approximated but not exact. Last, since our study consisted of homogeneous ethnic background, application of the results to other populations requires interpretation in the proper context.
In cirrhotic patients with AMI, there were associated decreased coronary and cerebrovascular events. However, these patients also had higher all-cause mortality due to adverse liver outcomes and major bleeding. Further management with liver cirrhosis related complications should improve overall outcome in these patients.
The authors would like to thank Zoe Ya-Jhu Syu and Alfred Hsing-Fen Lin for the statistical assistance during the completion of this manuscript.
Acquisition of data: Victor Chien-Chia Wu, Shao-Wei Chen, Yu-Sheng Lin.
Aanalysis and interpretation of data: Michael Wu, Pei-Chi Ting, Shang-Hung Chang, Chao-Yung Wang.
Conceptualization: Victor Chien-Chia Wu, Shao-Wei Chen, Yu-Sheng Lin.
Data curation: Victor Chien-Chia Wu, Shao-Wei Chen, Yu-Sheng Lin.
Formal analysis: Victor Chien-Chia Wu, Shao-Wei Chen, An-Hsun Chou, Michael Wu, Pei-Chi Ting, Yu-Sheng Lin.
Investigation: Victor Chien-Chia Wu, Shao-Wei Chen, An-Hsun Chou, Michael Wu, Pei-Chi Ting.
Methodology: Shang-Hung Chang, Chao-Yung Wang.
Project administration: Shang-Hung Chang, Chao-Yung Wang.
Resources: Ming-Shyan Lin, Kuo-Chun Hung, I-Chang Hsieh, Pao-Hsien Chu, Cheng-Shyong Wu.
Supervision: Pao-Hsien Chu, Cheng-Shyong Wu, Yu-Sheng Lin.
Writing – original draft: Victor Chien-Chia Wu.
Writing – review & editing: Shao-Wei Chen, Yu-Sheng Lin.
. Harding DJ, Perera MTPR, Chen F, et al. Portal vein thrombosis in cirrhosis
: controversies and latest development. World J Gastroenterol 2015;21:6769–84.
. Aggarwal A, Puri K, Liangpunsakul S. Deep vein thrombosis and pulmonary embolism in cirrhotic patients. World J Gastroenterol 2014;20:5737–45.
. Leonardi F, Maria N, Villa E. Anticoagulation in cirrhosis
: a new paradigm? Clin Mol Hepatol 2017;23:13–21.
. Russo MW, Pierson J, Narang T, et al. Coronary artery stents and antiplatelet therapy in patients with cirrhosis
. J Clin Gastroenterol 2012;45:339–44.
. Abougergi MS, Karagozian R, Grace ND, et al. ST elevation myocardial infarction mortality among patients with liver Cirrhosis
: a nationwide analysis across a decade. J Clin Gastroenterol 2015;49:778–83.
. Wu VC, Chen SW, Ting PC, et al. Selection of β-Blocker in patients with cirrhosis
and acute myocardial infarction
: a 13-year nationwide population-based study in Asia. J Am Heart Assoc 2018;7:e008982.
. Wu VC, Chen SW, Chou AH, et al. Dual antiplatelet therapy in patients with cirrhosis
and acute myocardial infarction
- a 13-year nationwide cohort study. PLoS One 2019;14:e0223380.
. Cheng CL, Lee CH, Chen PS, et al. Validation of acute myocardial infarction
cases in the national health insurance research database in Taiwan. J Epidemiol 2014;24:500–7.
. Lin YT, Wu PH, Lin CY, et al. Cirrhosis
as a risk factor for tuberculosis infection—a nationwide longitudinal study in Taiwan. Am J Epidemiol 2014;180:103–10.
. Wu CS, Lai MS, Gau SS, et al. Concordance between patient self-reports and claims data on clinical diagnoses, medication use, and health system utilization in Taiwan. PLoS One 2014;9:e112257.
. Cheng CL, Chien HC, Lee CH, et al. Validity of in-hospital mortality data among patients with acute myocardial infarction
or stroke in National Health Insurance Research Database in Taiwan. Int J Cardiol 2015;201:96–101.
. Wu CY, Chen YJ, Ho HJ, et al. Association between nucleoside analogues and risk of hepatitis B virus-related hepatocellular carcinoma recurrence following liver resection. JAMA 2012;308:1906–14.
. Cheng CL, Kao YH, Lin SJ, et al. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan. Pharmacoepidemiol Drug Saf 2011;20:236–42.
. Hsieh CY, Chen CH, Li CY, et al. Validating the diagnosis of acute ischemic stroke in a National Health Insurance claims database. J Formos Med Assoc 2015;114:254–9.
. Chen CC, Chen SW, Tu PH, et al. Outcomes of chronic subdural hematoma in patients with liver cirrhosis
. J Neurosurg 2018;130:302–11.
. Austin PC. Statistical criteria for selecting the optimal number of untreated subjects matched to each treated subject when using many-to-one matching on the propensity score. Am J Epidemiol 2010;172:1092–7.
. Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat 2011;10:150–61.
. Normand ST, Landrum MB, Guadagnoli E, et al. Validating recommendations for coronary angiography following an acute myocardial infarction
in the elderly: a matched analysis using propensity scores. J Clin Epidemiol 2001;54:387–98.
. Austin PC. Comparing paired vs non-paired statistical methods of analyses when making inferences about absolute risk reductions in propensity-score matched samples. Stat Med 2011;30:1292–301.
. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509.
. Howell WL, Manion WC. The low incidence of myocardial infarction in patients with portal cirrhosis
of the liver: a review of 639 cases of cirrhosis
of the liver from 17,731 autopies. Am Heart J 1960;60:341–4.
. Vanecek R. Atherosclerosis and cirrhosis
of the liver. Bull World Health Organ 1976;53:567–70.
. Berzigotti A, Bonfiglioli A, Muscari A, et al. Reduced prevalence of ischemic events and abnormal supraaortic flow patterns in patients with liver cirrhosis
. Liver Int 2005;25:331–6.
. An J, Shim JH, Kim SO, et al. Prevalence and prediction of coronary artery disease in patients with liver cirrhosis
: a registry-based matched case-control study. Circulation 2014;130:1353–62.