Key-words:
Database - female - liver transplantation - outcomes - Takotsubo cardiomyopathy
Introduction
Takotsubo cardiomyopathy (TCMP) is an acquired cardiomyopathy with transient systolic
and diastolic dysfunction characterized by apical left ventricular dysfunction or
other noncoronary distribution wall motion abnormalities. It is classically associated
with physical and emotional stress. However, it is now recognized that TCMP can also
occur in the setting of acute medical illness and after surgery.[[1]],[[2]] Transthoracic echocardiography typically demonstrates left ventricular apical ballooning
with basal hyperkinesis and is associated with new electrocardiographic abnormalities
such as transient ST-segment elevation with a notable absence of obstructive coronary
disease on cardiac catheterization.[[3]] TCMP has been rarely reported in liver transplantation recipients (LTRs), with
only few cases reported in the literature.[[4]],[[5]],[[6]],[[7]] Approximately 20 cases of TCMP were recorded in the LT registry of Toronto General
Hospital, Ontario, Canada, from approximately 1000 LTs performed in the last 10 years
(personal communication). This gives a prevalence of TCMP of approximately 2% in LTRs
from a single center.
Severe cases of TCMP may require vasopressor or inotropic therapy, oxygenation and
mechanical ventilation, or the placement of an intra-aortic balloon pump or extracorporeal
membrane oxygenation,[[1]],[[3]],[[8]] resulting in a significant burden on health-care utilization. Due to rarity of
this condition and reported cases, no specific intra-LT or post-LT predictors have
been reported to increase the risk for TCMP.[[9]] Epidemiology, predictors, and outcomes in LTRs who subsequently develop TCMP are
lacking. However, the use of extensive health-care resources due to TCMP in an already
tenuous LTR demand further study of trends, prevalence, predictors, and clinical outcomes.
Methods
The Nationwide Readmissions Database (NRD) 2010–2014 was queried using specific International
Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes.
The NRD, drawn from the Health-care Cost and Utilization Project, consists of 21 State
Inpatient Databases containing patient linkage numbers that can be used to track patients
through multiple admissions. The database also contains information on patient demographics,
hospital characteristics, procedures performed, and discharge diagnoses that can be
extrapolated to provide national estimates.
TCMP was identified by the ICD-9 code 429.83[[10]],[[11]] while LTRs were identified using ICD-9-CM procedural codes for LT. Codes for a
history of LT were avoided intentionally to ensure capture of TCMP after LT. Decompensated
cirrhosis was defined by the presence of an ICD-9-CM code for cirrhosis in addition
to an ICD-9-CM code for either ascites, encephalopathy, esophageal bleeding, or spontaneous
bacterial peritonitis [[Supplementary Table 1]]. As we aimed to measure TCMP development after LT, “cases” were regarded as patients
with index admissions with a procedural code for LT who developed TCMP, whereas “controls”
were patients who underwent LT without developing TCMP through either the index admission
or the calendar year. Patients with multiple organs transplantation (e.g., liver with
lung, heart, pancreas, or kidney) and pregnant females were excluded. Early TCMP was
defined as the development of TCMP during the index admission of LT while late TCMP
was defined as TCMP after the initial hospitalization. Outcomes examined were mortality
and hospital resource utilization, including length of stay (LOS) and hospital costs.
Predictors of TCMP development within the LTRs were analyzed.
Supplementary Table 1: International Classification of Diseases, Ninth Revision, Clinical modification codes
used for data extraction and analysis from the Nationwide Readmission Database 2013
Statistical analysis
All statistical analyses accounted for the NRD weights and stratified sampling design
(discharge weights, stratification, and clustering). Univariate comparisons of patient
characteristics between patients with and without TCMP were made by the Chi-square
test for categorical variables and linear regression for continuous variables. A weighted
logistic regression model was used to identify the variables associated with developing
TCMP at index admission or readmission. All variables from [[Table 1]] with a P < 0.05 were included in the multivariate model. The presence of specific
agency for Healthcare Research and Quality Comorbidities were compared by the Chi-square
test. Linear temporal trends were tested by linear regression on the yearly frequency
estimates weighted by the inverse variance of those estimates. A significance level
of α = 0.05 was used for all analyses. All analyses were performed using the SAS version
9.4 (Cary, NC) by a biostatistician.
Table 1: Demographics, hospital characteristics, and interventions in Takotsubo cardiomyopathy
in liver transplant recipients
Results
Trends of Takotsubo cardiomyopathy
A total of 28,067 patients underwent LT between the years 2010 and 2014. Of these,
TCMP occurred in 0.5% (141/28,067) of all LTRs. Early TCMP developed in 115 (82.1%)
while late TCMP in 38 (17.9%.) LTRs. The trends of both LTs and cases of TCMP annually
between 2010 and 2014 remained stable [[Supplementary Table 2]].
Supplementary Table 2: Temporal trends of Takotsubo cardiomyopathy in liver transplant recipients between
2010 and 2014, Nationwide Readmission Database
Univariate analysis
Patients with TCMP [[Table 1]] were more likely to be older (57.5 ± 1.3 vs. 55.1 ± 0.2 years, P < 0.001), females
(53.3% vs. 33.3%, P = 0.004), with 4 or more Elixhauser comorbidities (90.2 vs. 62.8%,
P < 0.001) as compared to LTRs without TCMP. Other predictors of TCMP development
included decompensated cirrhosis (77.1% vs. 52.9%, P < 0.001) with ascites (73.4%
vs. 48.3%, P < 0.001), hepatic encephalopathy (54.4% vs. 27.2%, P < 0.001), and portal
hypertension (66.8% vs. 39.3%, P < 0.001). TCMP developed more frequently in LTRs
in smaller bed size hospitals (19.7% vs. 6.7%, P < 0.001).
After LT, TCMP patients required higher resource utilization in the form of mechanical
ventilation, hemodialysis, vasopressors, and intra-aortic balloon pumps (IABP). TCMP
patients not only experienced a longer LOS (median: 26.6; interquartile range [IQR]:
20.4–43.3 vs. median: 10.8 days; IQR: 6.9–23.1, P < 0.001) but also incurred a higher
cost on index hospital admission (median: $148,274; IQR: 110,983–226,570 vs. $89,879;
IQR: 65,858–136,135; P < 0.001). Moreover, in patients who were readmitted at least
once, TCMP patients had significantly higher readmission costs ($35,886; IQR: 11,801–64,002
vs. $18,428; IQR: 7,781–45,472, P < 0.001).
Although index admission mortality did not differ significantly between cases and
controls (7.0% vs. 4.0%, P = 0.20), mortality during readmissions within the calendar
year was higher in LTRs with TCMP (5.4% vs. 1.7%, P = 0.02). Specific comorbidities
that were commonly seen in TCMP patients are shown in [[Table 2]].
Table 2: Significant elixhauser comorbidity variables comparing liver transplant recipients
with and without takotsubo cardiomyopathy
Multivariate analysis
Female gender (adjusted odds ratio [aOR]: 2.27; confidence interval [CI]: 1.2–4.27)
and 4 or more Elixhauser comorbidities (aOR: 2.36; CI: 1.15–4.83) were the independent
predictors of development of TCMP in LTRs [[Table 3]]. LT at a medium sized hospital was associated with lower odds of developing TCMP
(aOR: 0.17; CI: 0.03–0.88). Decompensated cirrhosis (aOR: 1.58; CI: 0.93–2.67), presence
of portal hypertension (aOR: 1.70; CI: 0.97–2.97), receiving mechanical ventilation
(aOR: 1.96; CI: 0.90–4.26), and hemodialysis (aOR: 1.98; CI: 0.94–4.15) were found
to be marginally associated with the development of TCMP. Patients with TCMP utilized
increased hospital resources in the form of vasopressors (aOR: 3.62; CI: 2.01–6.51)
and IABP (OR: 83.3; CI: 25.2–274.8) after LT as compared to those without it.
Table 3: Multivariable logistic regression analysisa of predictors and resource utilization in Takotsubo cardiomyopathyb
Discussion
Takosubo cardiomyopathy has been reported to be associated to with significant morbidity
and mortality in the general population. The American Heart Association incorporated
it under the class of acquired cardiomyopathies in 2006.[[12]] To date, no statistical analysis on its prevalence, predisposing factors, hospital
resource utilization, and mortality is available in LTRs. Therefore, we analyzed TCMP
in LTRs using a large national database. The prevalence of TCMP in LT was found to
be 0.5%, with the majority of cases occurring early during the index admission for
LT. Older females with ≥4 Elixhauser comorbidity index were at increased risk of developing
TCMP after LT. Increased resource utilization, in the form of invasive interventions
such as vasopressors and IABP, was observed in patients with TCMP. We also demonstrated
that TCMP patients have increased LOS and cost of index admission for LT and ultimately
suffer higher calendar year mortality.
Previous reports using a national inpatient sample database showed that TCMP was diagnosed
in about 0.02% of all hospitalizations in the United States.[[13]] Comparatively, the prevalence of TCMP in LTRs is nearly 25 times higher in our
study. As TCMP is thought to be related to stress and precipitated in part or entirely
by excessive catecholamine stimulation of the myocardium,[[1]],[[2]] increased occurrence could be due to the extreme physical stress and subsequent
catecholamine release as a result of LT. This is endorsed by our personal communication
from a single transplant center in Canada where they found TCMP in approximately 2%
of their LTRs. The trend of TCMP is thought to be increasing, although likely due
to increased recognition of this disease entity and increased severity of liver disease
in LTRs rather than true increased incidence.[[14]],[[15]]
Conforming to the typical scenario of TCMP, females with LT were more likely to develop
TCMP than their male counterparts. Whereas previous studies showed women accounting
for upward of 90% of all TCMP cases, we found that only 53% of TCMP were females.[[14]],[[15]] However, female gender more than doubled the odds of TCMP with LT. The cause of
apparent female predisposition for TCMP is unknown but could be related to gender
differences in myocardial sensitivity to catecholamine toxicity,[[16]] which may make less of a difference in the setting of a significant stress event
such as LT.
We found an increased (≥4) number of Elixhauser comorbidity to increase the odds of
the development of TCMP. In studies on the general population, obesity, hypertension,
uncomplicated diabetes, and pulmonary disease were reported to be associated with
TCMP.[[17]] In our study, the presence of congestive heart failure, coagulopathy, electrolyte
imbalances, and valvular disease were increasingly associated with TCMP development.
Common comorbidity in our and previously reported studies on TCMP was chronic renal
disease and malignancy.[[17]],[[18]],[[19]]
TCMP drastically increases the resource utilization during hospitalization. In 2014,
the total average cost of a LT was approximately $739,100,[[20]] posing an already significant burden on the health-care system. We found the median
cost of LTRs with TCMP almost doubled when compared to LTRs without TCMP. While evidence
is lacking in the TCMP setting, a previous European study found the cost of IABP support
in acute myocardial infarction to be more than medical therapy alone.[[21]] LOS has been used as a valid surrogate for hospital costs.[[22]] The median LOS for TCMP patients was more than double that of non-TCMP patients.
Moreover, more than one-fourth of our patients developed TCMP on a nonindex admission,
with the added cost of readmissions to be upward of $40,000 on average which represents
a significant health care resource expenditure.[[23]]
Cardiovascular disease is a common cause of mortality and morbidity after LT.[[24]] Inpatient mortality from all cardiovascular events following LT has been previously
found to be 4%.[[25]] Comparatively, our results show LTRs with TCMP suffered from an increased calendar
year mortality rate as compared to patients without TCMP. Moreover, TCMP in the overall
population had a mortality rate of 2%–4%.[[10]],[[26]] In contrast, we found the occurrence of TCMP had 5.4% mortality in the post-LT
setting.
Our study has multiple limitations inherent to administrative database analysis research.
First, this study relies on ICD9-CM codes for establishing diagnoses. Theoretically,
under/miscoding can lead to misclassification bias. However, NRD has been used to
study readmission in several medical and surgical conditions.[[27]],[[28]] It is possible that that few cases of cirrhotic cardiomyopathy were not diagnosed
before the LT and resulted in decompensated heart failure in the postoperative time
period. However, TCMP is a distinct entity from cirrhotic cardiomyopathy with characteristic
wall motion pattern and is unlikely to be confused with cirrhotic cardiomyopathy.
Furthermore, cirrhotic cardiomyopathy tends to get better after LT. Due to the high
positive predictive value of the TCMP ICD-9-CM code used in a European study,[[29]] we tend to believe that most cases captured in our study represent true TCMP cases.
The Mayo diagnostic criteria and diagnostic criteria by Kawai et al. for TCMP require
coronary angiography to exclude coronary artery obstruction.[[2]],[[30]] Furthermore, LTRs undergo a substantial cardiac work up, including electrocardiogram,
echocardiogram, and coronary angiography before LT, with abnormal findings precluding
patients to transplant. The NRD is limited to specific data and notably lacks laboratory
values, specific imaging and procedure results, and medication usage. We were unable
to present the Model for End-Stage Liver Disease score which inform us about the severity
of cirrhosis. However, we used codes for decompensated cirrhosis reported in previous
publications.[[31]] Furthermore, data extraction from the NRD relies on the use of ICD-9-CM codes,
which can vary in quality of diagnosis coding or documentation. Nevertheless, the
large sample size provided by the NRD increases statistical power to analyze this
uncommon condition.
Conclusion
In conclusion, this is the first study from the US reporting prevalence and trends
of TCMP and its risk factors in the post-LT population. Predisposing factors leading
to the development of TCMP in post-LT patients include older age, female gender, decompensated
cirrhosis, and a higher number of comorbidities. With the number of LTs increasing[[32]] along with increasing selection of more critical patients, the number of TCMP cases
in LTRs is likely to increase, calling for further research into early identification
of TCMP and improving clinical outcomes in this rarely reported post-LT complication.
The presence of TCMP was associated with increased health-care utilization and cost
which translated into a possibly increased mortality. Further research should elucidate
optimal prevention strategy and management of TCMP in post-LT recipients.
Author's contributions
All authors have approved the final version of the manuscript.
Kishan Patel: study concept and design, acquisition of data, analysis and interpretation
of data, drafting of the manuscript, critical revision of the manuscript for important
intellectual content.
Salman Bhatti: critical revision of the manuscript for important intellectual content.
Sylvester Black: critical revision of the manuscript for important intellectual content.
Kyle Porter: acquisition of data, analysis and interpretation of data, statistical
analysis, critical revision of the manuscript for important intellectual content.
James Hanje: critical revision of the manuscript for important intellectual content.
Khalid Mumtaz: study concept and design, acquisition of data, analysis and interpretation
of data, drafting of the manuscript, critical revision of the manuscript for important
intellectual content.
Reviewers:
Hussien Elsiesy (Tucson, AZ, USA)
Editors:
Elmahdi A Elkhammas (Columbus OH, USA)
