Journal Logo

Original Clinical Science—Liver

A Propensity-matched Survival Analysis: Do Simultaneous Liver-lung Transplant Recipients Need a Liver?

Freischlag, Kyle1; Ezekian, Brian MD2; Schroder, Paul M. MD, PhD2; Mulvihill, Michael S. MD2; Cox, Morgan L. MD2; Hartwig, Matthew G. MD2; Knechtle, Stuart MD2

Author Information
doi: 10.1097/TP.0000000000002529

Abstract

Simultaneous liver-lung transplantation (LLT) can be a lifesaving procedure when used in select patients with both end-stage liver and end-stage lung disease who may not be expected to survive isolated, single-organ transplantation. The primary indication for this procedure is cystic fibrosis but has also included patients with alpha-1-antitrypsin and separate liver and lung etiologies.1-3 Overall, LLT recipients have increased survival compared to patients who remain on the waitlist.4 However, the literature on outcomes in this rare cohort is limited to several case series.1-3,5-9

Proposed benefits to the LLT procedure include comparable survival to lung-alone recipients, and the possibility of protection against rejection due to the relative tolerogenicity of liver allografts.10 However, the long-term outcomes are significantly worse than isolated liver transplant.4,11 Survival after LLT is largely determined by the lung allograft function, as the most common causes of mortality in the LLT population are pulmonary infection and chronic lung rejection.1-3,9 Additionally, due to the rarity of this procedure, there has been significant variability in studies regarding patient selection, patient demographics, and long-term outcomes.1-3 Therefore, debate exists whether the long-term outcomes of this procedure warrant dual allocation of both liver and lung allografts to a single patient.

In this rare population, dual-organ allocation is primarily driven by the lung allocation score (LAS), because many of the patients in the LLT population have relatively lower Model for End-stage Liver Disease (MELD) than would be expected with isolated liver transplantation.4,12 Due to the low level of liver dysfunction, it is currently unknown whether patients who underwent LLT would have instead benefited to a similar degree from single-organ transplant. We hypothesized that LLT recipients would have improved survival compared to matched single-organ lung recipients with a similar degree of liver dysfunction.

PATIENTS AND METHODS

Study Design

The institutional review board at Duke University Medical Center approved this retrospective analysis of the United Network for Organ Sharing (UNOS) database. Eligible candidates were older than 18 years at the time of listing for lung alone or concurrent liver and lung transplantation. Eligible recipients were older than 18 years at the time of transplantation and either underwent single-organ lung transplant or concurrent liver and lung transplantation in the United States from January 1, 2006, to December 31, 2016.13 Liver-lung transplantation and single-organ lung transplants before 2006 were excluded due to the differences in waitlist mortality and outcomes after the introduction of the LAS.14 Any patient receiving additional allografts at the time of transplant was excluded (Figure 1).

FIGURE 1.
FIGURE 1.:
CONSORT diagram for lung-alone recipients and simultaneous lung-liver recipients.

Study Endpoints

The primary endpoint was overall survival. Secondary endpoints included short-term perioperative and graft outcomes including acute rejection episodes, retransplantation, and 30-day postoperative mortality.

Statistical Analysis

Before matching, waitlist mortality probability and transplant probability were calculated for all candidates listed for single-organ lung or LLT using cumulative incidence functions. A proportional hazards model was generated to examine waitlist mortality and an odds ratio (OR) was used to examine transplant probability.15 For transplanted patients, baseline characteristics were compiled and analyzed using the Kruskal-Wallis test for continuous variables and Pearson χ2 test for categorical variables. Because a MELD score and international normalized ratio (INR) value at transplant were unavailable for single-organ lung transplant recipients, a MELD-XI was calculated for each patient. MELD-XI is a predictive formula for 90-day pretransplant mortality incorporating bilirubin and creatinine but omitting INR, which has been validated in patients with thoracic disease with liver dysfunction.16-19 The diagnostic groups used are described and defined by UNOS (UNOS Policy 10.1.F.i).

From significant differences at baseline, a propensity score was formulated, and LLT recipients were matched with single-organ lung recipients based on recipient age, recipient BMI, recipient ethnicity, recipient gender, MELD-XI, UNOS diagnostic group, expiratory volume for 1 second as a percentage of the forced vital capacity (FEV1%), and donor BMI. The propensity match used a 1:2 nearest-neighbor matching algorithm to form matched pairs of LLT and single-organ recipients.20-23 Calipers of width 0.2 of the standard deviation of the logit of the propensity score were used in matching.24 The effectiveness of the reduction of bias within the model after matching was assessed by standardized mean differences. Cohen’s suggested threshold of 0.2 was used to denote a meaningful imbalance in the standardized mean difference of baseline covariates.20

After matching, baseline characteristics and unadjusted outcomes were treated as paired data, and McNemar’ test for categorical variables and the Wilcoxon signed-rank test for continuous variables were used.25 Kaplan-Meier method with log-rank test compared survival between groups.26 Because the proportional hazards assumption was not fulfilled until 6 months posttransplant, we conducted separate analyses for the periods before and after this time point. To avoid overfitting due to the limited number of events within the first 6 months, a univariate logistic regression analysis was undertaken to estimate the impact of LLT on mortality within 6 months after transplantation.27 A proportional hazards regression was used to calculate risk-adjusted mortality after 6 months. Covariates in the proportional hazards regression were chosen using bidirectional stepwise elimination.28,29 A complete list of candidate covariates tested for model inclusion is provided in Materials and Methods (SDC, http://links.lww.com/TP/B655).

A P value of <0.05 was deemed statistically significant. Statistical analysis was performed using R version 3.4.3 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Unmatched Waitlist Candidate Demographics, Mortality Risk, and Transplant Odds

A total of 110 candidates were listed for LLT and 22 063 candidates were listed for single-organ lung transplant (Figure 1). Candidate demographics are detailed in Table 1. While on the waitlist, the risk of mortality for LLT versus lung alone candidates was significantly higher (hazards ratio [HR], 3.2, 95% confidence interval [CI], 2.34-4.39; P < 0.001). Odds of transplant were significantly lower for LLT versus lung alone candidates (OR, 0.255, 0.175-0.373, P < 0.001; Figure 2).

TABLE 1.
TABLE 1.:
Unmatched and unadjusted baseline LLT and Lung-alone candidate characteristics
FIGURE 2.
FIGURE 2.:
Cumulative incidence of death and transplant for lung-alone vs LLT waitlist candidates.

Unmatched Patient Demographics

A total of 40 LLT recipients and 13 060 lung-alone recipients were included for analysis (Figure 1). Compared to isolated lung transplant recipients, LLT recipients were younger (mean 34.92 vs 54.60 y, P < 0.001) and had lower recipient BMI (21.11 vs 24.89, P < 0.001), fewer unilateral lung transplants (5.0% vs 33.1%, P < 0.001), and lower donor BMI (22.87 vs 25.81, P < 0.001). Notably, LLT recipients had significantly lower FEV1% (27.93 vs 37.66, P = 0.003) and higher MELD-XI (11.85 vs 9.91, P < 0.001) than the single-organ population. Within the LLT cohort, mean calculated MELD-XI was comparable to MELD score at transplant (10.67 vs 10.89, P = 0.7). Total bilirubin was elevated for LLT recipients compared to lung alone (2.79 vs 0.62, P < 0.001). The primary pulmonary indication for transplantation was different between the lung-alone population and the LLT population (P < 0.001). For LLT, the most common primary diagnosis was cystic fibrosis (62.5%, n = 25) and second was idiopathic pulmonary fibrosis (15.0%, n = 6). For lung alone, the primary indication for transplant was idiopathic pulmonary fibrosis (35.0%, n = 4573) followed by chronic obstructive pulmonary disease/emphysema (25.8%, n = 3365) and cystic fibrosis (12.2%, n = 1593). Lung allocation score at the time of transplantation between the groups was not significantly different (49.16 vs 46.71, P = 0.364; Table 2).

TABLE 2.
TABLE 2.:
Unmatched and unadjusted baseline LLT and lung-alone recipient and donor characteristics

Matched Patient Demographics

After a propensity score was generated, 38 LLT patients were matched to 75 single-organ lung recipients on recipient and donor demographic characteristics, etiology, and MELD-XI at transplant (Figure 3). No differences in baseline recipient characteristics, donor characteristics, lung function, or liver function were seen between the matched cohorts (all P > 0.05) (Table 3).

TABLE 3.
TABLE 3.:
Matched Lung-alone and LLT recipient and donor characteristics
FIGURE 3.
FIGURE 3.:
Standardized differences in baseline covariates comparing LLT and single-organ recipients, before and after propensity score matching.

Matched Patient Unadjusted Outcomes

Length of stay was significantly higher in LLT recipients compared with isolated lung recipients (45.89 days vs 22.44 days, P < 0.001). Acute rejection rate, percent undergoing retransplantation, percent with a functioning lung allograft at last known follow-up, and 30-day mortality were not significantly different (all P > 0.05) (Table 4).

TABLE 4.
TABLE 4.:
Posttransplant outcomes between LLT recipients and matched controls

Matched Patient Survival Analysis and Adjusted Outcomes

Median survival was not significantly different (LLT, 4.83 y vs isolated lung, 4.72 y; P = 0.90). Overall survival probability was not significantly different between LLT and matched isolated lung transplant (Figure 4: 1 y, 89.5% vs 86.7%; 3 y, 81.0% vs 76.5%; 5 y, 67.0% vs 64.6%; P = 0.20). After adjustment, patients receiving LLT did not show a decreased risk of mortality compared with matched lung-alone recipients before 6 months (OR, 1.353; 95% CI, 0.358-5.116; P = 0.66) or after 6 months (Table 5; HR, 0.556; 95% CI, 0.274-1.127; P = 0.10).

TABLE 5.
TABLE 5.:
Survival analysis before and after 6 months posttransplant for LLT mortality
FIGURE 4.
FIGURE 4.:
Kaplan-Meier survival analysis of simultaneous liver-lung transplant recipients compared to matched isolated lung recipients.

DISCUSSION

Because of the national scarcity of organs, it is unclear in the field of transplantation whether it is justified to allocate 2 organs to a single individual. This ambiguity is especially relevant in the LLT population because allocation is driven primarily by the LAS, resulting in liver transplant in patients with relatively lower MELD than would be expected in isolated liver transplantation.4,12 This observation may be related to factors other than the LAS; namely, the urgency of lung failure being greater than the urgency of liver failure in this patient population, and the current policies that permit the allocation of a second organ when a match is made for the first organ. Before this study, it was unknown whether patients who underwent LLT would benefit similarly from isolated lung transplantation. Our work used a propensity score analysis of the LLT population and is the first to show that long-term survival between LLT and matched single-organ controls is similar. Similarly, in our adjusted analysis, receiving a liver did not significantly lower the risk of mortality compared with matched isolated lung controls who did not receive a liver. Therefore, our study calls for reevaluation of patient selection and national guidelines for this procedure. With no significant difference in survival between LLT recipients and single-organ lung recipients with equivalent liver dysfunction, national guidelines are needed to clarify patient selection for LLT.

Proposed benefits of LLT include comparable survival to isolated lung and better survival compared with remaining on the waitlist. Previous case series have estimated 1-year survival at 56% to 80% and 3-year survival at 62% to 70%.1-8,30 However, most of these analyses included patients from 1988 to 2016, which encompasses multiple eras of transplantation and immunosuppression. Overall, these rates are similar to those of single-organ bilateral lung transplants (1-y survival 87.7%, 3-y survival 71.6%, and 5-y survival 58.4%).31 For LLT, our data showed 90% survival at 1 year and 67% survival at 5 years with similar survival in the matched isolated lung cohort. Overall, our long-term survival in both cohorts was similar to previous LLT reports and national lung alone data.

Wolf et al4 examined the UNOS database and found 42 LLT recipients from 1987 to 2010. They found that candidates on the LLT waitlist had significantly worse survival compared with isolated lung waitlist candidates, suggesting that candidates for LLT are indeed sicker and less likely to survive without transplantation. Our study supported these results in the modern era after the introduction the LAS. We showed waitlist mortality in LLT candidates was significantly higher than single-organ lung candidates and that odds of transplant were significantly lower for LLT candidates. Notably, the LAS for LLT candidates and recipients was not notably different from lung alone candidates. If LLT patients are sicker at baseline, the current allocation system is not adequately capturing degree of illness in this population. Although Wolf et al compared waitlist survival, survival analysis of LLT versus all single-organ lung transplant recipients did not evaluate similar populations. To understand whether receiving a liver conferred a survival benefit in the LLT population, it was necessary to match the populations on propensity score. Our study shows that in matched lung transplant recipients with liver dysfunction, there was no survival difference between those who also received a liver and those who did not.

The ethics of dual organ allocation to a single recipient are complex. One of the largest critiques of LLT comes from the principle of utilization or that dual allocation to a single recipient is not warranted when single organ allocation of the allografts to multiple recipients might save more lives. However, utility also evaluates overall outcomes such as patient and graft survival as well as quality of life. An ethical argument for continuation of dual allocation based on utility can be made in select groups of patients if it significantly improves long-term outcomes.4 Additionally, the principle of justice has been used to defend simultaneous thoracic and abdominal transplant. Justice argues that patients in need of dual allocation have greater medical urgency than patients with single-organ transplant due to increased medical risk from multiple failing organ systems. If LLT candidates had both higher waitlist mortality and better survival than matched lung-alone controls, then both the principles of utility and justice would support continuation of this practice. In our study, the waitlist mortality of candidates listed for LLT was higher, therefore, defending the principle of justice. Additionally, the similar long-term outcomes between the matched cohorts defended the principle of utility. Our overall results, however, require a more nuanced review.

Our data show that LLT patients had equivalent survival to lung alone recipients when matching for patient characteristics and degree of liver dysfunction. One possible conclusion is that dual allocation of organs was not helpful, was associated with longer hospital length of stay and potential morbidity and may not have the equipoise to justify allocation 2 organs to 1 patient. However, this conclusion ignores the higher waitlist mortality in this population, which may represent differences in baseline illness that are not captured by the variables reported in the UNOS database. This conclusion also discounts clinical judgment. To warrant allocation of 2 organs to a single individual, 2 separate clinical teams first needed to agree that transplantation of both liver and lung was necessary. Based on previous case series, indication for liver transplantation was most commonly mild but chronic, symptomatic liver disease.1-8,30,32,33 The nuances of the clinical picture prompting liver transplant in each case are ambiguous and not clearly captured in national databases or in allocation algorithms.

A second possible conclusion is that patients listed for LLT may be clinically sicker at baseline and projected to do worse than single-organ lung transplant. Combined LLT candidates have been described as those who would not be able to survive lung transplant alone.1,3,9 Our data, therefore, which can also be interpreted as improvement in LLT, have brought equivalent outcomes compared with isolated lung transplant patients. If so, our outcomes may suggest dual allocation successfully increased survival to equivalence with single-organ transplant. The difference in cohorts at baseline is supported by higher waitlist mortality in the first year after listing, but it is not reflected in a higher LAS or captured in any pretransplant clinical measures. Proponents of LLT express concern that isolated lung transplantation may lead to liver failure in those with preexisting liver dysfunction. However, LLT recipients did not show a clear difference in survival or long-term outcomes compared with single-organ recipients who had equivalent demographic characteristics, transplant etiology, lung function, and laboratory markers of liver function.

A probable third conclusion is that there is a cohort of LLT candidates that significantly benefit from dual allocation and also a cohort of LLT candidates that might have benefited equally from single-organ lung transplant. To elucidate and improve patient selection for LLT, future studies should place particular emphasis upon establishing guidelines for transplant clinicians who are considering dual listing of a patient. These guidelines may include data not captured in national databases, such as liver biopsy results, liver imaging for fibrosis, cirrhosis, severity of portal hypertension, and/or clinical symptoms of end-stage liver disease. Regardless, national guidelines are needed to better inform patient selection.

Our study has several limitations. This is a retrospective review of the national UNOS database. As such, it may suffer from selection bias for LLT or coding errors inherent in the database. A prospective randomized controlled trial in this population with similar numbers is functionally impossible due to the rarity of the procedure. Therefore, a national analysis remains the best way to attempt to answer this question. The modification of the MELD score, the MELD-XI, was used due to the absence of pretransplant recipient INR in the national lung data. The implications of our data are limited because is not currently known how closely MELD-XI approximates MELD score in patients without known liver disease. However, our data showed similar calculated MELD-XI versus reported MELD score (10.67 vs 10.89, P = 0.7) in our LLT cohort, and our results did not have significant differences in markers of liver function. Total bilirubin at candidate listing is not captured in the database, which limits our ability to conduct a similar survival analysis among candidates with liver dysfunction. No meaningful measures of pulmonary dysfunction are reported in the UNOS liver database, and thus, a comparison cannot be made to single-organ liver recipients without introducing unmeasured confounders. Our study cannot capture differences in immunosuppressive regimens, pretransplant evaluation, or posttransplant care with granularity due to the nature of the national database. By limiting our study to only adult patients after 2006, we attempted to limit dramatic differences in transplant care.13 The authors recognize that the database is not granular enough to say the cohorts were unquestionably matched. There likely exists the possibility of unmeasured confounders that may have clarified the need for LLT rather than isolated lung transplant. Therefore, improved guidelines and greater elaboration of common practices is needed to improve patient selection and appropriate organ allocation.

CONCLUSIONS

Survival for recipients of simultaneous LLT was comparable to isolated lung transplantation recipients, even after matching for patient characteristics and level of liver dysfunction. Candidates for LLT had significantly worse waitlist outcomes and LLT recipients had a significantly longer length of stay. However, they had otherwise similar short-term and long-term outcomes to single-organ lung transplant. Although LLT may be the only possible treatment for some transplant candidates, still others might benefit equally from single-organ lung transplant instead. Therefore, our study calls for reevaluation of patient selection and national guidelines for this procedure.

REFERENCES

1. Arnon R, Annunziato RA, Miloh T, et al. Liver and combined lung and liver transplantation for cystic fibrosis: analysis of the UNOS database. Pediatr Transplant. 2011;15:254–264.
2. Grannas G, Neipp M, Hoeper MM, et al. Indications for and outcomes after combined lung and liver transplantation: a single-center experience on 13 consecutive cases. Transplantation. 2008;85:524–531.
3. Yi SG, Burroughs SG, Loebe M, et al. Combined lung and liver transplantation: analysis of a single-center experience. Liver Transpl. 2014;20:46–53.
4. Wolf JH, Sulewski ME, Cassuto JR, et al. Simultaneous thoracic and abdominal transplantation: can we justify two organs for one recipient? Am J Transplant. 2013;13:1806–1816.
5. Ceulemans LJ, Monbaliu D, Verslype C, et al. Combined liver and lung transplantation with extended normothermic lung preservation in a patient with end-stage emphysema complicated by drug-induced acute liver failure. Am J Transplant. 2014;14:2412–2416.
6. Couetil JP, Houssin DP, Soubrane O, et al. Combined lung and liver transplantation in patients with cystic fibrosis. A 4 1/2-year experience. J Thorac Cardiovasc Surg. 1995;110:1415–1422; discussion 1422–1423.
7. Praseedom RK, McNeil KD, Watson CJ, et al. Combined transplantation of the heart, lung, and liver. Lancet. 2001;358:812–813.
8. Barshes NR, DiBardino DJ, McKenzie ED, et al. Combined lung and liver transplantation: the United States experience. Transplantation. 2005;80:1161–1167.
9. Freischlag KW, Messina J, Ezekian B, et al. Single-center long-term analysis of combined liver-lung transplant outcomes. Transplant Direct. 2018;4:e349.
10. Yi SG, Lunsford KE, Bruce C, et al. Conquering combined thoracic organ and liver transplantation: indications and outcomes for heart-liver and lung–liver transplantation. Curr Opin Organ Transplant. 2018;23:180–186.
11. Kim WR, Lake JR, Smith JM, et al. OPTN/SRTR 2015 annual data report: liver. Am J Transplant. 2017;17S1174–251.
12. Elwir S, Lake J. Current status of liver allocation in the United States. Gastroenterol Hepatol (N Y). 2016;12:166–170.
13. Freischlag K, Schroder PM, Ezekian B, et al. Improved survival in simultaneous lung-liver recipients and candidates in the modern era of lung allocation. J Surg Res. 2018;231:395–402.
14. Egan TM, Edwards LB. Effect of the lung allocation score on lung transplantation in the United States. J Heart Lung Transplant. 2016;35:433–439.
15. Scrucca L, Santucci A, Aversa F. Regression modeling of competing risk using R: an in depth guide for clinicians. Bone Marrow Transplant. 2010;45:1388–1395.
16. Wernly B, Lichtenauer M, Franz M, et al. Model for end-stage liver disease excluding INR (MELD-XI) score in critically ill patients: easily available and of prognostic relevance. PLOS One. 2017;12:e0170987.
17. Deo SV, Al-Kindi SG, Altarabsheh SE, et al. Model for end-stage liver disease excluding international normalized ratio (MELD-XI) score predicts heart transplant outcomes: evidence from the Registry of the United Network for Organ Sharing. J Heart Lung Transplant. 2016;35:222–227.
18. Grimm JC, Shah AS, Magruder JT, et al. MELD-XI score predicts early mortality in patients after heart transplantation. Ann Thorac Surg. 2015;100:1737–1743.
19. Heuman DM, Mihas AA, Habib A, et al. MELD-XI: a rational approach to “sickest first” liver transplantation in cirrhotic patients requiring anticoagulant therapy. Liver Transpl. 2007;13:30–37.
20. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 1988.2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates Publishers.
21. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat. 1985;39:33–38.
22. Austin PC, Grootendorst P, Anderson GM. A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study. Stat Med. 2007;26:734–753.
23. Austin PC. Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations. Biom J. 2009;51:171–184.
24. Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat. 2011;10:150–161.
25. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46:399–424.
26. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53:457–481.
27. Peduzzi P, Concato J, Kemper E, et al. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49:1373–1379.
28. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007;165:710–718.
29. Lin DY, Wei LJ. The robust inference for the Cox proportional hazards model. J Am Stat Assoc. 1989;84:1074–1078.
30. Corno V, Dezza MC, Lucianetti A, et al. Combined double lung-liver transplantation for cystic fibrosis without cardio-pulmonary by-pass. Am J Transplant. 2007;7:2433–2438.
31. Organ procurement and transplantation network, organ procurement and transplantation network lung Kaplan-Meier patient survival rates for transplants performed: 2008–2015. 2017.
32. Zimmerman AA, Howard TK, Huddleston CB. Combined lung and liver transplantation in a girl with cystic fibrosis. Can J Anaesth. 1999;46:571–575.
33. Backman S, Javela K, Koivusalo AM, et al. Successful liver and lung transplantation in patients with severe IgA deficiency, high anti-IgA concentration and a history of anaphylactic transfusion reaction. Transfus Med. 2014;24:251–253.

APPENDIX

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.