The negative health implications of tobacco use in the nontransplant setting are well recognized. Smokers are at increased risk of a wide range of pathologies including cardiovascular disease, stroke, cancer, and peptic ulcer disease.1 In addition, although it is less well known, smokers demonstrate an increased susceptibility to infection and are at increased risk of chronic kidney disease and possibly insulin resistance.2–4 As a result, smoking is the world's leading cause of premature mortality responsible for an estimated 5 million deaths each year.5
Prevalence of tobacco use among the general population is between 20% and 30%, whereas in liver transplant recipients, it may be as high as 40%.5, 6 Nevertheless, the consequences of smoking in this group specifically remain unclear. Therefore, our aim was to examine the effect of smoking on morbidity and mortality following liver transplantation.
This was a single-center retrospective case-note study of 136 consecutive patients who underwent elective liver transplantation between January 1, 1996 and December 31, 2000. The records of patients transplanted for acute hepatic failure were not reviewed. Smoking status was based on documentation at the time of transplant assessment. Four patients did not have smoking status available and were excluded from further analysis. Therefore, a total of 132 patients were included in the study. Mean follow-up time was 8.8 years (range 6.4 to 11.3 years).
Immunosuppression consisted of a calcineurin inhibitor, azathioprine, and prednisolone in most patients. Midway through the specified time period, the unit policy for the calcineurin inhibitor changed from cyclosporin to tacrolimus. Prednisolone was usually discontinued by 3 to 6 months post transplant unless otherwise indicated. Deviation from the protocol occurred only in the setting of an adverse event or graft rejection. Acute rejection was usually managed with 1 g of methyl prednisolone intravenously for 3 days followed by re-introduction of oral steroids with or without an increased dose of, or switch to, an alternative calcineurin inhibitor. Chronic rejection was managed with the latter, and in a proportion of patients, mycophenolate was prescribed. One patient was changed to sirolimus following the development of renal dysfunction.
Formal smoking cessation advice was not given either at the time of transplant assessment or during the follow-up period. In addition, it was not routine practice within our unit to recommend smoking cessation between 1996 and 2000.
Collected baseline demographic data consisted of age at listing, gender, ethnicity, etiology of liver disease, Child-Pugh score and Model for End-Stage Liver Disease score at listing, severity of circulatory dysfunction at listing (creatinine, glomerular filtration rate, and sodium), time on transplant waiting list, and the presence of significant comorbidity (diabetes, treated hypertension, treated dyslipidemia, and cardiovascular disease). The glomerular filtration rate was estimated with the Modification of Diet in Renal Disease Study 6-variable equation.7
The main outcomes examined were graft failure (death as a result of graft dysfunction or retransplantation), all-cause mortality, cardiovascular-related mortality (fatal coronary or cerebrovascular event), sepsis-related mortality, and malignancy-related mortality. In addition, data were collected on the following posttransplant variables: calcineurin inhibitor, need for postoperative renal replacement therapy, development of diabetes, treated hypertension, treated dyslipidemia, coronary artery disease, cerebrovascular disease, peripheral vascular disease, significant malignancy (excluding in situ malignancy), pneumonia, significant bacterial/fungal infection (requiring intravenous antibiotics or hospital admission), cytomegalovirus (CMV) infection (CMV viremia), hepatic artery thrombosis, at least one episode of early acute rejection (<3 months post transplant), at least one episode of late acute rejection (>3 months post transplant), chronic rejection, and disease recurrence.
Normally distributed continuous variables and nonparametric continuous variables were compared with the Student t test and Mann-Whitney test, respectively. Chi-squared analysis was used for comparison of categorical data. The effect of smoking at listing on the main outcomes was assessed with Kaplan-Meier plots and log-rank test for differences. Backwards stepwise Cox proportional hazards models, verified with forwards models, were used to determine the independent predictive factors for each outcome, with only those variables with P < 0.10 being included in the multivariate analysis. P < 0.05 was considered significant. Data were analyzed with the SPSS 15 package.
All values are expressed as mean and standard deviation or median and interquartile range as appropriate.
Of the 55 patients (42%) with a positive smoking history, 31 were active smokers and 24 were exsmokers at the time of transplant assessment, whereas the remaining 77 patients (58%) reported life-long nonsmoking. Therefore, the prevalence of active smoking was 23%, which is slightly less than the national average of 27% to 28% during the time period of 1994 to 2001.8 The median reported period of abstinence from tobacco use among the exsmokers was 9 years (range 0.12 to 24 years).
Characteristics of Smokers, Exsmokers, and Nonsmokers
Baseline demographic data are documented in Table 1. There was no difference in gender, race, severity of liver disease, or presence of relevant pretransplant comorbidity between the 3 groups. However, patients who smoked at the time of transplant assessment were younger than those who did not (median age 48 years versus 53 years, P = 0.051). Furthermore, active smokers were more likely to have alcoholic liver disease (35% versus 13%, P = 0.008) when compared with life-long nonsmokers. In fact, the prevalence of a positive smoking history among those with alcohol-related liver disease was twice the prevalence among those without alcohol-related liver disease (63% versus 36%, P = 0.012).
Table 1. Baseline Demographic Data
Active Smokers (n = 31)
Exsmokers (n = 24)
Nonsmokers (n = 77)
Abbreviations: eGFR, estimated glomerular filtration rate; MELD, Model for End-Stage Liver Disease.
Immunosuppression was similar, with 43%, 36%, and 35% of smokers, exsmokers, and nonsmokers, respectively, receiving cyclosporine therapy (P = 0.706). Moreover, during the posttransplant period, there was no difference in the proportion of patients treated for hypertension (52% versus 47% versus 48%, P = 0.935), dyslipidemia (10% versus 11% versus 12%, P = 0.964), or diabetes (16% versus 17% versus 27%, P = 0.327) between the 3 groups.
Influence of Smoking on Outcomes
Graft and Patient Survival
Smoking at the time of transplant assessment did not appear to influence graft survival. The estimated 1- and 10-year graft survival rates were 90% and 85%, respectively, for active smokers and 91% and 83% for life-long nonsmokers (Fig. 1; P = 0.880). Furthermore, there was no difference in the proportion of smokers and nonsmokers who were diagnosed with hepatic artery thrombosis (10% versus 9%, P = 0.924), early acute rejection (42% versus 40%, P = 0.873) or chronic rejection (7% versus 5%, P = 0.796), and disease recurrence (21% versus 27%, P = 0.467).
In contrast, smoking was associated with increased all-cause mortality following liver transplantation. The estimated 1-, 5-, and 10-year survival for active smokers was 94%, 68%, and 54%, respectively, versus 94%, 83%, and 77% for life-long nonsmokers (Fig. 1; P = 0.040). This increased risk of death was not maintained in exsmokers, who had an estimated 5-year survival of 75% and 10-year survival of 57% (P = 0.134 with respect to nonsmokers). On univariate analysis (Table 2), additional factors associated with all-cause mortality were need for perioperative renal replacement therapy (P = 0.009) and sodium at listing when considered as a continuous variable (P = 0.013). Notably, age, gender, a diagnosis of alcoholic liver disease, severity of liver disease, and the presence of diabetes were not risk factors. On multivariate Cox proportional hazards analysis, only active smoking at the time of transplant assessment (hazard ratio 2.23, P = 0.030) and perioperative renal replacement therapy (hazard ratio 2.78, P = 0.006) were independent predictors of death.
Table 2. Factors Associated with All-Cause Mortality Post Transplant: Univariate and Multivariate Analyses
Univariate HR (95% CI)
Multivariate HR (95% CI)
NOTE: The reference group (relative risk 1.00) is characterized as follows: male gender, liver disease not secondary to alcohol, no hepatocellular carcinoma, life-long nonsmoker, no perioperative RRT, and no pretransplant or posttransplant diabetes.
To determine the reason for increased mortality among active smokers with respect to life-long nonsmokers, disease-specific mortality was then assessed.
The proportion of active smokers compared with nonsmokers who developed coronary artery disease (16% versus 13%, P = 0.669), cerebrovascular disease (16% versus 8%, P = 0.195), or peripheral vascular disease (6% versus 1%, P = 0.140) during the follow-up period was no different.
However, active smokers did demonstrate increased cardiovascular-related mortality, with an estimated 13% of smokers and 2% of life-long nonsmokers experiencing a fatal cardiovascular event by 10 years post transplant (P = 0.010). On univariate analysis, alcoholic liver disease (P = 0.047) was also associated with cardiovascular-related mortality, whereas sodium (P = 0.064) and perioperative renal replacement therapy (P = 0.089) demonstrated a trend toward an association (Table 3). On multivariate analysis, the only independent predictor of cardiovascular-related death was active smoking at the time of transplant assessment (P = 0.033).
Table 3. Factors Associated with Cardiovascular-Related Mortality Post Transplant: Univariate and Multivariate Analyses
Univariate HR (95% CI)
Multivariate HR (95% CI)
NOTE: Factors not associated with cardiovascular-related mortality post transplant on univariate analysis are as follows: age, gender, pretransplant diabetes, posttransplant diabetes, hypertension, dyslipidemia, creatinine at listing, estimated glomerular filtration rate at listing, and calcineurin inhibitor. The reference group (relative risk 1.00) is characterized as follows: liver disease not secondary to alcohol, life-long nonsmoker, and no perioperative RRT.
During the immediate postoperative period, active smokers were more likely than nonsmokers to develop pneumonia (48% versus 18%, P = 0.006). However, there was no difference in the rates of significant bacterial/fungal infection (58% versus 44%, P = 0.191) or CMV viremia (6% versus 10%, P = 0.523), and both the need for renal replacement therapy (13% versus 17%, P = 0.607) and survival to discharge (94% versus 97%, P = 0.337) were also similar. Post discharge, the proportions of smokers compared with nonsmokers who developed pneumonia (21% versus 12%, P = 0.258), significant bacterial/fungal infection (31% versus 20%, P = 0.231), and CMV viremia (17% versus 24%, P = 0.456) requiring hospital admission were no different.
Nevertheless, smokers did demonstrate increased sepsis-related mortality. An estimated 26% of active smokers and 4% of life-long nonsmokers experienced a fatal septic event by 10 years post transplant (P = 0.022). The only other variable associated with sepsis-related mortality on univariate analysis was early acute rejection (P = 0.044; Table 4). On multivariate analysis, both active smoking (P = 0.041) at the time of transplant assessment and early acute rejection (P = 0.048) were independent predictors of death.
Table 4. Factors Associated with Sepsis-Related Mortality Post Transplant: Univariate and Multivariate Analyses
Univariate HR (95% CI)
Multivariate HR (95% CI)
NOTE: Factors not associated with sepsis-related mortality post transplant on univariate analysis are as follows: age, gender, alcoholic liver disease, pretransplant diabetes, posttransplant diabetes, late acute rejection, chronic rejection, and calcineurin inhibitor. The reference group (relative risk 1.00) is characterized as follows: life-long nonsmoker and no early acute rejection.
The proportions of smokers and nonsmokers who developed significant de novo malignancy (0% versus 12%, P = 0.057) or demonstrated recurrence of tumor (3% versus 3%, P = 0.857) were no different.
Furthermore, tobacco use was not predictive of malignancy-related mortality, with an estimated 4% of active smokers at time of assessment and 8% of life-long nonsmokers dying of malignancy by 10 years post transplant (P = 0.605). Moreover, when death from de novo malignancy was analyzed specifically, there was no difference between smokers and nonsmokers (estimated 0% versus 5% by 10 years, P = 0.310).
Increasing surgical experience and improved immunosuppression have extended graft longevity in liver transplant recipients. As a result, in recent years, transplant physicians have become more focused on late morbidity and mortality post transplant. It is now well recognized that most deaths in patients surviving beyond 1 year are non–graft-related, with malignancy, cardiovascular disease, and infection being the most common extrahepatic causes of death.9, 10 Immunosuppressive therapy both directly and indirectly, via an increased risk of hypertension, dyslipidemia, and diabetes, is acknowledged to be a chief contributor to this increased nonhepatic mortality.11 However, smoking, an additional modifiable risk factor for many of these disease states, has received little attention from the transplant community. More specifically, the effects of tobacco use on the risk of non–graft-related morbidity and mortality as well as graft function post transplant remain unclear.
We have shown for the first time that smoking at the time of liver transplant assessment is associated with increased all-cause mortality post transplant. Smokers are twice as likely as nonsmokers to die during a follow-up period of 10 years. This supports findings in both renal and heart transplant recipients and in the general population of an association between smoking and premature death.12, 13
In our cohort, this increased mortality rate was not explained by graft dysfunction: a similar proportion of patients demonstrated rejection or recurrence during the follow-up period, and there was no difference in risk of graft failure. Notably, in contrast to the recent study by Pungpapong et al.,14 we did not demonstrate any predisposition to hepatic artery thrombosis. Such an association is highly plausible as smokers have increased fibrinogen levels, impaired fibrinolysis, and increased platelet thrombus formation.15 We do not have an explanation for this discrepancy, but because of the retrospective nature of our study, it is possible that smoking cessation prior to transplantation may have influenced our findings. On the other hand, the prothrombotic effects of tobacco are not immediately reversible, and as a result, it is our opinion that this is unlikely to have been an important factor. Furthermore, it has been shown in renal transplant recipients that most smokers at time of assessment continue to smoke post transplant, whereas few nonsmokers adopt the habit.16 An additional explanation could therefore be that a proportion of our smokers were not “heavy” smokers. Unfortunately, assessment documentation did not allow accurate estimation of pack years, although the Scottish Health Survey of 2003 found that the mean number of cigarettes smoked per day by current smokers was 15.9 and 14.7 for men and women, respectively.17
The increased mortality rate post liver transplant appears instead to be explained by non–graft-related death. Despite relatively small patient numbers, smokers demonstrated an increased risk of cardiovascular-related and sepsis-related mortality. The association between smoking and atherosclerosis is well established. In addition to the aforementioned prothrombotic effects, smoking causes endothelial dysfunction, potentiation of inflammation, and an altered lipid profile, which are important in the initiation and progression of atherothrombotic disease.18 In contrast, the link between smoking and infection is less recognized. Smoking has been shown to cause local structural changes within the respiratory tract, including peribronchiolar inflammation and fibrosis, impaired mucociliary clearance, and altered pathogen adherence. Moreover, smokers have evidence of an impaired systemic immune response with suppressed immunoglobulin levels, an altered CD4 to CD8 cell ratio, and reduced phagocyte activity.2 These factors may explain the relationship between smoking and community-acquired pneumonia, postoperative pneumonia, wound infection and invasive pneumococcal disease, and the increased fatality of bacteremic smokers.19–23 In the immunosuppressed liver transplant recipient, it is conceivable that the susceptibility to infection of smokers is exaggerated and the potential for adverse outcome is greater.
Surprisingly, we did not demonstrate that smoking was associated with increased malignancy or malignancy-related mortality. Previous studies have found that smoking liver transplant recipients are more likely to develop noncutaneous de novo malignancy.24, 25 However, despite this, Herrero et al.24 did not find that smoking was an independent predictor of malignancy-related mortality. Perhaps an explanation for this lack of association is that transplant recipients who smoke are dying of an alternative pathology.
In this study, the increased risk of death among active smokers did not appear to persist in patients who had a previous smoking history but had managed to quit by the time of transplant assessment. This supports findings in renal transplant recipients, which suggest that patients who stop smoking before transplantation have survival similar to that of life-long nonsmokers.16 However, it has also been shown that renal transplant patients who quit more than 5 years before transplantation do not significantly reduce their risk of cardiovascular disease and that the number of pack years smoked may be a greater predictor of death.12, 16 In our cohort, only a small number of patients were exsmokers. Furthermore, we had no information regarding posttransplant smoking status, and these patients may be more at risk of tobacco use resumption. Therefore, it is difficult to draw any conclusions regarding long-term survival in these patients specifically.
In conclusion, smoking is an independent predictor of all-cause mortality, cardiovascular-related mortality, and sepsis-related mortality post liver transplantation. Although the risk for exsmokers remains unclear, it is likely that the successful cessation of tobacco use will have a beneficial effect on long-term outcomes. As a result, smoking should be considered alongside other modifiable risk factors such as hypertension, dyslipidemia, and diabetes and aggressively targeted by transplant physicians.
The authors thank Kirsty Martin, database manager, and Cat Graham, statistician, for their help and advice.