The lung transplantation candidate population is heterogeneous and survival benefit has not been established for all patient groups. UK data from a cohort of 1997 adult (aged ≥ 16), first lung transplant candidates (listed July 1995 to July 2006, follow-up to December 2007) were analyzed by diagnosis, to assess mortality relative to continued listing. Donor lungs were primarily allocated according to local criteria. Diagnosis groups studied were cystic fibrosis (430), bronchiectasis (123), pulmonary hypertension (74), diffuse parenchymal lung disease (564), chronic obstructive pulmonary disease (COPD, 647) and other (159). The proportion of patients in each group who died while listed varied significantly (respectively 37%, 48%, 41%, 49%, 19%, 38%). All groups had an increased risk of death at transplant, which fell below waiting list risk of death within 4.3 months. Thereafter, the hazard ratio for death relative to listing ranged from 0.34 for cystic fibrosis to 0.64 for COPD (p < 0.05 all groups except pulmonary hypertension). Mortality reduction was greater after bilateral lung transplantation in pulmonary fibrosis patients (p = 0.049), but not in COPD patients. Transplantation appeared to improve survival for all groups. Differential waiting list and posttransplant mortality by diagnosis suggest further use and development of algorithms to inform lung allocation.
Although International Registry data indicate that survival after lung transplantation for patients with end-stage pulmonary disease has improved over time (1) there is uncertainty whether lung transplantation confers prognostic benefit. Lung transplantation survival rates are lower than those for heart and other solid organ transplants (1–4) and survival curves have an accelerated attrition rate due mainly to chronic allograft dysfunction in the form of bronchiolitis obliterans syndrome (5). In addition, lung transplant recipients represent a heterogeneous population, with different diagnostic groups having different survival rates, both while awaiting and posttransplantation (1). Thus patients with different lung pathologies may derive varying levels of benefit from transplantation. Furthermore, some patients may be considered suitable for either single lung transplantation (SLT) or bilateral lung transplantation (BLT). In these patients SLT could be considered a partial treatment, with a lower treatment effect both in term of survival and functional benefit (6). Issues of donor organ scarcity may mean that SLT allows more patients to be treated and waiting times to be shorter. However, the relative benefit of accepting a SLT or remaining on the waiting list until a suitable pair of lungs becomes available is not established.
Previous US cohort studies have suggested that lung transplantation does not produce survival benefit for a large proportion of patients with COPD (7) (albeit considering an earlier era) and for children with CF (8). UK and European studies have been more encouraging but were based on small numbers of transplant candidates (9–13), with limited follow- up (9) or have combined lung with heart-lung transplant candidates (14,15).
This study was conducted to estimate the effect on survival of lung transplantation for the main diagnostic groups treated, using data from a national (UK) data base, taking into account the effect of transplant type (SLT vs. BLT).
Materials and Methods
This cohort study included all consecutive, first-time, adult (age ≥16) lung transplant registrations in the UK between July 1995 and July 2006, monitored to December 2007 to ensure at least 18 months follow-up. Heart-lung transplant candidates and those aged <16 were excluded. Patients registered retrospectively (after transplantation) were also excluded since their waiting time was unclear (n = 60). In the UK, lung transplantation was only performed in government-designated centers (n = 7 adult centers). Patients were listed for lung transplantation according to UK national criteria (see Appendix 1). Lungs were allocated first by geographical proximity, with each center responsible for allocating the organs to suitable recipients, according to ABO compatibility and size. If they could not be used locally they were offered to each other center on a rotational basis. There was no central urgency scheme or allocation system, although this may have occurred locally. All designated centers had a mandatory requirement to provide registration and follow-up audit data. Audit data, collected at listing, at transplant, at 3 months posttransplant and annually thereafter until the patient's death, were processed by UK Transplant and submitted to the UK Cardiothoracic Transplant Audit (UKTCA) monthly. Data validation occurred via a computer checking system and intermittent case note review. Initially data were collected on the basis of presumed consent but more recently informed consent has been sought when listing the patient for transplantation. In the UK, audit projects did not require separate research ethics committee (institutional review board) approval (16).
The following diagnostic groups were considered separately: Cystic Fibrosis (CF), Bronchiectasis (BE), Pulmonary Hypertension (PH), Diffuse Parenchymal Lung Disease (DPLD) and Chronic Obstructive Pulmonary Disease (including emphysema and α1 antitrypsin deficiency) (COPD). All other patients were combined into a heterogeneous group named ‘Other’. PH included primary pulmonary hypertension, Eisenmenger's syndrome and other congenital heart/lung diseases. DPLD included a range of fibrosing lung diseases, the details of which were not recorded on the national data base. The most common diagnosis in the ‘Other’ group was Sarcoidosis (n = 50, 31%).
The study outcome was time to patient death. Survival status was ascertained for patients at listing, at transplantation and at 3 months and annually after transplantation. Death before or after transplantation was recorded in the UKCTA database, otherwise survival time was censored at last contact. Noninformative censoring was assumed. Patients who were temporarily suspended from the list due to acute illness were included in the analysis from registration to transplant, death or December 2007. Exploratory analysis showed that patients permanently removed from the waiting list for nontransplant reasons could not be assumed to be randomly censored; therefore survival status of these patients was traced using the Office for National Statistics data base. They were included in the survival analysis from registration to death or December 2007. Thus for 100% of patients included in the study survival status was known within, at most, 1 year of the end of the study, and for recent patients within 3 months. For the 5 patients who had a second transplant during the study period, total patient survival was used.
Patient characteristics were summarized as mean (standard deviation) or number (percentage of within-group total). Within each of the diagnostic groups a nonproportional hazards survival analysis was undertaken. The hazard (or risk of death) at time t was modeled as follows:
where δtx = 1 is a marker for lung transplantation having been completed at or before time t, twait is the time spent on the waiting list before transplantation and λ0, λ1 and λ2 are coefficients that determine the shape of the relative hazard at different times after transplantation (17). Specifically, exponential(λ0+λ1) is the relative hazard at the time of transplantation, exponential(λ0) is the level to which the relative hazard converges and λ2 determines the rate at which the relative hazard converges. This model was chosen since the fit to the observed data was significantly better for all diagnostic groups (according to likelihood ratio statistics) when compared to (i) a proportional hazards model in which the hazards changed at transplantation, and (ii) a proportional hazards model in which the hazards changed twice, at transplantation and at 30 days posttransplantation. The time after transplant at which the immediate postoperative risk of death fell below the preoperative risk (crossover point), and the time after transplant at which the increased risk immediately posttransplant was outweighed by the subsequent lower risk (equity point) were estimated. The equity point was calculated as the time when the cumulative relative hazards sum to zero. Patient resampling (1000 bootstrap samples) was undertaken to estimate confidence intervals of quantities such as the coefficients λ0, λ1 and λ2, the crossover point and the equity point. In exploratory Cox regression analysis age, sex, body mass index (BMI), FEV1, donor age and ischemic time were assessed for their association with death. Models considered both pre- and posttransplantation separately, and nonproportional hazards models (for posttransplantation), stratified for diagnosis group. Continuous covariates were split approximately into tertiles (high, medium and low values). Initially covariates were added separately to the models, likelihood ratio tests were used to select covariates for further study, with either a p-value of less than 0.05 or a p-value close to 0.05 and supporting evidence from the Akaike Information Criterion (i.e. a lower AIC compared with the next simpler model). Overall, this analysis showed that survival was related to age, sex and body mass index (BMI), but not FEV1, donor age or ischemic time, and this was generally confirmed in separate analyses for each group, although statistics varied due to the differing number of patients in each group. The final Cox nonproportional hazards models reported in the results section included these variables as covariates on the baseline (pretransplant) hazards and on the posttransplant hazards. For COPD and DPLD groups, after adjusting for age, sex and BMI, the effect of transplant type (SLT vs. BLT) was investigated by allowing the parameters λ0, λ1 and λ2 to differ. This model was compared with the model that assumed the post-crossover point survival rate was the same for SLT and BLT (same λ0) using likelihood ratio tests. After adjusting for age, sex, BMI (and transplant type for COPD and DPLD groups) there was no significant change in survival over calendar time. Maximum likelihood estimates of the parameters were found using optimization programs written in the freely available statistical software R (18). Survival on the waiting list and posttransplant survival were calculated using Kaplan–Meier estimates with median posttransplant survival time taken from the point at which this curve reached the 50% line. Comparison of survival rates between diagnostic groups used likelihood ratio tests from the Cox regression models.
The national lung transplantation data base listed 1999 adult, first-transplant candidates between July 1995 and July 2006 (Figure 1). Inadequate diagnostic data excluded two patients. The remaining 1997 were included in the study, of whom 1143 underwent lung transplantation. Of 854 patients not transplanted, 705 died prior to transplantation and the remaining 149 were alive in December 2007. Of 1143 transplant recipients 540 had died by the end of the study.
The most common diagnoses of patients listed for transplant were COPD, DPLD and CF (Table 1). COPD patients were more likely to receive a transplant (crude comparison of proportions p < 0.001) while DPLD patients had the shortest wait for transplant.
Table 1. Waiting list and transplant activity in the UK July 1995 to July 2006 by diagnostic group, with follow-up to December 2007
*Time until end of study, transplant or death on the waiting list.
NA = upper limit cannot be calculated from the current data set.
Median time listed & 95% CI (days)*
Median survival posttransplant & 95% CI (days)
Descriptive characteristics are summarized by diagnosis in Table 2. As expected CF patients were listed at a much younger age than other groups and received lungs from younger donors. Mal-absorption problems that are common in CF are reflected by the low mean body mass index. The mean FEV1 was substantially less than 1 liter in the CF, BE and COPD groups.
Table 2. Patient characteristics by diagnostic group
SLT was the main operative treatment in the DPLD group and for approximately half the COPD group. SLT was not used in septic lung disease patients with the exception of a single case of CF that underwent SLT with contralateral pneumonectomy.
All subsequent analyses were adjusted for age, sex and BMI. There were significant differences in waiting list survival according to diagnostic group. Patients with DPLD were at greater risk and those with COPD were at lower risk than other groups (p < 0.001, data not shown).
Following transplantation patients with DPLD continued to be at higher risk of death than other groups and CF patients had the best survival postoperatively (p < 0.001, Figure 2). In addition the PH group had higher mortality posttransplant but the survival rate was not measured precisely due to small numbers.
Figure 3 shows the estimated risk of death over time, relative to continued listing for the six groups studied. All groups had high initial posttransplant risk that dropped below the line representing ‘risk equal to pretransplant’ (hazard ratio = 1) within the first year after transplantation. There was some variation between the groups immediately after transplantation and after the crossover point (p < 0.001).
Beyond the crossover point, the posttransplant risk, relative to pretransplant risk (exponential(λ0)), is displayed in Figure 4. All groups had a significant decrease in risk with the exception of the smallest group (PH) for which the hazard ratio was estimated imprecisely. Hazard ratios ranged from 0.34 for CF to 0.64 for COPD.
There was variation in the time-point at which the risk of dying posttransplant fell below the risk of dying on the waiting list (crossover point—Table 3). The initial increase in risk (compared to continued waiting) passed within two months for all groups with the exception of the COPD group, for which the crossover point occurred at 4.3 months. The proportion of patients that survived to the crossover point ranged from 68% for PH to 91% for CF.
Table 3. Post-crossover hazard ratio (exponential(λ0)), crossover and equity points by diagnosis
CF (n = 430)
BE (n = 123)
PH (n = 74)
DPLD (n = 564)
COPD (n = 647)
Other (n = 159)
∞ Indicates that there are insufficient data to identify the upper limit of the confidence interval.
NA = means that this upper limit could not be calculated from the current data set.
Crossover point in days (95%CI)
46 (27, 131)
68 (30, ∞)
49 (26, ∞)
59 (39, 110)
130 (94, 222)
43 (24, 175)
Percentage surviving to crossover point
Post-crossover hazard ratio (95%CI) (exponential(λ0))
0.34 (0.23, 0.51)
0.44 (0.22, 0.88)
0.45 (0.16, 1.26)
0.36 (0.26, 0.51)
0.64 (0.48, 0.86)
0.39 (0.22, 0.69)
Equity point in days (95% CI)
160 (87, 468)
286 (110, ∞)
1380 (394, ∞)
170 (103, 330)
905 (551, 2251)
264 (140, 1277)
Median survival posttransplant (95%CI) (days)
In addition, the point at which the posttransplant initial increase in risk was outweighed by subsequent reduction in posttransplant risk is reported (equity point—Table 3). Equity of risk occurred within the first year for CF, BE, DPLD and the ‘Other’ group. Patients with COPD and PH took longer to achieve this point on average. All groups had median posttransplant survival in excess of the equity point (Table 3). For three of the six groups (CF, DPLD and ‘Other’) the confidence interval on the equity point does not encompass the observed median survival and these are the groups with the greatest risk reduction beyond the crossover point (hazard ratios 0.34, 0.36 and 0.39, respectively).
The COPD and DPLD groups were large enough, and contained a sufficient proportion of SLT, to compare transplant types. For the DPLD group, adjusting for age, sex and BMI, survival immediately after transplantation was not significantly different for SLT and BLT patients. However, for COPD patients SLT had significantly poorer survival immediately after transplant (p = 0.03, Wald test for λ1 parameter). For DPLD and COPD groups the post-crossover point hazard ratios for SLT were slightly higher than those for BLT (exponential (λ0) –Table 4).
Table 4. Post-crossover hazard ratio (exponential(λ0)), crossover and equity points: Comparison of single and bilateral lung transplants
Bilateral lung transplant
Single lung transplant
*p-values are not calculated for these comparisons since the crossover and equity points are complicated functions of other parameters.
The crossover point is the time-point at which the risk of dying posttransplant fell below the risk of dying on the waiting list.
The equity point is the time after transplant at which the increased risk immediately posttransplant was outweighed by the subsequent lower risk.
Number of transplants (%)
Diffuse parenchymal lung disease (n = 564)
Chronic obstructive pulmonary disease (n = 647)
Hazard ratio (95% confidence interval)
Diffuse parenchymal lung disease (n = 564)
0.19 (0.08, 0.41)
0.44 (0.31, 0.62)
Chronic obstructive pulmonary disease (n = 647)
0.55 (0.37, 0.80)
0.69 (0.50, 0.94)
Crossover point in days (95% confidence interval)
Diffuse parenchymal lung disease (n = 564)
62 (17, 191)
50 (27, 146)
Chronic obstructive pulmonary disease (n = 647)
177 (116, 392)
106 (70, 274)
Equity point in days (95% confidence interval)
Diffuse parenchymal lung disease (n = 564)
176 (75, 1388)
156 (90, 608)
Chronic obstructive pulmonary disease (n = 647)
786 (456, 2180)
1098 (582, 6799)
This study suggests that lung transplantation within the UK lung allocation system confers a disease-specific survival benefit in patients with advanced lung pathology including cystic fibrosis, chronic obstructive pulmonary disease and diffuse parenchymal lung disease.
Chronic obstructive pulmonary disease
The survival benefit afforded to patients with COPD, was significant, (Table 3) albeit less than for other diagnostic groups, and this was consistent with earlier UK and European studies (14,15). In addition, 83% survived to the crossover point at which they might expect their risk of death to be less than continued waiting (Table 3). Some studies have challenged the use of lung transplantation for COPD. Stavem et al. conducted an analysis of the adult lung transplant programe in Norway and concluded that there was no survival advantage for patients with COPD, but for non-COPD patients there was an advantage for BLT but not for SLT (19). Similarly, in the early study by Hosenpud et al. for 1274 adult US transplant candidates with COPD the posttransplant risk of death never fell below the pretransplant risk during the study period (7). One reason for this lack of survival gain included the high-survival rate while listed, 80% and 70% at 2 and 3 years after listing. The low pretransplant mortality in our study is consistent with these data and suggests that it is more difficult to assess prognosis in this disease category as patients may have maintained good survival despite very poor lung function and related poor quality of life. In addition, in the Hosenpud study there may also have been bias due to selective removal from the waiting list (7).
In our study CF patients most rapidly gained a net survival benefit after lung transplantation and 91% of the patients survived to the crossover point when their risk of death fell below that of remaining on the list. In addition, this group had the greatest risk reduction after the crossover point (Table 3). Recently, there has been controversy around the value of transplantation for children with CF (8). A relatively large US study by Liou et al. used a proportional-hazards survival model that assumed a switch in risk of death at transplantation with the ratio of post-:pretransplant risk constant over time thereafter (8). Pediatric patients had higher risk of death posttransplant (hazard ratio 1.89) and only 5 of 248 pediatric lung transplant recipients were found to have benefited. A preliminary analysis of results for adult CF patients given in supplementary information to Liou et al. demonstrated a hazard ratio of 1.49, again indicating worse survival after lung transplantation (8). One reason for lack of effect in this report may have been that the pediatric CF patients were transplanted earlier in their disease natural history; the mean FEV1 was higher (31% predicted) compared with patients listed in our study (24% predicted). Previous studies from the same group recommended transplantation for adults with less than 50% predicted 5-year survival and the absence of both B. cepacia colonization and CF-related arthropathy (20). Patients in all subgroups had better survival without transplantation (20). Further limitations of the study by Liou et al. included the use of the proportional hazards model, which is dominated by the early posttransplant hazards, the exclusion of 70 (of 584) patients due to missing covariate values that introduces bias if records were not missing at random (21), and the poor 5-year survival rate of less than 40% (8).
Diffuse parenchymal lung disease
Patients with DPLD had the poorest survival rates both pre- and posttransplant but, despite this, had significantly better post-crossover point survival relative to continued listing. The overall risk reduction was 56% (Table 3) which is consistent with earlier studies in the United Kingdom (14), France (13), the United States (7) and the Eurotransplant region (15).
The smallest group studied had PH and although the post-crossover point hazard ratio was not significantly less than 1 (Table 3), it was very close to the estimates for other diagnostic groups. Since this is consistent with previous studies in the United Kingdom (14) and the Eurotransplant region (15), it is likely to be robust to the accumulation of further data.
Bronchiectasis and other
The candidates with BE formed a small group with post-crossover point hazard ratio (exponential(λ0)) similar to most other diagnostic groups (Table 3) and consistent with an early UK study that included a large proportion of heart-lung transplants (14). Patients with less common diseases were combined into a single group and were found to have similar post-crossover point risk reduction due to transplantation (exponential(λ0)), again consistent with Charman et al. (14) and De Meester et al. (15) This group was heterogeneous including both obstructive and restrictive lung diseases, with Sarcoidosis the most common (n = 50), and of the ‘Other’ group, 56% of the transplants were SLT.
Implications for case-selection and timing of transplantation
The large US studies that suggested that transplantation does not confer survival benefit for COPD (7) and CF (8) contained patients listed prior to the recently implemented lung allocation scheme (22). The new scheme uses objective models to estimate a lung allocation score, which incorporates both expected transplant benefit and estimated waiting list urgency for each transplant candidate (22,23) This system has been in operation since May 2005 and has both reduced the size of the active waiting list and made allocation of organs more efficient. It remains to be seen whether survival gain will improve. Within our national cohort, lung transplantation conferred a clear survival benefit for patients with CF, DPLD and COPD. However, differential waiting list mortality observed according to diagnosis, together with the different survival rates posttransplantation support the further use and development of algorithms to prioritize lung allocation either nationally or locally to maximize positive outcomes.
At listing, patients and clinicians can use the information in Table 3 to aid the decision strategy; for example, would a COPD patient risk a 17% chance of death within 19 weeks of transplant in order to reduce subsequent risk by one-third, and gain improved lung function and quality of life? Of course this decision may vary between patients and clinicians and will be influenced by a range of other factors specific to the individual under consideration, such as quality of life and comorbidities.
Strengths of our study
Analysis of results for this national cohort of lung transplant candidates over a 12-year period has allowed us to assess the survival benefit accrued with greater precision and with more realistic models than previous studies. Inclusion of all transplants undertaken nationally ensures that this analysis is representative of UK practice and that there is no reporting bias. This single nation approach was ideal for this study as within the UK there are only a small number of designated centers and an obligatory audit with central data collection. We were able to trace survival status for all patients who were listed prospectively for transplantation in this time period and so bias due to selective removals from the waiting list was minimized. For patients with COPD and BE there was a significantly higher mortality rate after removal from the list compared with those remaining on the list after similar waiting time. Therefore the common practice of censoring these cases at removal would have introduced bias by underestimating the death rate without transplantation and therefore the survival gain due to transplantation.
Limitations of our study
This and all previous studies of lung transplant benefit are observational in nature; no randomized controlled trials have been undertaken. Results should be interpreted in the context of clinical practice during the study. Although we have used realistically sophisticated survival models they are unlikely to completely encompass all variation in the experience of transplant candidates. In addition, our analyses were restricted to survival and do not include the possible functional benefits of transplantation.
In this audit-based study there was little detailed information available so that the diagnosis groups were necessarily broad and some misclassification was possible. This was particularly the case for the pulmonary hypertension group, which included some secondary hypertension cases, and for diffuse parenchymal lung disease, which is a heterogeneous group. No validation of reported diagnosis was possible. Further, although we were able to adjust for some covariates, a comprehensive risk adjustment was not possible. This is more likely to be a limitation in the comparison of BLT and SLT than for the assessment of transplant effect overall. The decision to transplant a single lung is multifactorial so that the two groups are not directly comparable and the clinical implication of the observed association between transplant type and survival is unclear.
Although BLT appeared more effective than SLT we cannot conclude that SLT should not be undertaken, since (i) risk adjustment was inadequate and (ii) the total lifetime gained due to treating two individuals with SLT, rather than one individual with BLT, may be justified in the context of a rationed (due to donor organ shortage) treatment. Additionally, in some organ donors only one lung is suitable for transplantation.
Finally, this study contained exclusively UK patients and depended on the organ allocation system in place during the study so that results may not be directly generalizable to other systems such as Eurotransplant and UNOS.
Lung transplantation as practiced in the UK during the study period improved survival for all diagnostic groups although the benefit was lower for SLT than for BLT. Differential waiting-list and posttransplant mortality according to diagnosis suggests further use and development of algorithms such as the UNOS lung allocation scheme (23), particularly for COPD patients.
Members of the UKCTA Steering Group and participating centers: Dr. NR Banner (Chairman) and Professor G. Dreyfus (Harefield Hospital, Middlesex), Professor RS Bonser (Queen Elizabeth Hospital, Birmingham), Mr. N Yonan (Wythenshawe Hospital, Manchester), Professor J.H. Dark (Freeman Hospital, Newcastle), Professor M. Elliot (Great Ormond Street Hospital, London), Mr. P. Braidley (Northern General Hospital, Sheffield), Mr. S. Tsui (Papworth Hospital, Cambridge), Mr. A.J. Murday (Golden Jubilee National Hospital, Glasgow), Dr. J. van der Meulen (Clinical Effectiveness Unit, The Royal College of Surgeons of England), Professor D. Collett (UK Transplant), Dr. W. Gutteridge (NHS National Commissioning Group). We also thank Professor Paul Corris (Freeman Hospital, Newcastle) for helpful comments and the patients who have allowed their data to be used.
Appendix 1 UK Transplant criteria for listing for lung transplantation December 2008
• Usually less than 60 years of age for a bilateral lung and heart-lung transplant, less than 65 years for single lung transplant, as there is increase in comorbidity illness with the ageing process. Outcome is less satisfactory. However, consider biologically fit older patients.
• The following conditions will be considered:
○ Idiopathic pulmonary fibrosis.
○ Occupational fibrotic lung disease.
○ Drug/toxic induced fibrotic lung disease.
○ Sarcoidosis, if burned out.
○ Obstructive lung disease.
○ Emphysema including alpha 1 antitrypsin deficiency.
○ Primary pulmonary hypertension.
○ Complex congenital heart disease and Eisenmenger's syndrome.
○ Cystic fibrosis.
Contraindications to Lung and Heart-Lung Transplantation
1 Absolute Contraindications
• Related comorbidity with advanced ageing.
• Severe right heart failure. Consider heart/lung transplantation.
• Active peptic ulcer or diverticulitis.
• Continued abuse of alcohol or other drugs.
• Irreversible secondary organ failure unless considered for a combined transplant.
• Psychiatric history likely to result in noncompliance and or persistent noncompliance with medical therapy.
2 Relative Contraindications
• HIV (subject to discussion with Transplant & Managing Director at UKT).
• Hepatitis B/C.
• Coronary artery disease, if for lung transplant only.
• Intubated and ventilator dependent.
• Obesity BMI >30.
• Chronic renal impairment with GFR <50 mL/min, unless candidate for combined renal transplant.
• Diabetes with target organ damage.
• Severe osteoporosis (bone mineral density >2 SDs less than predicted for age).