Study objectives
The DEVELOP-UK study is the first prospective multicentre study to be performed involving all of the adult cardiopulmonary transplant units across the UK. The objective was to assess the clinical effectiveness and cost-effectiveness of EVLP, a technology allowing objective assessment and reconditioning of unusable donor lungs, in increasing UK lung transplantation activity. Its strategic importance was recognised by the British Transplantation Society, the NHSBT, NHS specialist commissioners and by patient groups during the study design and funding application process.
The DEVELOP-UK study was designed as a non-randomised, non-inferiority observational study with an adaptive design, with two interim analyses planned for when one-third and two-thirds of total enrolment was reached. The planned interim analyses provided the opportunity to determine if the primary end point had been achieved, but also to calculate if any change in sample size was required. The original primary objective was to determine if the 12-month survival of recipients of ex vivo assessed and reconditioned donor lungs (EVLP intervention group) is non-inferior to 12-month survival in recipients of standard donor lungs (control group). The secondary objective was to measure key early clinical outcomes in recipients and changes in their HRQoL in the treatment and control groups in their first post-transplant year. These data were planned to be used in a within-study cost–utility analysis and a Markov model-based evaluation. The former comparison was to be a direct head-to-head comparison of outcomes over 12 months, and the latter was to model the change in availability of lungs as well as extrapolating over the expected lifetime of those needing a lung transplant. In addition, patients’ perceptions and understandings of EVLP-reconditioned donor lungs were evaluated in a qualitative substudy.
Timelines and targets
The official start date for the study was 1 January 2012 based on release of NIHR funds to the study team. To allow for local R&D approvals, research staff recruitment and subcontractor contracts with sites to be secured, a 3-month run-in period was proposed, anticipating that recruitment would have started in all sites by 1 April 2012. The actual start date of the study was therefore 1 April 2012. Study recruitment and enrolment was scheduled to run for 36 months, with data collection ending by 42 months, and the final study report was scheduled at 45 months in October 2015. The recruitment targets for the study were set based on official waiting list numbers across the UK in the five adult lung transplant centres. The aim was for a total of 600 patients from the lung transplant waiting list to consent to participate. As this was a non-randomised study, enrolment into the two arms of the study, as defined by undergoing lung transplantation (standard and EVLP transplant), occurred independently, and the study was powered on a predicted 3 : 1 (standard arm to EVLP arm) enrolment ratio. The target for enrolment as lung transplant recipients was 408 patients (306 in standard arm and 102 in EVLP arm).
Trial hypothesis
Had the study run to its planned duration in terms of recruitment, the tested hypothesis was to have been that survival during the first 12 months after transplantation in recipients of EVLP-assessed and -reconditioned donor lungs is non-inferior to that in recipients of standard donor lungs. The primary outcome measure was survival during the first 12 months after lung transplantation.
Consequences for the study analyses of the early closure of the study
The study was powered on survival during the first year post transplant and the target recruitment as 306 patients in the standard transplant arm and 102 in the EVLP transplant arm. This chapter reflects the analysis possible following the early closure of the study, with recruitment of patients stopping in early July 2014 on the advice of the Trial Steering Committee because of a combination of poor recruitment rates into the EVLP arm of the study, and also a safety signal from a higher than expected SAE rate resulting from the need for ECMO support in the EVLP arm. The analyses described below are appropriate to the achieved sample size and differ from that intended and described in the original protocol. The analysis to compare standard with EVLP transplant groups, as well as the analysis of overall survival of patients awaiting transplantation, are descriptive in nature and, as such, do not reflect the initial intention of testing for non-inferiority of EVLP to standard transplants. The originally planned interim analyses, intended to test for the possibility of stoping the study early if non-inferiority was achieved and to re-examine the sample size, did not take place, as the recruitment threshold to trigger the first of these (34 EVLP transplants) was never reached.
Planned timelines for study analysis
In light of the change of circumstances of the analysis, the intent has changed from one of conducting interim analyses to inform the continuation of the study while it was in progress, to one of a single main report of outcome data to the funder. The plan was that data should be available for this analysis from the end of May 2015. In practice, the collection and validation of data were delayed because of the large number of missing data and data queries to sites, meaning that the analysis started in October 2015 and continued into early 2016.
Longer-term analysis plans
It is important to recognise that, despite the early closure of the study, there remains a rich data set, particularly in respect of information on standard transplants and on the total cohort of donor lungs exposed to EVLP. Consideration of this alone was not part of the original comparative analysis plans and, as a result, this is outside the scope of the main study analysis.
Following completion of the main study analyses, and outside the scope of the report to the funder, it will be possible to consider further analysis of the data from the standard transplant arm. This did not form part of the trial statistical analysis plan, as it was not in scope of the originally planned study, but the information from this large contemporary cohort of 200 transplants, including extensive follow-up data, is likely to be useful to future study. Possible approaches include modelling of outcome variables using baseline clinical covariates to identify possible predictors of successful outcome at baseline.
In addition, a comprehensive sampling protocol was in place to collect perfusate, lung tissue and BAL from the donor lungs undergoing EVLP. This will provide a valuable assessment of events at a cellular and molecular level that can be correlated with clinical information within the main study data set. The work on the mechanistic understanding of EVLP falls outside this report (as it is not the subject of the NIHR HTA programme funding), but the tissue sample data will contribute to this subsequent analysis.
Recruitment
The study officially commenced on 1 January 2012, opened to recruitment on 1 April 2012, and closed to recruitment on 9 July 2014. There was a temporary halt in recruitment into the EVLP arm from 6 April 2013 until mid-July 2013, when the study activity in the EVLP arm recommenced with a modified protocol.
The timings for study recruitment in each individual site are shown in . The data analysed and presented in this report were downloaded from the MACRO database in October 2015. Additional data were assembled in Microsoft Excel® 2010 (Microsoft Corporation, Redmond, WA, USA) files for some of the outcome measures (e.g. SF-36 and some lung function measurements) not recorded on CRFs, and for the donor characteristics, which were imported from the NHSBT database.
Approvals and study commencement by site. Dates of NHS R&D approval and first patient consented across five study sites
A total of 593 patients were screened (from records) for eligibility, of whom 98 did not meet eligibility criteria and a further eight declined to participate. Reasons for not meeting the eligibility criteria included age, need for pre-transplant cardiorespiratory support, and listed for heart–lung transplantation or transplantion of lungs and another organ. The screening failure rate was, therefore, only 16.1%, and the refusal rate for participation was just 1.3%.
A total of 487 patients consented to participate or completed an EOI form while on the transplant waiting list, of whom 19 were subsequently removed from the waiting list because of a change in their transplant eligibility, leaving 468 participants eligible to be included in the study. The breakdown of patients consented per participating site is shown in , and the rate at which patient consents were accrued is shown in .
Recruitment (EOI/consent obtained)
Cumulative recruitment numbers across all five centres defined as signing EOI or consent forms.
By the end of the study, 158 participants remained on the waiting list for transplant; 74 had died while waiting, before transplant had occurred, and 34 were excluded after transplant as they did not reconfirm their consent, died before giving consent or were erroneously included after the recruitment cut-off date.
A total of 202 participants were included in the two transplant arms of the study, 184 in the standard transplant arm [60.1% of the target recruitment of 306, 95% confidence interval (CI) 55.4% to 65.7%] and 18 in the EVLP transplant arm (17.6% of target recruitment of 102, 95% CI 10.8% to 26.4%). A total of 53 EVLP assessments were performed, leading to the 18 transplants, giving a conversion rate of 34.0% (95% CI 26.6% to 42.0%). The transplant activity in the participating sites is shown in . It is the small number in the EVLP transplant arm that drives the need to restrict the comparative analysis to the use of descriptive statistics. A Consolidated Standards of Reporting Trials (CONSORT) diagram showing study activity is shown in .
Number of patients transplanted by centre
The CONSORT diagram: withdrawals.
The main delay to commencing recruitment to the study resulted from obtaining NHS R&D approvals across the study sites, which equated to 31 centre-months lost.
Two groups of patients were approached: patients already on a transplant waiting list and patients added to the waiting list during the study. As a result, some patients signed an EOI form only, some signed both (an EOI form followed by a consent form during a subsequent routine visit to the transplant centre or on the day of transplant), and some signed only a consent form on the day of transplant, having received study information previously.
No patients requested to be withdrawn from the study after transplantation, and none was withdrawn by study staff on safety grounds. All withdrawals were due to changes in patients’ eligibility for lung transplantation or to issues with completion of all necessary consent documents.
Analysis groups
Patients were analysed in groups defined by the type of transplant received [i.e. standard (control) or EVLP (intervention)]. Allocation was not random, and it was not possible to switch between groups. All donors who provided lungs assessed by EVLP (n = 53) were included in the descriptive analyses described in Identifying clinical predictors of successful ex vivo lung perfusion reconditioning. summarises the times at which the various data were collected for each patient.
Schedule of study activities and data collection points
Study population
Baseline patient characteristics
There are a number of donor-, recipient- and procedure-related variables that mean that lung transplant recipients constitute a heterogeneous group. The recipient or donor characteristics listed in and 6 have been identified from the ISHLT registry as those potentially influencing outcomes. summarises characteristics of the lung transplant recipients, split according to the type of transplant. The percentage of recipients who were male and the median age were higher in the EVLP group {72.2% and 56 years [interquartile range (IQR) 46–59 years], respectively} than in the standard transplant group [57.6% and 51 years (IQR 38–58 years), respectively]. About half of the patients in each group had been diagnosed with chronic obstructive pulmonary disease or CF; interstitial lung disease (in the EVLP group) or a combination of interstitial lung disease and emphysema (in the standard group) constituted a further 40% of the diagnoses. The 18 EVLP recipients did possess indicators to suggest that they may have been a higher-risk group, such as significant secondary pulmonary hypertension and requirement for non-invasive ventilation prior to transplant. No lung allocation score is used in the UK and so Lung Allocation Scores or other indicators of clinical urgency were not routinely recorded and cannot be reported.
Recipient characteristics in the EVLP and standard lung transplant groups
The percentage of recipients who were diabetic was similar in the EVLP (22.1%) and standard groups (18.1%), as were the median body mass index values, namely 21.6 kg/m2 (IQR 18.4–26.3 kg/m2) and 23.8 kg/m2 (IQR 20.5–26.5 kg/m2), respectively. The median FEV1 at baseline was 1.2 l (IQR 0.7–1.9 l) for the EVLP group and 0.9 l (IQR 0.6–1.4 l) for the standard group; the corresponding FEV1 percentage predicted was 29% (IQR 22–50%) in the EVLP group and 26% (IQR 20–44%) in the standard group. In both groups, > 80% of the lung transplants performed were bilateral. However, the observed percentage of transplants performed with the use of cardiopulmonary bypass was higher in the EVLP group than in the standard group (88.9% vs. 63.0%), although cardiopulmonary bypass status was missing for 12% of patients in the standard group.
The characteristics of the donors, again split between the EVLP and standard transplant groups, are shown in . The percentage of donors who were male was slightly higher in the EVLP group than in the standard group (55.6% vs. 46.7%) and donors in the former group were slightly older (median age 50.5 years and 44 years, respectively). DCD donor type was more common in the EVLP group (27.8% of donors) than in the standard group (16.9%). The numbers of left and right lungs were similar, reflecting – as indicated earlier – the high proportion of transplants that were bilateral. Within each transplant group, ischaemic times were similar for left and right donor lungs. However, total ischaemic times were much higher for EVLP transplants than for standard transplants, which reflects the nature of the procedure.
Compliance
Seventeen protocol violations were reported in 15 separate patients (7.4% of the total of 202 patients consented and transplanted). Five were major and were due to patients being transplanted with donor lungs that did not fully meet protocol criteria for transplant after EVLP assessment and reconditioning. In all these cases the decision to proceed to transplant was made by the supervising transplant surgeon on the basis of the balance of risks to the patient.
Twelve violations were minor, including approaching patients to give consent for retrospective data collection post standard transplant even though they had not returned an EOI form; failure to obtain full informed consent as the wrong consent form was signed; failure to obtain reconsent to continue on the night of transplant; and, on nine occasions, a > 24-hour delay in the submission of a SAE to the clinical trials unit. Seven of the protocol violations (all minor) were in standard transplant patients, whereas 10 (five major and five minor) were in EVLP transplant patients.
Serious adverse events
There were 42 SAEs affecting 38 patients (16 patients in Newcastle, four patients in London, four patients in Manchester, 11 patients in Cambridge and three patients in Birmingham). Fifteen (35.7%) of these SAEs (affecting 12 patients) occurred in EVLP transplant patients. Details of the SAEs reported are shown in .
Serious adverse event listing
Of the 42 SAEs, 18 (42.9%) were a result of death within 90 days of transplant; 14 (77.8% of all fatal SAEs) of these were in standard transplant patients and four (22.2%) were in EVLP patients. Of the total of 42 SAEs, four of these events (9.5%) arising in four patients were judged to be possibly causally related to study procedures; all of these serious adverse reactions were initially considered by the site PI to be possibly unexpected AEs. However, after review by the chief investigator, it was decided that all were in fact events that could be expected to occur after lung transplantation, such as PGD or infection.
Activity in the EVLP arm was halted temporarily on 9 April 2013 for just over 3 months to allow an independent review of the early study outcomes as part of a due diligence process. Before this point, four out of eight transplant recipients in the EVLP arm required ECMO support post-operatively; following resumption of EVLP transplants with a revised protocol, 3 out of 10 transplant recipients required ECMO, but in all cases ECMO duration was limited, all patients were successfully weaned from ECMO, and all except one were successfully discharged from the ITU.
Outcomes analyses
The original intention was that the statistical analysis be conducted in a number of parts: first, a comparison of outcomes between recipients of standard and EVLP transplants to establish non-inferiority; and, second, modelling of the effect of EVLP transplants on the overall survival of patients accepted for lung transplantation in the UK, in order to assess the impact on the service. Furthermore, additional analyses were also to be undertaken to identify clinical predictors with respect to donor characteristics of successful EVLP reconditioning.
The early closure of the study and low numbers in the EVLP arm mean that the analysis methods originally described in the protocol are no longer appropriate. Consequently, the comparative analysis of standard and EVLP transplant groups, as well as the analysis of overall survival of patients awaiting transplantation, reported in Primary outcome analysis, are descriptive in nature and, as such, do not reflect the initial intention of testing for non-inferiority of EVLP to standard transplants.
Missing data
From clinical experience of this patient group, it had not been anticipated that there would be significant numbers of dropouts or loss to follow-up of patients with respect to the primary outcome measure. Loss to follow-up or missing data on the secondary outcome measures was assessed but, because of the low numbers in the EVLP group, no imputation was performed for any outcome data. However, a significant number of data were missing because they were not collected at the study sites. Most sites worked very hard to keep data collection as complete as possible, but in one site the proportion of missing data was > 20%. The missing data included, but were not limited to, SF-36 questionnaires, detail from outpatient follow-up visits and some information collected during the EVLP procedure.
Primary outcome analysis
Survival in the 12 months following transplantation
The primary outcome of survival in the first 12 months post transplantation was compared in the EVLP and standard transplant groups. shows the Kaplan–Meier plot of the survival (in days) during the first 12 months post transplantation, split by study group. This analysis takes account of censoring; specifically, one patient (in the standard arm) emigrated during the follow-up, and has been included in the analysis up to the time of the last visit (at 30 days post transplantation).
Kaplan–Meier survival estimates post transplantation, by study group.
The numbers of patients who died, survived or were censored during the 12-month follow-up are shown in . The median follow-up was 365 days for both the EVLP and standard transplant groups.
Numbers of patients who died, survived or were censored during the first 12 months following transplantation
The Kaplan–Meier estimate of survival at 12 months was 0.67 (95% CI 0.40 to 0.83) for the EVLP arm and 0.80 (95% CI 0.74 to 0.85) for the standard arm. Based on Cox regression, the hazard ratio for all-cause mortality in the EVLP arm relative to the standard arm over the 12-month follow-up was 1.96 (95% CI 0.83 to 4.67). This equates to roughly a doubling of the risk of death in the EVLP arm relative to the standard arm, although with a wide CI that encompasses the possibility that mortality might be lower in the EVLP arm than in the standard arm. The width of this CI is influenced greatly by the small number of patients who received an EVLP transplant.
Of the 18 patients who received an EVLP transplant, eight received a transplant based on the hybrid protocol, and 10 received a transplant based on the Lund protocol (see ). Survival was then assessed by considering the EVLP protocol groups separately, and shows the Kaplan–Meier plot of survival separately for patients in the standard, EVLP-Lund and EVLP-hybrid groups.
Kaplan–Meier survival estimates post transplantation for the standard, EVLP-hybrid and EVLP-Lund study groups.
The Kaplan–Meier estimate of survival at 12 months was 0.80 (95% CI 0.41 to 0.95) for the Lund protocol patients and 0.50 (85% CI 0.15 to 0.77) for the hybrid protocol patients. Based on Cox regression, the hazard ratio for all-cause mortality in the EVLP-hybrid group relative to the EVLP-Lund group over the 12-month follow-up was 2.92 (95% CI 0.53 to 15.95). This wide CI reflects the small numbers of patients in these two EVLP protocol groups.
Secondary outcome measures
Survival at 30 and 90 days
Survival in the early post-operative period after lung transplantation is an important indicator of early complications and is widely used in audit and national and international registries of outcomes. The 30-day survival rates for the EVLP and standard transplant groups are shown in and the 90-day survival rates in .
Numbers of patients who died, survived or were censored during the first 30 days following transplantation
Numbers of patients who died, survived or were censored during the first 90 days following transplantation
The Kaplan–Meier estimate of survival at 30 days is 0.94 (95% CI 0.67 to 0.99) for the EVLP arm and 0.97 (95% CI 0.93 to 0.98) for the standard arm. In other words, survival at 30 days was similar for the two transplant groups. The Kaplan–Meier estimate of survival at 90 days was 0.78 (95% CI 0.51 to 0.91) for the EVLP arm and 0.94 (95% CI 0.89 to 0.97) for the standard arm.
Primary graft dysfunction
Primary graft dysfunction is the clinical syndrome of chest radiographic changes and poor oxygenation that represents early acute injury to the transplanted lung. The PGD scores used in the study were as defined in the ISHLT consensus definition.39 The distribution of the PGD score by study group, measured at baseline and 24, 48 and 72 hours after the transplant, is shown in . A score of 0 represents no evidence of PGD and a score of 3 represents the most severe form of PGD.
Primary graft dysfunction score by trial arm and time since transplant
The same information, but with the results for grades 0 and 1 combined, is shown in graph format in . The percentage of patients with PGD grade 3 at baseline was much higher in the EVLP group than in the standard group (88.9% vs. 30.2%). However, this difference narrowed as time passed, with 27.8% of patients in the EVLP group and 22.5% of those receiving a standard transplant having PGD grade 3 at 72 hours after transplant. Nevertheless, the percentages of patients with grade 0 remained fairly static over time and were higher in the standard group than in the EVLP group (22.5–27.4% and 5.6%, respectively).
Primary graft dysfunction score by trial arm and time since transplant.
Early intensive therapy unit management and duration of hospital stay
Data on several key aspects of the ITU management and hospital stay were collected for all patients transplanted and are presented in . The duration of invasive ventilation tended to be longer for patients receiving an EVLP transplant (median 72 hours, IQR 38–624 hours) than for those receiving a standard transplant (median 38 hours, IQR 19–140 hours). Similarly, ITU stay was longer for EVLP patients (median 14.5 days, IQR 5.4–20.6 days) than for patients with a standard transplant (median 4.3 days, IQR 2.1–10.8 days). However, the overall length of hospital stay was similar for both groups of patients (median of 28 days in both groups) and, among those patients readmitted to ITU, length of stay in ITU was similar in both groups [median 6 days (IQR 3–6 days) in the EVLP arm and 8 days (IQR 3–20.5 days) in the standard arm].
Summary statistics for duration of invasive ventilation, ITU stay, hospital stay before first discharge in days and (where relevant) ITU readmission
Post-operative infection
The number of patients with at least one post-operative infection, both at baseline and during subsequent follow-up periods, is shown in . The associated percentages are displayed graphically in . At baseline, just under half of patients in both the EVLP group and the standard group had at least one post-operative infection. In both groups, the percentage of patients with at least one post-operative infection dropped subsequently, as shown in . This percentage tended to be lower in the EVLP group than in the standard group. However, inferences are limited by the small numbers of patients with infections in the EVLP group.
Numbers of patients with at least one post-operative infection, at baseline and during subsequent periods
Percentage of patients with at least one post-operative infection, at baseline and during subsequent periods, by study group. Note that the number at risk is based on those at risk at the start of the relevant period.
The number of episodes and number of organisms involved in specific post-operative infections are detailed in . The most common organisms involved were Pseudomonas, Staphylococcus, Escherichia coli and Candida species. Owing to small numbers, it is difficult to compare any differences in the spectra of infections in the EVLP and standard groups.
Numbers of episodes and organisms involved in specific post-operative infections, at baseline and during subsequent follow-up periods
It is recognised that ischaemic injury to the donor lung could adversely affect the bronchus and lead to bronchial complications. It was therefore important to consider if there was any difference in rates of anastomotic complications between the study groups. In , the numbers of patients with anastomotic complications are presented by study group and time since transplant. None of the patients in the EVLP group had such complications at any of the follow-up times. In the standard group, the percentage of patients with these complications varied between 4.4% and 9.6% over the follow-up period.
Numbers of patients with anastomotic complications, by study group and time since transplant
The Couraud Classification of anastomotic healing provides a means to quantify the degree of healing that may be an indicator of low-level ischaemic injury.40 These scores are presented numerically by study group and time since transplant in , and as percentages in graphical format in . Among both EVLP and standard transplant patients, the percentage with grade 1 healing tended to increase over the period of the follow-up, from around 40% between baseline and 1 month to just over 80% between 6 and 12 months. Over the same period, the percentage of patients with grade 2 healing decreased, from around 50% between baseline and 1 month to roughly 16% between 6 and 12 months.
Anastomotic healing among patients (based on the Couraud Classification), by study group and time since transplant
Anastomotic healing among patients (based on the Couraud Classification), by study group and time since transplant.
Lung function measurements
Measurement of FEV1 in both absolute volume in litres and as a percentage of the patient’s predicted values based on age, sex, height and measurement of forced vital capacity (FVC) in litres is routinely performed as part of post-lung transplant follow-up. This information is presented in and also displayed in box plots in –. Each of these measures tended to increase with increasing length of follow-up. In addition, median values for the EVLP and standard transplant groups were generally similar.
Lung function measurements by study group and time since transplant
Box plot of FEV1 by time since transplant.
Box plot of FVC by time since transplant.
Box plot of FEV1% by time since transplant.
Abnormalities on chest radiographs
The number of patients with abnormalities on chest radiographs, both at baseline and at subsequent follow-up visits over the 12 months after transplant, is shown in . The associated percentages are also displayed graphically in . These percentages were lower at 6 and 12 months than at earlier times. There is also some suggestion that abnormalities were slightly less common in the EVLP group than in the standard group, although inferences are limited because of the small numbers of EVLP patients with abnormalities.
Numbers of patients with abnormalities on chest radiographs, by study group and time since transplant
Percentage of patients with abnormalities on chest radiographs, by study group and time since transplant.
The nature of the specific abnormalities on chest radiographs is shown in . Overall, effusion was the most common abnormality, followed by pneumothorax, consolidation, atelectasis and shadowing. Owing to the small numbers in EVLP group, it is not possible to compare the spectra of abnormalities between the EVLP and standard groups.
Numbers of specific abnormalities on chest radiographs, by study group and time since transplant
The number of episodes of allograft alloimmune injury was collected in both study groups. Data on acute vascular rejection by ISHLT score, presence of antibody-mediated rejection and lymphocytic bronchiolitis are presented in as numbers of rejection episodes by rejection type, study group and time period. Overall, these episodes were less frequent > 3 months after transplant than at earlier times. This decrease was particularly notable for A2+ episodes. The percentage of patients with at least one rejection episode was generally similar for the EVLP and standard groups.
Number of rejection episodes by rejection type, study group and time period
Short Form questionnaire-36 items health-related quality-of-life measure
The assessment of HRQoL in the DEVELOP-UK study was done using a well-validated questionnaire, the SF-36. The SF-36 is a measure of general health that generates eight dimensions and two summary scores from 36 different questions.46 In order to do so, each one of the 36 questions of the survey relates to a different pre-coded numeric value. In order to interpret the SF-36 data, the raw scores should be translated into a value from 0 (lowest or worst possible level of HRQoL) to 100 (highest or best possible level of HRQoL). These translated scores are then used to calculate the mean for each one of the following eight domains: physical functioning, role limitations due to physical health, bodily pain, general health perceptions, vitality, social functioning, role limitations due to emotional problems and general mental health. From these eight concepts, two summary measures of norm-based mental component score (MCS) and physical component score (PCS), with a mean of 50 and a standard deviation (SD) of 10 in a general population, can be constructed from all the emotionally and physically relevant items, respectively.47 The higher the value of the summary scores the higher the level of functionality of the patient.
Short Form questionnaire-36 items data analysis
The SF-36 questionnaires were completed while the recipients were still on the waiting list, as well as at 3 and 12 months post transplant, and their mean scores were converted into health-state utilities using the SF-6D algorithm48 described in Chapter 4. In the SF-36 data analysis, the mean (SD) and median (IQR) score for each of the eight domain scores were estimated for each one of the two study groups and for each time point. The SF-36 data were also used to estimate the mean, SD, median and IQR for MCS and PCS scores by study group. No hypothesis testing or modelling was undertaken and no imputation was performed other than that which forms part of the standard scoring system for the SF-36.
The SF-36 data, available at each time point of the study, as well as the number of participants with missing data for both the standard and EVLP trial arms, are summarised in . A significant number of participants did not complete the SF-36 at each of the three time points identified in the study protocol. The absolute number of patients for whom SF-36 data were collected at different stages of the study is shown in .
Number of SF-36 measurements at each time point of the study and by study group
Numbers of patients with SF-36 data available for each stage of the study
The detailed scores of the SF-36 questionnaires are presented in and . The mean, SD, median, IQR and range of the eight domain scores and the two summary scores, respectively, at each data collection point for each of the two transplant procedures are presented. These two tables also report the number of responses at each time point, together with the number of patients at risk (i.e. the number that would have been eligible to complete the SF-36).
Short Form questionnaire-36 items domain scores at each data collection point of the study, by study group
Short Form questionnaire-36 items measurements by study group at each time point
The results show a general increase in the mean scores of the eight SF-36 domains from baseline to 12 months post transplant (see ). The two domains that show the biggest increase in their scores are general health and vitality. Furthermore, although the physical functioning, role limitations owing to physical health and social functioning domains appear to show the same increase after 3 and 12 months from the day of the surgery in the standard arm of the study, they appear to increase at 3 months and then decrease at 12 months for the EVLP arm. The data are, however, too few for the relevance of this change to be sensibly interpreted. Similarly, the same pattern is seen for the general mental health and role limitations owing to emotional health dimensions. For the bodily pain domain there was a slight drop, on average, at 3 months and then an increase at 12 months for the EVLP arm. Nevertheless, the fact that few data are available for these three domains means that further interpretation must be done with caution.
The values of both the SF-36 summary measures, namely MCS and PCS, increase after transplant no matter which transplant procedure was performed. These data are presented in , and graphically in the box plots in and . In other words, the HRQoL of the patients improves throughout the follow-up of the lung recipients. In the standard procedure, the mean MCS score of the lung recipients increases from 43.5 at baseline to 51.0 at 3 months post transplant and 52.8 at 12 months post transplant, whereas the mean PCS score increases from 27.2 at baseline to 43.7 and 45.4 at 3 and 12 months, respectively. As far as the EVLP group is concerned, the mean MCS score changes from 44.2 at baseline to 57.2 at 3 months post transplant and 54.9 at 12 months post transplant, whereas the mean PCS score increases from 31.5 to 51.3 at 3 months after the transplant and drops slightly to 50.3 at 12 months’ follow-up. This slight decrease in the mean 12-month MCS and PCS scores in the EVLP arm of the study is consistent with the decrease in the scores for most of the eight domains at this time point. As mentioned above, any results regarding the effectiveness of the EVLP based on these scores given would not be reliable because of the very limited number of data available.
Box plot of SF-36 MCS scores, by study group and time point.
Box plot of SF-36 PCS scores, by study group and time point.
Examining the effect of ex vivo lung perfusion on the overall survival of patients awaiting transplantation
To capture the effects that an increased availability of donor organs due to EVLP might have on the survival of patients awaiting lung transplantation, waiting time in each of the two treatment groups was compared, and then waiting time in each group was compared with survival of those remaining on the waiting list. Waiting time is defined as the time from the date the participant is placed on the waiting list until the date transplant is performed. The waiting times until transplantation in the two study groups is shown in .
Summary of waiting time until transplant
The median waiting time for standard transplant was 197 days (IQR 95–373 days), whereas the median waiting time for a transplant using an EVLP donor was 142 days (IQR 60–199 days), as shown in . There was a median difference of 55 days between transplant groups, showing a reduction in waiting time if having a transplant using an EVLP-assessed donor organ. This is also illustrated by , which shows a maximum waiting time of 551 days for transplant using a EVLP donor compared with a maximum waiting time of 2143 days for a standard transplant. A log-rank test for difference in waiting times between transplant type using the Kaplan–Meier estimates gave a p-value of 0.042, which shows a significant difference in waiting times between the two groups. However, these findings should be interpreted with caution, in view of the small numbers of patients in the EVLP group.
Kaplan–Meier graph of waiting time until transplant by transplant type (each event in this graph indicates a transplant performed).
Survival from listing
To assess overall survival between transplant groups and those remaining on the waiting list, the time from being placed on the waiting list until 12 months post transplant or 1 May 2015, with censoring for death or loss to follow-up, is presented in . Waiting list dates were collated for all participants during recruitment to the study; however, for those remaining on the waiting list, some of this information was not collated. For these participants, waiting list dates were obtained from the NHSBT registry.
Summary of survival time from listing
For the purpose of this analysis, 1 May 2015 was chosen as the end date for those who did not have a transplant, since this is approximately 12 months after the last transplant was performed. Of the 232 participants remaining on the waiting list, 20 have been excluded from the analysis. Thirteen were excluded because they were recruited to another study in which they had a transplant, four were excluded because they no longer wanted to be part of the study and three were excluded because there is no record of the date they were removed from the waiting list. Thirty-nine participants who were included in the analysis have been censored: 11 of these were censored by the date they were removed from the waiting list or were lost to follow-up and 28 were censored by the date on which they had a transplant outside of this trial. There were 41 participants for whom we had been unable to confirm their status as of 1 May 2015. Since we have not received information regarding their death (which we would have expected), we have assumed these participants remained on the waiting list on 1 May 2015.
The Kaplan–Meier estimates of survival from being placed on the transplant list to transplant or death/censoring are shown in for the three study groups. The median survival time from listing was 536 days (IQR 429–703 days), 479 days (IQR 412–543 days) and 543 days (IQR 324–830 days) for standard transplant, EVLP transplant and waiting list groups, respectively (see ). The log-rank test of difference in survival times was significant (p = 0.007), implying that those having a standard lung transplant had better survival than those who remained on the waiting list and those having an EVLP transplant. However, these findings should be interpreted with caution in view of possible selection bias and the small number of patients in the EVLP group.
Kaplan–Meier graph of survival time from being placed on the transplant waiting list.
Identifying clinical predictors of successful ex vivo lung perfusion reconditioning
Between 1 April 2012 and 9 July 2014, lungs from 53 UK multiorgan donors were identified as unsuitable for immediate standard transplantation despite extensive donor management. These donors all satisfied the strict entry criteria for inclusion in the EVLP arm of the study (Boxes 1 and 2).
Criteria for EVLP assessment and reconditioning: using DBD or DCD lungs
Absolute contraindications to donor lung use for standard transplant or for EVLP
Of these 53 donors, 35 died from spontaneous intracranial haemorrhage (66%), eight from hypoxic brain injury (15%), five from traumatic brain injury (9%), three from thrombotic stroke (6%), one from an expansive brain tumour (2%) and one from meningitis (2%). The donor lungs were procured in a routine fashion and transported on ice to the accepting institution. A total of 27 (51%) were assessed in Newcastle, nine in Harefield (17%), seven in Manchester (13%), six in Birmingham (11%) and four in Papworth (8%). Fourteen donor lungs were from donors without a circulation (DCD) (Maastricht category III, 26%) and 39 were from brain-dead donors (DBD) (Maastricht category IV, 74%). The EVLP assessments were evenly distributed between donor sexes [26 female (49%) and 27 male (51%)] and the median donor age was 50 years (range 16–65 years). If one lung did not meet entry criteria and was deemed unusable because of severe consolidation or extensive contusion on inspection, or if the intended recipient was for a specific side (i.e. single-lung transplant), only one lung was procured. Fifty lungs (94%) were perfused as double lungs and three (6%) as singles (two right lungs and one left lung). They were assessed for a median time of 175 minutes (range 73–383 minutes) while being normothermically perfused on the Vivoline LS1 EVLP circuit.
The study protocol allowed for lungs that satisfied certain criteria, as outlined in Box 1, to be assessed using EVLP. The primary indications for EVLP assessment were grouped into three general categories: 35 donor lungs (66%) were found unsuitable for standard transplantation because of poor lung function with an optimised donor ratio of arterial partial pressure of oxygen (PaO2) – shown throughout the report as kPa (mmHg) – to FiO2 < 40 (300) and/or a selective pulmonary vein gas < 30 (225) with a FiO2 of 1.0 (100%) at procurement; 13 lungs (25%) were turned down as a result of abnormalities during inspection (pulmonary oedema, abnormal bronchoscopy, extensive atelectasis, etc.); and three lungs (6%) were turned down for standard transplantation for logistical reasons. Some lungs were excluded from EVLP as they had absolute contraindications to transplant (see Box 2).
Of the donor lungs turned down for logistic purposes, two had a suspicious mass in need of urgent histological evaluation before a decision on transplant was made. One was revealed to be cancerous and the lungs were discarded after an otherwise successful EVLP perfusion, and the other was confirmed benign and the lungs were transplanted. In one case, no theatre team was available as they had just started another transplantation. In this case, the lungs also remained stable or improved measurable physiological parameters during preservation on the EVLP circuit, but were in the end turned down for transplantation because of deteriorating oedema formation on inspection.
Of the 53 donors, 24 (45%) were current smokers, 18 (34%) had abnormal chest radiographs at the time of procurement and 22 (42%) had airway secretions deemed prohibitive of standard transplantation, predominantly purulent secretions. The median ventilation time for the EVLP donors before procurement was 2 days (range < 1–10.3 days) and 27 donors (51%) had positive microbiology cultures from sputum, BAL fluid or cerebrospinal fluid. The median optimised PaO2 : FiO2 ratio for the 53 EVLP donors at the time of procurement was 39.9 [299 (range 95–535 mmHg)] and after EVLP was 50.9 [381.5 (range 74–638 mmHg)]. The EVLP assessments followed one of two standardised perfusion protocols depending on when the lungs were entered into the study. Initially, between 1 April 2012 and 31 March 2013, a hybrid protocol combining elements of the Toronto and Lund protocols was used with an open left atrium, an acellular perfusate and a reduced perfusate flow to 40–60% of the donor’s calculated cardiac output. After a pause midway through the study because of concerns over the rate of ECMO use in transplants performed after assessment using the hybrid EVLP protocol, the perfusion strategy was altered and the study was restarted in August 2013 and the hybrid protocol replaced by the Lund protocol.
The Lund protocol uses a cellular perfusate (haematocrit 10–15%) and a full-flow perfusion strategy (100% of cardiac output), but is otherwise identical to the hybrid protocol. The assessment and ventilation strategies, airway and vascular pressure limits, and perfusate composition was otherwise unaltered ().
The DEVELOP-UK study EVLP protocols
Lung performance during ex vivo lung perfusion assessment
Transplant suitability was assessed as soon as the lungs had stabilised, at 37 °C and perfusing at full flow, and thereafter hourly until the end of perfusion. Lungs meeting transplant suitability at any time point were cooled and transplanted (Boxes 3 and 4). Lungs deemed to have futile prospects for improvement were taken off the circuit and discarded.
Criteria for transplant after successful EVLP assessment and reconditioning
Criteria for failed EVLP assessment and reconditioning: transplant will not proceed
The performance of the 53 donor lungs undergoing EVLP assessment in the study is reported in two different ways to give as robust a description as possible of this cohort.
First, clinical EVLP success was defined as any EVLP-assessed donor lung deemed clinically suitable for transplantation at the end of perfusion by the surgical team performing the assessment (n = 22). This included all EVLP lungs that were transplanted within the study (n = 18) and all lungs that were accepted for transplantation on the basis of EVLP performance, but had to be turned down due to unforeseen logistical reasons (n = 4) ().
Flow chart for clinical EVLP success (n = 22).
Of these four cases, one pair of lungs accepted for double lung transplant had to be discarded as the recipient presented with an active airway infection and was found to be at too high risk for the operation on arrival at the hospital. The lungs were offered on urgently, but turned down by all UK centres owing to logistics. One lung had an unreported irreparable left pulmonary artery laceration with haematoma formation from the retrieval surgery. The right lung was selectively perfused and deemed suitable for transplantation. There was, at this point, no available matching single lung recipient and the lung had to be discarded after having been offered on to all other UK centres. One left single lung accepted for transplantation for a matched recipient had to be aborted due to an emergency in theatres and lack of additional surgical capacity at the site. One lung was put on EVLP for logistical reasons while awaiting histology of suspicious masses found during the organ procurement. Evaluation of a liver nodule showed chronic lymphatic lymphoma. Although the lung biopsy was found benign, the transplant was aborted because of the elevated risk of tumour transmission to the recipient. The lungs were meanwhile successfully reconditioned on the circuit and deemed to be in a suitable condition for transplant had it not been for the histological liver findings.
Second, per-protocol success was defined as all perfused donor lungs meeting the complete set of predefined study criteria for EVLP success, regardless of clinical transplant result (n = 17). This included all EVLP lungs that were transplanted in the study while meeting all predefined transplant criteria (n = 13) and all non-transplanted EVLP lungs that met all transplant criteria but were discarded purely on logistical grounds (n = 4) ().
The flow chart for per-protocol EVLP success (n = 17).
Of the 18 EVLP lungs that were transplanted in the study, five had to be defined as per-protocol non-successful perfusions as, although deemed clinically suitable for transplantation, they failed to meet all predefined transplant criteria and were transplanted as a result of a protocol violation.
Two of these five lungs were transplanted even though their mixed pulmonary vein PaO2 : FiO2 ratio was 40.0 (< 300) during the assessment. One of these almost doubled its optimised retrieval PaO2 : FiO2 ratio during the perfusion, increasing from 22.0 (166) to 38.6 (290), and the other had a recorded mixed pulmonary vein PaO2 : FiO2 ratio of 35 (262), while all selective pulmonary vein gases were well above the study cut-off point [range 46.0–72.0 (345–540)]. Three EVLP lungs were transplanted while having one or two selective lower lobe pulmonary vein gases < 30 (225); however, their mixed pulmonary vein PaO2 : FiO2 met the study transplant criterion of > 40 (300). All five of these lungs were subsequently transplanted into recipients with satisfying post-transplant outcomes and a 100% 1-year survival. Of the 13 recipients receiving lungs that met all study criteria for transplantation, only seven (54%) remained alive 12 months post transplant.
Logistic regression analyses
A logistic regression approach was used to examine the association between successful reconditioning (defined on either of the bases explained above) and a number of potential predictors based on donor characteristics and indices measured during EVLP. As a result of the lower than planned numbers in the EVLP arm, the work here must be regarded as exploratory in nature rather than definitive.
In terms of the variables to be considered for the examination of potential predictors of successful EVLP, univariate exact logistic regression models were fitted with successful EVLP as the dependent variable, and each of the following in turn as the independent variable. These models were fitted using the exlogistic option in Stata® 13.1 (StataCorp LP, TX, USA):
donor {age, sex, cause of death, smoking history, ischaemic time, duration of ventilation, oxygenation, donor type [after brain death (DBD) or after circulatory death (DCD)]}
EVLP physiology (oxygenation, lung compliance and resistance, airway pressure, perfusion time).
The original intent was also to use similar approaches to examine EVLP donor lungs used in transplantation and the association between early clinical outcome measures in recipients and physiological indices measured during EVLP. However, having only 18 patients in the EVLP transplant group meant that such an analysis would not be statistically meaningful.
gives the results of the univariate analyses of predictors of clinical EVLP success. Here odds ratios are presented for different categories or, in the case of continuous variables, based on a unit increase in the variable. Although some of the point estimates for odds ratios varied somewhat from one, in all instances the associated 95% CI included one.
Univariate exact logistic regression analysis of predictors of successful EVLP reconditioning (clinical EVLP success)
shows the results from the corresponding analyses, based on per-protocol success. The odds ratios and CIs are very similar to those based on clinical EVLP success.
Univariate exact logistic regression analysis of predictors of successful EVLP reconditioning (per-protocol success)
In conclusion, there was no strong evidence to indicate specific predictors of successful EVLP reconditioning. However, the analyses were restricted by the relatively small numbers.
Archiving
The trial data were stored on the Newcastle Clinical Trials Unit’s MACRO database system, provided by Infermed’s MACRO software as a service system. The hardware was located at their hosting partner Rackspace’s secure premises in London, UK, and is managed and supported by the Rackspace team. All data were stored and transmitted securely. Data were hosted and backed up only in the UK and were never transferred overseas. Only authorised staff can grant and have control of access. Any snapshots of the database taken will be kept on the Newcastle University server, which is backed up daily.
Once all trial-related analysis and activities are completed, the database will remain on MACRO, with permissions removed. The data will be archived onto disk for each site file and also centrally in accordance with the Newcastle Clinical Trials Unit’s standard operating procedures (SOPs).
