If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Division of Respirology, Department of Medicine, Lung Transplant Program, Toronto General Hospital, University Health Network, Toronto, ON, CanadaDivision of Respirology, Adult Cystic Fibrosis Centre, St. Michael's Hospital, Toronto, ON, CanadaDivision of Respiratory Medicine, University Hospital of Lausanne, Lausanne, Switzerland
Division of Respirology, Department of Medicine, Lung Transplant Program, Toronto General Hospital, University Health Network, Toronto, ON, CanadaDivision of Respiratory Medicine, Department of Medicine, University of Calgary, Calgary, AB, Canada
Division of Respirology, Department of Medicine, Lung Transplant Program, Toronto General Hospital, University Health Network, Toronto, ON, CanadaDivision of Respirology, Adult Cystic Fibrosis Centre, St. Michael's Hospital, Toronto, ON, Canada
Mortality risk stratification is essential in lung transplantation (LTx) to allow listing, prioritization and mitigating strategies. In cystic fibrosis (CF) patients, predictors of post-LTx mortality are not established.
For this systematic review and meta-analysis, seven databases were searched until January 3, 2018 to identify predictors of post-LTx mortality in CF. We excluded studies of multi-organ transplantation, re-transplantation and graft survival. For multiple studies assessing the same population during overlapping time-periods, the largest one was analyzed. Risk of bias was assessed with the Newcastle-Ottawa scale (NOS). Pooled hazard ratios were calculated using random-effects models.
Fifty-four studies were included in the systematic review and 11 studies in the meta-analyses (low-to-moderate bias risk, NOS score ≥ 5). Among 10 factors assessed in the meta-analysis, B. cepacia complex (BCC) (N = 1451, unadjusted HR = 2.35, 95%CI:1.80–3.06; I2 = 20.4% and adjusted HR = 2.49, 95%CI:1.74–3.57; I2 = 46.2%) and ascending chronological year of LTx (N = 4207, unadjusted HR = 0.98, 95%CI:0.97–0.98, I2 = 4.8%) were predictors of post-LTx mortality. Male gender (N = 2903, adjusted HR = 1.12, 95%CI:1.0–1.26, I2 = 0%) and age in adults (N = 3677, unadjusted HR = 0.99, 95%CI:0.97–1.00; I2 = 64.1% and N = 2605, adjusted HR = 0.98, 95%CI:0.97–0.99; I2 = 34.3%) had borderline significant associations with post-LTx mortality. P. aeruginosa colonization, forced expiratory volume in one second (FEV1), pulmonary hypertension, body mass index (BMI), pancreatic insufficiency and CF-related diabetes (CFRD) were not predictors of mortality.
BCC was associated with a higher post-LTx mortality whereas FEV1, pulmonary hypertension, BMI, CFRD and female gender were not associated with post-LTx mortality. These findings indicate that CF-specific risk estimates of post-LTx mortality should be considered.
Prospective LTx candidates undergo a detailed evaluation to identify patients who would benefit from the procedure. This patient risk stratification is essential to allow listing, prioritization and targeted interventions for modifiable risk factors [
]. Due to the medical complexity of LTx candidates and the often contradictory published evidence, risk estimation can be challenging. A recent consensus document provides guidance for LTx referral and listing [
] but, in contrast to the LTx indications which are CF-specific, contraindications highlight areas of concern for all diagnoses reflecting the lack of clearly established, CF-specific risk factors for post-LTx mortality. Among the relative contraindications, progressive or severe malnutrition and colonization with highly resistant microorganisms are the most relevant in CF [
]. Moreover, as shown by a recent survey, it is often unclear whether parameters associated with higher mortality before LTx also increase post-transplant mortality and this becomes a barrier to LTx referral [
The aim of this systematic review and meta-analysis was to assess associations of pre-LTx factors with post-LTx mortality in CF and to consider a) whether factors associated with pre-LTx/waitlist mortality also confer a higher risk for post-LTx mortality and b) whether current contraindications for LTx apply to the CF population.
2.1 Search strategy and selection criteria
A systematic literature search was performed to identify studies on CF, LTx, mortality and survival. Ovid Medline, Ovid Medline In-Process and Other Non-Indexed Citations, Ovid EMBASE, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Clinical Trials, CINAHL and PubMed were searched from inception until January 3, 2018 for articles in English, French, or German. Citations specific to heart-lung transplantation, animal-only studies and book or conference materials were excluded. These searches were complemented by a manual search of reference lists of included articles. The search strategy is detailed in the Appendix A.
Two independent reviewers (AK, RV) screened the titles and abstracts of all retrieved references. The full-text of relevant articles and also the annual slides of the International Society for Heart and Lung Transplantation (ISHLT) were evaluated. Studies eligible for inclusion were: a) original research articles, b) articles assessing CF patients subjected to LTx and c) articles analyzing pre-operative factors in association with post-LTx mortality. Studies of multi-organ transplantation or re-transplantation were excluded unless they analyzed primary LTx. Studies using graft survival (instead of patient survival) were also excluded. Disagreements over inclusion were resolved by consensus, by contacting the study authors and/or by consulting a 3rd reviewer (LGS, ALS). Literature assessment was done in accordance with the PRISMA guideline [
Pre-LTx factors were extracted and categorized as microbiological and non-microbiological. Donor, intra-operative and post-operative factors were excluded. For each pre-LTx factor, we identified all articles reporting descriptive and/or numeric results. When two or more studies assessed the same patient population during an overlapping time-period, the study with the largest sample size was used for data synthesis. If available, ISHLT registry results were reported but were not included in the meta-analysis due to probable duplicate data with other cohorts. Study authors were contacted when there was uncertainty about patient characteristics, definitions of pre-LTx factors or outcomes, and when the hazard ratio (HR) and 95% confidence interval (CI) were not published. Factors evaluated in ≥2 distinct populations were included in the meta-analysis.
2.2 Risk of bias assessment
The risk for bias in included studies was assessed by two independent reviewers (AK, RV) using the Newcastle-Ottawa scale (NOS) for cohort studies [
]. The scale assesses 3 domains: selection, comparability and outcome (details provided in the appendix A - Table S2 footnote). The score ranges from 0 (very high risk of bias) to 9 (very low risk of bias) [
]. Disagreements were resolved by discussion between the two reviewers and/or by consulting a 3rd author (ALS, LGS, JS). Extracted variables included a) patient characteristics (e.g. age), b) pre-LTx factors, c) outcome of interest (e.g. global mortality and mortality at specific time-points if results differed from those of global mortality, duration of follow-up) and d) summary measures [such as HR (95%CI), odds ratios, risk ratios or data from which they could be extracted]. When the HR (95%CI) was not available but a Kaplan-Meier curve was published, we extracted the HR (95% CI) from the Kaplan-Meier curve [
Clinical heterogeneity between studies was assessed by evaluating a) the study design [e.g. cohort or other, registry or LTx center(s)], b) the population characteristics (adult or pediatric), c) the definitions of factors (discrepant or identical) [
]. Random-effects models were used to calculate the pooled HR (95% CI) using R version 3.3.0 (metafor version 2.0 package). Univariate and multivariable HR reported in the published studies were pooled separately and forest plots were generated. Results of the meta-analyses were reported next to results from the ISHLT registry, if available.
We screened 3260 references and assessed the full-text of 290 relevant articles, 54 of which were included in the systematic review and 11 in the meta-analysis. Fig. 1 presents the PRISMA flow-diagram [
] of study selection and Table S1 shows the study characteristics. Twenty-eight (52%) were registry studies and 5 (9%) analyzed data from two or more countries. They included 12 to 7245 transplanted CF patients between 1987 and 2016. Eighteen (33%) studies assessed only adults and 3 (6%) only pediatric patients. All studies had a NOS score ≥ 5 indicating a low-to-moderate risk for bias and 38 (70%) had a score ≥ 7. The most common potential source of bias was the adequacy of follow-up which could not be ascertained for 39 (72%) studies (Table S2).
Of the 54 articles, 4 (7%) assessed microbiological factors, 37 (69%) non-microbiological factors and 13 (24%) assessed both. Tables S3-S5 summarize these factors, the process of study identification for the meta-analysis and the final CF sample size. For 5 [
] of the 11 articles included in the meta-analysis, unpublished results were provided by the authors (Tables S6 and S7). For 10 parameters, for which data from ≥2 studies were available, meta-analysis was performed: 2 microbiological [B. cepacia complex (BCC), P. aeruginosa] and 8 non-microbiological [age, gender, year of LTx, forced expiratory volume in one second (FEV1), pulmonary hypertension (PH), body mass index (BMI), pancreatic insufficiency and CF-related diabetes (CFRD)]. Table 1 provides information about the qualitative assessment of these factors.
Table 1Qualitative assessment for the 10 predictors which were used for the meta-analysis (qualitative results for predictors not included in the meta-analysis of the Appendix A).
], the reference category was not given and we considered it was female as in the other papers (sensitivity analysis using male as the reference category provided similar results). Male gender had a borderline significant association with post-LTx mortality only in the adjusted analysis (3 studies [
], N = 2903, adjusted HR = 0.98, 95%CI:0.97–1; I2 = 52.5%; Fig. 3). The largest ISHLT study assessed year of LTx before and after 2006 as a categorical covariate. LTx before 2006 was associated with an increased post-LTx mortality in the univariate but not in the multivariable analysis [
Table S9 summarizes non-microbiological factors with insufficient data for a meta-analysis. Those associated with a higher risk of post-LTx mortality in a multivariate analysis were infectious/inflammatory lesions on chest imaging [
This is the first systematic review and meta-analysis assessing transplanted CF patients. We synthesized results from 54 cohort studies reporting pre-LTx factors in association with post-LTx mortality. Pooled estimates of risk were calculated for 10 parameters (BCC, P. aeruginosa, age, gender, chronological year of LTx, FEV1, PH, BMI, pancreatic insufficiency and CFRD) derived from 11 studies. Of these parameters, only BCC was consistently associated with an increased mortality after LTx. Earlier years of LTx were associated with post-LTx mortality only in the univariate synthesis, whereas younger age in adults and male gender had borderline associations with post-LTx mortality. Considering these results collectively, factors such as low FEV1, PH, low BMI, female gender and CFRD known to be associated with pre-LTx/waitlist mortality were not associated with post-LTx mortality, suggesting that risk estimates of post-LTx mortality based on these factors may underestimate the LTx survival benefit in CF. Among other factors, specific strains of B. cenocepacia, B. gladioli, panresistant bacteria other than BCC, mechanical ventilation, infectious/inflammatory lesions in chest imaging, hypoalbuminemia and a poor socio-economic status may be associated with higher post-LTx mortality based on limited evidence.
BCC infection is associated with an increased risk for pre-LTx mortality [
], and, as shown by the meta-analysis, post-LTx mortality. Although species and strain level data were not available, the geographic distribution of the 3 studies included in the meta-analysis suggests B. cenocepacia as the prominent species, with PHDC and Midwest strains being common in US [
] may also be associated with worse post-LTx survival. The study of Murray et al. demonstrated that identification of BCC should not in itself be viewed as a sufficient contraindication to LTx and that identification at a strain level is necessary for risk estimation [
]. Pooled risk estimates did not show significant associations between post-LTx mortality and any of these factors. For FEV1 and PH, this is not surprising since LTx leads to their improvement. For gender, results are of particular interest and indicate that the gender gap conferring a worse survival for women before LTx [
]. Finally, CFRD diagnosed before LTx was not associated with post-LTx mortality in the meta-analysis. Although no study has assessed the effect of pre-LTx glycemic control on post-LTx outcomes, a recent study evaluating diabetes after LTx showed that CF patients have worse glycemic control than patients with other diagnoses and that poor glycemic control is associated with increased mortality [
]. Lederer et al. studied adult CF and non-CF LTx recipients and reported a higher post-transplant mortality risk for underweight and overweight patients. However, for CF, only 2% patients were overweight and being underweight was associated with 5-year but not 1-year mortality [
]. Two potential explanations for this lack of observed association could be a biphasic effect of BMI on post-LTx mortality which cannot be captured by the pooled risk estimates and a patient selection bias resulting from the exclusion of severely malnourished patients from LTx. However, the low incidence of overweight CF LTx candidates mitigates the effect of the former. Concerning the latter, although we cannot exclude that the low proportion of severely malnourished patients could contribute to the absence of statistical significance, the BMI range of the studies synthesized in the meta-analysis [
] (Table S7) shows that patients with severe malnutrition were included in the studied population. For the pediatric population, limited nutritional status data did not allow a meta-analysis but an ISHLT registry study did not find significant associations between the BMI percentile or the grade of thinness (International Obesity Task Force cut-offs) and post-LTx survival in children with CF. [
] These results indicate that BMI may not be a sufficiently sensitive marker of nutritional status to allow risk stratification in this setting. More importantly, since patients with low BMI are at high risk for waitlist mortality but not necessarily post-LTx mortality, they may have an increased survival benefit from LTx. Thus, delaying listing on the basis of low BMI alone may be associated with an increased risk of death on the waitlist for these patients.
The two factors with significant heterogeneity in the meta-analysis were age at LTx and pancreatic insufficiency. Results for age in pediatric patients were not significant but hampered by study heterogeneity. For adults, the meta-analysis indicated a protective effect of older age, similar to the results reported from the ISHLT registry [
]. However, the borderline significant association and the adjustment for different factors between studies in the multivariable analysis does not allow for definite conclusions. Regarding pancreatic insufficiency, the pooled estimates of risk did not reach significance and only one study provided adjusted HR for this factor showing a higher mortality risk for pancreatic sufficient patients [
]. Considering the very high prevalence of pancreatic insufficiency in CF LTx candidates (exceeding 90%) and potential interactions of this factor with age, malnutrition and CFRD, interpretation of this finding is challenging [
]. Patient selection bias, publication bias and residual confounding may be present. The selection bias is particularly relevant for LTx studies which include only patients accepted for LTx. Moreover, acceptance criteria vary across centers and across allocation systems. We could not assess publication bias in our review due to small number of studies in each analysis. However, based on the exhaustive search and discussion with prominent investigators in the field, we are confident that we have not missed significant studies in our review. In order to assess adjusted estimates, we pooled data from studies which reported adjusted HR. This should be interpreted with caution as there were differences in the covariates used in various studies. Additionally, all studies being observational in nature, residual confounding due to unmeasured confounders always remains. The largest populations in the meta-analysis were from North America, possibly reducing the generalizability of the results. Finally, despite the many factors retrieved from the literature, data on several key variables that may affect post-LTx outcomes (e.g. psychosocial issues or rare microbial pathogens such as Scedosporium spp) were lacking or described in case reports only. Moreover, none of the previously studied factors could recapitulate the multidimensional assessment performed at listing. Concerning donor and/or intraoperative factors, although these were excluded from the systematic review, they could also affect patient outcomes. However, their interpretation can be more difficult considering that recipients' characteristics may also influence decisions about donor and/or intra-operative factors. These limitations were considered during the interpretation of our findings.
The main strengths of this systematic review and meta-analysis are the rigorous methodology for study selection, analysis and heterogeneity assessment, the low risk for bias of the included studies and their large sample size. The main difficulties encountered were the lack of harmonized definitions for many factors, studies including multi-organ transplantation or re-transplantation, different outcomes (patient survival, graft survival), studies assessing the same patient population and partially published results. Our methodology addressed each of these issues and, in many cases, study authors were contacted for clarification and unpublished results, facilitating data synthesis and interpretation. This emphasizes the need for standardized reporting in studies of predictive models as proposed by the published TRIPOD statement [
]. We believe that these recommendations should be a requirement for future publications on this subject.
In conclusion, risk stratification of LTx candidates is a complex integration of patient-specific data, evidence from the literature and clinical expertise from the multidisciplinary LTx team. This systematic literature review and meta-analysis is a comprehensive summary of current evidence on pre-LTx factors and post-LTx mortality in CF. Although a large number of parameters have been studied in this setting, data sufficient for a meta-analysis were available only for 10 factors. These factors are currently used in clinical practice during the evaluation of LTx candidates but they cannot recapitulate the comprehensive assessment performed by the LTx and CF teams. Concerning microbiological parameters, pre-LTx BCC colonization was associated with increased post-LTx mortality, whereas, for P. aeruginosa, colonization did not confer an increased mortality risk but limited evidence suggested that panresistant isolates may be associated with worse post-LTx outcomes. The main finding was that parameters, such FEV1, PH, BMI, CFRD and female gender, associated with increased mortality before LTx do not confer a higher risk for post-LTx mortality. Risk estimates of post-LTx mortality based on these factors may underestimate the LTx survival benefit in CF, and the current use of BMI as a relative contraindication may need to be adjusted for the CF-population. These findings support the use of CF-specific estimates of risk for LTx listing and prioritization. Future research focusing on more granular information (e.g. BCC strains, glycemic control, malnutrition measures other than BMI), potentially modifiable factors, multidimensional factors such as frailty, as well as separate analyses for adult and pediatric populations may help refine patient risk stratification, optimize access to LTx and improve post-LTx outcomes. Moreover, to understand independent impact of these factors on outcomes, an individual patient data meta-analysis using either retrospective or prospective data would help us to refine the quantitative role of individual predictors. Finally, standardized reporting should be a requirement for publications on this subject to facilitate future integration and interpretation of the results.
AK contributed to the study concept and design, study selection, data extraction, quality assessment, manuscript drafting and revision. AK had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. RV contributed to study selection, data extraction, quality assessment and revision of the manuscript. JS contributed to data analysis and revision of the manuscript. AOC designed the literature search strategy, performed the literature search and contributed to the revision of the manuscript. PSS provided methodological and statistical advice and revised the manuscript for important intellectual content. CC and ET revised the manuscript for important intellectual content. LGS and ALS contributed to the study design and supervision, and revised the manuscript for important intellectual content. All authors reviewed the study findings, read and approved the final version before submission.
During this study, AK received grants from the Swiss National Science Foundation (grant P300PB_164733) and the University Hospital of Lausanne (Fond de perfectionnement). The funding sources had no role in study design, data collection, data analysis or writing of the report.
Conflict of interest
We declare no competing interests.
We would like to thank all the authors who replied to our enquiries, and especially Dr. Fred Adler, Dr. Hubert Chen, Dr. Libor Fila, Dr. Denis Hadjiliadis, Dr. Jan Havlin, Dr. Don Hayes, Dr. Markus Hofer, Dr. Naritaka Kimura, Dr. Theodore G. Liou, Dr. Robert Lischke, Dr. Michael Parkins, Dr. Julia Pritchard, Dr. Antoine Roux, Dr. Nadim Srour, Dr. Dmitry Tumin, Dr. Lucie Valentova-Bartakova and Dr. Stijn Verleden for so kindly providing additional results and information for their studies that helped us perform this systematic review and meta-analysis.