enLanguage
Home / Knowledge / Content

Knowledge

Survival For Waitlisted Kidney Failure Patients Receiving Transplantation Versus Remaining On Waiting List: Systematic Review And Meta-analysis Ⅰ

May 14, 2024

Abstract 

Objectives To investigate the survival benefit of transplantation versus dialysis for waitlisted kidney failure patients with a priori stratification. Design Systematic review and meta-analysis. 

Data sources Online databases MEDLINE, Ovid Embase, Web of Science, Cochrane Collection, and ClinicalTrials.gov were searched between database inception and 1 March 2021. 

Inclusion criteria All comparative studies that assessed all causes of mortality for transplantation versus dialysis in patients with kidney failure waitlisted for transplant surgery were included. Two independent reviewers extracted the data and assessed the risk of bias in the included studies. Meta-analysis was done using the DerSimonian-Laird random effects model, with heterogeneity investigated by subgroup analyses, sensitivity analyses, and meta-regression. 

Results The search identified 48 observational studies with no randomized controlled trials (n=1245850 patients). In total, 92% (n=44/48) of studies reported a long-term (at least one year) survival benefit associated with transplantation compared with dialysis. However, 11 of those studies identified a stratum in which transplantation offered no statistically significant benefit over remaining on dialysis. In 18 studies suitable for meta-analysis, kidney transplantation showed a survival benefit (hazard ratio 0.45, 95% confidence interval 0.39 to 0.54; P<0.001), with significant heterogeneity even after subgroup/ sensitivity analyses or meta-regression analysis. 

Conclusion Kidney transplantation remains the superior treatment modality for most patients with kidney failure to reduce all causes of mortality, but some subgroups may lack a survival benefit. Given the continued scarcity of donor organs, further evidence is needed to better inform decision-making for patients with kidney failure. 

Study registration PROSPERO CRD42021247247.


cistanche benefits for kidney


HOW LONG DOES IT TAKE FOR CISTANCHE TO WORK FOR KIDNEY PATIENTS?

Introduction

Transplantation is established as the gold standard therapy of choice for patients with kidney failure because of improvements in mortality and quality of life for most eligible candidates. However, not all patients with kidney failure are deemed eligible for kidney transplant surgery owing to their personalized risk-benefit determination. For example, the latest UK Renal Registry report (up to 31 December 2019) shows that 28303 patients with kidney failure are managed on hemodialysis or peritoneal dialysis,1 with a greater but undefined number of patients living with advanced chronic kidney disease. Yet as of 31 March 2021, only 3525 patients were actively awaiting kidney transplant surgery (with 4321 waitlisted but

suspended).2 This suggests that not every patient with kidney failure is deemed suitable for the rigors of general anesthesia, kidney transplant surgery, and/or complications associated with a long term need for immunosuppression.


Although many studies have published their findings, Tonelli and colleagues were the last, a decade ago, to do a systematic review of cohort studies comparing adult kidney transplant recipients against patients on chronic dialysis.3 From 110 eligible studies, including 1922300 participants from 27 different countries, they found a significantly lower

mortality risk associated with kidney transplantation and an increase in the relative magnitude of the survival benefit over time (P<0.001). However, patient and/or study-level factors associated with greater or lesser benefit from transplantation were not identified. Additionally, because of the broad approach, most of the cohort studies included compared transplantation with all patients on dialysis, which introduces a significant selection bias.

1

Although many patients with kidney failure have contraindications prohibiting them from being considered candidates for kidney transplantation, comparing the survival of waitlisted patients with kidney failure who proceed with transplantation versus those who remain on dialysis allows a meaningful comparison. In a subset of studies (n=10), which compared transplant recipients with waitlisted patients on dialysis, Tonelli and colleagues found that the benefits of transplantation remained significant but were less pronounced.3 Moreover, although studies from across the globe were included, data were not stratified based on geographical region, meaning that concerns about external validity remained.

In the context of continued disparity in the supply of and demand for donors, understanding which subgroups of patients with kidney failure may not attain survival benefits after kidney transplantation is important for counseling and clinical decision-making. Any recommendations about the risk versus benefit of kidney transplant surgery for eligible kidney transplant candidates should be based on the best available evidence. Evidence has been gained across several cohorts, but a comprehensive and contemporary review using recommended analytical

techniques is lacking. We, therefore, systematically reviewed the survival benefit of kidney transplantation versus remaining on dialysis for waitlisted patients with kidney failure, with a meta-analysis of studies reporting eligible empirical data.


Methods

The study protocol was prospectively registered on the PROSPERO database (CRD42021247247) and conducted by PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses).4


7

Eligibility criteria

We included all studies comparing mortality between patients with kidney failure deemed suitable for transplant surgery (that is, waitlisted) receiving transplantation versus those remaining on dialysis with a minimum of one year of follow-up. We excluded studies based solely on pediatric populations (age <16 years) or multi-organ transplant recipients. We included only studies of primary kidney transplantation, as re-transplants represent a small fraction of transplant candidates with a different risk "phenotype" about survival outcomes.5 6 We also excluded studies with small cohort sizes (<30 patients), as Tonelli and colleagues did, to optimize work without appreciable loss of power and inflation of bias. We excluded reviews, expert opinions, editorials, correspondences, and case reports.

For eligibility criteria of the meta-analysis, we excluded studies reporting data before the 1980s as, although informative for a narrative synthesis, they do not reflect current practice for pooled analysis. We also reviewed studies for data sources, time periods, population structures, and geographical regions. Where studies presented overlapping populations, concerning periods and geographical regions, we gave priority for study selection to the more contemporary or largest data cohort. Similarly, in the context of two competing studies with overlapping patient cohorts (for example, age), we gave selection priority to the study with a general patient cohort (for example, all adults) over specific patient cohorts (for example, older

adults only). No two studies with potential overlapping patient data could be selected together.


Search strategy

Two reviewers (DC and AC) independently searched the following databases: MEDLINE, Ovid Embase, Web of Science, Cochrane Collection, and ClinicalTrials.gov. Searches were performed from inception to 1 March 2021, using a diverse amalgamation of Medical Subject Heading (MeSH) terms and search terms tailored to the tree structure and descriptors of each database to

improve their reach (see supplementary table A). We placed no limits on language or year of publication.



Selection of studies

Two reviewers (DC and AC) independently screened all records retrieved from the databases for relevancy on the hierarchical basis of title, abstract, and finally full-text review for eligible studies. For studies reported in a non-English language, we sought appropriate translators before assessment. Any disagreements were resolved through discussion or arbitration
with a third reviewer (AS). We followed up studies without accessible abstracts or full text by contacting the British Library (through interlibrary loan), along with the corresponding primary authors of the study. After failing to gain accessibility after these steps, we excluded studies. After full-text analysis, we screened reference lists of all eligible publications to flag additional studies

9


Data extraction and outcome measures

The two reviewers (DC and AC) extracted data independently from eligible studies, with consultation with a third reviewer (AS) as needed. We collected extracted data by using a standardized piloted data collection spreadsheet (Excel, Microsoft Corp, WA, US). Data extracted included study characteristics (study identifier, year of publication, country, source

of funding, data source, period of inclusion, follow-up duration, statistical approach adopted for survival analysis, sample size of waitlisted and transplanted patients, and special subgroups of population), population characteristics (age, sex, ethnicity, cause of primary kidney disease, and comorbidities), treatment characteristics (dialysis type, donor type, and immunosuppressive therapy regimen), and outcomes. The main outcome measure was all-cause mortality. Owing to heterogeneity highlighted during scoping searches, we deemed all forms of patient survival data acceptable, including but not limited to unadjusted hazard ratios or risk ratios, adjusted hazard ratios or risk ratios, survival curves, and crude dichotomous event rates. Adjusted hazard ratios and their respective confidence intervals were the primary outcome measures of interest for the meta-analysis; we noted and summarised the remaining measures to aid narrative synthesis. Where studies reported subgroup data for different comorbidities, age groups, or donor types, we extracted data for each stratum.



Risk of biased quality scoring

Two reviewers (DC and AC) assessed the quality of studies and their risk of bias independently by using the Newcastle-Ottawa Assessment Scale (NOS) for comparative non-randomized studies (cohort or case-control).7 8 The NOS consists of three quality parameters: selection of study participants (4 stars), quality of adjustment for confounding (2 stars), and ascertainment of the exposure or outcome of interest (3 stars). For the confounding criteria, 1 star was allocated if groups were comparable on variables of age and sex, and an additional star if groups were comparable on the cause of primary renal disease and comorbidity burden. Therefore, the maximum available score was 9, representing the highest methodological quality. Despite lacking any formalized criteria for what score constitutes a "high-quality study," many papers have conventionally regarded a NOS score ≥7 as the threshold.9 For this study, we adopted a more stringent approach, with a total score of ≤5 considered low, 6-7 considered moderate, and 8-9 deemed high quality. Any discrepancies between reviewers were resolved by discussion or arbitration with a third reviewer (AS).


14

Data synthesis and statistical analysis

We summarised the results of the systematic review both qualitatively and quantitatively. In the absence of individual patient data, we followed the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions and did the meta-analysis of time-to-event data by pooling reported long-term mortality hazard ratios from studies.10

Where possible, we chose the most adjusted overall estimate of transplantation versus dialysis (that is, multivariate regression over univariate regression). If a study reported regression estimates only by subgroup (for example, by donor type), we entered estimates separately into the analysis. If studies reported Cox regression results as relative risks, we regarded them

as hazard ratios. When hazard ratios and their 95% confidence intervals were unavailable, we estimated the hazard ratios, provided sufficient information was present (for example, log-rank test P values or Kaplan-Meier curves), using the statistical procedures described by Parmar et al in 1998 and Tierney et al. in 2007.11 12 To aid in this later step, Kaplan-Meier curves were digitalized with Digitizelt.13

After the calculation of the natural logarithmic hazard ratio and its standard error for each viable study, the DerSimonian-Laird random effects model to do the meta-analysis.14 We selected the random effects model a priori, as we expected significance between-study heterogeneity. We assessed inter-study heterogeneity with the Cochran Q test and quantified it by using the I2 statistic. For the Q statistic, we considered a P value <0.1 to be a statistically significant indicator of heterogeneity. I2 values considered representative of low, moderate,

and high risk of heterogeneity were <25%, 26-50%, and >50%, respectively.15 We assessed publication bias by using funnel plot analysis, with an evaluation of asymmetry by visual inspection followed by Egger's test.16

We planned a priori analyses stratified by geographical region, donor type, and population type. As estimating hazard ratios from Kaplan-Meier curves are lower down the hierarchy of evidence, we did a sensitivity analysis for the primary outcome by limiting studies to only those presenting adjusted hazard ratios with confidence intervals. Additionally, as we regarded Cox regression results reported as relative risks to be hazard ratios, assuming the authors had used inappropriate statistical terms, we did a sensitivity analysis of studies by restricting studies to only those that explicitly stated their results as hazard ratios. We did further sensitivity analyses by looking at studies based on the median point of case recruitment before and after the year 2000, outlier analysis (removing studies with confidence intervals that do not overlap with the confidence intervals of the pooled effect), leave-one-out influence analysis
to assess the effect of individual studies on the pooled effect,17 and construction of a graphical display of heterogeneity (GOSH plot). This plot is an exploratory combinatorial method that helps with the visualization of study heterogeneity by doing all possible meta-analyses in all the different study subsets (that is, the effect size of 2n−1 subsets).18 Lastly, to further

investigate the heterogeneity and effect of continuous study moderators, we also planned meta-regression (using mean age, maximum duration of follow-up, and median period of case recruitment as covariates).

We used R 4.0·4 for all analyses, with packages including tidyverse, meta, and metaphor.19-21 We defined statistical significance for a treatment effect as P<0.05, and all tests were two-sided.



Patient and Public Involvement

Although this research involved no direct patient or public involvement, the research question was informed from national kidney patient group meetings in which the risk versus benefit of kidney transplantation compared with remaining on dialysis has been discussed as a key question.































You Might Also Like