Objectives: To evaluate the clinical and cost-effectiveness of machine perfusion (MP) compared to cold storage (CS), as a means of preserving kidneys prior to transplantation. Transplantation of kidneys from both heart-beating donors (HBDs) and non-heart-beating donors (NHBDs) is considered. Finally to review whether the use of MP can allow valid testing of kidney viability prior to transplantation.
Data sources: Fifteen electronic bibliographic databases were searched. The reference lists of relevant articles and sponsor submissions were hand searched and various health service research-related resources were consulted via the Internet.
Review methods: A literature search was undertaken to identify relevant studies and a meta-analysis performed on the studies that had appropriate comparator groups and reported sufficient data. A structured review examined tests of viability of kidneys on MP. Economic modelling was used to determine the cost-effectiveness and cost-utility of MP.
Results: The meta-analysis suggested that the use of MP, as compared with CS, is associated with a relative risk of delayed graft function (DGF) of 0.804 (95% confidence limits 0.672 to 0.961). There was no evidence to suggest that this effect is different in kidneys taken from HBDs as opposed to NHBDs. Meta-analysis of 1-year graft survival data showed no significant effect, but the studies, even when aggregated, were severely underpowered with respect to the likely impact on graft survival. The size of effects demonstrated were in line with those predicted by an indirect model of graft survival based on the association of DGF with graft loss. The economic assessment indicated that it is unlikely that in the UK health setting complete cost recovery will be obtained from a reduction in the incidence of DGF. The probability that MP is cheaper and more effective than CS in the long term was estimated at around 80% for NHBD recipients and 50-60% for HBD recipients. Flow characteristics of the perfusate of kidneys undergoing MP may be an indicator of kidney viability, but data were inadequate to calculate the sensitivity and specificity of any test based on this. The concentration of alpha-glutathione-S-transferase (a marker of cell damage) in the perfusate may be the basis of a valid test. A threshold of 2800 micrograms/100 g gave a sensitivity of 93% and specificity of 33% (and hence a likelihood ratio of 1.41).
Conclusions: The baseline analysis indicated that in the long-term MP would be expected to be cheaper and more effective than CS for both HBD and NHBD recipients. A definitive study of the clinical benefit of MP in order to establish its effect on DGF and longer term graft survival would be valuable, together with an economic evaluation of the benefits. While direct evidence relating to improvements in graft survival would be preferable, the small predicted improvement indicates that a very large sample size would be required. In addition to seeking direct evidence of the impact on DGF, research quantifying the impact of DGF on graft survival in this technology is required. Research is also needed to establish whether a valid test (or combination of tests) of kidney viability can be developed.