OBJECTIVES: The aim of this review is to determine the clinical effectiveness and cost-effectiveness of enzyme replacement therapy (ERT) in the treatment of symptomatic Gaucher's disease. DATA SOURCES: Major electronic databases were searched from their inception to August 2003; and updated from January 2003 to July/August 2004. REVIEW METHODS: Databases were searched for studies that met the criteria and selected data were extracted and evaluated. Studies were assessed for their relevance to the UK context and the review objective. The bibliographic databases were also searched to identify existing cost studies, economic evaluations and models. A Markov decision model was constructed based on patients moving between states defined by the modified Severity Score Index (SSI). Most of the parameters were derived from the published literature. ERT was assumed to restore patients to full health in the base case. RESULTS: Sixty-three studies were included, all suggestive of benefit with ERT. However, the way in which the effects translate into patient well-being and survival or the need for services and resources has not been reliably estimated. Quality of life improvements with ERT have been reported. Nonetheless, studies based on the Short Form 36 (SF-36) indicate that patients treated with ERT continue to have reduced health-related quality of life (HRQoL) compared with the general population. No study attached utility values to quality of life measures for ERT-treated patients. Thirty-one studies relevant to the natural history of the disease were found. Sixteen looked at multiple clinical characteristics of a cohort of patients with type I Gaucher's disease. There was considerable within-study and between-study heterogeneity, but all showed that Gaucher's disease was a progressive condition. Some suggested that the disease may become more indolent in adulthood; however, studies were discrepant on this point. Most disease is diagnosed in adulthood, although about one-quarter presented in childhood, these patients having the most severe symptoms and greatest rate of progression. Modelling of natural history was undertaken using the five papers that reported the SSI for each patient, along with patient-level data on age, age at diagnosis, splenectomy status and genotype, to address the question of whether disease stabilises in adulthood and the degree of correlation between phenotype and genotype. Analysis of the available data suggested that disease progression is likely to slow markedly in adulthood and that genotype is a useful predictor of clinical expression of the disease. Five studies looked at quality of life. Data on this topic were also obtained from the registries. The evidence suggests that the vast majority of the clinical characteristics of type I Gaucher's disease have little impact on subjective HRQoL and that therefore for the majority of people with type I Gaucher's disease this may not be a severe condition. Bone and skeletal symptoms contribute most to the morbidity of the disease and can lead to severe pain and immobility. The mean cost per patient treated was approximately pounds sterling 86,000 per annum in England and Wales. The cost per patient varied considerably by dose. Four existing economic evaluations were found, all of which calculated a very high cost per quality-adjusted life-year (QALY). Using the Markov decision model, ERT was assumed to restore patients to full health in the base case. The estimated incremental cost per QALY [incremental cost-effectiveness ratio (ICER)] in the base case ranged from pounds sterling 380,000 to pounds sterling 476,000 per QALY, depending on genotype. Univariate sensitivity analyses examined ERT not restoring full health, more severe disease progression in the untreated cohort, and only treating the most severely affected patients. These produced ICERs of approximately pounds sterling 1.4 million, pounds sterling 296,000 and pounds sterling 275,000 per QALY, respectively. The base-case unit cost of the drug is pounds sterling 2.975. The unit cost would have had to be reduced ten-fold, to pounds sterling 0.30, to obtain an ICER of pounds sterling 30,000 per QALY. At a unit cost of pounds sterling 1 the ICER would be pounds sterling 120,000 per QALY. CONCLUSIONS: Although ERT for treating the 'average' Gaucher's disease patient exceeds the normal upper threshold for cost-effectiveness seen in NHS policy decisions by over ten-fold, some argue that since orphan drug legislation encouraged the manufacture of Cerezyme, and Gaucher's disease can be defined as an orphan disease, the NHS has little option but to provide it, despite its great expense. More information is required before the generalisability of the findings can be determined. Although data from the UK have been used wherever possible, these were very thin indeed. Nonetheless, even large errors in estimates of the distribution of genotype, genotype--phenotype associations, effectiveness and numbers of patients will not reduce the ICER to anywhere near the upper level of treatments usually considered cost-effective. Further research could help to clarify the many uncertainties that exist. However, although doing so will be of clinical interest, it is questionable whether, within the current pricing environment, such research would have any substantive impact on policy decisions. It is highly improbable that, whatever the findings of such research, the ICER could be brought down by the orders of magnitude required to make ERT an efficient use of health service resources. (The possible exception to this would be investigating the most efficient alternative treatment strategies for using ERT in a paediatric population only.) Moreover, if under equity considerations for orphan diseases the NHS feels it is important to provide this drug, regardless of its cost-effectiveness, then refining the precision of the ICER estimate also becomes superfluous.