Fabry disease is a rare X-linked lysosomal storage disorder characterised by deficiency of alpha-galactosidase and the accumulation of its substrate, globotriaosylceramide (Gb3).1 The cardiovascular manifestations, which include myocardial hypertrophy, fibrosis and ischaemia, and arrhythmia, are common, affecting more than 50% of males and females, and represent the leading cause of death.2 3

Disease-specific therapy, including enzyme replacement therapy (ERT) or chaperone therapy (migalastat), is recommended for patients exhibiting cardiovascular involvement, but its efficacy, optimal timing of initiation and cost-effectiveness are yet to be fully elucidated.

Previous systematic reviews are limited by the exclusion of chaperone therapy, which has been part of routine clinical practice for over 5 years, and the exclusion of non-randomised studies, which account for much of the published data.4 5 Furthermore, by evaluating effectiveness in Fabry disease in general, evaluation of the cardiovascular impact of disease specific therapy has been relatively limited.

This study aimed to systematically review and evaluate the effectiveness of disease-specific therapy, compared with placebo, and to no intervention, for the cardiovascular manifestations of Fabry disease.

This systematic review evaluated the effectiveness of disease-specific therapy compared with placebo, and to no intervention, for the cardiovascular manifestations of Fabry disease.

The included studies were heterogeneous in design, size, comparator, risk of bias and outcome. The ERT RCTs (n=5) and chaperone therapy RCTs (n=2) were small, and in the case of ERT, under-represented females (ERT: 183 males and 13 females, migalastat: 49 males and 75 females), but were not at serious risk of bias. The 16 non-randomised studies of ERT with a comparator group were all at serious risk of bias. Comparator groups included patients not requiring disease-specific therapy, who presumably had milder phenotypes although treatment criteria varied considerably, previously published datasets, ‘natural history’ cohorts and interrupted time-series analyses. There were no non-randomised studies of chaperone therapy with a comparator group. The remaining 49 studies (ERT 45; chaperone therapy 4) were non-randomised studies of intervention without a comparator group, and all were at serious risk of bias.

Studies were predominantly small and single-centre, although the non-randomised ERT studies included large international registries. Inclusion criteria (eg, all-comers vs exclusively males,24 exclusively females,30 or patients with pre-existing cardiac or renal disease),8 12 duration (20 weeks to 10 years), and analysis methodology (eg, within-group comparisons, between-group comparisons, stratification by variables such as age,24 sex or LVH,13 open-label extensions) were highly variable, and many reported salient levels of missing data.

Outcome measurements were particularly heterogeneous. Clinical cardiovascular events were seldom reported and were infrequent when they were reported, precluding meaningful analysis. LV mass assessments comprised the most common endpoints (65 studies), although these were also inconsistent, including LV mass, LV mass indexed to body surface area, LV mass indexed to height12 13 19 and surrogates of LV mass including septal, posterior wall and segmental thickness (online supplemental reference 1).30 The level of heterogeneity made effectiveness of disease-specific therapy difficult to assess.

In a series of publications with increasing follow-up duration relating to a 20-week placebo RCT, ERT was consistently associated with a reduction in cardiac endothelial Gb3 (online supplemental references 2 and 3), although this is the only study to evaluate the impact of disease-specific therapy on cardiac endothelial Gb3. ERT was not associated with a reduction in myocardial Gb3 (online supplemental reference 2).12

Data regarding the impact of disease-specific therapy on cardiac phenotype are inconsistent. A number of studies found ERT and chaperone therapy to reduce or slow the progression of measurements of LV mass, predominantly in patients with LVH at baseline. However, other studies did not identify a significant impact of disease-specific therapy on measurements of LV mass. The reasons for the variable results are unclear, but may include differences in participant characteristics, duration of therapy and differential treatment response. In most studies, assessment of LV mass was made using echocardiography. The limited accuracy and high variability of echocardiography-derived LV mass, particularly in the context of variable ventricular geometry, such as is evident in Fabry disease, is well described (online supplemental reference 4), and measurement of LV wall thickness, even when performed using CMR images, is highly variable (online supplemental reference 5). Together with the small sample sizes and often short duration, many studies may not have been sensitive to relatively small changes in LV mass, especially considering the slowly progressive nature of Fabry disease.

In other conditions, tissue characterisation with CMR has identified myocardial injury and disease expression in advance of changes in ‘macro’ structure and function, such as LV mass or ejection fraction (online supplemental reference 6). In line with this, myocardial T1 relaxation time is a putative non-invasive biomarker of Gb3 accumulation. The available studies suggested that ERT may possibly be associated with an improvement in T1 relaxation time, however, data were very limited and somewhat inconsistent.16 21 Data regarding the impact of disease-specific therapy on LGE, a measure of focal myocardial fibrosis, were also very limited (online supplemental references 7 and 8).16

Previous systematic reviews focus exclusively on ERT and have been variable in their conclusions. A Cochrane review of RCTs concluded that the long-term influence of ERT on risk of morbidity and mortality remains to be established.5 Two recent systematic literature reviews incorporating observational data concluded that in males: ‘data published in adult male patients with Fabry disease demonstrates that the effect of ERT on plasma Gb3 levels, eGFR, and cardiac outcomes is strongest and substantiated by a wide range of publications, showing consistent, dose-dependent reductions in Gb3 accumulation, a reduced decline in eGFR, and improvements in cardiac outcomes’. Whereas in females: ‘ERT in adult female patients with Fabry disease has a beneficial effect on Gb3 levels and cardiac outcomes’ (online supplemental references 9 and 10). A meta-analysis concluded that ERT did have a beneficial effect on the course of LV mass when compared with untreated groups (online supplemental reference 11). Specifically, in males with LVH at baseline LV mass remained stable, whereas in males without LVH at baseline the rate of LVH progression was lower than in untreated patients. In females with LVH at baseline LV mass decreased, and in females without LVH at baseline, LV mass remained stable compared with an increase in untreated patients. Importantly however, this meta-analysis included data from only 6 of 64 studies that report LV mass (online supplemental reference 11).

Given the marked heterogeneity of study design and outcome measurements, the relatively small sample size of most studies and low reported clinical event rates, the risk of study bias and the rare and slowly progressive nature of the condition, it remains unclear whether disease-specific therapy sufficiently impacts the cardiovascular manifestations of Fabry disease, particularly in the context of the not inconsiderable associated cost.

In future, large, statistically powered, multi-centre, prospective studies assessing the efficacy of therapy using standardised outcomes are required. The impact of disease-specific therapy on pre-defined ‘hard’ clinical cardiovascular outcomes, such as sudden cardiovascular death or malignant ventricular arrhythmia must be prioritised and additional secondary phenotypic outcomes derived from contemporary imaging, circulating biomarker and heart rhythm techniques should be measured using contemporary techniques. Moreover, if these secondary phenotypic outcomes are to be meaningful, their prognostic significance must be understood. In particular, the prognostic significance of a change in LVMI and native myocardial T1 relaxation time over time requires further elucidation (online supplemental reference 12). This future work is likely to require international collaboration, expert consensus, patient and patient-group involvement and impartial funding.

A limitation of the current study is that a meta-analysis was not performed; however, the heterogeneity of published studies precluded statistical pooling and meta-analysis. Agalsidase alfa and beta were not considered separately in keeping with most studies. Measurements such as LVMI are subject to variability based on methodological approach (inclusion or exclusion of papillary muscles). Patients receiving disease-specific therapy are generally more severely affected than those who are untreated. Direct or indirect industry involvement is common in Fabry disease studies, reflecting the nature of rare disease research. Study funders are listed in the tables summarising study characteristics (online supplemental tables 1–3). A formal Tool for Addressing Conflicts of Interest in Trials (TACIT) is being developed under the auspices of the Cochrane Bias Methods Group.