The renin-angiotensin system (RAS) has been well documented to contribute to the pathophysiology of ALI/ARDS by increasing vascular permeability . ACE is a key enzyme of the RAS that converts inactive angiotensin I to the vasoactive and aldosterone-stimulating peptide angiotensin II that metabolizes kinins and many other biologically active peptides . ACE is found in varying amounts on the surface of lung epithelial and endothelial cells. Activation of the RAS can stimulate production of tumor necrosis factor alpha in cardiac fibroblasts . Angiotensin II induces apoptosis of lung epithelial and endothelial cells and is a potent fibrogenic factor [33–36]. Based on these biological properties of ACE, there is considerable interest in its potential involvement in ALI/ARDS.
Despite conflicting findings among the genetic association studies, a recent meta-analysis  suggested a positive association between the ACE I/D polymorphism and ARDS susceptibility among Caucasians. Here, we have extended this meta-analysis to a much larger sample size (532 ALI/ARDS patients, 3032 healthy controls, and 1432 patients without ALI/ARDS), and failed to detect an association between the ACE I/D polymorphism and ALI/ARDS risk. The discrepancy may be because of the different methodology used and the different sample size. We believe that our result is more convincing because we added two studies to those used by Hu et al.  and also used three genetic models, as recommended in the literature . Furthermore, we extended our meta-analysis to examine mortality, and detected a significant association between the ACE I/D polymorphism and the mortality risk of ALI/ARDS in Asian populations.
ALI/ARDS patients had a higher frequency of the D/D genotype than controls did (Additional file 1: Table S1). The ACE I/D polymorphism has been associated with 28–47% of the variance in circulating ACE levels in healthy subjects, and high plasma ACE levels are associated with the D/D genotype [6, 7]. It is therefore possible that the high plasma ACE levels produced by having the D/D genotype affect the mortality of ALI/ARDS. Recently, ACE inhibitors have been reported to attenuate ALI induced by bleomycin, acid, and endotoxin in animal models [37–39]. Taken together, patients at risk of mortality from ALI/ARDS, as determined by ACE I/D polymorphism screening, may be good candidates for treatment with ACE inhibitors. Furthermore, because our results suggest that the ACE I/D polymorphism genotype might influence treatment outcome through differences in drug metabolism and activity, it will be important to investigate the pharmacogenomics of ALI/ARDS and gene–gene interactions in the RAS to clarify the role of different RAS genes in the pathophysiology and treatment response of ALI/ARDS.
We did not find an association between the ACE I/D polymorphism and ALI/ARDS mortality in the total population. However, there were significant heterogeneities among the ORs. This heterogeneity may be derived from: (1) different ancestries, (2) incomplete genotyping or genotyping error differences among the studies, and (3) a relatively small total sample size (196 survivors and 210 non-survivors). Differences in the genotype distributions have been reported between Caucasians and Asians, which may support the above explanations (the studies of Villar et al. and Adamzik et al. reported that the major allele was “Deletion”, but other Asian studies showed that the major allele was “Insertion”). Therefore, we stratified the population into Caucasian and Asian subjects, thereby eliminating the significant heterogeneity among the ORs for both ethnicities. After stratification, we detected a significant association between the ACE I/D polymorphism and ALI/ARDS mortality in Asian populations. D/D was the risk genotype for mortality in Asian ALI/ARDS patients.
Several limitations of our meta-analysis remain, e.g., a small sample size and differences in the quality control among the studies. Neither the mode of inheritance nor the heritability of ALI/ARDS is known. Therefore, we used three genetic models to obtain more statistical power than in previous individual studies. Based on the common disease-common variant hypothesis, and assuming ALI/ARDS to be a complex disease, at least 4000 samples ( i.e., 2000 cases and 2000 controls) are required to obtain sufficient statistical power (GRR = 2.0 and P = 5 × 10-7) . Given that our study is under-powered because of its small sample size, we may have failed to detect a genuine association between the ACE I/D polymorphism genotype and ALI/ARDS susceptibility. A replication study using a larger sample size, and/or samples from other populations, is required to obtain conclusive results.