Ten renin-angiotensin system-related gene polymorphisms in maximally treated Canadian Caucasian patients with heart failure

Abstract

AIMS

Racial differences in survival outcomes point towards a genetic role in the pathophysiology of heart failure. Furthermore, contemporary evidence links genetics to heart failure (HF) predisposition. We tested for a difference in prevalence of 10 renin-angiotensin-aldosterone system (RAAS)-related gene polymorphisms between a homogenous population of HF patients and healthy controls.

METHODS

One hundred and eleven healthy volunteers and 58 HF patients were included in this study. The healthy control group consisted of males aged between 18 and 35 years old. The HF group consisted of patients (89.7% male) who were 63.8 ± 7.9 years old, were in New York Heart Association (NYHA) class II-III and had a documented left ventricular ejection fraction (LVEF) ≤ 40% within the previous 6 months. Despite being treated maximally for their condition with angiotensin-converting-enzyme (ACE)-inhibitors and β-adrenoceptor blockers, they continued to be symptomatic and, as such, were a highly specialized and homogeneous patient population. Both groups were composed of Canadian Caucasians. The analyzed polymorphisms were: ACE (I/D), angiotensin-II-receptor-type-1 (AGTR1)(A1166C), angiotensinogen (AGT)(M235T and T174M), endothelial-nitric-oxide-synthase (eNOS)(T-786C and Glu298Asp), adrenergic-receptor-â2 (ADRB2)(Gln27Glu), bradykinin-receptor-β2 (BDKRB2)(+9/−9), aldosterone-synthase (CYP11B2)(T-344C) and adducin-1 (ADD1)(Gly460Trp).

RESULTS

The AGT (T235) allele (P = 0.0025, OR 2.02, 95% CI 1.24, 3.30) was found to be more prevalent in our HF group. The AGT (174M)-AGT (235T) haplotype was also associated with the HF phenotype (P = 0.0069). Exploratory evaluation of gene-gene combinations revealed an indicative association of the AGT (T235)/ACE(D) combined polymorphisms in the HF group (P = 0.02, OR 2.12, 95% CI 1.11, 4.06).

CONCLUSIONS

This study demonstrates that the SNPs of AGT may be associated with HF in our population and that the AGT/ACE gene combination may play an important role in disease predisposition.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• The progression and pharmacological response of heart failure-affected patients are subject to interindividual variability.

• t is also acknowledged that the genotype frequency of certain gene polymorphisms varies across different ethnic groups and that a difference in gene polymorphism frequencies between healthy and heart failure patients seems to exist.

WHAT THIS STUDY ADDS

• This study investigated associations between 10 gene polymorphisms of RAAS-related genes with an individual's susceptibility to heart failure.

• Our data suggest that the angiotensinogen (AGT) 235 single nucleotide polymorphism (SNP) may be associated with heart failure in our population and that the AGT(M174)-AGT(T235) haplotype, as well as the AGT/angiotensin-converting enzyme (ACE) gene combination, may play an important role in disease predisposition.

Introduction

Heart failure (HF) is a major cause of morbidity and mortality [1]. It remains an increasingly prevalent illness where only half of the patients survive beyond 5 years after diagnosis [2]. Despite being progressively debilitating, this condition does not affect everybody equally and its evolution and pharmacological response are subjected to heterogeneity among individuals [3]. Indeed, racial differences in survival outcomes point toward a genetic role in the pathophysiology of HF [4]. Great attention has therefore been drawn toward the genetic makeup of the renin-angiotensin-aldosterone system (RAAS), a neurohormonal pathway that is activated in HF patients. The RAAS plays a major role in the pathophysiology of HF and its chronic stimulation negatively affects the already failing heart through, most importantly, increased vasoconstriction, sodium and water retention, heart remodelling and myocardial fibrosis [5]. Moreover, its blockade is strongly advocated by the current American College of Cardiology/American Heart Association (ACC/AHA) Heart Failure management guidelines through the use of angiotensin-converting-enzyme inhibitors (ACEIs), angiotensin II type 1 receptor blockers (ARBs) and aldosterone antagonists [6]. Given the value of RAAS as a therapeutic target in this disease, variance in the genetic constitution of this system may represent a predisposing factor to HF and be implicated in the risk of disease manifestation in certain patients.

Whereas several studies found an association between certain RAAS-related gene polymorphisms and HF [7–9], others did not [10–12]. The most promising results actually came from studies considering major risk factors for HF, such as by Cambien et al.[13], which found that the DD genotype of the ACE gene, coding for the angiotensin-converting enzyme, was significantly more prevalent in patients with myocardial infarction than in controls. That finding was further substantiated in a meta-analysis by Samani et al.[14] and subsequently deemed of relatively modest significance in an updated meta-analysis that included an additional large case-control study [15]. Furthermore, numerous studies implicated the DD genotype in other pathological conditions considered as key risk factors for HF, such as hypertrophic cardiomyopathy [16, 17] and hypertension [18, 19], albeit with inconsistent results. In the context of HF predisposition, the other substantially studied genetic variant of the RAAS-related genes is the angiotensinogen (AGT) M235T single nucleotide polymorphism (SNP). Most studies did not find an association between that SNP and an increased risk of HF [12, 20]. Nonetheless, Goldbergova et al.[21] did reveal a disease-related haplotype (GGMT). Furthermore, numerous studies demonstrated a connection between the AGT 235TT genotype and cardiac hypertrophy in Asian populations [22, 23], where incidentally the T allele has a higher frequency.

The RAAS is composed of numerous components, where consequences of one polymorphism may be counterbalanced by compensatory effects of another. The aim of the present study was to analyze associations between 10 polymorphisms within RAAS-related genes (Figure 1) and risk of HF in a homogenous, maximally treated population of patients, and to compare gene frequencies between Canadian Caucasians with other ethnic groups. A supplementary objective was to perform an exploratory evaluation of the association of certain gene-gene combinations within the studied 10 RAAS-related genes with the risk of HF. The gene combinations were selected according to their pathological relevance and based on the rational that the possible deleterious effect of one polymorphism may be further increased by a subsequent polymorphism that lies in line relatively to the studied pathway; whereas the allele selected for each combination was the one deemed heart-detrimental from earlier reports [24–35]. Due to the limited number of HF patients, we restrained our analysis to a selection of 11 combinations of two gene polymorphisms, and two triple combinations consisting of the ACE(D)/AGTR1(1166C)/CYP11B2(C) and ATG (235T)/ACE(D)/CYP11B2(C) alleles.

Figure 1.

Renin-angiotensin-aldosterone pathway and other related systems

Methods

Study population

The HF group was derived from patients participating in a multicentre clinical trial that agreed to participate in a pharmacogenetic substudy. Briefly, that multicentre clinical study was a randomized, double-blind, placebo-controlled trial investigating the effects of candesartan in patients with stable systolic HF already on an ACE inhibitor treatment. Patients included were ≥18 years old and had symptomatic HF of the New York Heart Association (NYHA) functional class II-IV for at least 3 months prior to randomization. The patients also had a documented left ventricular ejection fraction (LVEF) ≤40% within the previous 6 months. Recruited patients were required to receive a stable dose of an ACE inhibitor for at least 1 month prior to inclusion. Major exclusion criteria were: treatment with an ARB within 8 weeks prior to randomization, known hypersensitivity to an ARB, serum creatinine ≥200 μmol l⁻¹, serum potassium ≥5 mmol l⁻¹, a history of ACE inhibitor induced hyperkalaemia, bilateral renal stenosis, symptomatic hypotension and/or systolic blood pressure ≤85 mmHg, stroke, acute MI or open heart surgery within the last 4 weeks before randomization, significant liver disease, connective tissue disease of chronic inflammatory condition, and any other condition that in the opinion of the investigator would jeopardize the evaluation of the efficacy or the safety of candesartan. The healthy control group was comprised of 18–35 years old male volunteers that had no history of cardiovascular disease.

All participating HF patients signed an informed consent form specific to the genetic substudy, which was approved by all the participating centres, in addition to signing the informed consent form to participate in the main trial. The volunteer group also signed an informed consent form relevant to genetic research that was approved by the ethics committee of the Quebec Heart Institute.

Genotyping

A 7 ml blood sample was collected in an EDTA-containing tube at anytime during the main study or after its completion and was kept at room temperature (for a maximum of 24 h) or at 4°C (for a maximum of 7 days) until the deoxyribonucleic acid (DNA) was isolated. DNA extraction from whole blood was performed with the GenElute Blood Genomic DNA Kit (NA2000 Sigma, USA). Extracted DNA was then stored at −20°C until genotyping was performed. The studied gene polymorphisms were amplified by polymerase chain reaction (PCR) and processed by restriction enzymes (when needed). The processed PCR products were then analyzed by electrophoresis on agarose gels and visualized by ethidium bromide staining under an UV light. Because of the expected small number of patients participating in the genetic substudy, examination of genetic polymorphisms was limited to those with known high allele frequencies. Therefore, genotyping was performed for the following gene polymorphisms according to previously published protocols: ACE I/D [36] (rs1799752), AGT T174M [37] (rs4762), AGT M235T [38] (rs699), angiotensin II type I receptor (AGTR1) A1166C [39] (rs5186), aldosterone synthase (CYP11B2) C-344T [40] (rs1799998), bradykinin β₂ receptor (BDKRB2) +9/−9 insertion/deletion [41] (rs5810761), α-adducin (ADD1) Gly460Trp [42] (rs4961), endothelial nitric oxide synthase (eNOS) Glu298Asp [43] (rs1799983), eNOS T-786C [44] (rs2070744), β₂-adrenergic receptor (ADRB2) Gln27Glu [32] (rs1042714). To eliminate the possibility of mistyping subjects heterozygous for the ID alleles of the ACE gene as D homozygous due to the preferential amplification of the D allele and the inefficiency in amplification of the I allele of the ACE gene [45], an additional PCR using insertion-specific primer pair, which recognizes only the insertion sequence, was performed (DD check) [46]. In addition, because the ATR1 A1166C gene polymorphism was not in Hardy–Weinberg equilibrium (P = 0.03) in the HF group, all heterozygotes were regenotyped with an additional technique to exclude genotypic errors [47].

Statistical analysis

To test for Hardy–Weinberg equilibrium, the expected genotype frequencies were calculated from the allele frequencies and deviation from the observed genotype frequencies was determined using the chi-square statistic and 10 000 replicates were used for exact P value computations. Genetic association was tested by comparing allele and genotype frequencies between the HF group and the healthy group using a contingency table and a chi-squared analysis or the Fisher's exact test if necessary. The Armitage trend test was substituted for the allele test in the analysis of AGTR1 as Hardy–Weinberg equilibrium was not met. Haplotype association was assessed using an omnibus test performed over all haplotypes with a likelihood ratio statistic testing the null hypothesis of no likelihood ratio differences between cases and controls, using EM haplotype estimates and by computing exact P values with 100 000 Monte Carlo replicates. Linkage disequilibrium was evaluated at AGT and eNOS using the D′ statistic with EM-inferred haplotypes and the composite linkage disequilibrium statistic was used to test for departure from the null expectation of no linkage disequilibrium using the chi-square test. Statistical analyses were conducted with SAS v.9.1.3 (SAS Institute Inc., Cary, NC, USA). All tests were two-sided with significance threshold set to 0.05. Exploratory statistics investigating the role of gene-gene combination of variant alleles were conducted with the statistical program NCSS (Hintze, J; 2001. NCSS and PASS. Number Cruncher Statistical Systems. Kaysville, Utah. http://www.ncss.com). All exploratory tests were two-sided with significance threshold set to 0.05. Group sample sizes of 58 cases and 111 controls achieve 80% power to detect an allele frequency difference of 0.16 (odds ratio 1.98) for an allele of 35% population frequency, using a two-sided Chi-square test with continuity correction and with a significance level of 0.05.

Results

The HF group consisted of 58 individuals (Table 1). These patients were mostly male (89.7%) Caucasians (100%), aged 63.8 ± 7.9 years old, with a mean LVEF of 27.3% ± 7.5 and with mainly an ischaemic subjacent cause to their HF (89.7%). Almost a third were diabetic (31%). All patients were receiving an ACE inhibitor and 93.1% were also on β-adrenoceptor blocker therapy. The unmatched, heart-healthy, control group consisted of 111 young Caucasian males aged between 18 and 35 years old, who were exclusively of French-Canadian descendants.

Table 1.

Characteristics (mean ± SD) of the heart failure population

Variables	n = 58
Men (%)	89.7
Caucasians, %	100
Age (years)	63.8 ± 7.9
Ischaemic HF (%)	89.7
LVEF (%)	27.3 ± 7.5
NYHA class II/III/IV (%)	56.9/43.1/0
Diabetes (%)	31.0
Atrial fibrillation (%)	24.1
Pharmacological treatment (%)
ACE inhibitors	100
β-adrenoceptor blockers	93.1
Digoxin	62.1
Furosemide	75.9
Spironolactone	27.6
Lipid-lowering agent	89.7

All of the studied genotype distributions were compatible with Hardy–Weinberg expectations in both cases and controls, with the exception of the AGTR1 A1166C gene polymorphism in the HF group (P = 0.03). After re-genotyping all of the heterozygotes for that polymorphism with another genotyping protocol [47], no patient was in need of reclassification, making genotyping errors as a cause for the Hardy–Weinberg deviation improbable. The genotype frequencies in the healthy male French-Canadian population, as well as their comparison with the HF group, are shown in Table 2.

Table 2.

Comparison of genotype frequencies between the healthy and the heart failure patients

Gene (wt/mut)	Healthy wt/wt	mut/wt	mut/mut	Heart failure wt/wt	mut/wt	mut/mut	Genotype test P value	Allele test P value
ACE (I/D)	18	47	45	5	31	22	0.2800	0.7265
	16.4%	42.7%	40.9%	8.6%	53.5%	37.9%
AGTR1 (A1166C)	48	49	13	23	33	2	0.1205	0.7081†
	43.6%	44.6%	11.8%	39.7%	56.9%	3.5%
AGT (M235T)	51	50	10	16	28	14	0.0074	0.0025
	46.0%	45.1%	9.0%	27.6%	48.3%	24.1%
AGT (T174M)	88	23	0	38	19	1	0.0483	0.0446
	79.3%	20.7%		65.5%	32.8%	1.7%
eNOS (T-786C)	41	53	17	15	34	9	0.3128	0.3498
	36.9%	47.8%	15.3%	25.9%	58.6%	15.5%
eNOS (Glu298Asp)	40	57	14	22	23	12	0.2731	0.6367
	36.0%	51.4%	12.6%	38.6%	40.4%	21.1%
ADRB2 (Gln27Glu)	26	65	20	21	27	10	0.2195	0.2490
	23.4%	58.6%	18.0%	36.2%	46.6%	17.2%
BDKRB2 (+9/−9)	26	59	26	23	23	11	0.0802	0.0670
	23.4%	53.2%	23.4%	40.4%	40.4%	19.3%
CYP11B2 (T-344C)	34	54	23	23	22	12	0.3872	0.4248
	30.6%	48.7%	20.7%	40.4%	38.6%	21.1%
ADD1 (Gly460Trp)	69	38	4	37	21	0	0.4452	0.6707
	62.2%	34.2%	3.6%	63.8%	36.2%

^†Armitage trend test.

We noted a significant difference of the AGT T235 allele (P = 0.0025) in the HF patients with frequencies of 48.3% in HF patients vs. 31.5% in the controls. The AGT M174 allele was also present in 18.1% of HF patients vs. 10.4% of healthy controls (P = 0.0446). Those two polymorphisms were in prominent linkage disequilibrium (LD) together in the HF group (P < 0.003) and in the healthy group (P < 0.005) given that the T235 allele was more common in individuals with at least one M174 allele (D′ = 0.731). We conducted a haplotype association test with the alleles of the AGT gene. The haplotype consisting of the variant alleles at both loci, AGT M174 and AGT T235 was present in 18.1% of HF patients vs. 8.4% of healthy controls. The likelihood ratio statistic using exact P value computation found a strong association with P = 0.0069.

With regards to the nine other gene polymorphisms studied, none attained the threshold of significance, though the BDKRB2 +9/−9 polymorphism reached borderline significance (P = 0.0670) with 39.5% of patients carrying the insertion vs. 50.0% of controls.

Subsequently, we conducted an exploratory study comparing the distribution of selected gene combinations in the HF and healthy control groups. There was an increased frequency of patients having at least one ACE(D) allele in combination with at least one ATG (T235) allele in the HF group when compared with the frequency of patients having that combination in the healthy group (P < 0.02) (OR 2.12, 95% CI 1.11, 4.06). Other gene combinations tested yielded no significant results (Table 3).

Table 3.

Frequency of individuals with selected allele combinations

	Healthy (%)	Heart failure (%)	P value	Odds ratio (95% CI)
≥1 allele AGT 235T and	45.0	63.8	0.02	2.12
≥1 allele ACE D				(1.11, 4.06)
≥1 allele AGT 235T and	30.9	44.8	0.07	1.81
≥1 allele AGTR1 C				(0.94, 3.47)
≥1 allele AGT 235T and	37.8	44.8	0.38	1.33
≥1 allele CYP11B2 C				(0.70, 2.53)
≥1 allele ACE D and	47.7	53.4	0.48	1.25
≥1 allele AGTR1 C				(0.67, 2.36)
≥1 allele ACE D and	57.7	55.2	0.76	0.90
≥1 allele CYP11B2 C				(0.48, 1.70)
≥1 allele AGTR1 C and	40.0	36.8	0.69	0.88
≥1 allele CYP11B2 C				(0.46, 1.69)
≥1 allele ACE D and	64.9	56.9	0.31	0.72
≥1 allele BDKRB2 + 9				(0.38, 1.36)
≥1 allele ACE D and	54.1	65.5	0.15	1.60
≥1 allele eNOS −786C				(0.83, 3.07)
≥1 allele ACE D and	57.6	58.6	0.90	1.04
≥1 allele eNOS 298Asp				(0.55, 1.97)
≥1 allele BDKRB2 + 9 and	49.5	48.3	0.88	0.95
≥1 allele eNOS −786C				(0.51, 1.79)
≥1 allele BDKRB2 + 9 and	47.7	39.7	0.32	0.72
≥1 allele eNOS 298Asp				(0.38, 1.37)
≥1 allele ACE D and	32.4	39.7	0.35	1.37
≥1 allele AGT 235T and				(0.71, 2.63)
≥ 1 allele CYP11B2 C
≥1 allele AGT 235T and	26.1	37.9	0.11	1.72
≥1 allele AGTR1 C and				(0.88, 3.38)
≥1 allele ACE D

We retrieved published genotype frequencies of our tested gene polymorphisms for various ethnic groups (Table 4) for descriptive comparison purposes that are intended to be qualitative in nature only. Compared with other Caucasians, our population had a similar distribution of allele and genotype frequencies. One could argue for a slight difference in the genotype frequency of the AGT M235T and the AGTR1 A1166C SNPs. However, the data representing Caucasians in this table was retrieved from various sources which included diverse populations (e.g. Germans, French, English, Scandinavians, etc.). Conversely, the Asian population differs widely from our population with respect to practically every gene polymorphism studied, except for the AGT T174M SNP. The most striking difference lies in the BB₂R⁺-polymorphism, where the Asian population is solely composed of –/– homozygotes. Moreover, the genotypic distribution of certain SNPs was reversed when compared with Caucasian populations (e.g. wild-type/wild-type > variant/variant for Caucasians and variant/variant > wild-type/wild-type for Asians), such as in the case of the AGT M235T, ACE I/D and the ADD1 Gly460Trp gene polymorphisms. Finally, the African-American and the African populations, which were virtually similar in terms of their genotypic frequencies, also appeared extensively divergent from our population.

Table 4.

Comparison of genotype frequencies (as %) between the French-Canadian population and other ethnic groups

	ACE (I/D)	AGTR1 (A1166C)	AGT (M235T)	AGT (T174M)	CYP11B2 (T-344C)	ADD1 (Gly460Trp)	ADRB2 (Gln27Glu)	BDKRB2 (+/–)	eNOS (T-786C)	eNOS (Glu298Asp)
Africans (Black)	DD 29.2–44	AA 90	MM 4	TT 91	TT 62	GG 94	GnGn 69	n/a‡	n/a‡	n/a‡
	ID 41.5–49.6	AC 10	MT 27	MT 9	TC 34	GT 6	GnGu 28
	II 14.5–21.2	CC 0	TT 69	MM 0	CC 4	TT 0	GuGu 3
African-Americans	DD 34	AA 88	MM 3	TT 86–90	TT 52	GG 76	GnGn 66	++: 35	TT 67	GG 79
	ID 48	AC 11	MT 28	MT 9–14	TC 40	GT 21	GnGu 30	+−: 45	TC 31	GA 20
	II 18	CC 1	TT 69	MM 0	CC 8	TT 3	GuGu 4	—: 19	CC 2	AA 1
Asians	DD 17–29	AA 84	MM 8	TT 86	TT 49	GG 25	GnGn 85	++: 0	TT 83	GG 86
	ID 34–36	AC 15	MT 42	MT 14	TC 44	GT 49	GnGu 14	+−: 0	TC 16	GA 13
	II 36–49	CC 1	TT 50	MM 0	CC 7	TT 27	GuGu 1	—: 100	CC 1	AA 1
Caucasians	DD 25–41	AA 52	MM 28–40	TT 72	TT 29	GG 57.0–74	GnGn 36	++: 22	TT 30	GG 41
	ID 39–57	AC 39	MT 41–53	MT 26	TC 52	GT 20–37.0	GnGu 48	+−: 50	TC 47	GA 46
	II 17–25	CC 9	TT 13–28	MM 2	CC 18	TT 2.6–8	GuGu 16	—: 28	CC 24	AA13
French Canadians	DD 40.9	AA 43.6	MM 45.9	TT 79.3	TT 30.6	GG 62.2	GnGn 23.4	++: 23.4	TT 36.9	GG 36.0
	ID 42.7	AC 44.6	MT 45.1	MT 20.7	TC 48.7	GT 34.2	GnGu 58.6	+−: 53.2	TC 47.8	GA 51.4
	II 16.4	CC 11.8	TT 9.0	MM 0	CC 20.7	TT 3.6	GuGu 18.0	—: 23.4	CC 15.3	AA 12.6

^‡n/a = not available.

Discussion

Our results report an increased frequency of the AGT T235 allele and the AGT 235TT genotype in HF patients. In addition, the AGT M174 allele was also significantly more prevalent in the HF group. The haplotype consisting of these variant alleles at both loci (AGT M174/AGT T235) was more prevalent in patients compared with controls, as was the ACE(D)/AGT (T235) allele combination. It is important to underline the fact that our results stem from a maximally treated patient population who represented a narrow NYHA class (II and III only), were all white and were drawn from the same geographical basin, and as such, represent a homogenous population of HF affected individuals.

The consistent result of both studied AGT SNPs being significantly more prevalent in the HF patients reflects their prominent linkage disequilibrium, a finding also shown by others [48, 49]. Whereas most studies did not find an association between the angiotensinogen M235T SNP and the development of HF [12, 50, 51], our outcome is, on the contrary, highly significant with a P value of 0.0025.

Mindful of the deleterious effects of the RAAS on the prognosis of HF and the importance of angiotensinogen as the initial step of this pathway, any variation increasing its concentrations could be considered as detrimental. Indeed, the angiotensinogen concentrations are increased in a stepwise manner with the number of T235 alleles [26, 52]. Additionally, animal data provided evidence that angiotensinogen concentrations influence blood pressure, and hypertension is the major risk factor for HF. Furthermore, transgenic animals overexpressing the AGT gene had an elevation in blood pressure and, conversely, AGT-knock-out mice had a reduction in blood pressure [53, 54]. Human data also suggest a connection of the T235 variant with hypertension [55–59]. However, the majority of studies evaluating the role of this variant in the context of myocardial remodelling have not been able to identify any association [60–64].

The ACE I/D and the AGTR1 A1166C gene polymorphisms are probably the most studied genetic variants in the field of cardiovascular disease. Since the products of those two genes are directly implicated in the pharmacological modulation of the RAAS, they evidently rouse great interest. We noted no difference in genotype frequency between the healthy and the HF groups, neither in the ACE, nor in the AGTR1, gene polymorphisms. However, the AGTR1 A1166C gene polymorphism was not in Hardy–Weinberg equilibrium in the latter group. Since regenotyping produced no patient reclassification, hence making genotypic errors unlikely, a survival bias could be evoked. Indeed, although not statistically significant, the HF group had a lower proportion of patients with the CC genotype (3.5%) when compared with the healthy group (11.8%). On the other hand, susceptibility to HF may only be present in particular situations, such as the concomitant presence of other risk factors for that disease. Indeed, Hindorff et al.[65] demonstrated that although the AGTR1 A1166C SNP was not associated with any incident cardiovascular events in the overall studied population, treated hypertensives with the CC genotype were more likely to develop HF and ischaemic stroke during their follow-up. Similarly, the incidence rate of HF increased with the number of d-alleles of the ACE gene only in hypertensive, and not normotensive, patients in a study by Schut et al.[66]. In those cases, hypertension seems to act as a trigger that renders the ACE D and the AGTR1 C alleles significant in the predisposition to HF. Similarly, one could also envisage a genetic variance in the upstream genes being that trigger and fuelling the whole pathway, such as the AGT M235T gene polymorphism for instance. To our knowledge, however, our study is the first to reveal an indicative association of the AGT (T235)/ACE(D) combined polymorphisms with HF predisposition. Nevertheless, a fine example of such gene–gene interaction was presented for a hypertension model by Rankinen et al.[67] demonstrating an impaired blood pressure training response in only those men who carried both the AGT 235TT genotype and at least one ACE D allele. These findings are along the lines of a subcohort study of the Framingham Heart Study showing that the increased risk of hypertension associated with the AGT 235TT genotype is further elevated in subjects also carrying the ACE DD genotype [59]. Since hypertension is a direct major risk factor for HF (also indirectly through myocardial infarction), one can stipulate that down the road the AGT M235T-ACE I/D interaction can indeed augment the risk of HF.

On the other hand, certain gene polymorphisms may indeed be more important in the progression of a disease rather than in its predisposition. This in turn may cause a survival bias when analyzing results from association studies. For example, Andersson et al.[68] illustrated that HF patients with the ACE DD genotype had a significantly higher 5-year mortality than their II/ID counterparts. In this case, since the II/ID patients had a longer longevity, they presumably also had a higher chance of being recruited in a study, artificially inflating their composition in frequency analysis. Therefore, studies having missed the ACE DD association with HF and reported similar genotype frequencies between HF patients and controls may indeed have fallen victim to the selective survival of II/ID patients and selective mortality for the DD patients.

One major contribution of this study resides in the characterization of genotype frequencies for 10 polymorphisms related to the RAAS in a healthy young male French-Canadian population, which served as our control group. The advantage of this control group is that it represents the basal frequency of the polymorphisms in our population, before survival selection supposedly occurs. Furthermore, phenotype characterization with such a group is not an issue, whereas in age–matched genetic association studies of older individuals, a patient could one day be included in the control group, and the next, after a heart attack for example, in the case group. The selection of a random control group for the case–control association tests, however, does not isolate HF as well as it would have been possible with a similarly ascertained population matched for important covariates. As such, the interpretation of results with respect to HF has to be considered with care. However, we consider that limitation, when weighed against survival selection and phenotype misclassification, a reasonable trade-off. Furthermore, we praise the method of carefully selecting the HF population in order to avoid any dilution of associations, a shortcoming encountered frequently when the studied group is not uniform. We therefore decided to make our studied group as homogenous as possible with regards to their disease stage, their pharmacological response, their ethnic background as well as their geographical location. The obvious drawback of choosing such a tight population is that they are in a short supply and are therefore difficult to recruit, which clearly limited our number of study subjects. Indeed, the modest number of participants recruited is clearly the principal limitation of this study. We therefore advocate that these results be replicated in a larger trial. Nevertheless, given our allele frequencies for M235T, we had 83% power to detect the resulting difference in spite of the limited number of subjects. We believe that selecting a uniform group of individuals allows for a gene effect to concentrate and is consequently easier to detect, which is reflected in our highly statistically significant results.

In conclusion, our study indicates that the AGT M235T polymorphism may very well be associated with HF predisposition. Also, the same may apply to the AGT (235T)/ACE(D) gene-gene polymorphism combination. Although the majority of RAAS-related gene polymorphisms studied in this trial were not associated with a risk for predisposition to HF, other polymorphisms that we have not included in our analysis may indeed play that role. Furthermore, due to our limited sample size, we cannot exclude that these variants do indeed have an impact, each individually minor, on the risk of HF development. HF is probably caused by many genetic factors that are all components of larger complex systems that interact with environmental factors. Besides the need for larger studies to examine the effects of single genetic variants and haplotypes, future studies also need to focus on the complexity of these systems and study gene–gene interactions and gene–environment interactions.