Prevalence of left ventricular diastolic dysfunction in European populations based on cross-validated diagnostic thresholds

Cardiovascular risk factors underlie the first stage of diastolic heart failure (HF). This stage evolves into asymptomatic left ventricular diastolic dysfunction (LVDD) characterized by impaired relaxation and increased left ventricular (LV) stiffness, and finally progresses to clinically overt diastolic HF [1]. Recently published community-based studies making use of conventional and tissue Doppler echocardiographic imaging (TDI) revealed a high prevalence of LVDD, ranging from 11.1% up to 34.7% [2–8]. In the Flemish Study on Environment, Genes and Health Outcomes (FLEMENGHO), the frequency was 27.3% [9]. One issue making the interpretation of the aforementioned reports difficult is that the prevalence of LVDD cannot be easily compared, partially because of differing diagnostic criteria and/or divergent distributions of cardiovascular risk factors in the sampled populations (for review see Additional file 1: Tables S1 and S2). In the FLEMENGHO study, [9] we derived age-specific criteria for the classification of LVDD based on a reference sample. Since the publication of our previous report, [9] echocardiographic examinations continued in the FLEMENGHO cohort and started in four centers taking part in the European Project on Genes in Hypertension (EPOGH). In the present study, we compared age-specific echocardiographic criteria for LVDD across populations as well as the correlates and prevalence of this condition. By showing consistency across populations, we aimed to propose a diagnostic classification that might be useful in clinical practice.

Epidemiological studies performed in the general population provide an unbiased estimation of the prevalence and prognostic significance of health-related states or events (LVDD in our study). Epidemiology is highly regarded in evidence-based medicine for identifying risk factors for disease and determining optimal treatment approaches for clinical practice. Because the process of myocardial remodeling starts before the onset of symptoms, we place special emphasis on the detection of subclinical (asymptomatic) LV systolic and diastolic dysfunction and the timely identification of patients who are at risk for developing overt HF. Thus, the present report focused on the prevalence and comparison of echocardiographic criteria for LVDD between populations. Mean age of the FLEMENGHO and EPOGH participants was 51.1 and 44.5 years and the prevalence of LVDD was 24.7% and 20.2%. The age-standardized estimated were 22.4% and 25.1%, respectively. The key finding was that age-specific cut-off limits for the transmitral E/A ratio and the threshold for the E/e' ratio, which we used for the classification of LVDD, were consistent and reproducible across two independently recruited population cohorts. The same also applied to the correlates of the E/A, e'/a' and E/e' ratios. These observations lend support to the clinical applicability of the criteria for the classification of LVDD as proposed in this report.

The reported prevalence of LVDD depends on several factors, such as the characteristics of the population under study, imaging techniques, and the criteria applied to diagnose or to grade LVDD (Additional file 1: Table S1). Redfield [2] and Abhayaratna [3] used similar, but not identical pulsed wave Doppler and TDI criteria. Three studies^4-6 defined LVDD exclusively based on transmitral E/A ratio cutoffs. Fisher [7] and Azevedo [8] applied different combinations of the criteria proposed by the European Working Group on Diastolic Heart Failure in 1998 [17]. The criteria proposed in the recently published recommendations [18] for the evaluation of LV diastolic function by echocardiography require further testing for their applicability in the general population.

In all aforementioned studies participants were randomly recruited from populations (Additional file 1: Table S2). Age at recruitment was more than 45 years in 3 reports [2, 4, 8] and more than 55 years in the Rotterdam Study [5]. The Augsburg study [7] encompassed an age range from 25 to 75 years, whereas the Canberra study [3] enrolled only elderly from 60 to 86 years. The prevalence of hypertension and diabetes ranged from 28.2% to 75.0% and from 3.3% to 54.7%, respectively. The prevalence of any grade of LVDD ranged from 11.1% in Augsburg [7] to 34.7% (7.3%) in Canberra [3] (Additional file 1: Table S1). In our current study, the frequency of any grade of LVDD was 23.0%. This estimate has to be interpreted, keeping in mind that age averaged 48.6 years (range, 14.2-89.5 years), and the proportions of women and patients with hypertension or diabetes mellitus were 42.5%, 42.4% and 3.1%, respectively.

Our current study showed that 3.5% and 2.8% of the FLEMENGHO and EPOGH participants had a low E/A ratio in the presence of an elevated E/e' ratio. To our knowledge, this group of patients was not described in previous reports [2, 3] with the exception of our previous FLEMENGHO report [9]. In the current report, we confirmed the existence of this type of LVDD in EPOGH participants. The underlying mechanisms require clarification. Parallel increases in the pressures in the LA and LV during diastole might contribute to this type of LVDD.

Recent community-based studies [2, 4, 19, 20] explored the prognostic role of classical Doppler indexes and the new TDI-derived parameters. In the Olmsted study [2] mild LVDD (hazard ratio, 8.31; P < 0.001) and moderate or severe diastolic dysfunction (10.2; P < 0.001) predicted all-cause mortality, while controlling for age, sex, and EF. However, in this study, the authors did not adjust the models for other important cardiovascular risk factors. In 1036 participants enrolled in the Copenhagen City Heart study [19], low systolic and A' velocities derived from colour Doppler imaging and averaged from 16 myocardial segments independently predicted total mortality [19]. The authors did not explore whether the new TDI indexes captured prognostic information over and beyond classical Doppler measurements of diastolic function as it was shown in previous studies [4, 20]. For instance, In the Cardiovascular Health Study,[20] the adjusted risk of development of overt HF was highest at the extremes of the distribution of the transmitral E/A ratio. In the Strong Heart Study,[4] all-cause and cardiac mortality had a U-shaped relation with the E/A ratio.

In the absence of an outcome-driven age-specific diagnostic reference frame, averaging the 2.5 and 97.5 percentiles for the E/A ratio in subjects free from cardiovascular diseases included in the FLEMENGHO and EPOGH cohorts, and rounding the resulting boundaries to the closest integer value, produced working definitions of normal LV diastolic function, which can be easily remembered. Following this approach, absolute values for the normal lower limit of the E/A ratio, would be higher than 1.2, 1.0, 0.9, 0.8, and 0.7 in subjects aged < 30, 30-39, 40-49, 50-59, and > 60 years, respectively. The lower boundaries of the age-specific thresholds for the E/A ratio decreased approximately by 0.10 per decade of age. In the FLEMENGHO and EPOGH studies, the 97.5th percentiles of E/e' ratio in the reference groups were 8.6 and 8.5, respectively. These findings are in line with the data reported from invasive studies, that an E/e' ratio of less than 8 accurately predicts normal LV filling pressure [17, 21].

Because NT-proBNP values vary with the degree of LVDD,²³ we also compared the levels of NT-proBNP among the LVDD groups. In line with our previously published FLEMENGHO findings,[9] we observed higher levels in subjects with impaired relaxation pattern (group 1) and in subjects with elevated LV filling pressure (group 2). However, in subjects who had an elevated E/e' ratio and an abnormally low age-specific E/A (group 3), the NT-proBNP levels were not different from those in subjects with normal LV function.

Our study has to be interpreted within the context of its potential limitations and strengths. First, the Doppler blood flow measurements and the TDI velocities are prone to measurement error. In the present study, one experienced observer in each center recorded all Doppler images using a common highly standardized imaging protocol [9]. Furthermore, all digitally stored images were centrally post-processed by one observer. Second, because NT-proBNP was measured by different methods in FLEMENGHO and EPOGH, we reported z-scores instead of the measured values of NT-proBNP. However, within cohorts the associations between the classification of LVDD and circulating NT-proBNP levels were consistent. Third, the patterns of transmitral flow and mitral annulus velocities depend to some extent on the compliance and contractile function of the left atrium. Thus, the evaluation of diastolic function using the proposed cut-offs might not be applicable to patients with sustained atrial fibrillation. Fourth, in continuous analyses, the correlates of LVDD in two cohorts were the same and in line with previous reports [3–5]. This observation might be considered as a validation of our current results.

In conclusion, LVDD has a prevalence of over 20% in European populations. To classify different grades of LVDD, we applied age-specific diagnostic thresholds based on a combination of Doppler velocities ratios derived from transmitral blood flow and mitral annular movement. We demonstrated that these thresholds were consistent and reproducible across population cohorts. This diagnostic classification needs validation in prospective studies in terms of progression of disease and as predictor of cardiovascular complications. Data collection to meet these goals is currently in progress in the FLEMENGHO and EPOGH cohorts.