Biomarkers of cardiovascular injury and stress are associated with increased frequency of ventricular ectopy: a population-based study

This is a substudy of the Akershus Sleep Apnea Project (ASAP), which is a community-based, cross-sectional study conducted in south Eastern Norway. In a two-phased design, we first randomly drew 30 000 persons between 30 and 65 years of age from the National Population Register (Fig. 1). All participants received the Berlin Questionnaire, a standardized questionnaire for classifying the risk for obstructive sleep apnea (OSA) [18], by mail. Of the 16302 (55.7%) responders, 1772 persons were randomly drawn as a pool for inclusion in phase 2. A 2:1 ratio of high risk:low risk participants according to the Berlin questionnaire in the Phase 2 clinical sample resulted in oversampling of participants with risk factors for obstructive sleep apnea, i.e. snoring, daytime sleepiness, obesity and hypertension. In addition, persons with previous CHD, otitis media surgery or diabetes were oversampled. Predefined age-dependent strata were made to ensure gender equilibrium and a 2:1 ratio of high risk:low risk participants according to the Berlin Questionnaire. Invitations were made by telephone, and 585 of the invited refused to participate, 202 were not reached after three attempts, and 28 persons were excluded. Exclusion criteria of the study included use of continuous positive airway pressure, physical impairment, pregnancy and insufficient Norwegian language skills. In the end, 535 (46.7% of the contacted) persons participated in the clinical phase of the study, while 442 persons from the pool of 1772 were not contacted as the predefined strata were completed. Further details of the recruitment protocol have been described previously [19].

Fasting blood samples were collected from an antecubital vein in the morning, and were centrifuged within one hour. Serum samples were stored at – 20°C for a maximum of 20 days, before they were transported to the central biobank and frozen at −80°C pending later measurements. Sampling and storing failed for 4 participants. hs-TnI was measured using the ARCHITECT STAT hs Troponin assay (Abbott Diagnostics, Illinois, USA) as previously reported [20]. NT-proBNP was measured by electrochemiluminescent immunoassay (Roche Diagnostics, Basel, Switzerland) with a lower detection limit of 5 pg/ml. The hs-CRP measurements were performed using a latex immunoassay kit on a COBAS INTEGRA400 (Roche Diagnostics, Basel, Switzerland), with a lower detection limit of 0.1 pg/mL. Analyses of biomarkers were not performed for 9 participants due to lack of available plasma. Creatinine, glucose and lipids were measured by standard laboratory methods. The creatinine clearance was estimated by the Cockcroft-Gault formula [21].

All participants underwent an in-hospital 5-channel ambulatory electrocardiography (Medilog AR12, Oxford Instruments Medical, Surrey, United Kingdom) for assessment of cardiac arrhythmias. Recordings with a shorter duration than 10 h, or with technical failure, were excluded from the analysis (n = 10). The mean length of the recordings was 18.3 h. The electrocardiography recordings were automatically analyzed by a software engine (Medilog Darwin, ScanMed Medical, Gloucestershire, United Kingdom) according to predefined criteria. A premature ventricular complex (PVC) was defined as a N-V complex >15% shorter than the previous 3 N-N complexes. All PVC’s were counted and dichotomized at a cut off of 5/h. We defined complex ventricular ectopy as any bigeminy (≥3 consecutive V-N complexes), trigeminy (≥3 consecutive V-N-N complexes), or nonsustained ventricular tachycardia (≥3 consecutive V complexes). One researcher, blinded for the participants’ medical background and biomarker levels, manually reviewed and re-scored all recordings. A second researcher manually reviewed 10% of the recordings. The inter-rater correlation was 0.99. Examples of frequent PVC and complex ventricular ectopy are shown in Fig. 2.

The participants were asked about history of CHD (myocardial infarction or coronary artery revascularization), history of congestive heart failure, use of antihypertensive medication, diabetes mellitus and current smoking habits and underwent a standard physical examination. Blood pressure was measured in the sitting position after 15 min rest by an automatic device (Dinamap, ProCare 400, GE HealthCare, Milwaukee, WI, USA). The mean of the two last out of three recordings was calculated and used in analysis. The Sokolow-Lyon criteria were used for estimation of left ventricle hypertrophy [22]. Body mass index (weight in kg/ height in meters squared) was calculated. The participants underwent in-hospital polysomnography, as previously reported [19]. Data were manually scored using the Somnologica 3.2 software (Flaga-Medcare, Buffalo, New York), 13 registrations were not analyzed due to technical failure or short sleep duration. The apnea-hypopnea index (AHI) was calculated using the average number of apneas (airflow <10% of reference in more than 10s) plus hypopneas (airflow <30% in more than 10s and subsequent ≥4% fall in oxygen saturation) per hour of sleep.

Due to skewed distribution of data, we transformed NT-proBNP, hs-TnI, CRP and AHI concentrations logarithmically prior to the analysis. Categorical data are presented as absolute numbers and percentages, continuous data as mean (standard deviation) or median (interquartile range). The associations between biomarkers and ventricular arrhythmias were investigated by univariate and multivariate logistic regression analyses. According to a previous report from the same study, OSA may be associated with increased risk of PVC’s [23], consequently we included the AHI as covariate in addition to the conventional cardiovascular risk factors (age, gender, history of CHD, history of diabetes, history of hypertension, body mass index, systolic blood pressure, total cholesterol:HDL cholesterol ratio). Covariates significantly (p < 0.05) associated with arrhythmias in univariate analyses were first entered into multivariate models. The cardiovascular biomarkers were subsequently added to make separate multivariate models to evaluate whether hs-TnI, NT-proBNP or hs-CRP added significant incremental information to the regression models. The results are reported as odds ratio (OR) and 95% confidence intervals (CI) and a p value (2-sided) < 0.05 was considered statistically significant for all analyses. All statistical analyses were performed with the SPSS statistical software package, version 20 (SPSS Inc., Chicago, USA).