Methods
Patients and Study Design
This study was approved by the North West – Liverpool East Research and Ethics Committee (11/NW/0810) and registered with clinicaltrials.gov (NCT02106286). Written informed consent was obtained from all patients. Between April 2012 and August 2013 we recruited consecutive, unselected patients who attended our vascular laboratory for routine AAA surveillance. We approached all patients aged >18 yrs undergoing active AAA surveillance (AAA <5.5 cms) who were able to perform a CPET, and had a WHO performance status of ≤2. We excluded patients who had a known contraindication to beta-blockers, severe ischaemic heart disease, stage IV or V chronic kidney disease, uncontrolled hypertension, those unable to consent, and absolute contraindications to CPET based on the ATS/ACCP guidelines.
Patients were classified into 2 cohorts: (i) chronically beta-blocked group (patients on long-term beta-blocker therapy at baseline) or (ii) acutely beta-blocked group (patients not on long-term beta-blockers at baseline). All patients underwent two CPETs; one on and one off beta-blockers at least one week apart. Patients in the 'chronic' group underwent their first CPET whilst on their present medication; the beta-blocker was then stopped for at least one week before a second CPET. The 'acute' group underwent the first CPET without any alteration in medical therapy, then, at least one week later, a second CPET was performed after the commencement of bisoprolol once daily, 48 h before the test (minimum two doses, adjusted for body weight: 1.25 mg if the patient weighed 50–75 kg, 2.5 mg if 75–100 kg and 3.75 mg if >100 kg). All patients were informed of potential risks of the alteration in their medical therapy at the time of consent. The recruitment of both types of patients replicated real clinical practice where acute commencement or withdrawal of beta blockade before a perioperative CPET is routinely encountered.
Measurements
CPET (Geratherm Respiratory GmbH; Love Medical Ltd, Manchester, United Kingdom) followed a standard protocol described elsewhere. CPET was reported by two experienced clinicians (MW and PW) with an experienced clinician scientist (SJ) resolving any differences. All 3 clinicians were blind to patient group allocation and beta-blocker treatment. Patient characteristics recorded included age, gender, height, weight, BMI, smoking status, haemoglobin concentration, aneurysm ultrasound details, medication details (type and dose of beta-blocker therapy) and co-morbidity, including prior patient self-reported diagnoses of diabetes, ischaemic heart disease, myocardial infarction, cerebrovascular disease, chronic obstructive pulmonary disease or heart failure. Resting flow-volume loops were used to derive Forced Expiratory Volume over 1 s (FEV1) and Forced Vital Capacity (FVC). Ventilation and gas exchange variables included oxygen uptake (V̇O2) both absolute and weight adjusted, ventilatory equivalents for oxygen and carbon dioxide (V̇E/V̇O2; V̇E/V̇CO2), oxygen pulse (V̇O2/heart rate), work rate and heart rate; all measured both at estimated lactate threshold (
) and at peak exercise. Our primary objective was to assess the change in these variables caused by beta blockade after adjustment for confounders. V̇ O2 at
and V̇ O2 at peak were primary outcome variables with V̇ E/V̇ CO2, workload, O 2 pulse, absolute V̇ O2 and heart rate both at
and peak treated as exploratory outcomes.
Statistical Methods
We estimated that 32 patients undergoing AAA surveillance were required in order to detect a 10% difference in V̇O2 at
. This estimate was based on a standard deviation of 1.5, using a two-tailed paired t-test with 90% power and allowing for a 25% drop out.
Continuous variables are summarized in terms of mean (sd) or median (IQR) if non-normal and categorical variables are presented as frequency (%). Univariate statistical comparisons of patient characteristics between beta-blockade groups were conducted; for continuous variables, using a two-sample t-test or a Mann-Whitney U-test when non-normal and for categorical variables, using a χ test or a Fisher's Exact test when cell frequencies were insufficient.
For the primary analysis, a linear mixed-effects model was fitted in order to investigate any difference in CPET variables when on and off a beta-blocker. A linear regression model with robust standard errors was used where the fit was inadequate and a log-transformation of O2 pulse was performed before analysis. Models were adjusted for beta-blockade group (chronic or acute) and a two-way interaction was investigated. Sensitivity analysis was conducted adjusting for age, gender, BMI, smoking status and presence of cardiac disease (myocardial infarction, heart failure or ischaemic heart disease). Model fit was assessed using residual and Q-Q plots and the adjusted estimates have been stated for each analysis. The significance level for the primary endpoint was <0.05. The P-values displayed for the exploratory endpoints were corrected for multiple comparisons using Simes's procedure and declared as significant if a P-value <0.02 was achieved.
Exploratory regression techniques were used to investigate the impact of heart rate and absolute V̇O2 on the change in O2 pulse. Estimates produced from modelling each CPET variable separately were combined to explore how much of the change in O2 pulse was as a result of a change in heart rate and how much was as a result of a change in V̇O2.
Results were considered significant if a P-value <0.05 was achieved. All statistical analyses were performed using Stata version 13 (StataCorp. 2013. Stata Statistical Software: release 13. College Station).