TexGen is a collaborative, prospective genetic registry which enrolls patients with any personal or family history of cardiovascular disease who seek care at several institutions within the Texas Medical Center including the University of Texas Health Science Center, The University of Texas M.D. Anderson Cancer Center, Baylor College of Medicine and their affiliated hospitals, and St. Luke’s Episcopal Hospital at the Texas Heart Institute. The cohort includes patients admitted with acute coronary syndromes, strokes, transient ischemic attacks, those undergoing percutaneous coronary interventions or CABG, as well as those undergoing vascular surgical procedures.
For the current analyses, we restricted the population to TexGen patients undergoing CABG (with or without valve surgery procedures) from September 2001 through September 2008, who self-reported their race as Caucasian to avoid the issue of population stratification.
Pre-operative, intra-operative and postoperative characteristics of patients were obtained through a database maintained at the St. Luke’s Episcopal Hospital at The Texas Heart Institute. This research database prospectively collects information on patients enrolled in TexGen. The database has written documentation for each field that is coded and has systems in place to ensure reliability of the data, including range edit limits for every data field, consistency checks between fields, and periodic inter-rater reliability. Data from patient charts were abstracted by 3 trained abstractors, with 95% agreement among abstractors. Monthly electronic quality-control checks were conducted to detect logical errors. When verification coding (recoding the same admission) was performed on a minimum of 10 admissions per month, it found that 95% of the data collected was correct.
The variables used for our analysis included age, sex, history of acute coronary syndromes, hypertension, diabetes mellitus, concomitant valve surgery, prior history of AF, reduced left ventricular ejection fraction (defined as <0.35), New York Heart Association functional class at the time of surgery, pre-operative BB, anti-arrhythmic medication, or statin use, and aortic cross-clamp and total bypass time.
Prospective follow-up of outcomes included annual follow-up phone calls by the TexGen research nurses as well as annual surveys mailed to each patient. In addition, any hospitalization for patients enrolled in the database was also verified using hospitalization records.
Genotyping of rs2200733 and rs10033464 variants was performed using validated TaqMan assays (Applied Biosystems). These two polymorphisms on chromosome 4q25 were chosen a priori
as these polymorphisms are the ones extensively studied in the literature thus far.1–9
PCR product was amplified utilizing 0.9 μM each of the forward primer and reverse primers, 0.2 μM each of the FAM and VIC MGB (minor groove binder) sequence-specific probes, 3 ng DNA, 5.0 mM MgCl2
, and 1X TaqMan Universal PCR Master Mix containing AmpliTaq Gold DNA Polymerase in a 5.5 μl reaction volume. Both the SNPs had a call rate of greater than 99%. Hardy Weinberg (HW) equilibrium of P>0.05 was calculated using a chi-square goodness-of-fit model. QC concordance for 37 samples was 100%.
Our primary outcomes of interest included the development of postoperative AF after CABG surgery and long term development of AF after discharge. Postoperative AF was defined as the occurrence of any AF during the hospitalization for the index surgery. All patients were on continuous telemetry (either in the intensive care units or the telemetry floors) throughout their hospital stay, and any episode of AF was included in the analysis. Long term development of AF after the index hospitalization for CABG included any new AF which either necessitated hospitalization or was noted on an electrocardiogram for a hospitalization for a reason other than AF. We also performed exploratory analyses to assess whether the associations between these polymorphisms and the risk of postoperative AF could be modulated by age, gender or the type of pre-operative therapy (pre-operative BB, statin, or anti-arrhythmic) used.
Our secondary outcomes of interest included the development of postoperative stroke, development of stroke at long term follow-up, as well as long term survival. Stroke was defined as clinical evidence of a focal neurologic deficit along with a radiologic defect on computed tomogram or magnetic resonance imaging scan of the brain. Vital status on these patients was ascertained using data from the Texas State Bureau of Vital Statistics.
Categorical and continuous variables were compared between the groups using Pearson chi square test or Wilcoxon rank sum test, respectively. Hardy-Weinberg equilibrium expectation of allelic frequencies was tested using a chi-square goodness-of-fit model.
For the development of postoperative AF, we initially performed univariate analyses to study the associations between each SNP, rs2200733 and rs10033464, and the risk for postoperative AF development. Logistic regression analyses were subsequently performed using a limited adjustment model that included age and gender, followed by a more extensive adjustment model that included age, gender, prior history of AF, hypertension, pre-operative beta-blocker, statin or anti-arrhythmic use, and concomitant valve surgery as covariates to ascertain whether the polymorphisms of interest were independent predictors of postoperative AF in the fully adjusted models.
For long term risk of AF, we initially carried out log rank test to study the association between rs2200733, rs10033464 and time to development of AF. Cox proportional-hazards models (adjusting for age, gender, and hypertension) were subsequently used to determine whether polymorphisms in rs2200733 and rs10033464 were independently associated with long term development of AF. Similarly, logistic and Cox regressions analyses were used to determine whether rs2200733 and rs10033464 were associated with development of postoperative stroke or long term development of stroke, respectively. The associations between rs2200733, rs10033464 and survival were analyzed using log rank test.
For all the above mentioned analyses, a dominant model was used to study the associations between polymorphisms in rs2200733 and rs10033464 and outcomes of interest. Statistical analyses were performed using SAS statistical software version 9.1 (SAS Institute, Inc.; Cary, NC). All analyses were performed using two-tailed tests for significance. Because we were evaluating these polymorphisms on an a priori basis, a p value <0.05 was considered statistically significant and adjustments were not made for multiple comparisons.