In hypertrophic cardiomyopathy (HC), atrial fibrillation (AF) is an important determinant of clinical deterioration due to heart failure or embolic stroke. This study characterizes left atrial (LA) structural and functional parameters to establish markers predictive of AF risk, using cardiovascular magnetic resonance (CMR) imaging. We studied 427 consecutive patients with HC in sinus rhythm with CMR (age 44 ± 18 years), including 41 who developed clinically overt AF after study entry (2.6 ± 2.1 years), 49 patients with AF before CMR, 337 patients with HC but without AF, and 244 normal controls. LA chamber was assessed for absolute and indexed end-diastolic volume (LAEDV), end-systolic volume, and percent ejection fraction (LAEF). In the 41 prospectively studied patients with HC who developed AF during follow-up, LAEDV was significantly greater than in patients without AF (146 ± 48 vs 107 ± 37 ml) or in normal controls (81 ± 24 ml, p <0.001). Percent LAEF was lower in patients developing AF (36 ± 10%) than without AF (46 ± 12%) or controls (55 ± 9%, p <0.001). Multivariate analysis identified LAEF (<38%), LAEDV (≥118 ml), and age (≥40 years) as independently associated with AF occurrence. In conclusion, CMR measures of LA remodeling and dysfunction reliably identified patients with HC at risk for future development of AF. Decrease in LAEF represents a strong novel marker of susceptibility to AF in this disease.
Atrial fibrillation (AF) is the most common sustained arrhythmia in hypertrophic cardiomyopathy (HC), occurring in about 20% of patients and potentially impacting the natural history and prognosis of HC by promoting progressive heart failure and increasing the risk for embolic stroke. Early recognition of susceptibility to AF would be advantageous for longitudinal surveillance and timely prophylactic interventions and management strategies in HC. Tomographic high spatial resolution cardiovascular magnetic resonance (CMR) offers a contemporary opportunity for characterizing the functional dynamics and structural characteristics of the left atrial (LA) chamber with respect to the clinical course of HC. To this purpose, we report a large HC cohort studied with CMR imaging, including a prospective subset that developed AF after study entry, to establish potential markers for future development of AF.
Methods
The Hypertrophic Cardiomyopathy Center (Minneapolis Heart Institute Foundation) database was accessed for patients with HC and CMR studies. Of the 427 patients consecutively studied with CMR from 2003 to 2010, 90 (21%) had at least one 12-lead electrocardiogram documented clinically overt symptomatic episode of AF requiring acute medical care, including pharmacologic or direct-current shock cardioversion. Ambulatory (Holter) electrocardiographic recordings were not performed routinely for the purpose of identifying asymptomatic episodes of AF.
Patients with HC were categorized as: (1) sinus rhythm without a history of AF (n = 337), (2) history of ≥1 AF episodes with the first occurring 1 week to 7 years (mean 2.6 ± 2.1 years) after CMR and study entry (n = 41; 37 paroxysmal and 4 permanent), and (3) ≥1 AF episodes occurring before CMR and study entry (n = 49; 40 paroxysmal and 9 permanent). An independent control group of subjects without evidence of cardiovascular disease by CMR (age 42 ± 18 years, men 62%, and body surface area [BSA] 2.0 ± 0.3 m 2 ) was assembled for comparison.
All patients were in normal sinus rhythm during the CMR study. Imaging was performed with a 1.5-T clinical magnetic resonance imaging scanner (Avanto; Siemens Medical, Erlangen, Germany). Cine true fast imaging with steady state precession (FISP) sequences were performed in standard 4- and 2-chamber views. Horizontal long-axis, vertical long-axis, steady-state free precession cine images: echo time/time to repetition (TE/TR), 1.5/3.0 ms; flip angle, 60°; slice thickness, 10 mm; in-plane resolution, 1.5 × 1.5 mm; temporal resolution, 45 ms; and breath hold duration, 14 to 17 heartbeats at end expiration. Timing of volume measurement in the scan was precisely determined by visual assessment of the maximal and minimal volumes in the temporal image sequence corresponding to the cardiac cycle.
CMR image analysis was performed on a commercial imaging workstation (Vitrea; Vital Images, Minnetonka, Minnesota). LA volume was calculated using a biplane area-length method previously described and indexed for BSA. Endocardial borders of the LA were traced manually in vertical and horizontal long-axis planes. Interobserver variability for this method has been shown to have favorable agreement for measurement of LA chamber volumes and LA ejection fraction (LAEF).
Maximum LA volume (LA end-diastolic volume [LAEDV]) was measured at the end of ventricular systole, and minimum LA volume was measured at the end of atrial systole (LA end-systolic volume [LAESV]). LA appendage and pulmonary veins were excluded from LA volume measurements. LA stroke volume was calculated as LAEDV − LAESV and LAEF as LA stroke volume/LAEDV. All volumes were indexed to BSA. The LA transverse diameter was measured in horizontal long-axis images from the level of the mitral annulus (point of leaflet coaptation) to posterior margin of the LA.
A short-axis stack was obtained parallel to the atrioventricular groove covering the entire left ventricle (LV). LV volumes (LV end-systolic and end-diastolic volumes) were measured using automated endocardial and epicardial border recognition tracing of successive short-axis slices at end-diastole and end-systole. Contour tracing was validated by reviewing cine images with contours attached and manually correcting if judged necessary by the operator. In normal control patients, CMR-derived LA and LV volumes were consistent with those of Maceira et al. Interobserver variability for measuring chamber volumes with CMR has been shown to be low.
Late gadolinium enhancement (LGE) images were acquired 10 to 15 minutes after intravenous administration of 0.2 mmol/kg gadolinium diethylene triamine pentaacetic acid (DTPA) (Magnevist; Schering, Berlin, Germany) using a breath-held, segmented, inversion-recovery sequence acquired in the same orientations as cine images. Mean signal intensity of normal myocardium was calculated, and the threshold ≥6 SD exceeding the mean was used to define LGE.
Seventy-three patients were screened with laboratory-specific methods for identification of mutations in the protein-coding exons and intron-exon boundaries sequencing the 8 myofilament genes that most commonly cause HC. DNA variants were judged to be pathogenic based on standard criteria.
Study entry was defined as the time of the patient’s first CMR study. Descriptive statistics are displayed as means and SDs for continuous variables and proportions for categorical variables. When continuous variables had skewed distributions, as determined by Shapiro-Wilk normality test, data were transformed to achieve normality. Continuous variables were assessed for statistical significance using the 1-way analysis of variance or Student t test; chi-square or Fisher’s exact tests were used for categorical variables.
Univariate analyses were performed to assess associations between individual predictors and clinical outcomes. Independent predictors of AF risk were assessed using multivariate logistic regression including the following variables: age, percent LAEF, LAEDV, LV volume, and LV ejection fraction. Only those variables with p <0.05 for univariate associations were entered into stepwise multivariate logistic regression models. In this model, variables with p <0.1 were retained. Only patients with complete data on all covariates were included in the multivariate analysis. Probabilities for the development of AF were then derived from variables retained in the final model.
Receiver operating characteristic (ROC) curves were generated from logistic regression models. Sensitivity, specificity, positive and negative predictive values, and accuracy thresholds were calculated. Areas under the curve were generated to determine the optimal variables for identification of risk for AF. Cut points for age, LAEF, and LV end-diastolic volume were chosen based on Youden’s index. Analyses were performed with Stata, version 11.2 (StataCorp LP, College Station, Texas).
Results
The 427 study patients were 44 ± 18 years of age at the time of CMR study; 303 (71%) were men ( Table 1 ). At study entry, most patients were asymptomatic in New York Heart Association class I (295; 69%); 91 were mildly symptomatic in class II, and 41 were severely symptomatic in classes III and IV. LV outflow obstruction (gradient ≥30 mm Hg) was present in 89 patients (21%). Percent LV ejection fraction by CMR was 74 ± 8%, with 286 patients (67%) <50%. Follow-up period was 3.4 ± 2.1 years.
| Parameter | All HC (n = 427) | HC: Without AF (n = 337) | AF: After CMR ∗ (n = 41) | AF: Before CMR (n = 49) | p Value |
|---|---|---|---|---|---|
| Age at CMR (yrs) | 44 ± 18 | 41 ± 18 | 50 ± 14 | 57 ± 13 | <0.001 |
| Men (%) | 303 (71) | 241 (72) | 27 (66) | 35 (71) | 0.75 |
| BSA (m 2 ) | 2.02 ± 0.29 | 2.01 ± 0.3 | 2.06 ± 0.28 | 2.04 ± 0.23 | 0.47 |
| NYHA class, initial visit | |||||
| I/II | 386 | 307 | 35 | 44 | 0.002 |
| III/IV | 41 | 30 | 6 | 5 | |
| NYHA class, last visit | |||||
| I/II | 389 | 310 | 37 | 42 | 0.002 |
| III/IV | 38 | 27 | 4 | 7 | |
| Maximum LV thickness (mm), CMR (range) | 22 ± 6 (10–41) | 22 ± 6 (11–41) | 23 ± 5 (14–37) | 21 ± 4 (10–32) | 0.29 |
| LV EF (%), CMR (range) | 74 ± 8 (23–91) | 74 ± 8 (23–90) | 77 ± 8 (51–91) | 72 ± 9 (40–88) | 0.004 |
| AF | |||||
| Paroxysmal | 77 | n/a | 37 | 40 | — |
| Chronic/persistent | 13 | n/a | 4 | 9 | — |
| LVOT gradient ≥30 mm Hg (%), rest (echo) | 89 (21) | 66 (20) | 11 (27) | 12 (24) | 0.45 |
| LA transverse dimension (echo; mm) | 41 ± 8 (20–63) | 40 ± 7 (20–62) | 46 ± 7 (31–60) | 46 ± 8 (34–63) | <0.001 |
| Surgical myectomy | 57 | 42 | 9 | 6 | 0.23 |
| RFA/Maze | 22 | — | 8 | 14 | |
| End stage (EF <50%) | 16 | 10 | 1 | 5 | 0.05 † |
| Coronary artery disease | 22 | 18 | 2 | 2 | 1.0 † |
| Time: from CMR to AF onset (range; mean) | — | n/a | 2.6 ± 2.1 yrs (1 wk–7 yrs) | n/a | — |
| Heart transplant | 6 | 2 | 2 | 2 | 0.02 † |
| Embolic stroke | 5 | 2 | 1 | 2 | 0.06 † |
| Noncardiac death | 9 | 5 | 2 | 2 | 0.12 † |
| HC death | 18 | 16 | 1 | 1 | 0.82 † |
| Heart failure | 3 | 3 | 0 | 0 | |
| Sudden death event | 14 | 12 ‡ | 1 § | 1 § | |
| Operative | 1 | 1 | 0 | 0 |
∗ No patient had myectomy, RFA, or Maze surgical procedure before CMR and study entry.
† Fisher’s exact test used to assess statistical significance.
‡ Includes 3 sudden deaths, 3 out-of-hospital cardiac arrest survivors, and 6 appropriate ICD interventions.
Of the 90 patients who experienced AF (77 paroxysmal and 13 chronic and/or permanent), age was 54 ± 13 years and 28 (31%) were ≥60 years (range 18 to 84). LV outflow obstruction (≥30 mm Hg) was present at rest in 23 patients (26%); 1 patient (1%) had heart failure progression with systolic dysfunction (ejection fraction <50%).
Percent LAEF was significantly lower in 41 patients with HC and AF after CMR (36 ± 10%) than 337 patients with HC without AF (46 ± 12%) or 244 normal controls (55 ± 9%, p <0.001) but similar to 49 patients with HC and AF before CMR (36 ± 17%, p = 0.97; Figures 1 and 2 and Table 2 ).
| Characteristic | Controls (n = 244) | HC Without AF (n = 337) | HC: AF After CMR (n = 41) | HC: AF Before CMR (n = 49) | p Value ∗ |
|---|---|---|---|---|---|
| Age at CMR (yrs) | 42 ± 18 | 41 ± 18 | 50 ± 14 | 57 ± 13 | 0.003 |
| Men %, (n) | 62 (151) | 72 (241) | 66 (27) | 71 (35) | 0.16 |
| BSA (m 2 ) | 2.0 ± 0.3 | 2.0 ± 0.3 | 2.1 ± 0.3 | 2.0 ± 0.2 | 0.30 |
| LAEF (%) | 55 ± 9 | 46 ± 12 | 36 ± 10 | 36 ± 17 | <0.001 |
| LAEDV (ml) | 81 ± 24 | 107 ± 37 | 146 ± 48 | 152 ± 70 | <0.001 † |
| LAEDV indexed (ml/m 2 ) | 40 ± 11 | 54 ± 20 | 72 ± 28 | 76 ± 39 | <0.001 † |
| LAESV (ml) | 37 ± 15 | 59 ± 29 | 95 ± 38 | 105 ± 72 | <0.001 † |
| LAESV indexed (ml/m 2 ) | 19 ± 7 | 30 ± 16 | 47 ± 22 | 52 ± 39 | <0.001 † |
| LA SV (ml) | 43 ± 13 | 48 ± 17 | 51 ± 18 | 47 ± 18 | 0.33 |
| LA SV indexed (ml/m 2 ) | 22 ± 6 | 24 ± 8 | 25 ± 9 | 23 ± 10 | 0.52 |
| LVEDV (ml) | 153 ± 38 | 153 ± 41 | 160 ± 49 | 166 ± 35 | 0.38 † |
| LVEDV indexed (ml/m 2 ) | 76 ± 16 | 77 ± 19 | 78 ± 26 | 82 ± 17 | 0.83 † |
| LVESV (ml) | 51 ± 20 | 41 ± 19 | 37 ± 17 | 48 ± 23 | 0.066 † |
| LVESV indexed (ml/m 2 ) | 26 ± 9 | 21 ± 10 | 18 ± 9 | 24 ± 11 | 0.022 † |
| LV SV (ml) | 102 ± 24 | 112 ± 31 | 123 ± 37 | 118 ± 26 | 0.046 † |
| LV SV indexed (ml/m 2 ) | 51 ± 10 | 56 ± 14 | 60 ± 19 | 58 ± 12 | 0.11 † |
| LV mass (g) | 120 ± 33 | 191 ± 80 | 210 ± 73 | 189 ± 56 | 0.20 |
| LV LGE present %, (n) | — | 146 (43) | 24 (59) | 25 (51) | 0.064 |
| LV LGE (%), median (25th–75th percentile) | — | 5.2 (2.5–10.4); n = 142 | 2.7 (1.6–6.9); n = 22 | 6.7 (3.3–18.2); n = 22 | 0.047 ‡ |
| NYHA functional class | |||||
| Class I/II % (n) | — | 91 (308) | 85 (35) | 90 (44) | 0.21 |
| Class III/IV % (n) | — | 9 (29) | 15 (6) | 10 (5) | |
| LVOT gradient ≥30 mm Hg, rest (echo) % (n) | 20 (66) | 27 (11) | 24 (12) | 0.28 |
∗ p Value compares HC without AF (n = 337) with HC and AF after CMR (n = 41).
† t Test performed on transformed variables due to non-normal distribution.
Absolute LAEDV was significantly greater in 41 patients with HC and AF after CMR (146 ± 48 ml) than 377 patients with HC without AF (107 ± 37 ml) or controls (81 ± 24 ml, p <0.001), but similar to 49 patients with HC and AF before CMR (152 ± 70 ml, p = 1.0). Absolute LAESV was significantly greater in 41 patients with HC and AF before CMR (95 ± 38 ml) than 377 HC patients without AF (59 ± 29 ml) or controls (37 ± 15 ml, p <0.001) but similar to 49 patients with HC and AF before CMR (105 ± 72 ml, p = 1.0). When indexed to BSA, LAEDV and LAESV subgroup relations did not differ from the absolute nonindexed values.
Univariate predictors of subsequent AF onset were determined to be LAEDV, LAESV, and percent LAEF (p <0.001 for each), age (p = 0.003), and LV systolic volume (0.046). Because percent LAEF is a linear combination of LAEDV and LAESV, we excluded LAESV from the multivariate model.
The model was limited to 377 patients with HC but without AF and those 41 patients with AF occurring after study entry ( Table 3 ). By stepwise selection, 3 variables associated with subsequent onset of AF were identified: LAEF (p = 0.002), LAEDV (p = 0.001), and age (p = 0.058). The c-statistic from the multivariate model was 0.814, and Hosmer-Lemeshow test indicated good fit (chi-square = 10.96, p = 0.20).
| Effect | Coefficient (SE) | Odds Ratio (95% CI) | p Value |
|---|---|---|---|
| Intercept | −2.728 (1.123) | — | 0.015 |
| Age ∗ | 0.021 (0.011) | 1.046 (0.998–1.098) | 0.058 |
| LAEF (%) | −0.051 (0.016) | 0.950 (0.920–0.981) | 0.002 |
| LAEDV † | 0.014 (0.004) | 1.062 (1.026–1.104) | 0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
