Influence of Concentric Left Ventricular Remodeling on Early Mortality After Aortic Valve Replacement

Transcription

1 Influence of Concentric Left Ventricular Remodeling on Early Mortality After Aortic Valve Replacement Andra Ibrahim Duncan, MD, Boris S. Lowe, MD, Mario J. Garcia, MD, Meng Xu, MS, A. Marc Gillinov, MD, Tomislav Mihaljevic, MD, and Colleen G. Koch, MD, MS Departments of Cardiothoracic Anesthesia, Outcomes Research, Cardiovascular Medicine, Thoracic and Cardiovascular Surgery, and Quantitative Health Sciences, Cleveland Clinic Foundation, Cleveland, Ohio; and Division of Non-Invasive Cardiac Imaging, Mount Sinai Heart, Departments of Medicine and Radiology, Mount Sinai School of Medicine, New York, New York Background. Severe left ventricular (LV) hypertrophy compared to patients with nonconcentric geometries by increases risk for adverse outcome after aortic valve propensity and logistic regression modeling. Also, prognostic ability of RWT and LV mass was compared. replacement. Whether LV geometry influences mortality risk after aortic valve replacement is unclear. And, Results. Nine hundred sixty-four patients with concen- geometry were propensity-matched to 964 patients whether LV mass or relative wall thickness (RWT) bettertric predicts risk for adverse postoperative outcomes is unknown. The purpose of this investigation was to examine[3.9%] versus 18 [1.9%]; p 0.007), cardiac morbidity (33 with nonconcentric geometry. In-hospital mortality (38 the influence of LV geometry and LV hypertrophy on[3.4%] versus 17 [1.8%]; p 0.022), and prolonged intubation (85 [8.8%] versus 58 [6.0%]; p 0.019) were higher morbidity and in-hospital mortality after aortic valve replacement, and to determine whether LV mass or RWTin patients with concentric versus nonconcentric geometry. Increasing RWT, not LV mass, was associated with had better prognostic ability. Methods. Between January 1996 and June 2004, 5,083 adverse outcomes. patients underwent aortic valve replacement. Preoperative echocardiographic data was used to calculate LV increased risk for in-hospital mortality after aortic valve Conclusions. Concentric geometries are associated with mass and RWT. Left ventricular geometry was classified replacement. Increased RWT is associated with adverse into one of four categories on the basis of LV mass outcomes. Preoperative risk stratification should include indexed to body height and RWT: (1) concentric hyper-assessmenttrophy, (2) eccentric hypertrophy, (3) concentric remodeling, of LV hypertrophy and LV geometry. and (4) normal. Postoperative mortality and multisystem (Ann Thorac Surg 2008;85:2030 9) morbidities of patients with concentric geometries were 2008 by The Society of Thoracic Surgeons Increases in myocardial mass and alterations in leftmass in uncomplicated hypertensive patients is a reliable ventricular geometry (LV) help limit systolic wall predictor for long-term fatal cardiovascular events, inde- of arterial pressure, the presence of coronary stress and preserve ejection fraction in response to ven-pendentltricular pressure or volume overload. Although limiting artery disease, and other risk factors [3 5]. The geometry wall stress, LV hypertrophy (LVH) is a pathologic condi-otion associated with increasing myocardial connective LVH heralding poor outcome [1, 3]. Concentric LVH in the left ventricle further stratifies risk, with concentric tissue and fibrosis [1]. Left ventricular geometry may be hypertensive patients is associated with a twofold increase in cardiovascular deaths compared with patients classified into the following structurally determined categories based on LV mass and relative wall thicknesswith eccentric hypertrophy [3]. Furthermore, medically (RWT): (1) concentric hypertrophy, (2) eccentric hypertrophy, (3) concentric remodeling, and (4) normal LVH are at highest risk of cardiovascular morbidity and treated hypertensive patients with persistent concentric geometry [2]. mortality compared with patients with eccentric LVH, The development of LVH in hypertensive patients is associated with adverse long-term outcomes, including myocardial infarction, congestive heart failure, stroke, and death [3] and may be the most important prognostic indicator of long-term survival [1, 4]. Increased LV even with similar values of LV mass index [6]. Severe LVH is also a risk factor for adverse outcome in surgical patients after aortic valve replacement surgery (AVR), including higher in-hospital mortality Accepted for publication Feb 25, Address correspondence to Dr Duncan, Department of Cardiothoracic Anesthesia, Cleveland Clinic Foundation, 9500 Euclid Ave, G30, Cleveland, OH 44195; duncana@ccf.org. Dr Garcia discloses that he has a financial relationship with Phillips, Vital Images, and Pfizer; Dr Gillinov with Edwards Lifesciences, Medtronic, and St. Jude by The Society of Thoracic Surgeons /08/$34.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg DUNCAN ET AL 2008;85: CONCENTRIC REMODELING AND SURGICAL RISK 2031 [7 9] and more frequent postoperative complications, such as congestive heart failure, low-output syndrome, respiratory failure, renal insufficiency, and atrial and ventricular arrhythmias [8]. Although LV geometry influences outcomes in the nonoperative setting, the influence of type of LV geometry on outcomes in surgical patients remains unclear. Although some reports conclude that concentric LVH in patients with aortic stenosis increases risk for adverse outcome after AVR [7, 10], others reported no difference with wall thickness-to-radius ratio measurements on post-avr outcomes [11]. Furthermore, eccentric ventricular hypertrophy is thought to increase risk for adverse outcome because of its association with congestive heart failure, ventricular arrhythmias [12], and depressed LV functional response to stress [13]. Thus, the extent to which type of ventricular geometry influences outcomes after AVR remains unclear. Various measures of LVH severity, including increasing RWT [7, 10], which denotes LV concentric remodeling, and increased LV mass [8, 9], have been used to examine the influence of LVH on outcomes in patients after AVR. Increased RWT was associated with increased mortality in patients undergoing AVR [7, 10], whereas other reports found increased LV mass to predict adverse outcomes [8, 9]. Whether increased LV mass or RWT best identifies increased perioperative risk in patients having AVR has yet to be determined. Risk stratification of patients who require AVR may be improved by more complete preoperative assessment of LV geometry and identification of which LVH descriptor best predicts adverse outcomes. Furthermore, timing of surgical correction may be improved by attention to specific measures of LVH. We therefore tested the hypothesis that in-hospital mortality after AVR is greater in patients with concentric geometries than in those with nonconcentric geometries. In addition, we compared the ability of RWT versus LV mass to predict increased risk of mortality in patients after AVR. Patients and Methods Patient Population and Clinical Data Collection Patient data were obtained from the Cardiothoracic Anesthesia Patient Registry of the Department of Cardiothoracic Anesthesia and from the Cardiovascular Information Registry of the Department of Thoracic and Cardiovascular Surgery at the Cleveland Clinic using methods that have been reported previously [14]. Between January 1996 and June 2004, 5,083 patients underwent AVR with or without concomitant procedures. All clinical data were collected daily, concurrent with patient care, by experienced and specifically trained research personnel. Data that did not conform within a range of expected results was rejected and reevaluated. The use of the registries for research purposes was approved by the institutional review board. The institutional review board granted a waiver of the need for individual patient consent for this investigation. Outcome variables as described by Higgins and colleagues [14] included (1) mortality (all-cause in-hospital mortality); (2) cardiac morbidity defined as a combination of postoperative myocardial infarction or low cardiac output with a requirement for intraaortic balloon pump, ventricular assist device, or extracorporeal membrane oxygenation (Postoperative myocardial infarction was defined by specific electrocardiographic findings consistent with myocardial infarction [15] with a creatine phosphokinase myocardial band of at least 50 IU or aspartate aminotransferase level of at least 80 U/L. Low cardiac output was defined as a cardiac index less than 1.8 L min 1 m 2 despite adequate fluid replacement and high-dose inotropic agents for more than 4 hours.); (3) neurologic morbidity defined as new postoperative focal deficit (aphasia, decrease in limb function, or hemiparesis confirmed by clinical findings or computed tomographic scan) or global neurologic deficit (diffuse encephalopathy with greater than 24 hours of severely altered mental status, or failure to awaken postoperatively); (4) prolonged intubation defined as a requirement for mechanical ventilation or endotracheal intubation of greater than 72 hours; (5) renal morbidity defined as postoperative anuria or oliguria (urine output less than 400 ml/24 hours) or institution of renal dialysis or ultrafiltration; (6) infection morbidity, including culture-proven pneumonia, mediastinitis, wound infection, or septicemia with appropriate clinical findings; and (7) overall morbidity defined as the incidence of one or more of the above morbidities, including death, since early death precludes observation of morbidity. Echocardiographic Methods Echocardiographic indices were obtained from the Adult Echocardiography Laboratory Database at the Cleveland Clinic. Studies were performed using commercially available echocardiographic imaging systems equipped with 3.0- to 3.5-MHz transducers with M-mode, twodimensional, pulsed, continuous, and color-flow Doppler capabilities. Correct orientation of imaging planes and measurements of LV internal dimension and septal and posterior wall thicknesses were measured at enddiastole, according to the American Society of Echocardiography recommendations [16]. The parasternal long-axis and short-axis views were used to record twodimensional and M-mode recordings of the LV internal diameter and wall thickness at, or just below, the tips of the mitral leaflets. When optimal perpendicular orientation of the LV M-mode to the ultrasound beam could not be obtained, correctly oriented two-dimensional linear dimensions were made by the American Society of Echocardiography leading-edge convention. Left ventricular mass was calculated according to the methods described by Devereux and Reichek [17]: LV mass 1.04 [(LVID VST PWT) 3 (LVID) 3 ] 13.6 where LVID is left ventricular internal dimension, VST is left ventricular septal thickness, and PWT is left ventricular posterior wall thickness. Left ventricular mass indexed to height (g/m) was used to normalize LV mass measurements according to body size since adjustment by body surface

3 2032 DUNCAN ET AL Ann Thorac Surg CONCENTRIC REMODELING AND SURGICAL RISK 2008;85: area may not recognize obesity-related hypertrophy. Upper limits of normal for LV mass was defined as 143 g/m for men and 102 g/m for women, corresponding to two standard deviations above the mean values for LV mass in a healthy reference group [2, 18]. Relative wall thickness was calculated as RWT PWT/ (0.5) LVID, and RWT of 0.45 or less was considered normal. Left ventricular geometry was classified into one of the following four mutually exclusive groups on the basis of LV geometry: (1) concentric hypertrophy (increased LV mass and increased RWT); (2) eccentric hypertrophy (increased LV mass and normal RWT); (3) concentric remodeling (normal LV mass with increased RWT); and (4) normal geometry (normal LV mass and normal RWT). Standard methods were used to calculate LV ejection fraction [19]. Statistical Methods The SAS 8.2 software (SAS Institute Inc, Cary, NC) statistical software was used for statistical analysis. Results are expressed as mean standard deviation, median (25th, 75th percentiles), and frequencies as percentages. Differences between two groups were assessed by the Student s t test or Wilcoxon rank-sum test for continuous variables and 2 or Fisher s exact test for categorical data as appropriate; analysis of variance was performed for comparison among multiple groups. Transformation was made as needed to meet the linearity assumptions for logistic regression analysis. A probability value of less than 0.05 was used as the significant criterion for each comparison. Univariate and Propensity Matching Patients with concentric geometries (concentric hypertrophy and concentric remodeling) were compared with patients with nonconcentric geometries (eccentric hypertrophy or normal geometry). Before propensity matching a parsimonious explanatory model was developed whereby variables found to be significantly associated with concentric geometry were identified. Confounding variables were balanced in the two groups with propensity analysis [20]. A propensity score was calculated for each patient from a logistic model that included 48 variables listed in Tables 1 and 2. Patients with missing variables necessary for calculation of the propensity score were excluded from this analysis. No interaction terms were used. The C statistic for the propensity model was Patients with concentric geometries were compared with patients with normal geometry. Confounding variables were balanced in the two groups with propensity analysis. A propensity score was calculated for each patient from a logistic model that included 48 variables listed in Tables 1 and 2. The C statistic for the propensity model was For both of the above propensity analyses, patients were 1:1 matched on propensity scores with greedy matching techniques [22]. Morbidity and mortality outcomes were compared between matched pairs with 2, Fisher s exact, and Wilcoxon rank-sum tests. Logistic Regression Logistic regression models were used to examine the relationship between LV geometry and mortality. In addition, further logistic regression analyses evaluated the relationship between RWT and LV mass on postoperative morbidity and mortality. Stepwise-selection logistic regression models were built from bootstrapping the data 1,000 times using the same 48 potential confounding variables as in the propensity analysis, but excluding the variable of interest, which differed in each analysis (LV geometry, RWT, or LV mass to height ratio). A bagging algorithm was used to summarize the results. Final logistic regression models for outcomes were built by using confounding variables that appeared in 50% of all models from the bootstrap procedure, then adding the variable of interest (LV geometry, RWT, or LV mass). Results were analyzed with SAS 8.2 software (SAS Institute Inc). Results Risk Profiles Five thousand eighty-three patients were categorized into one of four groups (concentric geometry, n 3,046 [59.9%] concentric remodeling, n 398 [7.8%]; eccentric hypertrophy, n 1,441 [28.3%]; normal geometry, n 198 [3.9%]). Baseline and operative characteristics for patients categorized into concentric versus nonconcentric geometry (concentric, n 3,444; nonconcentric, n 1,639) undergoing AVR are found in Tables 3 and 4. Mean RWT for patients with concentric hypertrophy, concentric remodeling, eccentric hypertrophy, and normal geometry was (mean standard deviation) , , , and , respectively. Left ventricular mass for patients with concentric hypertrophy, concentric remodeling, eccentric hypertrophy, and normal geometry was (mean standard deviation) , , , and g/m, respectively. Patients with concentric geometry were more likely to be female and older, and to have a history of hypertension, diabetes mellitus, carotid artery disease, or stroke, and have a better preserved left ventricular function and a larger aortic transvalvular gradient. Severe aortic insufficiency, mitral insufficiency, congestive heart failure, myocardial infarction, cardiogenic shock, and previous cardiac surgery were more prevalent among patients with nonconcentric LV geometry. Additional perioperative variables, baseline laboratory values, and surgical procedures were significantly different between groups as described in Tables 3 and 4. Outcomes: All Patients Univariate analysis for all patients demonstrated higher mortality in patients with concentric geometry compared with those with nonconcentric geometry (110 of 3,444 [3.2%] versus 30 of 1,639 [1.8%], respectively; p 0.006). Univariate analysis demonstrated that unmatched patients with concentric geometry had similar cardiac morbidity (69 [2.0%] versus 32 [2.0%]; p 0.90), prolonged

4 Ann Thorac Surg DUNCAN ET AL 2008;85: CONCENTRIC REMODELING AND SURGICAL RISK 2033 Table 1. Categorical Baseline and Operative Characteristics for the Propensity-Matched Patient Population a Categorical Variable (n 964) (n 964) p Value Male sex 632 (65.6%) 649 (67.3%) 0.41 Clinical history Hypertension 561 (58.2%) 564 (58.5%) 0.89 Congestive heart failure 437 (45.3%) 437 (45.3%) 0.99 Myocardial infarction 247 (25.6%) 248 (25.7%) 0.96 Cardiogenic shock 1 (0.1%) 2 (0.2%) 0.99 Left main disease 26 (2.7%) 24 (2.5%) 0.77 Left ventricular function 0.09 Normal (EF 0.60) 370 (38.4%) 377 (39.1%) Mild dysfunction (EF ) 205 (21.3%) 219 (22.7%) Mild moderate dysfunction (EF ) 9 (0.9%) 21 (2.2%) Moderate dysfunction (EF ) 87 (9.0%) 72 (7.5%) Moderate severe dysfunction (EF ) 73 (7.6%) 54 (5.6%) Severe dysfunction (EF 0.35) 220 (22.8%) 221 (22.9%) Pulmonary hypertension 154 (16.0%) 150 (15.6%) 0.80 Chronic obstructive pulmonary disease 107 (11.1%) 106 (11.0%) 0.94 Smoking 517 (53.6%) 528 (54.8%) 0.62 Diabetes mellitus 0.92 Insulin-dependent 68 (7.1%) 69 (7.2%) Non insulin-dependent 106 (11.0%) 108 (11.2%) Diet-controlled 26 (2.7%) 31 (3.2%) Carotid artery disease 174 (18.1%) 176 (18.3%) 0.91 Stroke 64 (6.6%) 68 (7.1%) 0.72 Peripheral vascular disease 106 (11.0%) 126 (13.1%) 0.16 Renal failure requiring dialysis 17 (1.8%) 19 (2.0%) 0.74 New York Heart Association class 0.72 I 111 (11.5%) 125 (13.0%) II 521 (54.1%) 511 (53.0%) III 250 (25.9) 240 (24.9%) IV 82 (8.5%) 88 (9.1%) Previous cardiovascular surgery history Previous cardiac surgery 313 (32.5%) 335 (34.8%) 0.29 Carotid surgery 41 (4.3%) 52 (5.4%) 0.24 Major vascular (other than carotid) procedure 33 (3.4%) 44 (4.6%) 0.20 Severity of aortic valve insufficiency 0.62 None 312 (32.4%) 338 (35.1%) (21.7%) 191 (19.8%) (19.6%) 183 (19.0%) (18.2%) 165 (17.1%) 4 79 (8.2%) 87 (9.0%) Mitral insufficiency ( 3 ) 169 (17.5%) 163 (16.9%) 0.72 Perioperative factors Emergency procedure 6 (0.6%) 5 (0.5%) 0.76 Preoperative use of intraaortic balloon pump 4 (0.4%) 3 (0.3%) 0.99 Inotropic drug requirement before cardiopulmonary bypass 4 (0.4%) 3 (0.3%) 0.99 Packed red blood cell transfusion 537 (55.7%) 543 (56.3%) 0.78 Blood component therapy b 183 (19.0%) 185 (19.2%) 0.91 Aortic valve replacement procedure Valve type 0.97 Bioprosthesis 713 (74.0%) 717 (74.4%) Mechanical 143 (14.8%) 142 (14.7%) Allograft 108 (11.2%) 105 (10.9%) Aortic annulus enlargement 15 (1.6%) 18 (1.9%) 0.60

5 2034 DUNCAN ET AL Ann Thorac Surg CONCENTRIC REMODELING AND SURGICAL RISK 2008;85: Table 1. Continued Categorical Variable (n 964) (n 964) p Value Concomitant cardiac surgical procedures Coronary artery bypass grafting 470 (48.8%) 466 (48.3%) 0.86 Left internal mammary artery graft 190 (19.7%) 197 (20.4%) 0.69 Right internal mammary artery graft 9 (0.93%) 12 (1.2%) 0.51 Saphenous vein graft 416 (43.2%) 404 (41.9%) 0.58 Mitral valve replacement 125 (13.0%) 122 (12.7%) 0.84 Mitral valve repair 118 (12.2%) 114 (11.8%) 0.78 Pulmonic valve repair or replacement 0 (0%) 0 (0%)... Tricuspid valve repair or replacement 87 (9.0%) 89 (9.2%) 0.87 Maze procedure 24 (2.5%) 26 (2.7%) 0.77 Myomectomy 7 (0.7%) 6 (0.6%) 0.78 Aortic surgery 80 (8.3%) 85 (8.8%) 0.68 a Results are shown as number (%). b Blood component therapy refers to the administration of fresh-frozen plasma, platelets, or cryoprecipitate. EF ejection fraction. intubation (266 [7.7%] versus 111 [6.8%]; p 0.23), renal morbidity (84 [2.4%] versus 33 [2.0%]; p 0.34), neurologic morbidity (119 [3.5%] versus 41 [2.5%]; p 0.069), serious infection (151 [4.4%] versus 61 [3.7%]; p 0.27), and overall morbidity (361 [10.5%] versus 158 [9.6%]; p 0.35) compared with patients with nonconcentric geometry, respectively. Outcomes: Propensity-Matched Patients Using the variables listed in Tables 1 and 2, 964 (28.0%) patients with concentric geometry (concentric LVH, n 869; concentric remodeling, n 95) were propensity matched to 964 (58.8%) nonconcentric geometry (eccentric LVH, n 824; normal, n 140) patients. Propensity matching resulted in a similar distribution of baseline and operative variables (Tables 1, 2). Mortality was significantly higher in patients with concentric geometry compared with those with nonconcentric geometry (38 [3.9%] versus 18 [1.9%]; p 0.007). Cardiac morbidity (33 [3.4%] versus 17 [1.8%]; p 0.022) and prolonged intubation (85 [8.8%] versus 58 [6.0%]; p 0.019) were also higher in patients with concentric geometry. Renal, neurologic, infection, and overall morbidity did not differ significantly in patients with concentric and nonconcentric geometries (Table 5). Patients with concentric geometries were compared with patients with normal geometries. One hundred forty-nine patients with normal geometry were propensity matched to 149 patients with concentric geometries. Neurologic morbidity was higher in patients with concentric geometry (9 [6.0%] versus 2 [1.3%]; p 0.032]. Mortality (8 [5.4%] versus 4 [2.7%]; p 0.24), cardiac morbidity (6 [4.0%] versus 2 [1.3%]; p 0.28), and prolonged intubation (18 [12.1%] versus 9 [6.0%]; p 0.069) were higher in patients with concentric geometry, but did not reach statistical significance. Renal, infection, and overall morbidity were not different between two groups. Table 2. Continuous Baseline and Operative Characteristics for the Propensity-Matched Patient Population a Continuous Variables (n 964) (n 964) p Value Demographics Age (y) 70 (58, 77) 70 (59, 77) 0.85 Body surface area (m 2 ) 1.9 (1.7, 2.1) 1.9 (1.7, 2.1) 0.87 Preoperative laboratory values Hematocrit (%) 40.2 (36.8, 42.9) 40.0 (36.7, 42.7) 0.59 Serum creatinine (mg/dl) 1.0 (0.9, 1.3) 1.0 (0.9, 1.3) 0.60 Serum albumin (g/dl) 4.2 (3.8, 4.4) 4.1 (3.8, 4.4) 0.90 Serum bilirubin (mg/dl) 0.6 (0.5, 0.9) 0.6 (0.5, 0.9) 0.40 Perioperative variables Preoperative mean gradient across aortic valve 35 (24, 47) 34 (23, 46) 0.16 Aortic cross-clamp time (min) 84 (62, 109) 84 (61, 108) 0.74 Indexed orifice area (cm 2 /m 2 ) b 1.8 (1.6, 2.1) 1.8 (1.6, 2.1) 0.53 a Results are shown as median (25th, 75th percentiles). b Indexed orifice area (cm 2 /m 2 ) of the implanted aortic valve was calculated from prosthesis internal orifice diameter assuming a circular orifice shape divided by patient s body surface area [21].

6 Ann Thorac Surg DUNCAN ET AL 2008;85: CONCENTRIC REMODELING AND SURGICAL RISK 2035 Table 3. A Comparison of Baseline and Operative Characteristics for the Unmatched Patient Population Undergoing Aortic Valve Replacement With or Without Concomitant Procedures a Categorical Variable (n 3,444) (n 1,639) p Value Male sex 2,071 (60.1%) 1,147 (70.0%) Clinical history Hypertension 2,188 (63.5%) 923 (56.3%) Congestive heart failure 1,216 (35.3%) 721 (44.0%) Myocardial infarction 602 (17.5%) 388 (23.7%) Cardiogenic shock 2 (0.1%) 5 (0.3%) Left main disease 82 (2.4%) 33 (2.0%) 0.41 Left ventricular function Normal (EF 0.60) 1,965 (57.1%) 572 (34.9%) Mild dysfunction (EF ) 853 (24.8%) 392 (23.9%) Mild moderate dysfunction (EF ) 29 (0.8%) 30 (1.8%) Moderate dysfunction (EF ) 163 (4.7%) 137 (8.4%) Moderate severe dysfunction (EF ) 124 (3.6%) 102 (6.2%) Severe dysfunction (EF 0.35) 310 (9.0%) 406 (24.8%) Pulmonary hypertension 493 (14.3%) 242 (14.8%) 0.67 Chronic obstructive pulmonary disease 361 (10.5%) 161 (9.8%) 0.47 Smoking 1,864 (54.1%) 855 (52.2%) 0.19 Diabetes mellitus Insulin-dependent 232 (6.7%) 89 (5.4%) Non insulin-dependent 390 (11.3%) 148 (9.0%) Diet-controlled 103 (3.0%) 35 (2.1%) Carotid artery disease 661 (19.2%) 221 (13.5%) Stroke 289 (8.4%) 109 (6.7%) Peripheral vascular disease 442 (12.8%) 184 (11.2%) 0.10 Renal failure requiring dialysis 51 (1.5%) 29 (1.8%) 0.44 New York Heart Association class I 509 (14.9%) 262 (16.1%) II 1847 (54.0%) 849 (52.2%) III 838 (24.5%) 375 (23.0%) IV 227 (6.6%) 142 (8.7%) Previous cardiovascular surgery history Previous cardiac surgery 879 (25.5%) 518 (31.6%) Carotid surgery 162 (4.7%) 68 (4.2%) 0.37 Major vascular (other than carotid) procedure 150 (4.4%) 63 (3.8%) 0.39 Severity of aortic valve insufficiency None 1336 (38.8%) 413 (25.2%) (22.7%) 243 (14.8%) (19.3%) 284 (17.3%) (13.4%) 384 (23.4%) (5.7%) 315 (19.2%) Mitral insufficiency ( 3 ) 376 (10.9%) 346 (21.1%) Perioperative factors Emergency procedure 13 (0.4%) 19 (1.2%) Preoperative use of intraaortic balloon pump 11 (0.3%) 8 (0.5%) 0.36 Inotropic drug requirement before cardiopulmonary bypass 13 (0.4%) 16 (1.0%) Packed red blood cell transfusion 1980 (57.5%) 844 (51.5%) Blood component therapy b 625 (18.2%) 330 (20.1%) Aortic valve replacement procedure Valve type Bioprosthesis 2782 (80.8%) 1086 (66.3%) Mechanical 344 (10.0%) 257 (15.7%) Allograft 318 (9.2%) 296 (18.1%) Aortic annulus enlargement 79 (2.3%) 24 (1.5%) 0.050

7 2036 DUNCAN ET AL Ann Thorac Surg CONCENTRIC REMODELING AND SURGICAL RISK 2008;85: Table 3. Continued Categorical Variable (n 3,444) (n 1,639) p Value Concomitant cardiac surgical procedures Coronary artery bypass grafting 1649 (47.9%) 678 (41.4%) Left internal mammary artery graft 798 (23.2%) 286 (17.5%) Right internal mammary artery graft 41 (1.2%) 22 (1.3%) 0.65 Saphenous vein graft 1390 (40.4%) 580 (35.4%) Mitral valve replacement 339 (9.8%) 210 (12.8%) Mitral valve repair 232 (6.7%) 267 (16.3%) Pulmonic valve repair or replacement 0 (0%) 3 (0.2%) Tricuspid valve repair or replacement 240 (7.0%) 149 (9.1%) Maze procedure 106 (3.1%) 54 (3.3%) 0.68 Myomectomy 89 (2.6%) 9 (0.6%) Aortic surgery 351 (10.2%) 193 (11.8%) a Results are shown as number (%). b Blood component therapy refers to the administration of fresh-frozen plasma, platelets, or cryoprecipitate. EF ejection fraction. Outcomes: Logistic Regression Because the propensity analysis matched only 38% of the patient population (964 patients in each group of the total 5,083 patients), logistic regression modeling was performed to adjust for confounding variables (listed in Tables 1 and 2) to compare outcomes in all patients. In addition, logistic regression techniques were able to compare each type of geometry individually. After adjustment for confounding variables, LV wall geometry was associated with an increased mortality risk (p 0.009). Concentric hypertrophy increased risk of mortality compared with eccentric hypertrophy (odds ratio [OR], 1.98; 95% confidence interval [CI], 1.21 to 3.20) and concentric remodeling increased risk of mortality compared with eccentric hypertrophy (OR, 3.23; 95% CI, 1.56 to 6.67). Risk of mortality was similar between patients with concentric hypertrophy and patients with concentric remodeling (OR, 0.62; 95% CI, 0.34 to 1.13). Eccentric hypertrophy versus normal (OR, 0.69; 95% CI, 0.24 to 1.99), concentric hypertrophy versus normal (OR, 1.36; 95% CI, 0.50 to 3.68), and concentric remodeling versus normal (OR, 2.21; 95% CI, 0.72 to 6.77) were not significantly different. Relative wall thickness, but not LV mass, was associated with increased mortality, cardiac, renal, neurologic, and overall morbidity (Table 6). Comment This investigation finds evidence to suggest that preoperative LV geometry influences outcome after AVR surgery. Patients with concentric hypertrophy and concentric remodeling had worse outcomes than patients with nonconcentric LV geometry. Further, increased RWT, not Table 4. A Comparison of Baseline and Operative Characteristics for Unmatched Patients a Continuous Variables N Median (25th, 75th percentiles) N Median (25th, 75th percentiles) p Value Demographics Age (y) 3, (63, 78) 1, (53, 75) Body surface area (m 2 ) 3, (1.7, 2.1) 1, (1.7, 2.1) Preoperative laboratory values Hematocrit (%) 3, (36.6, 42.8) 1, (36.8, 42.9) 0.14 Serum creatinine (mg/dl) 3, (0.8, 1.2) 1, (0.9, 1.2) Serum albumin (g/dl) 3, (3.8, 4.4) 1, (3.9, 4.4) 0.30 Serum bilirubin (mg/dl) 3, (0.4, 0.8) 1, (0.5, 0.9) Perioperative variables Preoperative mean gradient across aortic valve 3, (31, 55) 1, (21, 44) Aortic cross-clamp time (min) 3, (59, 106) 1, (64, 111) Indexed orifice area (cm 2 /m 2 ) b 3, (1.6, 2.1) 1, (1.6, 2.2) a Results are shown as median (25th, 75th percentiles). b Indexed orifice area (cm 2 /m 2 ) of the implanted aortic valve was calculated from prosthesis internal orifice diameter assuming a circular orifice shape divided by patient s body surface area [21].

8 Ann Thorac Surg DUNCAN ET AL 2008;85: CONCENTRIC REMODELING AND SURGICAL RISK 2037 Table 5. Postoperative Morbidity and Mortality Outcomes for the Propensity-Matched Pairs a Variables (n 964) (n 964) OR (95% CI) p Value Mortality 38 (3.9%) 18 (1.9%) 2.2 (1.2, 3.8) Morbidity Cardiac 33 (3.4%) 17 (1.8%) 2.0 (1.1, 3.6) Prolonged intubation 85 (8.8%) 58 (6.0%) 1.5 (1.1, 2.1) Renal failure 33 (3.4%) 21 (2.2%) 1.6 (0.9, 2.8) Neurologic 31 (3.2%) 28 (2.9%) 1.1 (0.7, 1.9) 0.69 Infection 49 (5.1%) 34 (3.5%) 1.5 (0.9, 2.3) Overall 116 (12.0%) 93 (9.7%) 1.3 (1.0, 1.7) a Results are presented as number (%) and (95% confidence intervals). CI confidence interval; OR odds ratio. increased LV mass, was associated with an increased risk of early death after AVR surgery. Propensity modeling was used in our investigation to compare postoperative outcomes in patients by type of LV geometry. For the purpose of the propensity analysis, a dichotomous variable describing LV geometry was created such that patients were labeled as either concentric geometry or nonconcentric geometry. To form a dichotomous variable, patients with eccentric hypertrophy and normal geometry were combined into a single group. This technique allowed us to compare concentric geometry with all other geometries, and we found that patients with concentric geometries had an increased risk for early mortality after AVR. Further, logistic regression techniques were performed, which compared individual geometries and their association with adverse outcomes. The results from both analyses were consistent in that patients with concentric LVH and concentric remodeling were at increased risk for mortality compared with eccentric hypertrophy. This analysis compared outcomes related to ventricular geometry, rather than type of aortic valve disease. The reasons for this comparison include the fact that other conditions (eg, other valvular disease, previous myocardial infarction, hypertension, and neurohormonal factors) in addition to type of aortic valvular disease contribute to ventricular remodeling and the development of concentric versus eccentric geometries. Certainly, more than 30% of patients with aortic stenosis have been found to have eccentric (rather than concentric) hypertrophy [23]. Thus, our objective was to evaluate the relationship between concentric LV geometry and outcomes after cardiac surgery rather than a comparison of aortic stenosis versus aortic insufficiency. Outcomes were also compared between patients with concentric geometry and those with normal geometry. However, the number of patients with normal geometry was small only 3.9% of patients with aortic valve disease have normal geometries. Our findings were consistent with our initial analysis: mortality, cardiac morbidity, and prolonged intubation in patients with concentric geometry were more than twice that of patients with normal geometry. However, because of the small sample size with normal geometry, this analysis lacked power to find a difference, and statistical significance was not achieved. It is also possible that some patients with normal geometry may have experienced acute aortic valvular dysfunction related to endocarditis or aortic dissection, before the development of chronic LV remodeling. These events may have increased risk for postop- Table 6. In-Hospital Postoperative Mortality and Morbidities Related to Increasing Relative Wall Thickness or Left Ventricular Mass a Increasing Relative Wall Thickness b Increasing LV Mass c Outcome OR (95% CI) p Value OR (95% CI) p Value Mortality 1.3 (1.1, 1.6) (0.8, 1.3) 0.99 Morbidity Cardiac 1.4 (1.1, 1.6) (0.7, 1.4) 0.96 Prolonged intubation 1.1 (1.0, 1.3) (0.9, 1.4) 0.18 Renal failure 1.3 (1.0, 1.5) (0.8, 1.4) 0.63 Neurologic 1.2 (1.0, 1.4) (0.7, 1.2) 0.66 Infection 1.1 (1.0, 1.3) (0.7, 1.1) 0.21 Overall 1.1 (1.0, 1.3) (0.8, 1.1) 0.59 a Results are presented as odds ratio (95% confidence intervals). b Per 0.2 unit increase. c Per 100 g/m increase. CI confidence interval; OR odds ratio.

9 2038 DUNCAN ET AL Ann Thorac Surg CONCENTRIC REMODELING AND SURGICAL RISK 2008;85: erative complications in patients with normal geometries and resulted in a smaller difference in outcomes. Because our investigation indexed LV mass to height, rather than to body surface area, obesity was less likely to underrecognize increased LV mass and falsely classify patients with LVH as normal LV mass [2, 18]. Our findings are consistent with other studies showing the association of LVH with adverse outcome after AVR surgery, including higher in-hospital mortality and morbidity [7 9]. Various descriptors of LVH severity have been used in other reports. As in our study, Orsinelli and associates [7] and Bech-Hanssen and coworkers [10] reported that patients with increased RWT and concentric hypertrophy were at increased risk of early postoperative mortality after AVR, although their investigations included only patients with aortic stenosis. In addition, Aurigemma and colleagues [24] reported that specific echocardiographic findings in patients, including morepronounced hypertrophy, smaller cavities, and higher ejection fractions, were associated with increased postoperative mortality. These reports are in agreement with our findings that RWT predicts adverse outcome after AVR. In contrast, other investigations used LV mass index, rather than RWT, to assess severity of LVH and evaluate its influence on outcomes after AVR [8, 9]. In contrast to our findings, Mehta and associates [8] reported that increased LV mass index was associated with adverse outcome after AVR with low cardiac output syndrome being the most common mode of death. Although Fuster and coworkers [9] reported that LV mass index was a predictor of early mortality and morbidity after AVR, considerably higher cutoff points were used to define increased LV mass (277 g/m 2 in males, 251 g/m 2 in females). Although cumulative mortality was increasingly higher for LV mass index values around 180 g/m 2, patients with mild to moderate increase (cutoff points of 134 g/m 2 in males, 110 g/m 2 in females) in LV mass did not have an increase in mortality [9]. The cutoff points of increased LV mass in our investigation were at conventional levels [2, 18]. Thus, many of the patients that were classified as increased LV mass in our investigation would not have fit criteria for increased LV mass according to Fuster and colleagues [9]. Therefore, our results, which find that increased LV mass, conventionally defined, does not predict mortality after AVR do not contradict Fuster and coworkers [9]. Further study by Fuster and associates [25] found that severely increased LV mass index was associated with mortality only in patients with low LV ejection fraction. Consistent with this theory, Milavetz and colleagues [11] did not find that the degree of LVH, conventionally defined, influenced postoperative outcome after AVR. An increased risk for early mortality in patients with concentric LV geometries undergoing AVR may be related to postoperative hemodynamic abnormalities, myocardial ischemia, and diastolic dysfunction. Patients with concentric LVH may be particularly susceptible to hemodynamic abnormalities in the immediate postoperative period as a result of the acute afterload reduction related to the removal of a stenotic aortic valve in addition to a stiff, noncompliant ventricle with impaired diastolic filling. Furthermore, an underfilled hypertrophied LV could be susceptible to dynamic LV outflow tract obstruction during systole [24]. Patients with concentric LVH are also at increased risk for preoperative myocardial ischemia as a consequence of structural alterations of the intramyocardial coronary vasculature, increased coronary resistance [26], coronary flow abnormalities [27], insufficient growth of coronary arteries relative to the degree of hypertrophied myocardial mass, and impaired subendocardial blood flow owing to increases in LV end-diastolic pressure. Further, myocardial oxygen demand is increased because of increased myocardial mass and higher wall tension, although myocardial efficiency is decreased [28] as increased interstitial fibrosis in the hypertrophied ventricle increases oxygen consumption, without contributing to LV minute work. Intraoperatively, patients with severe LVH are also at increased risk for suboptimal myocardial protection related to cardioplegic techniques [29], which may result in increased risk for development of myocardial dysfunction or arrhythmias. Diastolic dysfunction associated with concentric LVH [30] may also increase risk for adverse postoperative outcome. The presence of LVH amplifies the diastolic dysfunction that occurs with ischemia and reperfusion [31], which may further compromise cardiac output. These many factors may explain the increase in risk for adverse outcome in patients with concentric geometries after AVR. Study Limitations Because of its retrospective nature, the clinical data on the patients may be incomplete, and some determinants of outcome may not have been captured. Clinical care was not standardized; thus, selection bias and effects of unmeasured confounding variables cannot be excluded. Additionally, results from a single academic institution may not apply elsewhere because patient characteristics and care protocols will differ at least to some extent. However, a large, consecutive series from a single institution involving patients undergoing the same operation also reduces variability, thus making it easy to identify outcome patterns. Conclusions This investigation suggests that patients with concentric hypertrophy and concentric remodeling have worse outcomes than those with nonconcentric geometries. Increasing RWT, but not increasing LV mass, is associated with increased risk of early postoperative death. Further studies are required to address whether postoperative outcomes could be improved by AVR before a significant increase in RWT occurs. References 1. Kahan T. The importance of left ventricular hypertrophy in human hypertension. J Hypertens 1998;16(Suppl 7):S23 9.

10 Ann Thorac Surg DUNCAN ET AL 2008;85: CONCENTRIC REMODELING AND SURGICAL RISK Krumholz HM, Larson M, Levy D. Prognosis of left ventricular geometric patterns in the Framingham Heart Study. J Am Coll Cardiol 1995;25: Koren MJ, Devereux RB, Casale PN, et al. Relation of left ventricular mass and geometry to morbidity and mortality in uncomplicated essential hypertension. Ann Intern Med 1991;114: Levy D, Garrison RJ, Savage DD, et al. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med 1990; 322: Devereux RB. Left ventricular geometry, pathophysiology and prognosis. J Am Coll Cardiol 1995;25: Muiesan ML, Salvetti M, Monteduro C, et al. Left ventricular concentric geometry during treatment adversely affects cardiovascular prognosis in hypertensive patients. Hypertension 2004;43: Orsinelli DA, Aurigemma GP, Battista S, et al. Left ventricular hypertrophy and mortality after aortic valve replacement for aortic stenosis. J Am Coll Cardiol 1993;22: Mehta RH, Bruckman D, Das S, et al. Implications of increased left ventricular mass index on in-hospital outcomes in patients undergoing aortic valve surgery. J Thorac Cardiovasc Surg 2001;122: Fuster RG, Argudo JA, Albarova OG, et al. Left ventricular mass index in aortic valve surgery: a new index for early valve replacement? Eur J Cardiothorac Surg 2003;23: Bech-Hanssen O, Wallentin I, Houltz E, et al. Gender differences in patients with severe aortic stenosis: impact on preoperative left ventricular geometry and function, as well as early postoperative morbidity and mortality. Eur J Cardiothorac Surg 1999;15: Milavetz DL, Hayes SN, Weston SA, et al. Sex differences in left ventricular geometry in aortic stenosis. Chest 2000;117: Levy D, Anderson KM, Plehn J, et al. Echocardiographically determined left ventricular structural and functional correlates of complex or frequent ventricular arrhythmias on one-hour ambulatory electrocardiographic monitoring. Am J Cardiol 1987;59: Blake J, Devereux RB, Borer JS, et al. Relation of obesity, high sodium intake, and eccentric left ventricular hypertrophy to left ventricular exercise dysfunction in essential hypertension. Am J Med 1990;88: Higgins TL, Estafanous FG, Loop FD, et al. Stratification of morbidity and mortality outcome by preoperative risk factors in coronary artery bypass patients: a clinical severity score. JAMA 1992;267: Surawicz B, Knilans TK. Myocardial infarction and electrocardiographic patterns simulating myocardial infarction. In: Surawicz B, Knilans TK, eds. Chou s electrocardiography in clinical practice: adult and pediatric, 5th ed. Philadelphia: WB Saunders, 2001; Schiller NB, Shah PM, Crawford M, et al. Recommendations for quantitation of the left ventricle by two-dimensional echocardiography. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms. J Am Soc Echocardiogr 1989;2: Devereux RB, Reichek N. Echocardiographic determination of left ventricular mass in man: anatomic validation of the method. Circulation 1977;55: Levy D, Savage DD, Garrison RJ, et al. Echocardiographic criteria for left ventricular hypertrophy: the Framingham Heart Study. Am J Cardiol 1987;59: Gudmundsson P, Rydberg E, Winter R, Willenheimer R. Visually estimated left ventricular ejection fraction by echocardiography is closely correlated with formal quantitative methods. Int J Cardiol 2005;101: Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70: Blackstone EH, Cosgrove DM, Jamieson WR, et al. Prosthesis size and long-term survival after aortic valve replacement. J Thorac Cardiovasc Surg 2003;126: Parsons LS. Reducing bias in a propensity score matchedpair sample using greedy matching techniques. Proceedings of the Twenty-Sixth Annual SAS Users Group International Conference. Cary, NC: SAS Institute Inc; Available at Accessed April 30, Antonini-Canterin F, Huang G, Cervesato E, et al. Symptomatic aortic stenosis: does systemic hypertension play an additional role? Hypertension 2003;41: Aurigemma G, Battista S, Orsinelli D, et al. Abnormal left ventricular intracavitary flow acceleration in patients undergoing aortic valve replacement for aortic stenosis: a marker for high postoperative morbidity and mortality. Circulation 1992;86: Fuster RG, Argudo JAM, Albarova OG, et al. Left ventricular mass index as a prognostic factor in patients with severe aortic stenosis and ventricular dysfunction. Interact Cardiovasc Thorac Surg 2005;4: Muiesan ML, Rizzoni D, Salvetti M, et al. Structural changes in small resistance arteries and left ventricular geometry in patients with primary and secondary hypertension. J Hypertens 2002;20: Polese A, De Cesare N, Montorsi P, et al. Upward shift of the lower range of coronary flow autoregulation in hypertensive patients with hypertrophy of the left ventricle. Circulation 1991;83: Laine H, Katoh C, Luotolahti M, et al. Myocardial oxygen consumption is unchanged but efficiency is reduced in patients with essential hypertension and left ventricular hypertrophy. Circulation 1999;100: Ascione R, Caputo M, Gomes WJ, et al. Myocardial injury in hypertrophic hearts of patients undergoing aortic valve surgery using cold or warm blood cardioplegia. Eur J Cardiothorac Surg 2002;21: de Simone G, Kitzman DW, Chinali M, et al. Left ventricular concentric geometry is associated with impaired relaxation in hypertension: the HyperGEN study. Eur Heart J 2005;26: Mochizuki T, Eberli FR, Ngoy S, et al. Effects of brief repetitive ischemia on contractility, relaxation, and coronary flow: exaggerated postischemic diastolic dysfunction in pressure-overload hypertrophy. Circ Res 1993;73:550 8.