Results
Baseline CRP and Participant Characteristics
Among the 1,773 subjects who were enrolled in MESA baseline ancillary tagging study, 851 participants underwent repeated tagged MRI. Among them, 62 cases were excluded because of noise, misregistration of tag, different tagging sequence, and unreadable images by analyzing software. Four cases were lost due to no CRP data. Finally, a total of 785 cases had both CRP value and good quality mid-LV slice strain curve from both baseline and follow-up. The mean age was 64 ± 9 years, and 386 participants were women (49%). The median CRP concentration was 1.7 mg/L (interquartile range, 0.8–3.8 mg/L), and women had higher CRP than did men (2.42 mg/L vs1.25 mg/L, P < .001). The mean interval between the baseline and follow-up MRI examinations was 1,713 ± 95 days. During the follow-up period, 6 patients had myocardial infarction (time to interim myocardial infarction 1,390 ± 444 days) and 18 underwent coronary revascularization (time to revascularization 1,176 ± 506 days), with 3,690 person-years of follow-up. Table I shows the baseline characteristics and LV indices for each sex-specific CRP quartile. The higher CRP quartile subgroup was associated with higher body mass index, higher prevalence of hypertension, smokers, African Americans, Hispanics, and higher LV mass. Chinese Americans and statin users were associated with lower CRP quartiles.
Longitudinal Myocardial Functional Changes
Temporal change in circumferential myocardial shortening was used to index changes in myocardial function across time. Negative ΔEcc values reflect functional improvement, whereas positive ΔEcc indicates functional deterioration. The mean midventricular Ecc did not significantly change for the entire study group during the follow-up period (from −17.8% ± 2.6% to −17.9% ± 2.8%, P = .26), and average ΔEcc (ie, Ecc at follow-up − Ecc at baseline) was −0.1% ± 3.3%. Therefore, we considered ΔEcc >3.2% (Ecc deterioration >1 SD) as definite myocardial functional deterioration and ΔEcc <−3.4% as definite Ecc improvement. The definite deterioration group (n = 112) had higher baseline CRP. The prevalence of current smokers was significantly higher, and the prevalence of Chinese was significantly lower in the definite deterioration group (Table II). Higher body mass index was related to definite deterioration group in men (odds ratio [OR] 1.1, 1.03–1.18, P = .004) but not in women (1.04, 0.99–1.11, P = .13). Hypertension (OR 1.64, 1.09–2.48, P = .018) and angiotensin-converting enzyme inhibitor users (OR 2.02, 1.21–3.38, P = .008) were significantly related to definite Ecc improvement.
Predictors of Regional Myocardial Functional Deterioration
Baseline CRP was weakly but significantly correlated with change in myocardial shortening (ΔEcc, r = 0.11, P = .002), indicating that increased CRP is associated with myocardial functional reduction. In univariate analysis, higher CRP was significantly correlated with definite myocardial functional deterioration (OR 1.07 per 1-mg/L increase, 95% CI 1.03–1.11, P < .001). Because ΔEcc was significantly correlated with baseline Ecc (r = −0.59, P < .001), we reanalyzed the data after adjusting for baseline Ecc. Higher CRP, indexed LV mass, male gender, current smoking, and diastolic blood pressure were all significantly correlated with the myocardial functional deterioration after adjustment for baseline Ecc values both in categorical and continuous variable analyses (Table III). Participants with CRP >3 mg/L had a higher prevalence of Ecc deterioration compared with participants with CRP ≤3mg/L (20% vs 12%, P = .003).
C-reactive Protein as an Independent Predictor of Regional Myocardial Functional Deterioration
In linear regression analysis, higher CRP was related to functional deterioration independent of age, gender, and ethnicity (B = 0.08; ΔEcc change per 1-mg/L CRP change, 95% CI 0.04–0.13, P < .001, model 1). In model 2, which included systolic blood pressure, heart rate, diabetes, smoking status (never, former, current), body mass index, current medication use (antihypertensive therapy, aspirin, and statins), and glomerular filtration rate in addition to model 1 covariates, higher CRP was also related to Ecc deterioration (B = 0.10, 0.05–0.14, P < .001). In model 3, we included indices of subclinical cardiovascular disease (LV mass, presence of coronary calcium) as well as interim myocardial infarction and coronary revascularization in addition to model 2 covariates. Higher CRP remained related to Ecc deterioration (B = 0.09, 0.05–0.14, P < .001). Moreover, the relationship remained statistically significant after adjustment for baseline Ecc in model 4 (B = 0.06, 0.02–0.10, P = .002). Even after systolic blood pressure was replaced by diastolic blood pressure, relation between CRP and Ecc change remained significant. Finally, in a sensitivity analysis, after exclusion of participants with interim myocardial infarction or coronary revascularization, higher CRP was still significantly related to Ecc deterioration. Importantly also, in logistic regression analysis, higher CRP was significantly correlated with definite myocardial deterioration using the same models (Table IV).