Carbohydrate antigen-125 (CA-125) has recently been reported to correlate with the severity of systolic heart failure (HF). However, the association between this marker and HF with preserved ejection fraction (HFpEF) remains elusive. We studied 158 consecutive women with preserved ejection fraction, who were categorized into 3 groups: those with HF (HFpEF group, n = 35), those with ≥1 cardiovascular risk (at-risk group, n = 78), and 45 normal controls (n = 45). All subjects underwent echocardiography with serum N-terminal pro-brain type natriuretic peptide (NT-ProBNP), CA-125 level, and other tumor markers obtained. HFpEF group showed significantly greater baseline levels of CA-125 and NT-ProBNP than both normal and at-risk groups (p <0.05). In addition, the serum CA-125 level correlated with the maximum left atrial volume (r = 0.24, p = 0.002). During a mean follow-up of 828.1 days (interquartile range 38 to 1,504.5), those with CA-125 levels >17.29 U/ml had a greatest incidence of HF hospitalization (hazard ratio 6.2, p <0.01) and remained an independent prognosticator in the multivariate Cox models. CA-125 superimposed on NT-ProBNP successfully expanded the receiver operating characteristic curve further in predicting HF hospitalization (area under curve 0.72 to 0.82, c-statistic 0.0049). In conclusion, serum CA-125 might serve as a novel biomarker for HFpEF and predicting HF hospitalization in women.
Heart failure (HF) with preserved ejection fraction (HFpEF), defined by the clinical presentation of HF without an obvious decline in ejection fraction, has been shown to cause morbidity and mortality comparable to HF with a reduced ejection fraction, leading to substantial socioeconomic burden. Few reliable and simple tests have yet been proved to be useful in screening such disease or in risk prediction among high-risk patients. Brain-type natriuretic peptide and N-terminal Pro-BNP (NT-ProBNP) are the most widely adopted biomarkers for the detection of HF, either HFpEF or HF with a reduced ejection fraction. However, some limitations exist in the clinical interpretation of such serum markers, including a nonspecific increase in patients with renal failure or in the elderly and female population. Therefore, exploring other novel biomarkers could be of great clinical value in understanding the fundamental pathophysiologic mechanisms and might further aid in HFpEF diagnosis or screening. Recently, increasing evidence has demonstrated that carbohydrate antigen (CA)-125 levels reflect the severity of systolic HF in terms of functional class and could be a good predictor of HF rehospitalization. However, its association with HFpEF and its clinical predictive value in this population remain unclear. We thus sought to investigate this issue in the present study.
Methods
A derivation cohort of 169 consecutive women, aged 20 to 75 years, was initially enrolled in the present study from cardiovascular outpatient clinics from March to June 2006. They were categorized into 3 groups: (1) HFpEF group, subjects with the symptoms and signs of HF or HF history according to Framingham’s criteria; (2) at-risk group, subjects who did not have any symptoms or signs of HF but had ≥1 cardiovascular risk factor, such as diabetes, hypertension, coronary artery disease, previous stroke, or body mass index >25 kg/m 2 ; (3) normal group, subjects who were healthy without known cardiovascular risk factors or systemic disease. Another out-of-sample validation study was performed of 224 subjects obtained from those who underwent an annual health survey or were referred by physicians from local outpatient clinics. The baseline characteristics and anthropometric data (i.e., age, height, weight, and body mass index). The data from plain chest radiography, body surface 12-lead electrocardiography, echocardiography were recorded. Also, blood tests for biochemical data (e.g., hepatic, renal, and lipid profiles, uric acid, fasting blood glucose, insulin), complete blood cell counts, creatine phosphokinase, high-sensitivity C-reactive protein, and NT-ProBNP, and tumor markers, including CA-125, CA-199, CA-153, neuron-specific enolase, α fetoprotein, and carcinoembryonic antigen were performed. The cuff-defined blood pressure after the patient rested for 10 to 15 minutes was measured by experienced medical staff. Hypertensive subjects were defined as those with a history of hypertension, taking antihypertensives, or with systolic blood pressure of ≥140 mm Hg and/or diastolic blood pressure ≥90 mm Hg. A history of coronary artery disease referred to those with diagnosed coronary artery stenosis or a history of myocardial infarction with left ventricular hypertrophy by electrocardiography, defined using Sokolow-Lyon criteria. Subjects with a left ventricular ejection fraction <50%, significant primary valvular heart disease (regurgitation or stenosis), pulmonary hypertension (systolic pulmonary arterial pressure >60 mm Hg), severe lung disease, or congenital heart disease by echocardiography were excluded from the study. Those with a known cardiovascular surgery history, rheumatic heart disease, atrial fibrillation, previous pacemaker implantation, overt renal insufficiency (creatinine >2.5 mg/dl), acute or symptomatic coronary events, or a known malignancy of any site were also excluded from our study. The local ethics committee in accordance with the Declaration of Helsinki approved the present study.
We collected echocardiographic parameters such as M-mode–based left atrial (LA) diameter, right ventricular diameter, aortic root diameter, diastolic and systolic left ventricular volumes using biplane Simpson’s method, left ventricular mass and left ventricular mass index, midwall fractional shortening with and without stress correction, circumferential wall stress, and systolic pulmonary arterial pressure derived from the maximum velocity of tricuspid regurgitant flow using the modified Bernoulli equation: systolic pulmonary arterial pressure = 4Vmax 2 + right atrial pressure, where right atrial pressure approximated 10 mm Hg. The LA volume was estimated using biplane Simpson’s method: LA volume = 0.8 × A1 × A2/L, where A1 and A2 are areas measured in 2-chamber and 4-chamber views and L (LA length) represent the linear measurement acquired in 2 different views, either parallel to the atrial septum or perpendicular to the mitral annulus. For the LA volume and length, the maximum and minimum volume were both calculated. The percentage of the LA ejection fraction and length change were further defined as 100 × (maximum value − minimum value)/maximum value.
Sample processing and testing was performed in a standard laboratory with international accreditation (ISO-15189). All subjects were requested to fast for >10 to 12 hours before the collection of venous blood, according to the standard requirements dictated by the Clinical Laboratory Standards Institute guidelines (Specimen Choice, Collection, and Handling; Approved Guideline H18-A3). A Hitachi 7,170 Automatic Analyzer (Hitachi, Hitachinaka Ibaraki, Japan) was used to measure the biochemistry test with the Coulter Gen · S Automatic Analyzer for complete blood count test. CA-125, CA-199, CA-153 and other associated markers were measured with an electrochemiluminescence immunoassay (used on the Roche Elecsys 1010/2010 and Modular Analytics E170 [Elecsys module] immunoassay analyzers; Roche Diagnostics, Mannheim, Germany). High-sensitivity C-reactive protein levels were determined using a highly sensitive, latex particle-enhanced immunoassay Elecsys 2010 (Roche Diagnostics) with NT-ProBNP measured using an electrohemiluminescence immunoassay electrochemiluminescence immunoassay (Roche E170, Roche Diagnostics). Renal function was assessed using the Modification of Diet in Renal Disease formula.
Of the 169 participants initially entered into the derivation cohort, 9 were excluded because of a known record of malignancy, with 2 excluded because of a diagnosis of salpingitis or endometriosis, which has been shown to affect the plasma CA-125 level. The interquartile range and the mean clinical follow-up were calculated. The detailed medical history and admission records were examined carefully for clinical events and any interventional procedure. Incident HF hospitalization was defined as evidence of HF incidence by clinical symptoms (aggravated dyspnea or exercise intolerance) and signs (lower leg edema and pulmonary congestion from the chest radiographic findings) with diuretics prescribed after admission that were adjudicated clinically by experienced physicians.
Continuous data are expressed as the mean ± SD and were compared using the t test or Mann-Whitney U test for unpaired data, as appropriate. Categorical data are presented as ratios and were compared using the chi-square test or Fisher’s exact test. Data violating the principle of normal distribution across ordered groups were compared using a nonparametric trend test (Wilcoxon rank-sum test). One-way analysis of variance was used for an unadjusted comparison of the 3 groups with post hoc Bonferroni’s correction for multiple comparisons among groups. The receiver operating characteristic curve was constructed with the area under curve determined for the optimal cutoff from the largest summation of sensitivity and specificity in predicting incident HF hospitalization. Cox proportional hazards ratios were computed for the probability distribution between categorized NT-ProBNP and CA-125 by the optimal cutoff and time to HF or were performed with conventional risk factors, including age and body mass index as covariates. The overall event-free survival rates were estimated using Kaplan-Meier analysis, and the relative risks between NT-ProBNP and CA-125 strata were compared further using the log-rank test. All data were analyzed by commercialized software, Stata, version 8.2, package (StataCorp, College Station, Texas). The significance of the p level (α value) was 2-sided with p <0.05 considered statistically significant.
Results
After exclusion, 158 women entered our study. The subsequent follow-up data with detailed baseline characteristics and demographic data are listed Table 1 . From the normal to HFpEF groups, age, blood pressures, waist circumference, body mass index, and serum insulin level all increased in an ordered fashion (all p for trend ≤0.01) with decreasing high-density lipoprotein cholesterol level observed across the 3 groups (all p for trend <0.001). The fasting glucose and triglyceride levels showed a borderline increase across the 3 groups (trend p = 0.09 and p = 0.06, respectively). The prevalence of hypertension, coronary artery disease, and left ventricular hypertrophy from the electrocardiographic findings were significantly greater in the HFpEF group (both p <0.05).
Variable | Normal (n = 45) | At Risk (n = 78) | HFpEF (n = 35) | p Value for Trend |
---|---|---|---|---|
Age (years) | 48 ± 11 | 52 ± 10 | 57 ± 12 ⁎ | 0.01 |
Systolic blood pressure (mm Hg) | 109 ± 11 | 121 ± 20 ⁎ | 142 ± 16.2 ⁎ † | <0.001 |
Diastolic blood pressure (mm Hg) | 68 ± 8 | 73 ± 10 ⁎ | 73 ± 10 | 0.01 |
Heart rate (min −1 ) | 65 ± 10 | 65 ± 10 | 70 ± 12 | 0.94 |
Waist (cm) | 73 ± 7 | 80 ± 10 ⁎ | 82 ± 12 ⁎ | <0.001 |
Body mass index (kg/m 2 ) | 21.3 ± 2.3 | 24.8 ± 4.2 ⁎ | 27.4 ± 4.6 ⁎ † | <0.001 |
Hemoglobin (g/L) | 13.4 ± 0.8 | 13.3 ± 0.9 | 12.9 ± 1.5 | 0.16 |
White blood cell count (×10 3 /μl) | 5.384 ± 1.239 | 5.585 ± 1.336 | 6.000 ± 1.887 | 0.32 |
Fasting glucose (mg/dl) | 91 ± 6 | 98 ± 23 | 105 ± 35 ⁎ | 0.09 |
Insulin (IU/ml) | 3.59 ± 2.68 | 4.93 ± 3.75 | 6.17 ± 4.04 ⁎ | <0.001 |
Total cholesterol (mg/dl) | 199 ± 45 | 204 ± 36 | 203 ± 49 | 0.89 |
Triglyceride (mg/dl) | 94 ± 51 | 119 ± 75 | 131 ± 82 | 0.06 |
High-density lipoprotein cholesterol (mg/dl) | 70 ± 15 | 62 ± 17 | 60 ± 15 ⁎ | <0.001 |
Uric acid (mg/dl) | 4.8 ± 1.0 | 5.3 ± 1.2 | 5.5 ± 1.5 | 0.26 |
Estimated glomerular filtration rate (ml/min/1.73 m 2 ) | 91 ± 16 | 91 ± 19 | 83 ± 20 | 0.12 |
Hypertension | 0 | 29 (37.2%) | 15 (42.9%) | <0.001 |
Diabetes | 0 | 12 (15.4%) | 5 (14.3%) | 0.06 |
Known coronary artery disease | 0 | 5 (6.4%) | 7 (20.0%) | 0.003 |
Left ventricular hypertrophy | 0 | 9 (12.5%) | 12 (41.4%) | <0.001 |
The echocardiographic parameters, including septal wall thickness, left ventricular posterior wall thickness, left ventricular mass/mass index, left ventricular mass/volume ratio, estimated systolic pulmonary arterial pressure ( Table 2 ), and LA diameter and volumes ( Figure 1 ) all demonstrated a graded increased from the normal to HFpEF groups (all p for trend <0.05), with a reduction of midwall fractional shortening with and without stress correction observed across the 3 groups (trend p = 0.01). Although the LA diameter and minimum and maximum LA volumes all showed significant trends toward increasing across the groups (all trend for p <0.001; Figure 1 ), only the LA volume and maximum LA volume were significantly larger in the HFpEF group than in the at-risk group ( Figure 1 ). Although an increasing maximum LA volume showed a significant correlation with the serum CA-125 level (r = 0.24, p <0.001), the LA ejection fraction showed just a borderline association with CA-125 (r = 0.16, p = 0.065; Figure 1 ). However, the LA length, whether the minimum or maximum value, or percentage of length change, was not associated with the serum CA-125 level ( Figure 1 ).
Variable | Normal (n = 45) | At Risk (n = 78) | HFpEF (n = 35) | p Value for Trend |
---|---|---|---|---|
Aortic root diameter (mm) | 28.9 ± 3.8 | 29.9 ± 2.9 | 30.3 ± 4.0 | 0.06 |
Septal wall thickness (mm) | 9.2 ± 1.7 | 10.6 ± 1.8 ⁎ | 10.8 ± 2 ⁎ | <0.001 |
Posterior wall thickness (mm) | 9.2 ± 1.3 | 10.5 ± 1.4 ⁎ | 10.7 ± 1.4 ⁎ | <0.001 |
Left ventricular end-diastolic volume (ml) | 86.5 ± 18.5 | 85.4 ± 17.2 | 84.9 ± 18.5 | 0.55 |
Left ventricular end-systolic volume (ml) | 28.1 ± 8.8 | 27.8 ± 7.3 | 27.5 ± 6.4 ⁎ | 0.59 |
Left ventricular ejection fraction (%) | 67.6 ± 5.9 | 67.2 ± 6.3 | 67.5 ± 4.1 | 0.75 |
Right ventricular diameter (mm) | 14.3 ± 5.1 | 15.0 ± 5.0 | 16.0 ± 4.2 | 0.14 |
Left ventricular mass (g) | 131.5 ± 38.8 | 156.8 ± 37.3 ⁎ | 167.5 ± 34.2 ⁎ | <0.001 |
Left ventricular mass index (g/m 2 ) | 82.7 ± 25.4 | 89 ± 17.6 ⁎ | 99.3 ± 17.5 ⁎ | <0.001 |
Left ventricular mass/volume ratio | 1.53 ± 0.36 | 1.87 ± 0.41 ⁎ | 2.03 ± 0.5 ⁎ | 0.03 |
Midwall fractional shortening | 0.21 ± 0.02 | 0.21 ± 0.03 | 0.19 ± 0.02 ⁎ | 0.01 |
Stress-corrected midwall fractional shortening | 0.23 ± 0.03 | 0.23 ± 0.03 | 0.21 ± 0.02 ⁎ | 0.01 |
Circumferential wall stress (kdyne/cm 2 ) | 79.1 ± 15.1 | 81.6 ± 20.1 | 83.3 ± 16.2 | 0.27 |
Meridional wall stress (kdyne/cm 2 ) | 40.4 ± 8.56 | 41.3 ± 11.3 | 42.1 ± 9.1 | 0.39 |
Systolic pulmonary arterial pressure (mm Hg) | 27.3 ± 4.6 | 29.0 ± 4.1 | 32.5 ± 4.8 ⁎ † | <0.001 |

To further elucidate the usefulness of the blood markers in the detection of HFpEF, many tumor markers and biomarkers were subjected to analysis ( Table 3 ). Of these markers, only CA-125, NT-ProBNP, and high-sensitivity C-reactive protein were shown to demonstrate a stepwise, significant increase from normal to the at-risk to the HFpEF group (all p for trend <0.05). Of the 3 markers, only the CA-125 and NT-ProBNP levels were significantly greater in the HFpEF group (post hoc p < 0.05) compared to those in the at-risk group.
Variable | Normal (n = 45) | At Risk (n = 78) | HFpEF (n = 35) | p Value for Trend |
---|---|---|---|---|
Creatine kinase (U/L) | 97.0 ± 52.5 | 93.2 ± 44.3 | 103 ± 61 | 0.66 |
Carbohydrate antigen-125 (U/ml) | 11.8 ± 6.9 | 13.6 ± 5.7 ⁎ | 17.6 ± 10.2 ⁎ † | <0.001 |
Carbohydrate antigen-199 (U/ml) | 12.8 ± 8.9 | 13.8 ± 14.1 | 17.6 ± 14.1 | 0.15 |
Carbohydrate antigen-153 (U/ml) | 11.2 ± 6.2 | 11.4 ± 4.7 | 12.7 ± 5.8 | 0.13 |
Neuron-specific enolase (ng/ml) | 7.67 ± 2.13 | 7.86 ± 2.49 | 7.44 ± 1.51 | 0.83 |
α Fetoprotein (ng/ml) | 2.92 ± 1.71 | 2.80 ± 1.69 | 2.93 ± 2.45 | 0.29 |
Carcinoembryonic antigen (ng/ml) | 1.67 ± 0.77 | 2.11 ± 1.31 | 2.47 ± 1.83 ⁎ | 0.07 |
High-sensitivity C-reactive protein (mg/L) | 0.13 ± 0.18 | 0.21 ± 0.34 ⁎ | 0.31 ± 0.66 ⁎ | 0.01 |
N-terminal pro-brain type natriuretic peptide (pg/ml) | 40.4 ± 38.2 | 45.6 ± 38.7 | 104 ± 151 ⁎ † | 0.03 |
When testing the clinical variables and those blood markers on univariate analysis, we found that age, coronary artery disease, CA-125, and NT-ProBNP were all significant indicators in detecting the presence of HFpEF ( Table 4 ), with both NT-ProBNP and CA-125 remaining independently associated with HF after adjustment for clinical variables (p = 0.009 and p <0.001, respectively).
Clinical Variable | Univariate Model | Multivariate Model | ||||
---|---|---|---|---|---|---|
OR | 95% CI | p Value | OR | 95% CI | p Value | |
Age (years) | 1.9 | 1.25–2.88 | 0.002 | 1.05 | 1.00–1.10 | 0.052 |
Body mass index (kg/m 2 ) | 1.34 | 0.98–1.82 | 0.063 | 1.06 | 0.95–1.17 | 0.298 |
Systolic blood pressure (mm Hg) | 1.02 | 1.00–1.04 | 0.066 | 1.004 | 0.98–1.03 | 0.772 |
Hypertension | 2.36 | 0.97–5.76 | 0.059 | 1.15 | 0.40–3.34 | 0.792 |
Diabetes | 1.56 | 0.38–6.40 | 0.535 | 0.63 | 0.12–3.31 | 0.59 |
Coronary artery disease | 5.96 | 1.76–20.2 | 0.004 | 4.78 | 1.27–17.96 | 0.021 |
N-terminal pro-brain type natriuretic peptide (pg/ml) | 1.01 | 1.00–1.02 | 0.003 | 1.01 | 1.00–1.02 | 0.009 |
Carbohydrate antigen-125 (U/ml) | 1.08 | 1.03–1.14 | 0.004 | 1.14 | 1.06–1.21 | <0.001 |
Creatine kinase (U/L) | 1 | 1.00–1.01 | 0.408 | 1 | 1.00–1.01 | 0.762 |
Carbohydrate antigen-199 (U/ml) | 1.02 | 1.00–1.05 | 0.101 | 1.02 | 1.00–1.05 | 0.101 |
Carbohydrate antigen-153 (U/ml) | 1.05 | 0.98–1.12 | 0.178 | 1.04 | 0.96–1.12 | 0.348 |
Neuron-specific enolase (ng/ml) | 0.93 | 0.76–1.15 | 0.519 | 0.85 | 0.66–1.08 | 0.18 |
α Fetoprotein (ng/ml) | 1.02 | 0.84–1.24 | 0.84 | 0.99 | 0.79–1.23 | 0.91 |
Carcinoembryonic antigen (ng/ml) | 1.28 | 0.99–1.66 | 0.058 | 1.16 | 0.86–1.57 | 0.326 |
High-sensitivity C-reactive protein (mg/L) | 1.9 | 0.82–4.40 | 0.134 | 1.76 | 0.75–4.15 | 0.197 |

Stay updated, free articles. Join our Telegram channel

Full access? Get Clinical Tree


