Carbohydrate Antigen-125 and N-Terminal Pro-Brain Natriuretic Peptide Levels

Abstract

Carbohydrate antigen-125 (CA-125) is emerging as a prognostic biomarker of risk in heart failure. In a prospective study, we compared the prognostic values of CA-125 and amino-terminal pro-brain natriuretic peptide (NT-proBNP) in patients with stable heart failure.

We enrolled 102 consecutive chronic, stable, systolic-heart-failure patients (68 men and 34 women; median age, 71 yr) from November 2008 through February 2010. We measured baseline NT-proBNP and CA-125 levels and compared their prognostic values. The primary endpoint was all-cause death and other major adverse events, defined as hospitalization for decompensated heart failure or acute coronary syndrome.

During a mean follow-up period of 14 ± 2 months, 12 patients died and 35 others sustained major adverse events. We found that CA-125 level significantly correlated with New York Heart Association functional class, pulmonary artery pressure, microalbuminuria, creatine kinase–MB fraction, and hemoglobin, albumin, and NT-proBNP levels. Upon receiver operating characteristic curve analysis, CA-125 and NT-proBNP had similar accuracy in predicting major adverse events and death: for major adverse events, area under the curve (AUC) was 0.699 for CA-125 (P=0.002) and 0.696 for NT-proBNP (P=0.002); for death, AUC was 0.784 for CA-125 (P=0.003) and 0.824 for NT-proBNP (P=0.001). Multivariate Cox regression analysis showed that CA-125 levels greater than 32 U/mL and NT-proBNP levels greater than 5,300 pg/mL had independent prognostic value for major adverse events and death.

We conclude that baseline CA-125 and NT-proBNP levels are comparably reliable as heart-failure markers, and that CA-125 can be used for prognosis prediction in heart failure.

Heart failure is a major public health problem in Western countries and the developing world.¹ In most studies of heart-failure patients, the investigators have reported correlations between amino-terminal pro-brain natriuretic peptide (NT-proBNP) plasma levels and clinical, functional, and hemodynamic values. Therefore, NT-proBNP is almost universally considered to be the gold standard as a biomarker of heart failure.² In recent years, carbohydrate antigen-125 (CA-125), a glycoprotein synthesized by epithelial serosal cells, has emerged as a prognostic biomarker in heart failure and has shown very good correlations with clinical, hemodynamic, and echocardiographic variables indicative of the severity of the disease.³ In patients with chronic heart failure, increased levels of CA-125 upon hospital admission were predictive of death and rehospitalization after 6 months.⁴ After heart transplantation, fluctuations in serum levels of CA-125 after treatment were associated with long-term prognosis.⁵ In a prospective study, we sought to compare the prognostic values of NT-proBNP and CA-125 in patients with stable heart failure.

Patients and Methods

We enrolled 102 consecutive patients who had been admitted to the Duzce Medical Faculty cardiology outpatient clinic (68 men and 34 women; median age, 71 yr) from November 2008 through February 2010. Each patient had the diagnosis of stable, chronic systolic heart failure (New York Heart Association [NYHA] functional class I–III) and a left ventricular ejection fraction <0.40. Seventy-five patients had ischemic and 27 had nonischemic dilated cardiomyopathy, 96 continued to take angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and 77 continued to take β-blockers. No patient was undergoing implantable cardioverter-defibrillator or cardiac resynchronization therapy. The duration of clinical stability before study inclusion ranged from 3 to 180 months (median, 22 mo).

Exclusion Criteria. We excluded from the study patients with a diagnosis of acute heart failure, acute coronary syndrome, cancer, severe hepatic disease, chronic inflammatory disease, severe chronic renal failure, or infectious disease, and anyone who had been taking steroids or immunosuppressive drugs. Patients who had been hospitalized during the last 3 months were also excluded.

Each eligible patient provided written informed consent before study entry. The Ethics Committee of Duzce Medical School approved the study protocol. On the day of enrollment, each patient underwent physical examination, standard 12-lead electrocardiography, and complete echocardiographic evaluation and venous blood sampling. Demographic, clinical, and instrumental values for each patient were recorded, including age, sex, body mass index, NYHA functional class, systolic blood pressure, and cardiac rhythm. The patients' NT-proBNP and CA-125 levels were measured only at the beginning of the study.

Echocardiographic Evaluation. Each patient underwent a 2-dimensional Doppler echocardiographic examination (Vivid 3®, GE VingMed Ultrasound; Haifa, Israel) with use of a 2.5-MHz transducer. Left ventricular ejection fraction was calculated with use of the Simpson formula after end-diastolic and end-systolic volumes were measured in apical 4-chamber view. Pulmonary artery pressure was measured by apical 4-chamber-view Doppler estimation.

Plasma Natriuretic Peptide Measurement. Blood samples were collected into EDTA tubes on ice; the plasma was immediately separated and stored at −70°C. Plasma NT-proBNP levels were measured with use of a commercially available electrochemiluminescence immunoassay (Elecsys® proBNP assay, Roche Diagnostics Corporation; Indianapolis, Ind). Interassay coefficients of variation ranged from 8% to 15% intra-assay coefficients of variation ranged from 6% to 8%. It took approximately 36 min to obtain each patient's NT-proBNP value.

Carbohydrate Antigen-125 Measurement. Serum levels of CA-125 were determined with use of the commercially available Tumor Markers CA 125 AxSYM® System (Abbott Laboratories; Abbott Park, Ill). The upper normal limit of CA-125 is 35 U/mL. The analytical sensitivity of the AxSYM CA-125 was calculated to be 2 U/mL; the median intra-assay and interassay coefficients of variation were 5% and 7.5%, respectively. It took approximately 70 min to obtain each patient's CA-125 value.

Monitoring during the Study Period. The patients underwent monitoring through clinical visits or by telephone for 12 months (mean, 14 ± 2 mo). The primary endpoint, major adverse events (MAE), was defined as death from any cause, or hospitalization due to decompensated heart failure or acute coronary syndrome. The 1-year prognostic values of NT-proBNP and CA-125 were then compared.

Statistical Analysis

All values are given as mean ± SD. Statistical Package for the Social Sciences software version 10.0 (SPSS, now part of IBM Corporation; Somers, NY) was used for comparisons of demographic and clinical variables. The unpaired Student t test was used to compare normally distributed continuous variables between the groups. Abnormally distributed variables were compared by means of the Mann-Whitney U test. Categorical variables were compared with the χ² test. Spearman or Pearson correlation tests were used to evaluate the association between the clinical and laboratory variables and the heart-failure biomarkers (NT-proBNP and CA-125). Multivariate analyses based on the Cox proportional hazards model were performed to identify the most significant factors related to MAE-free survival. MedCalc® software (MedCalc Software; Mariakerke, Belgium) was used to calculate cutpoints for receiver operating characteristic curve analyses, sensitivity and specificity, and positive and negative predictive values. Any P value less than 0.05 was considered statistically significant.

Results

During follow-up, 12 patients died and 35 others sustained nonfatal MAE. Table I shows the comparative demographic and clinical variables of the patients who sustained MAE and those who did not. The variables were similar except for NYHA functional class upon study enrollment.

TABLE I. Comparison of Demographic and Clinical Variables in Patient Groups with Different Outcomes

Comparison of echocardiographic and laboratory variables revealed that patients who sustained MAE had significantly higher levels of CA-125 and NT-proBNP; however, their serum albumin levels were significantly lower (Table II).

TABLE II. Comparison of Echocardiographic and Laboratory Variables in Patient Groups with Different Outcomes

Correlation analysis showed that CA-125 had statistically significant correlations with NYHA class, mean pulmonary artery pressure, microalbuminuria, creatine kinase–MB fraction, and levels of hemoglobin, albumin, and NT-proBNP (Table III).

TABLE III. Significant Correlations between Clinical Variables and Prognostic Markers

The receiver operating characteristic curve analysis showed that CA-125 and NT-proBNP were similarly accurate in predicting MAE. For all MAE, area under the curve (AUC) was 0.699 (P=0.002) for CA-125 and 0.696 (P=0.002) for NT-proBNP; for death, AUC was 0.784 (P=0.003) for CA-125 and 0.824 (P=0.001) for NT-proBNP (Fig. 1). A baseline level of CA-125 greater than 32 U/mL could predict 1-year death with 83% sensitivity and 72% specificity, and a baseline level of NT-proBNP greater than 5,300 pg/mL could predict 1-year death with 82% sensitivity and 81% specificity. A multi-marker approach that used CA-125 and NT-proBNP together was analyzed to see whether diagnostic sensitivity and specificity increased. A baseline level of CA-125 greater than 32 U/mL plus an NT-proBNP level greater than 5,300 pg/mL could predict 1-year death with 92% sensitivity but only 63% specificity (AUC=0.772; P <0.001)—although sensitivity increased, specificity decreased significantly.

Fig. 1 Receiver operating characteristic curve analyses of carbohydrate antigen-125 (CA-125) and amino-terminal pro-brain natriuretic peptide (NT-proBNP) in predicting A) all major adverse events and B) death alone.

Figure 2 shows Kaplan-Meier survival curves of the probability of event-free survival according to CA-125 and NT-proBNP levels. Multivariate analyses in accordance with the Cox proportional hazards model were used to identify the most significant factors related to event-free survival. The variables entered into the model were age, CA-125 level, NT-proBNP level, NYHA class, cardiomyopathy type, ejection fraction, and uric acid level. Only CA-125 levels greater than 32 U/mL and NT-proBNP levels greater than 5,300 pg/mL had independent prognostic value for event-free survival (Table IV).

Fig. 2 Kaplan-Meier graphs show the probability of event-free survival according to A) carbohydrate antigen-125 (CA-125) and B) amino-terminal pro-brain natriuretic peptide (NT-proBNP) levels.

TABLE IV. Multivariate Cox Regression Analysis of Event-Free Survival

In this prospective study, we found that CA-125 was an accurate prognostic marker in stable heart failure, showing reliability comparable with that of NT-proBNP.

Discussion

A meta-analysis in 2006 showed that pooled sensitivity and specificity values for NT-proBNP were 92% and 65% in detecting heart failure, and that it was also an independent predictor of death and other cardiac composite endpoints in patients with heart failure.⁶ It has been shown that CA-125 correlates well with clinical, hemodynamic, and echocardiographic metrics indicative of the severity of heart failure.3–5,7–9 Of note, fluctuations in serum levels of CA-125 after treatment for heart transplantation have also been found, suggesting the usefulness of CA-125 in a serial and long-term therapeutic intervention evaluation.^4,5 In a univariate setting, increased CA-125 levels predicted death and rehospitalization 6 months after admission.⁴ Nunez and colleagues⁹ found significant differences in CA-125 levels in patients with and without heart failure: patients with acute heart failure exhibited a 7-fold increase in mean CA-125 serum levels in comparison with control subjects (105.2 vs 14.9 U/mL; P=0.001).⁹ Upon multivariate analysis, CA-125 levels predicted death independently of age, sex, diabetes mellitus, NYHA class, systolic blood pressure, and serum creatinine and hemoglobin values. Death risk increased markedly as CA-125 levels increased to 60 U/mL. Chen and associates¹⁰ also found that BNP and CA-125 had similar accuracy and positive predictive value in the diagnosis of heart failure. Our results confirm those of the 2 aforementioned studies. We have further shown prospectively that CA-125 levels can just as reliably predict the risk of MAE as can NT-proBNP in stable heart-failure patients.

Carbohydrate antigen-125 is actually a cancer marker, and its level has been increased in patients with pericardial, pleural, and peritoneal effusions.^11–13 Epiney and colleagues¹¹ reported that mesothelial cells from the peritoneum and pleura can synthesize CA-125. Nunes and associates⁹ showed an association between radiologic pleural effusion and CA-125 serum concentrations. Turk and co-authors¹⁴ showed that CA-125 values were highest in patients with pleural effusion, intermediate in those with no pleural effusion, and lowest in the control group, with statistically significant differences. This finding can be interpreted as indirect evidence to support the claim that serosal effusions, especially from the pleura, are the main source of increased CA-125 in heart failure. It is also possible that CA-125 is produced from activated mesothelial cells, even in the absence of classical stimuli (such as fluid retention) consequent to tissue-stretching caused by increased chronic left ventricular filling pressure and cytokine stimulation. Increased interleukin-6, another cytokine, has been shown to stimulate the proliferation of CA-125–producing cells.¹⁵ Increased serum interleukin-6 concentration in congestive heart failure may explain elevated serum CA-125 concentrations.¹⁶

Although a pathophysiologic basis underlying the observed association between death in heart failure and an increase in CA-125 is not yet clear, more data are accumulating in regard to the value of CA-125 as a prognostic screening tool. Another advantage of CA-125 is its price: we found that it was 10 times less expensive to perform a CA-125 test than an NT-proBNP test (2 vs 20 Turkish lira, or approximately U.S. $1.07 vs $10.74 as of mid-January 2012¹⁷).

Limitations of the Study

Although correlation analysis revealed significant correlations, the coefficients were relatively low, showing a loose association. The r value of the correlation between CA-125 and NT-proBNP was relatively small, although the P value was significant. The number of patients in our study was relatively small. Even if any 2 trials used the same methodology and achieved the same level of statistical significance, there could be a substantial difference in the fragility of the demonstrated P values. In small-cohort trials, a small number of patients with the primary outcome could change the P value to an insignificant level, whereas adding the same number of MAEs in large-cohort trials would have no meaningful impact on the P value, which would remain significant. Increasing the sample size in our study might have led to better cutoff values for predicting MAEs and maximized the prognostic value of CA-125. Finally, we measured pulmonary artery pressure with Doppler echocardiography, a method less reliable than right-sided heart catheterization.

Conclusion

In conclusion, we found that baseline CA-125 level was an accurate prognostic marker in stable heart failure. Direct comparison with NT-proBNP revealed that CA-125 is at least as reliable as NT-proBNP in detection of event-free survival and can be used for long-term prognostication.