Carbohydrate antigen 125 and clinical outcomes in heart failure: systematic review and meta-analysis.

In this study, we adhered to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) criteria, and the study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) (ID: CRD420251023141). This systematic review and meta-analysis (SRMA) followed a clear, pre-established protocol with specific guidelines for selecting studies, extracting data, and conducting the analysis. We included studies involving patients diagnosed with acute or chronic heart failure (AHF and CHF) that investigated the CA125 levels across the cases. The outcomes of interest were clinical events, including all-cause mortality, heart failure-related hospitalizations, disease progression, and correlations with established disease activity markers. The articles included were restricted to studies published in English. The unpublished manuscripts and conference abstracts were excluded. We conducted a thorough search of four electronic databases: PubMed, Scopus, Embase, and Web of Science, up to April 2025. The Medical Subject Headings (MeSH) terminology and keywords which we used were as follows: Heart failure, Ventricular Dysfunction, Carbohydrate antigen 125, Cancer antigen 125, and CA125 Antigen.The specific search terms and queries for each database are listed in the Supplemental file. To identify and remove duplicates, we utilized online reference management tools, including Rayyan and EndNote 21, to manage citations. The retrieved studies were selected by two reviewers (SMP and FOS) based on their abstracts and titles, and studies deemed irrelevant were excluded. The final selection process involved a thorough review of the entire text of potentially eligible articles to determine whether they met our eligibility requirements. To settle any disputes, the third investigator (KH) was consulted. We evaluated the quality of the included cohort and cross-sectional studies using the Newcastle-Ottawa Scale (NOS), adapted for each study design. For cohort studies, the NOS evaluates selection, comparability, and outcome, which are then classified as Good (3–4 stars in selection, 1–2 in comparability, 2–3 in outcome), Fair (2 stars in selection, 1–2 in comparability, 2–3 in outcome), or Poor (0–1 star in selection, 0 in comparability, 0–1 in outcome). For cross-sectional studies an adjusted NOS classifies quality into four categories: Very Good (9–10 points), Good (7–8 points), Satisfactory (5–6 points), and Unsatisfactory (0–4 points). Two authors (SMP and DS) independently evaluated all studies, and any disagreements were resolved by consulting with a senior author (KH). Data were extracted separately by two authors (SMP and FOS) using a pre-made format. The extracted data included general details such as the first author’s name, publication year, study title, design, sample size, baseline characteristics, and outcomes. Furthermore, we contacted the corresponding authors of the included studies to retrieve missing data. However, when critical data could not be obtained, those parameters were omitted from the analysis. A third author (KH) was consulted to resolve any differences. We used the hazard ratio (HR) as an effect size for time-to-event outcomes, such as mortality and hospitalization, and the relative risk (RR) for other binary outcomes, including New York Heart Association (NYHA) classes. For continuous outcomes, mean difference (MD) was chosen as an effect size. We pooled the effect sizes using a random-effects model. The Hartung-Knapp-Sidik-Jonkman method was applied to calculate the confidence interval for the pooled effect size. Between-studies heterogeneity was assessed using Cochran’s Q-test, and the restricted maximum likelihood method was used for variance estimation. Heterogeneity was classified based on the I 2 values as follows: low (I 2  < 25.0%), moderate (25.0% < I 2   75.0%). Subgroup analyses were performed using pre-defined variables such as HF type (AHF vs. CHF), Age (≤ 70 years vs. >70 years), follow-up duration (≤ 1 year vs. >1 year), sample size (≤ 500 cases vs. >500 cases), and location (Asia, Europe, International), whenever at least two articles were categorized in each subgroup. A sensitivity analysis based on the leave-one-out method was conducted for each reported outcome to evaluate the robustness of findings. Statistical analysis was performed in R version 4.3.3 using the “meta” package. Statistical significance was assessed using 2-tailed tests, with a threshold of P  < 0.05 to report statistical significance.

Following a systematic literature search, 1471 records were identified through PubMed ( n  = 218), Scopus ( n  = 213), Web of Science ( n  = 337), and Embase ( n  = 703) databases. Initially, 718 duplicate studies were excluded. Next, 753 articles were assessed by title and abstract for inclusion, and 81 records were selected for full-text evaluation. Ultimately, 29 studies [ 10 – 38 ] were included in the current systematic review and meta-analysis. Figure  1 represents the screening process and detailed results of the findings from each step. The included articles, published between 2003 and 2024, contain information on 20,458 cases, with an average age of 70.3 years. Fourteen articles [ 10 , 12 – 24 ] investigated the AHF, and fifteen examined the CHF [ 11 , 25 – 38 ]. In cohort ( n  = 22) and post-hoc ( n  = 2) studies, the median follow-up time ranged from 3 to 49 months. Based on the 20 studies that reported the CA125 cut-off value, the average cut-off level was 37.78 U/mL, with 35 U/mL being the most commonly used cut-off level for CA125. All included cohorts were classified as in the “Good” group based on the NOS for cohort studies. The overall scores of 9, 8, and 7 were achieved in six, eight, and three studies, respectively (See Supplemental Table 1). Regarding the QA of Cross-sectional articles, overall scores were 6 (“Satisfactory”) in four studies and 9 (“Very Good”) in one article (See Supplemental Table 2). Table  1 shows detailed information on the main characteristics and outcomes of the included articles. Fig. 1 PRISMA flow diagram of selection process results PRISMA flow diagram of selection process results Table 1 Summary of clinical studies evaluating the role of CA125 in acute and chronic heart failure: study characteristics, patient demographics, and key findings Study Patients Main findings First author Year Design Country Follow-up Population Age Sample Size Female (%) CA125 -cutoff Davutoglu [ 13 ] 2010 Cohort Turkey 6 AHF 65 100 41.00% 50 1 CA125 as a predictor of poor outcomes, independent of the presence of pleural effusion Kouris [ 15 ] 2006 Cohort Greece 15 AHF 70 95 0.00% 35 Serum level of CA125 as a marker of severity and risk of re-hospitalization in congestive HF Mansour [ 17 ] 2010 Cohort USA 40 AHF 55.8 170 38.90% 35 In African American cases with HF, CA125 levels were associated with 40-month all-cause mortality while not significantly associated with 18-month HF rehospitalization Monteiro [ 19 ] 2009 Cohort Portugal 13 AHF 53.5 88 29.55% 38 CA125 levels were a helpful marker in advanced HF cases (mean LVEF = 24.5%) Chen [ 12 ] 2020 Cohort China 12 AHF 70.02 213 46.48% 47.6 A combination of CA125 and NT-proBNP could improve risk stratification in AHF patients Docherty [ 14 ] 2023 Post-Hoc Multinational 24 AHF 67.25 3123 22.13% 35 CA125 was the predictor of poor outcomes in HFrEF cases treated with Dapagliflozin Lourenco [ 16 ] 2022 Cohort Portugal 12 AHF 75 363 48.48% 35 A decrease in CA125 levels from admission to discharge day was associated with better prognosis in cases with AHF who have more than 10 days length of hospital stay Yilmaz [ 23 ] 2011 Cohort Turkey 8 AHF 66 150 46.00% 35 LVEF, right ventricular dilatation, and the presence of pericardial effusion could determine the CA125 levels Yoon [ 24 ] 2019 Cohort Korea 24 AHF 64.1 413 48.18% 54.5 The combination of both CA125 and NT-proBNP significantly enhanced the prediction of mortality in decompensated patients with AHF Zhang [ 25 ] 2024 Cohort China 6 AHF 76 352 46.90% 39.7 CA125 added prognostic value to NT-proBNP, and their combination had a better predictive ability compared to each one Nunez [ 22 ] 2022 Cohort Europe 6 AHF 73.6 3231 48.10% 23 Individuals with AHF who had low CA125 levels upon admission may require less intensive monitoring after discharge Nunez [ 20 ] 2020 Cohort Europe 12 AHF 69 2356 26.10% 38.6 2 Elevated CA125 levels were significantly associated with the probability of the presence of congestion parameters such as peripheral edema, elevated JVP, hepatomegaly, and orthopnea Nunez [ 21 ] 2017 Cohort Europe 31.6 AHF 71 946 47.25% 35 The combined serial measurement of CA125 and NT-proBNP in a long-term course following the hospitalization due to the AHF improved the risk stratification of all-cause mortality Minana [ 18 ] 2020 Cohort Spain NA AHF 73.9 2949 48.90% 58 2 Clinical parameters of congestion and severity of tricuspid regurgitation were the most important factors in determining CA125 levels as a marker of right-sided HF Vizzardi [ 39 ] 2012 Cohort Italy 43 CHF 64.3 102 17.65% 30 In non-severe HF cases (mean LVEF = 34.4%), CA125 levels could predict HF hospitalization and cardiovascular events D’Aloia [ 26 ] 2003 Cohort Italy 6 CHF 69 286 57.34% 35 Serial measurement of CA125 could help monitor treatment in patients with HF Durak-Nalbantic [ 27 ] 2013 Cross-sectional Bosnia NA CHF 73.5 50 44.00% 35 The CA125 levels were significantly higher in HF cases with pleural and/or pericardial effusion compared to other HF patients Faggiano [ 28 ] 2004 Cross-sectional Italy NA CHF 67 191 54.97% 35 CA125 levels correlated with the severity of CHF, and the CA125 concentrations significantly altered in response to treatment Ferreira [ 29 ] 2024 Post-Hoc Multinational 12 CHF 70.1 1111 32.22% 20.77 CA125 concentrations may predict cardiovascular events and HF hospitalization in HFrEF cases, but it was not useful in patients with HFpEF Kouris [ 30 ] 2004 Cross-sectional Greece NR CHF 73 77 0% 35 Elevated CA125 levels are associated with more severe disease and higher fluid congestion in CHF patients Ma [ 31 ] 2013 Cohort China 6 CHF 91 2115 15.69% NR CA125 was the only tumor marker related to the severity of CHF in cases older than 85 years Mendez [ 32 ] 2014 Cohort Spain 17 CHF 72 156 37.00% 60 High CA125 levels are associated with poor survival in patients with stable HF Menghoum [ 33 ] 2024 Cohort Belgium 49 CHF 78 139 60% 35 CA125 levels were a significant and independent prognostic marker of HF-related hospitalization in cases with HFpEF Ordas [ 35 ] 2024 Cross-sectional Spain NA CHF 74 802 35% 14 CA125 was associated with extravascular volume overload variables (peripheral edema and lung B-lines), while NT-proBNP was associated with intravascular congestion signs and symptoms Ordu [ 34 ] 2012 Cohort Turkey 14 CHF 71 102 33.33% 32 Baseline levels of CA125 and NT-proBNP could be reliable markers in HF Pacho [ 36 ] 2018 Cohort Spain 12 CHF 82 522 57.08% 47 Although CA125 could not independently predict the 30-day primary endpoints, it could significantly predict 1-year all-cause mortality, outperforming NT-proBNP Turk [ 37 ] 2003 Cross-sectional Turkey NR CHF 61.65 36 36.11% NR Elevated CA125 was associated with pleural effusion in HF patients Varol [ 38 ] 2005 Cohort Turkey 3 CHF 66.3 44 63.51% 16.3 CA125 levels were significantly related to the severity of CHF and the presence of pericardial effusion Zhang [ 40 ] 2023 Cohort China 18 CHF 76 176 43.75% 65.7 In cases with stage D HF (NYHA classes III/IV), the elevated CA125 was associated with poor outcomes NR Not reported, AHF Acute heart failure, CHF Chronic heart failure, LVEF  Left ventricular ejection fraction, CA125  Carbohydrate antigen 125, NT-proBNP N-terminal pro–B-type natriuretic peptide, JVP  Jugular venous pressure, HFrEF  Heart failure with reduced ejection fraction, HFpEF  Heart failure with preserved ejection fraction 1 HR reported for each 50 U/mL increase in CA125 2 The mentioned values were the median of the CA125 levels across the patients Summary of clinical studies evaluating the role of CA125 in acute and chronic heart failure: study characteristics, patient demographics, and key findings NR Not reported, AHF Acute heart failure, CHF Chronic heart failure, LVEF  Left ventricular ejection fraction, CA125  Carbohydrate antigen 125, NT-proBNP N-terminal pro–B-type natriuretic peptide, JVP  Jugular venous pressure, HFrEF  Heart failure with reduced ejection fraction, HFpEF  Heart failure with preserved ejection fraction 1 HR reported for each 50 U/mL increase in CA125 2 The mentioned values were the median of the CA125 levels across the patients Meta-analysis of nine studies [ 11 , 14 , 21 , 24 , 27 , 31 , 32 , 37 , 38 ] that reported the risk of experiencing endpoint events (death or HF-related hospitalization) showed significantly higher risk (HR 2.23, 95% CI 1.69–2.93; p  < 0.01) among the patients with higher CA125 levels compared to cases with lower levels, using a random-effects model (Fig.  2 A). The observed heterogeneity between studies was high (I 2  = 75%). The sensitivity analysis, conducted using the leave-one-out method, revealed that the significance of the results would not change if any articles were omitted (See Supplemental Fig. 44). Further subgroup analyses were conducted, and higher risk remained statistically significant in studies with AHF phenotype ( N  = 2 studies, HR 1.88, 95% CI 1.54–2.30; p  < 0.01), CHF phenotype ( N  = 7 studies, HR 2.52, 95% CI 1.69–3.77; p  < 0.01), average age less than 70 years ( N  = 3 studies, HR 4.53, 95% CI 2.08–9.86; p  < 0.01) or higher than 70 years ( N  = 6 studies, HR 1.83, 95% CI 1.59–2.10; p  < 0.01), less than one-year follow-up duration ( N  = 4 studies, HR 2.06, 95% CI 1.24–3.42; p  < 0.01) or more than 1-year follow-up duration ( N  = 5 studies, HR 2.35, 95% CI 1.78–3.10; p  < 0.01), using 35 U/mL as CA125 cut-off ( N  = 3 studies, HR 2.81, 95% CI 1.73–4.56; p  < 0.01) or other cut-off values ( N  = 6 studies, HR 1.85, 95% CI 1.57–2.18; p  < 0.01). (See Supplemental Figs. 1–6). The detailed findings of all subgroup analyses are available in Supplemental Table 3. The pooled estimate of nine articles [ 12 , 14 , 16 , 17 , 19 , 21 , 23 , 31 , 38 ] investigating all-cause mortality revealed a significantly higher risk of death among cases with elevated CA125 than patients with normal CA125 levels (HR 2.28, 95% CI 1.73-3.00; p  < 0.01) (Fig.  2 B). The heterogeneity analysis demonstrated high between-study heterogeneity (I 2  = 80.5%). Although the leave-one-out sensitivity analysis showed that removing any articles did not significantly change the pooled results, omitting the Yoon et al. [ 23 ] study significantly reduced the observed heterogeneity across the included articles (I2 = 0.0%) (See Supplemental Fig. 45). Subgroup analysis of studies evaluating the AHF (HR 2.35, 95% CI 1.52–3.65; p  < 0.01) or CHF (HR 2.48, 95% CI 2.00-3.06; p  < 0.01) cases showed a similar significantly increased risk of death. However, the heterogeneity of the three studies in the CHF subgroup was low (I2 = 5.4%), while studies in the AHF subgroup were significantly heterogeneous (I2 = 78.1%). Further analysis indicated that higher CA125 was associated with increased mortality risk in both subgroups of studies with 35 U/mL (HR 2.59, 95% CI 2.16–3.10; p  < 0.01) as the CA125 cut-off and studies with other cut-off values (HR 2.29, 95% CI 1.37–3.81; p  < 0.01), while heterogeneity was very low (I² = 0.0%) in first subgroup. The findings of other subgroup analyses based on location, average patient age, follow-up duration, and sample size are shown in Supplemental Table 3 (See Supplemental Figs. 7–10). Fig. 2 Forest plot of random-effects meta-analysis comparing A endpoint and B all-cause mortality risk between elevated versus normal CA125 levels Forest plot of random-effects meta-analysis comparing A endpoint and B all-cause mortality risk between elevated versus normal CA125 levels A further meta-analysis based on the five studies [ 12 , 20 , 23 , 30 , 38 ] that compared the average CA125 levels between deceased and survived HF patients revealed significantly increased CA125 levels in the deceased compared to the survived (MD: 23.904 U/mL, 95% CI 1.634–46.174; p  = 0.04) (See Supplemental Fig. 11). The observed between-study heterogeneity was high (I 2  = 82.4%), while sensitivity analysis showed that excluding the Nunez et al. study could significantly reduce the heterogeneity (I 2  = 8.4%). Furthermore, the Leave-one-out analysis indicated a non-significant mean difference of CA125 levels between deceased and survivors upon removal of any included articles except the Nunez et al. [ 20 ] (See Supplemental Fig. 46). Further analyses restricting included articles to studies with AHF or CHF patients also did not reveal significant differences between the average CA125 level of deceased and survived patients. Findings of more subgroup analyses are presented in Supplemental Table 4 (See Supplemental Figs. 12 and 13). Meta-analysis of four studies [ 14 , 27 , 30 , 38 ] containing patients with CHF reported cardiovascular death risk across the cases with abnormal CA125 levels, indicating increased risk in patients with abnormal CA125 levels (HR 2.15, 95% CI 1.35–3.44; p  < 0.01) compared to others, while the observed heterogeneity was moderate (I 2  = 74.5%) (See Supplemental Fig. 14). The sensitivity analysis showed that between-study heterogeneity was significantly reduced by removing Ferreira et al. [ 27 ] (I 2  = 0.0%) while omitting none of the articles changed the significance of pooled results (See Supplemental Fig. 47). The findings of subgroup analyses based on the sample size and study design are available in Supplemental Table 3 (See Supplemental Figs. 15 and 16). Five studies [ 14 , 15 , 19 , 27 , 31 ] investigated the differences in the risk of hospitalization due to HF between cases with increased and normal CA125 levels. The meta-analysis of these included articles suggested a significantly higher risk among the subjects with abnormal CA125 levels (HR 1.79, 95% CI 1.08–2.94; p  = 0.02) with high observed heterogeneity between selected studies (I 2  = 87.4%) (See Supplemental Fig. 17). The sensitivity analysis showed that omitting the Monteiro et al. [ 19 ] study reduced the heterogeneity to a moderate class (I 2  = 58.6%), while the risk of hospitalization remained significantly higher across individuals with abnormal CA125 levels. However, the pooled RR of hospitalization after removing any of the other included studies was not statistically significant (See Supplemental Fig. 48). Supplemental Figs. 18–20 represent the forest plot of subgroup analyses of age, sample size, and study design. A meta-analytic pooling of six studies’ findings [ 14 , 17 , 18 , 27 , 30 , 31 ] regarding the NYHA class of individuals showed that abnormal CA125 levels were significantly associated with a higher rate of higher NYHA class (III/IV) (RR 1.40, 95% CI 1.16–1.68; p  < 0.01), yielding high between-study heterogeneity (I 2  = 71.1%) (See Supplemental Fig. 21). The sensitivity analysis based on the leave-one-out method revealed that removing none of the included studies would significantly change the pooled estimate (See Supplemental Fig. 49). The pooled RR was higher in subgroups of 4 studies, including cases with CHF (RR 1.51, 95% CI 1.35–1.68; p  < 0.01), compared to 2 studies involving AHF patients (RR 1.15, 95% CI 0.94–1.41; p  < 0.01). Five studies [ 11 , 15 , 26 , 28 , 36 ] reported the difference in the average CA125 level between cases with NYHA classes I/II and III/IV. Meta-analysis of these studies indicated significantly higher CA125 levels among the cases with NYHA classes III/IV compared to others (MD: 73.85 U/mL, 95% CI 16.01 to 131.69; p  < 0.01), while studies were significantly heterogeneous (I 2  = 96.3%, p  < 0.01) (See Supplemental Fig. 27). The sensitivity analysis using the leave-one-out method revealed that the pooled estimate MD between cases with different NYHA classes ranged from 61.25 U/mL (95% CI −7.62 to 130.11) by removing Faggiano et al. study [ 26 ] to 86.71 U/mL (95% CI 18.76 to 154.66) by removing Kouris et al. study [ 15 ] (See Supplemental Fig. 50). Six articles [ 13 , 15 , 25 , 28 , 29 , 35 ] reported the MD of CA125 between cases with and without pleural effusion. The between-study heterogeneity was significantly high (I 2  = 87.0%, p  < 0.01), and a meta-analysis using the random effects model was performed. The results yielded increased CA125 levels in patients with pleural effusion compared to cases without pleural effusion (MD: 74.62 U/mL, 95% CI 17.54 to 131.71; p  = 0.02) (See Supplemental Fig. 29). The leave-one-out analysis indicated that removing the Ma et al. study [ 29 ] could significantly reduce the heterogeneity of included studies to moderate (I 2  = 41.3%), while omitting any other article did not significantly change the pooled result (See Supplemental Fig. 51). Supplemental Table 4 reveals the findings of all other available subgroup analyses (See Supplemental Figs. 30–32). Meta-analysis of 4 articles [ 15 , 28 , 29 , 33 ] investigating the association of CA125 levels with peripheral edema did not show a significant MD between cases with and without peripheral edema (MD: 47.06 U/mL, 95% CI −26.93 to 121.07; p  = 0.13), while included studies were heterogeneous (I 2  = 95.4%) (See Supplemental Fig. 33). The sensitivity analysis indicated that the mean CA125 levels, after excluding the Ma et al. [ 29 ] study, were significantly higher in cases with peripheral edema than others (MD: 19.35 U/mL, 95% CI 4.41 to 34.29; p  = 0.03) with low between-study heterogeneity (I 2  = 0.0%) (See Supplemental Fig. 52). The results of subgroup analyses are represented in Supplemental Table 4 (See Supplemental Figs. 34 and 35). These results suggest that CA125 concentration was significantly elevated in cases with higher NYHA classes or pleural effusion. In contrast, CA125 levels did not significantly differ between patients with and without peripheral edema. A significant direct correlation was found between CA125 and pro-BNP levels ( r  = 0.42; 95% CI 0.30 to 0.54; p  < 0.01; I 2  = 96.9%; n  = 13 [ 10 , 13 , 14 , 17 , 18 ,  25 – 27 ,  29 , 32 , 34 ] (Fig.  3 A). The sensitivity analysis using the leave-one-out method represented non-significant changes in the pooled estimate result by removing any included article (See Supplemental Fig. 53). The correlation remained significant in both AHF ( r  = 0.39; 95% CI 0.01 to 0.67; p  = 0.04; I 2  = 98.3%; n  = 4) and CHF ( r  = 0.43; 95% CI 0.28 to 0.55; p  < 0.01; I 2  = 94.6%; n  = 9) subgroup analyses (See Supplemental Fig. 36). The findings of other subgroup analyses are provided in Supplemental Table 5 (See Supplemental Figs. 37–40). A meta-analysis of 6 articles [ 11 , 17 , 22 , 24 , 30 , 31 ] investigating the correlation between CA125 level and LVEF percentage showed a non-significant inverse correlation ( r = −0.12; 95% CI −0.28 to 0.05; p  = 0.13; I 2  = 63.1%) (Fig.  3 B). The sensitivity analysis demonstrated that omitting Mansour et al. study could significantly change the pooled correlation ( r = −0.15; 95% CI −0.26 to −0.04; p  = 0.01) and reduce the heterogeneity (I 2  = 34.2%) (See Supplemental Fig. 54). The observed change is likely explained by the smaller sample size of cases with available LVEF in the Mansour et al. ( n  = 39) study [ 17 ] compared to other articles. Supplemental Table 5 indicates the results of the available subgroup analyses (See Supplemental Figs. 41–43). Fig. 3 Forest plot representing meta-analysis of correlation between CA125 level and A pro-BNP level, B LVEF percentage Forest plot representing meta-analysis of correlation between CA125 level and A pro-BNP level, B LVEF percentage

In this meta-analysis, we showed that CA125 is a valuable prognostic biomarker in heart failure, associated with increased mortality, hospitalization risk, and clinical severity. It is also correlated with NT-proBNP and can complement it in assessing systemic congestion and right-sided HF. Given its low cost and availability, CA125 may be helpful in monitoring and directing HF therapy. However, variability in assay methods and cut-off values, as well as lack of prospective data, currently limit its clinical application. Standardization and further validation, especially for individuals with AHF, are required before routine implementation.

Our comprehensive systematic review and meta-analysis, which included 29 studies and over 20,000 HF patients, highlights CA125 as a reliable, non-invasive biomarker with both prognostic and clinical value in HF management. Our results consistently showed that elevated CA125 levels are associated with an increase in all-cause and cardiovascular mortality, hospitalization rates, and symptom burden (e.g., NYHA class III/IV and pleural effusion). Moreover, CA125 levels were positively correlated with NT-proBNP and showed potential value in monitoring fluid overload. The association between higher CA125 and endpoint events (death or HF-related hospitalization) was significant across subgroups, including AHF or CHF, different age groups, and follow-up periods. However, these results should be interpreted cautiously due to heterogeneity and limited data in some subgroups. Our results align with and extend previous findings. In 2018, Li et al. conducted a meta-analysis of 16 studies and 8401 patients with AHF. They found significantly higher all-cause mortality (HR 1.68), more HF-related readmissions (HR 1.77), and more severe fluid overload in AHF patients with high CA125 levels [ 39 ]. In another meta-analysis published in 2014, Zhuang et al. assessed 23 studies that included both AHF and CHF patients, demonstrating that CA125 levels correlated significantly with echocardiographic parameters, pleural effusion, and BNP/NT-proBNP levels. They further observed elevated CA125 levels in patients with higher NYHA class HF, which aligns with our findings regarding its application in diagnosing, stratifying, and prognosticating HF patients [ 40 ]. Complementary to these studies, our study included both CHF and AHF patients and analyzed the association of CA125 with peripheral edema, LVEF, and pro-BNP for the first time, providing a comprehensive and updated synthesis. CA125 is primarily produced by mesothelial cells in the pleura, pericardium, and peritoneum. In HF, elevated hydrostatic pressure, especially on the right side of the heart, leads to fluid overload and distension of serosal cavities, stimulating CA125 production through mesothelial shear stress [ 6 , 41 , 42 ]. Moreover, studies have reported that pericardial stretch due to atrial dilation can also increase CA125 levels. Our findings of a significant association between CA125 and pleural effusion, rather than peripheral edema, support the mentioned mechanism. Another explanation is that the inflammatory state in HF, induced by cytokines such as IL-6, IL-10, TNF-α, and IL-1, leads to higher CA125 expression [ 43 , 44 ]. These cytokines are correlated with the severity of HF [ 30 ]. Furthermore, venous congestion may facilitate the transfer of endotoxin into the circulation, sustaining a pro-inflammatory state that elevates CA125 levels [ 45 , 46 ]. Additionally, CA125 has been associated with cardiac remodeling through its effects on the extracellular matrix, suggesting it could even contribute to HF pathogenesis [ 5 , 42 , 47 ]. Together, these explanations support the role of CA125 as an available marker of systemic congestion, serosal involvement, and inflammatory burden in HF. Our analysis, consistent with previous studies, showed a moderate but significant correlation between CA125 and NT-proBNP ( r  = 0.42) [ 40 ]. Compared to this conventional biomarker, CA125 has several advantages in assessing HF, particularly in patients with systemic congestion and right-sided involvement [ 46 ]. However, our results showed an overall non-significant association between CA125 and LVEF, supporting the hypothesis that CA125 is more strongly associated with chronic fluid overload, serosal effusions, tricuspid regurgitation [ 18 , 48 ], IVC pressure, and signs of right heart dysfunction. At the same time, NT-proBNP is primarily associated with acute hemodynamic stress in the left heart [ 46 , 49 ]. Additionally, CA125 has a longer half-life, making it a more stable biomarker for chronic HF, and adding complementary information to the initial therapy response shown by BNP [ 32 ]. Several studies have evaluated the diagnostic accuracy of CA125 in predicting mortality and hospitalization compared to NT-proBNP. Findings indicate that CA125 may offer marginally superior accuracy over NT-proBNP in these predictive outcomes [ 12 , 20 , 32 ]. Some articles identified CA125 as the only independent predictive factor, while NT-proBNP was not significantly associated with endpoint outcomes [ 13 , 18 , 31 ]. Furthermore, all studies that reported results of risk stratification in the HF population using both NT-proBNP and CA125 showed that elevated CA125 levels were significantly associated with worse outcomes in subjects with normal NT-proBNP levels [ 12 , 23 , 24 , 31 , 50 ]. CA125 is also less affected by confounders such as age and renal function, making it a more reliable marker in older patients and those with renal impairment [ 18 , 20 ]. In a study involving 4595 patients with AHF by de la Espriella et al., the median CA125 concentration remained relatively consistent across the different glomerular filtration rates (GFR) categories. In contrast, NT-proBNP levels increased significantly in the lower GFR subclasses [ 51 ]. Our findings from the meta-analysis of the correlation between NT-proBNP and CA125 ( r  = 0.42) and a systematic literature review suggested that CA125 could be a complementary factor in HF populations, rather than just a surrogate marker of NT-proBNP. The combination of CA125 with NT-proBNP may provide a more comprehensive assessment of patient status, better prognostication, and risk stratification. Its wide availability and lower cost further support its integration into clinical practice [ 46 ]. On the downside, CA125 lacks disease specificity, as its levels may also be elevated in a range of non-cardiac conditions [ 42 , 52 ]. In cirrhosis, reduced clearance of CA125 due to impaired liver function results in markedly high CA125 levels (usually exceeding 200 U/mL) [ 53 ]. Similarly, studies on ovarian malignancies, especially serous carcinoma, and some benign ovarian conditions, such as endometriosis, have shown significantly increased CA125 concentrations [ 54 , 55 ]. In addition, a study on 1157 premenopausal women revealed that cases with non-ovarian cancer have 45% higher CA125 levels compared to others [ 56 ]. The elevated baseline concentrations of CA125 may negatively impact the accuracy of predicting HF outcomes and risk stratification in such populations. As a result, nearly all studies included in our meta-analysis excluded patients with known malignancy, chronic liver disease, or other major confounding conditions. Therefore, the current study was unable to assess the role of CA125 in these populations, and further research is needed to clarify its utility in these clinical conditions. On the other hand, the CA125 concentration, similar to NT-proBNP levels, may be reduced by obesity and high adipose tissue [ 57 ]. Several studies indicated lower body mass index (BMI) in patients with HF who had higher CA125 levels [ 19 , 27 , 31 , 58 ]. The observed difference may be explained by more plasma volumes, which could cause the hemodilution of soluble tumor markers, such as CA125 [ 59 ]. Additionally, smoking via accelerating CA125 clearance could falsely reduce the CA125 level [ 56 ]. Therefore, we recommend considering lower CA125 cut-off values in future studies involving HF patients with obesity or a history of smoking. Despite the mentioned variations, the majority of included studies relied on the conventional cut-off value of 35 U/mL, originally derived from oncological research. We recommend that future studies use population-specific median values or perform receiver operating characteristic (ROC) curve analysis, rather than employing a fixed cut-off. This approach is particularly important for predicting outcomes, such as mortality or HF hospitalization, where the CA125 level differences may be less significant compared to stratifications based on clinical severity, such as NYHA functional class. Additionally, several studies support the utility of serial CA125 measurements. Docherty et al. demonstrated that higher CA125 levels at 12 months compared to baseline were significantly associated with worse outcomes [ 14 ]. A study by Lourenco et al. compared CA125 levels at hospital admission and discharge in an AHF population and showed that cases whose CA125 levels did not decline during a hospital stay longer than 10 days had a significantly higher mortality rate during one-year follow-up, compared to those with decreasing values [ 16 ]. These findings support the integration of dynamic CA125 measurement, at baseline, discharge, and follow-up (e.g., at 12 months), into clinical evaluation, as this may better reflect congestion status, treatment response, and long-term prognosis than a single static cut-off. Although our results remained consistent in most subgroups, our analyses revealed variations in the prognostic strength of CA125 in some of those. In terms of HF type, CA125 had prognostic significance in both AHF and CHF. However, the robustness of some subgroup findings, particularly for cases with AHF, was limited due to high heterogeneity and a small number of studies. The higher heterogeneity in AHF studies may be the result of a higher inflammatory burden and variable timing of biomarker measurement in acute settings, which is often limited to a single baseline value obtained at admission, before stabilization or full decongestion. In contrast, CHF studies typically included more standardized, outpatient populations, allowing CA125 to reflect more stable and sustained congestion states. CA125 was significantly associated with NYHA class III/IV overall as well as in the subgroup of studies with a mean patient age over 70 years, but not in those with an average age under 70. This may reflect lower disease burden, milder congestion, or smaller sample sizes in the younger subgroup. Hospitalization outcomes were significant in the overall analysis but inconsistent across subgroups. This could be because of the subjective nature of hospitalization decisions, which vary based on the healthcare system, physician judgment, and patient comorbidities. Moreover, the limited number of studies in this subgroup analysis may have reduced the power to detect significant associations. These variations in the study populations, including disease subtypes (AHF vs. CHF), average age, and disease severity, could explain the heterogeneity observed between the included articles. Furthermore, variability in the CA125 cutoff values across studies, ranging from 14 U/mL to 65.7 U/mL, potentially contributed to the high observed heterogeneity. In particular, the subgroup of studies with a similar cut-off (35 U/mL) was homogeneous in the all-cause mortality analysis, while studies in another subgroup remained significantly heterogeneous. This finding supported the role of the variability of cut-offs in heterogeneity. Our sensitivity analysis revealed that excluding specific studies reduced heterogeneity. In the correlation analysis between CA125 and LVEF, removing the study by Mansour et al. [ 17 ] reduced heterogeneity and made the results significant, which may have resulted from including only African American patients with different HF characteristics, and a higher prevalence of HF with preserved LVEF [ 60 ]. Moreover, the observed effect of the Ma et al. study [ 29 ] on pleural effusion and peripheral edema analyses may be attributed to the inclusion of only patients with HF aged over 85 years. CA125 can be a valuable biomarker for guiding HF therapy. Adjusting diuretic dosage to reduce fluid overload without impairing renal function is one of the main challenges in HF treatment [ 46 , 61 ]. The CHANCE-HF trial showed the efficacy of CA125-guided therapy in improving short-term kidney function and reducing the rate of HF readmissions. In this approach, AHF cases were evaluated using clinical, echocardiographic, and laboratory tests, including CA125, at scheduled visits within the first 30 days after discharge, as well as at 1 month and 12 months post-discharge. They reduced the loop diuretic dose (LDD) for subjects with reduced CA125 levels below the normal level (35 U/mL), especially if they were treated with a furosemide equivalent dose (FED) greater than 120 mg per day. If CA125 levels decline but remain above 35 U/mL, the existing LDD was either maintained or, in cases where the FED was low (< 80 mg/day), the dose could be increased. Further, if CA125 levels show an increasing trend, they suggested higher LDD or additional diuretic agents and more frequent follow-up [ 62 – 64 ]. In contrast, NT-proBNP-guided management in HF patients did not yield improved outcomes compared to conventional therapy [ 65 ]. CA125’s quick kinetics, low cost, and accessibility also make it a potential biomarker for monitoring response to treatments such as diuretic therapy, heart transplantation, or peritoneal dialysis [ 5 , 66 – 68 ]. Overall, CA125 complements conventional biomarkers and may improve risk classification, diuretic titration, discharge readiness, and outpatient follow-up frequency [ 62 , 68 ]. However, prospective trials are required to validate CA125-guided management methods before their widespread use.

Several limitations should be considered. First, the observational design of most included studies limits causal interpretation. Second, there was considerable heterogeneity across most of our analyses, which could be due to variations in HF definitions, CA125 assays, and follow-up durations. Although we performed sensitivity and subgroup analyses to address this, residual confounding is still likely. Subgroup analyses involving cases with AHF or populations with an average age below 70 years were constrained by a limited number of studies and substantial heterogeneity, which may reduce the reliability of these findings and warrant further investigations focusing specifically on AHF and younger populations. Additionally, most studies were conducted in Europe and East Asia, which may limit the generalizability of our findings to other regions, including the North American region or Middle- and low-income countries. The geographical and racial variations are particularly important, as some studies on healthy adults have shown differences in baseline CA125 levels among individuals of different races/ethnicities [ 69 ]. Furthermore, an article by Lombardi et al. showed significant differences in HF characteristics and treatment outcomes across different geographic areas of Europe, supporting the potential variation in the role of CA125 in HF management and the need for further investigation in various populations [ 70 ]. Future research should prioritize standardizing assay methodologies, establishing dynamic and clinically context-specific CA125 cut-off values, including more diverse patient populations, and designing well-powered prospective trials to validate the efficacy of CA125-guided therapeutic strategies.

Heart failure (HF) is a clinical syndrome originating from any structural or functional heart condition affecting its function to fill with or pump blood efficiently. It is a major healthcare issue affecting over 64 million individuals globally. It accounts for 1–2% of all hospitalizations in Western countries and is associated with significant use of resources and healthcare costs. HF is characterized by specific signs, such as edema and rales, and symptoms, including dyspnea and fatigue. In most patients with HF, symptoms are caused by compromised myocardial function of the left ventricle [ 1 , 2 ]. Recently, various biomarkers have been examined to represent different processes in HF, and several of them have been proposed to assist in diagnosing and predicting outcomes in HF [ 3 , 4 ]. The carbohydrate antigen 125 serum (CA125) is a glycoprotein of high molecular weight that is generated by serous epithelium [ 5 , 6 ]. While CA125 is a commonly used marker for ovarian cancer, its serum concentrations can also be elevated in normal physiological states as well as in various non-cancerous medical conditions [ 7 , 8 ]. Furthermore, levels of CA125 are associated with volume overload, and CA125 may serve as a possible indicator of congestion in individuals with HF [ 9 ]. Recent investigations have revealed that elevated levels of CA125 are correlated with invasively measured filling pressures and established biomarkers of congestion, such as N-terminal pro–B–type natriuretic peptide (NT-proBNP). Additionally, these elevated concentrations serve as independent predictors of outcomes in patients hospitalized with worsening HF [ 9 – 11 ]. Several studies have shown a correlation between CA 125 serum levels and the presence of HF in patients. However, there is a lack of comprehensive meta-analyses synthesizing the prognostic value of CA125 across various heart failure populations and outcomes. This gap limits the integration of CA125 into clinical practice guidelines. To better understand this topic, we aimed to perform a systematic review and meta-analysis of existing studies to explore the contributory role of CA 125 in clinical applications.

Supplementary Material 1 Supplementary Material 1