Carbohydrate antigen 125 (CA125) following acute myocardial infarction: effects of empagliflozin and association with heart failure readouts in the EMMY trial

Abstract

Carbohydrate antigen 125 (CA125) is increasingly recognized as a biomarker in heart failure (HF). However, its response to sodium-glucose co-transporter 2 inhibitors (SGLT2i) after acute myocardial infarction (AMI) remains insufficiently explored. In this study, CA125 levels were measured from plasma samples collected in the EMMY trial, and robust linear mixed-effects models (R-LMEMs) were applied to assess the effects of empagliflozin on CA125 levels, their changes over a 26-week period, and their associations with established HF biomarkers. Our analysis showed that empagliflozin had no statistically significant effect on log-transformed CA125 levels. CA125 exhibited minor fluctuations at 6 weeks before declining at 26 weeks to a level below baseline, which had been measured shortly after AMI. Log-transformed CA125 was significantly associated with log-transformed N-terminal pro-B-type natriuretic peptide (NT-proBNP) and multiple echocardiographic parameters, including left ventricular ejection fraction (LVEF), end-diastolic and end-systolic dimensions, and volumes (LVEDD, LVEDV, LVESD, LVESV), and the ratio of early diastolic transmitral inflow velocity to early diastolic mitral annular velocity (E/e′). Notably, CA125’s associations with NT-proBNP and E/e′ were modified by empagliflozin. In conclusion, although CA125 showed significant associations with established HF biomarkers, it did not exhibit a parallel response to empagliflozin after AMI.

Introduction

Carbohydrate antigen 125 (CA125) is a transmembrane glycoprotein that is traditionally regarded as a marker for ovarian cancers, but it also increases in heart failure (HF), potentially in response to inflammatory and hemodynamic stimuli^1–3. Although not specific for HF, measuring CA125 could be beneficial for the identification of high-risk patients and for the guidance of therapy^4,5. In contrast to the N-terminal prohormone of brain natriuretic peptide (NT-proBNP), CA125 is not affected by age or renal impairment, which may be advantageous as a potential biomarker of HF in elderly patients and patients with renal insufficiency⁶.

CA125 levels are elevated following an acute myocardial infarction (AMI)⁷. Along with its role as a biomarker for acute HF following AMI^8,9, CA125 has shown strong potential as a marker for pulmonary congestion¹⁰. Additionally, CA125 holds promise for early risk stratification of patients with AMI as it can early predict the subsequent major cardiac events and mortality¹¹. Moreover, CA125 measured at the early presentation of an acute coronary syndrome is independently associated with cardiac structure and functional parameters after one year¹². In an analysis including 1,099 patients, CA125 ranked among the top three biomarkers for predicting outcomes of AMI, alongside the well-established markers NT-proBNP and growth differentiation factor 15 (GDF-15)¹³. Despite the evidence highlighting its importance, CA125 remains largely understudied in the context of AMI.

Sodium-glucose co-transporter 2 inhibitors (SGLT2i) are a central therapeutic pillar of HF therapy irrespective of left ventricular ejection fraction (LVEF)^14,15. The EMMY trial (EMpagliflozin in Acute Myocardial infarction) demonstrated that the SGLT2i, empagliflozin, improves NT-proBNP levels and functional as well as structural echocardiographic HF parameters in patients with recent AMI¹⁶. To the best of our knowledge, the impact of SGLT2i on CA125 levels following AMI has yet to be investigated.

In the present study, we evaluated the response of CA125 to empagliflozin, its serum level changes, and its associations with other heart failure markers in post-AMI patients from the EMMY trial cohort.

Results

The present analysis included 418 patients, of whom 207 received empagliflozin and 211 received placebo. No statistically significant differences were observed between the two groups with regard to the baseline characteristics, as detailed in Table 1. The measurements of the HF readouts, including CA125, NT-proBNP and a selection of echocardiographic parameters: LVEF, left ventricular end-diastolic diameter (LVEDD), left ventricular end-systolic diameter (LVESD), left ventricular end-diastolic volume (LVEDV), left ventricular end-systolic volume (LVESV), and the ratio of early diastolic transmitral inflow velocity to early diastolic mitral annular velocity (E/e′ ratio) are summarized in Table 2.

Table 1.

Participant baseline characteristics.

Characteristic	Overall N = 4181	Empagliflozin N = 2071	Placebo N = 2111	p-value2
Sex				0.930
Male	348 (83%)	172 (83%)	176 (83%)
Female	70 (17%)	35 (17%)	35 (17%)
Age	57 (52, 64)	57 (52, 64)	57 (51, 64)	0.767
Type 2 Diabetes, n (%)	51 (12%)	29 (14%)	22 (10%)	0.263
Active or former smoker	304 (73%)	152 (73%)	152 (73%)	0.872
Systolic blood pressure [mmHg]	125 (116, 131)	125 (116, 131)	125 (117, 131)	0.392
Diastolic blood pressure [mmHg]	78 (74, 85)	78 (74, 85)	78 (75, 85)	0.800
Pulse [beats/min]	72 (65, 80)	72 (65, 79)	72 (66, 81)	0.518
Body weight [kg]	85 (75, 94)	85 (75, 95)	84 (75, 94)	0.322
History of Carcinoma	19 (4.5%)	8 (3.9%)	11 (5.2%)	0.508
History of stroke	6 (1.4%)	5 (2.4%)	1 (0.5%)	0.119
Peripheral arterial disease	6 (1.4%)	5 (2.4%)	1 (0.5%)	0.119
Arterial Hypertension	182 (44%)	83 (40%)	99 (47%)	0.160
Coronary artery disease	43 (10%)	21 (10%)	22 (10%)	0.925
History of NSTEMI	20 (4.8%)	12 (5.8%)	8 (3.8%)	0.337
Depression	22 (5.3%)	13 (6.3%)	9 (4.3%)	0.356
Treatment
ACE/ARB	404 (98%)	200 (98%)	204 (98%)	0.750
ARNI	9 (2.2%)	2 (1.0%)	7 (3.3%)	0.175
Beta blocker	403 (97%)	196 (95%)	207 (98%)	0.094
Mineralocorticoid receptor antagonists	164 (39%)	77 (37%)	87 (41%)	0.421
Loop diuretics	45 (11%)	24 (12%)	21 (10.0%)	0.576
Digoxin	1 (0.2%)	0 (0%)	1 (0.5%)	> 0.999
Statin	409 (98%)	200 (97%)	209 (99%)	0.171
Ezetimibe	46 (11%)	23 (11%)	23 (11%)	0.931
Other lipid therapy	4 (1.0%)	2 (1.0%)	2 (0.9%)	> 0.999
Thiazide	17 (4.1%)	7 (3.4%)	10 (4.7%)	0.482
Calcium channel blocker	18 (4.3%)	7 (3.4%)	11 (5.2%)	0.356
Platelet aggregation inhibitors	416 (100%)	207 (100%)	211 (100%)	0.245
Anticoagulation therapy	32 (7.7%)	14 (6.8%)	18 (8.5%)	0.497
Metformin	36 (8.6%)	18 (8.7%)	18 (8.5%)	0.952
DPP-4 inhibitor	11 (2.6%)	6 (2.9%)	5 (2.4%)	0.736
Insulin	11 (2.6%)	5 (2.4%)	6 (2.8%)	0.785
Sulfonylurea	3 (0.7%)	2 (1.0%)	1 (0.5%)	0.621
GLP-1 RA	4 (1.0%)	2 (1.0%)	2 (0.9%)	> 0.999
Laboratory parameters
HbA1c [%]	5.60 (5.40, 6.00)	5.60 (5.40, 6.00)	5.70 (5.40, 6.00)	0.669
HbA1c [mmol/L]	38 (36, 42)	38 (36, 42)	39 (36, 42)	0.653
Creatine kinase [U/L]	1,690 (1,222, 2,457)	1,677 (1,154, 2,478)	1,709 (1,256, 2,456)	0.726
Creatine kinase MB [U/L]	158 (90, 236)	151 (91, 228)	164 (89, 241)	0.589
Troponin [µg/L]	3,597 (2,115, 5,691)	3,603 (2,027, 5,633)	3,548 (2,161, 5,957)	0.719
Aspartate Aminotransferase [U/L]	205 (124, 319)	206 (136, 320)	205 (120, 319)	0.583
Alanine Aminotransferase [U/L]	49 (37, 74)	49 (37, 76)	49 (38, 70)	0.755
Gamma-glutamyltransferase [U/L]	31 (21, 50)	29 (21, 50)	33 (21, 52)	0.499
C-reactive protein [mg/dL]	6 (2, 14)	5 (2, 13)	6 (2, 15)	0.347
CA125 (U/mL)	12 (9, 16)	12 (9.5, 16)	12 (9, 16)	> 0.999
Creatinine [mg/dL]	0.89 (0.79, 1.01)	0.89 (0.79, 1.01)	0.89 (0.78, 1.01)	0.809
eGFR [mL/min]	92 (79, 102)	92 (79, 101)	92 (79, 102)	0.980
Total cholesterol [mg/dL]	188 (162, 223)	188 (164, 225)	188 (161, 220)	0.689
Triglycerides [mg/dL]	126 (94, 177)	129 (98, 180)	123 (90, 176)	0.268
High-density lipoprotein [mg/dL]	43 (36, 52)	43 (36, 52)	42 (36, 52)	0.951
Low-density lipoprotein [mg/dL]	119 (92, 149)	118 (97, 150)	120 (89, 145)	0.659

¹n (%); Median (Q1, Q3)

²Pearson’s Chi-squared test; Wilcoxon rank sum test; Fisher’s exact test.

ACE, angiotensin-converting enzyme; ARB, angiotensin II receptor blockers; ARNI, angiotensin receptor-neprilysin inhibitors; CA125, carbohydrate antigen 125; DPP-4, dipeptidyl peptidase-4; eGFR, estimated glomerular filtration rate; GLP-1 – RA, glucagon-like peptide-1 receptor agonist; HbA1c, glycated hemoglobin; IQR, interquartile range; N, number of participants; NSTEMI, non-ST-elevation myocardial infarction.

Table 2.

Median and interquartile range (IQR) of heart failure readouts in the study.

	Baseline Empagliflozin Median (IQR)	Baseline Placebo Median (IQR)	6-week Empagliflozin Median (IQR)	6-week Placebo Median (IQR)	26-week Empagliflozin Median (IQR)	26-week Placebo Median (IQR)
CA125 (U/mL)	12 (9–16)	12 (9.5–16)	12 (9–17)	12 (9–16)	11 (8–15)	12 (9–16)
NT-proBNP (pg/mL)	1272.5 (773–2247.3)	1373 (754–2217)	498 (253.5–1029.5)	569 (287–1118)	202.5 (89.3–380.8)	224.0 (128–449)
LVEF (%)	48 (43–53)	49 (43–54)	52 (46–56)	51 (45–56)	53 (47–58)	52 (47–57)
LVEDD (mm)	49 (45–52.8)	49 (45–52)	50 (46.3–54.8)	51 (47–54)	50 (46–55)	50 (47–54)
LVEDV (mL)	119 (93–139)	114 (92–134)	126 (108.5–144.5)	122 (101.8–147.8)	122.0 (101.0–145.0)	120.0 (100.5–153.5)
LVESD (mm)	36 (32–40)	36 (32–40)	37 (33–40)	37 (33–41)	37 (32–41)	37 (33–42)
LVESV (mL)	61.1 (48.0–76.0)	60 (45.8–73)	62 (51–76)	60 (45–76)	60 (45.5–72.5)	59 (46–80.8)
E/e′ ratio	8.94 (7.45–10.91)	8.94 (7.54–10.74)	8.42 (6.93–10.20)	8.14 (6.75–10.51)	7.77 (6.62–9.33)	8.21 (6.93–9.88)

CA125; carbohydrate antigen 125; E/e′ ratio, ratio of early diastolic mitral inflow velocity to early diastolic mitral annular velocity; IQR, interquartile range; LVEDD, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; LVESV, left ventricular end-systolic volume; NT-proBNP, N-terminal pro B-type natriuretic peptide.

We used robust linear mixed-effects models (R-LMEM) to evaluate the impact of empagliflozin on CA125, assess the temporal dynamics of CA125, and examine its associations with other HF readouts. As shown in Fig. 1, empagliflozin had no statistically significant impact on log-transformed CA125 (p = 0.916). While the treatment and placebo groups displayed minor differences in their patterns over time, these differences were not statistically significant, as neither the treatment-visit interaction (p = 0.057) nor the pairwise comparisons reached the statistical significance. CA125 levels exhibited slight deviations from baseline at 6 weeks, before declining below baseline at week 26 (p < 0.001). The magnitude of these changes was very small, with differences of less than 1 U/mL. Baseline log-transformed CA125 values showed a significant negative interaction with the visit time point indicating that higher initial levels were associated with greater reductions over time (p < 0.001). Likewise, no statistically significant interaction was observed between treatment and LVEF (p = 0.12), indicating that baseline values of LVEF did not significantly modify the treatment effect.

Fig. 1.

CA125 levels in the placebo (green) and empagliflozin (red) groups, modeled by the robust linear mixed-effects model (R-LMEM). The model was adjusted for age, sex, BMI, smoking status, diabetes status, AMI types, and eGFR. CA125 values were log-transformed for modeling and subsequently reverse-transformed for visualization. A significant visit effect was detected (p_visit < 0.001), while treatment had no significant impact (p_Tx = 0.916). The treatment-visit interaction showed a non-significant trend (p_{Tx: visit} = 0.057). Pairwise comparisons with Holm’s adjustment revealed no significant differences between groups at any time point. Data are presented as mean ± CI.

The R-LMEM fixed-effect estimates for the associations between log-transformed CA125 and other HF markers are presented in Table 3. The association lines across the three time points are illustrated in Figs. 2 and 3. Notably, a statistically significant positive association between log-transformed CA125 and log-transformed NT-proBNP (estimate: 0.28; p < 0.001) was observed. Furthermore, a statistically significant positive association was identified between log-transformed CA125 and multiple echocardiographic parameters, including LVEDD (estimate: 0.68; p = 0.034), LVEDV (estimate: 4.03; p = 0.037), LVESD (estimate: 1.19; p = 0.003), LVESV (estimate: 5.42; p < 0.001), and the E/e′ ratio (estimate: 0.4; p = 0.013). On the other hand, a statistically significant negative association was identified between log-transformed CA125 and LVEF (estimate: −2.37; p < 0.001). These findings suggest that elevated CA125 levels are associated with increased cardiac stress and impaired systolic and diastolic function.

Table 3.

Adjusted robust linear mixed model analysis of log-transformed CA125 with cardiac markers. Overall: overall association between log-transformed CA125 and outcomes independent of visit and treatment effect. Baseline: association between log-transformed CA125 and outcomes at baseline. 6-week: association between log-transformed CA125 and outcomes at 6 weeks. 26-week: association between log-transformed CA125 and outcomes at 26 weeks. Placebo: association between CA125 and outcomes in placebo. Empagliflozin: association between CA125 and outcomes in empagliflozin. The model was adjusted for age, sex, BMI, smoking status, diabetes status, AMI types, and eGFR.

	NT-proBNP (pg/mL; log)	LVEF (%)	LVEDD (mm)	LVEDV (mL)	LVESD (mm)	LVESV (mL)	E/e′ ratio
	Coef. (95% CI)	Coef. (95% CI)	Coef. (95% CI)	Coef. (95% CI)	Coef. (95% CI)	Coef. (95% CI)	Coef. (95% CI)
Overall	0.28 (0.18, 0.38)	−2.37 (−3.30, −1.44)	0.68 (0.051, 1.31)	4.03 (0.243, 7.81)	1.19 (0.40, 1.98)	5.42 (3.00, 7.84)	0.40 (0.09, 0.71)
P-value	< 0.001	< 0.001	0.034	0.0370	0.003	< 0.001	0.013
CA125*Visit
Baseline	0.23 (0.09, 0.36)	−2.43 (−3.72, −1.15)	0.22 (−0.65, 1.09)	3.96 (−1.22, 9.15)	0.58 (−0.52, 1.68)	5.85 (2.52, 9.18)	0.40 (−0.06, 0.86)
6-week	0.30 (0.19, 0.41)	−2.13 (−3.12, −1.14)	0.83 (0.13, 1.54)	6.73 (2.61, 10.85)	1.68 (0.81, 2.56)	6.58 (4.02, 9.15)	0.60 (0.24, 0.96)
26-week	0.32 (0.18, 0.46)	−2.54 (−3.87, −1.21)	0.99 (0.08, 1.91)	1.39 (−3.94, 6.72)	1.30 (0.12, 2.49)	3.82 (0.39, 7.25)	0.19 (−0.27, 0.65)
p-value	0.451	0.808	0.260	0.142	0.172	0.270	0.257
CA125*Treatment
Placebo	0.18 (0.04, 0.33)	−1.56 (−2.95, −0.17)	0.36 (−0.58, 1.31)	2.46 (−3.09, 8.02)	0.46 (−0.75, 1.68)	4.11 (0.49, 7.73)	0.04 (−0.42, 0.51)
Empagliflozin	0.38 (0.25, 0.51)	−3.17 (−4.40, −1.95)	1.00 (0.17, 1.83)	5.59 (0.47, 10.71)	1.91 (0.89, 2.93)	6.73 (3.53, 9.93)	0.75 (0.33, 1.17)
p-value	0.055	0.088	0.322	0.417	0.072	0.286	0.025

CA125, carbohydrate antigen 125; CI, confidence interval; E/e′ ratio, ratio of early diastolic mitral inflow velocity to early diastolic mitral annular velocity; LVEDD, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; LVESV, left ventricular end-systolic volume; NTproBNP, N-terminal pro B-type natriuretic peptide.

Fig. 2.

The average difference (Coefdiff) in association slopes between log-transformed carbohydrate antigen 125 (CA125, U/mL) and log-transformed N-terminal pro-B-type natriuretic peptide (NT-proBNP, pg/mL), left ventricular ejection fraction (LVEF, %), left ventricular end-diastolic diameter (LVEDD, mm) and left ventricular end-diastolic volume (LVEDV, mL) between the empagliflozin and placebo groups. p = p-value for difference in association slopes between empagliflozin and placebo. Data are presented as mean ± CI.

Fig. 3.

The average difference (Coefdiff) in association slopes between log-transformed carbohydrate antigen 125 (CA125, U/mL) and left ventricular end-systolic diameter (LVESD, mm), left ventricular end-systolic volume (LVESV, mL) and the ratio of early diastolic mitral inflow velocity to early diastolic mitral annular velocity (E/e′ ratio), between the Empagliflozin and Placebo groups. p = p-value for difference in association slopes between Empagliflozin and Placebo. Data are presented as mean ± CI.

Empagliflozin showed a non-significant tendency to alter the overall association between log-transformed CA125 and log-transformed NT-proBNP (p = 0.055; Table 3); however, its impact at 6 weeks was statistically significant (coefficient difference: 0.29; p = 0.023), while no significant differences were observed at baseline or 26 weeks Fig. 2. In addition, empagliflozin significantly affected the overall association between log-transformed CA125 and the E/e′ ratio (p = 0.025; Table 3). When examined across the three time points, this effect was again statistically significant only at 6 weeks (coefficient difference: 0.91; p = 0.04; Fig. 3). In contrast, empagliflozin did not significantly influence the associations between CA125 and other echocardiographic parameters, including LVEF, LVEDD, LVEDV, LVESD, or LVESV Table 3; Figs. 2 and 3.

Discussion

The aim of this study was to evaluate the effect of empagliflozin therapy on CA125, the temporal dynamics of CA125, and the association between CA125 and a selection of established biomarkers of HF after AMI. The study demonstrated that within our cohort and observation period, the impact of empagliflozin on CA125 was not statistically significant. CA125 levels fluctuated slightly at 6 weeks before decreasing below baseline at 26 weeks. Log-transformed CA125 had a significant association with log-transformed NT-proBNP and various echocardiographic measures of HF, with empagliflozin modifying the association in some parameters.

In our study, the impact of empagliflozin on CA125 was not statistically significant. While SGLT2i have demonstrated clear benefits in patients with heart failure^14,15, their efficacy in post-AMI patients without preexisting heart failure or diabetes remains uncertain. Although the EMMY trial showed improvements in NT-proBNP and echocardiographic parameters¹⁶, secondary analyses found no benefit on inflammatory cytokines and even reported an increase in trimethylamine‑N‑oxide (TMAO) in response to empagliflozin^17,18. This variability in biomarker responses suggests that the effects of SGLT2i may be biomarker-specific rather than uniform. Such variability complicates the interpretation of biomarker-based studies and reinforces the importance of clinical outcomes in evaluating the benefits of SGLT2i in AMI. Interestingly, clinical trials with clinical outcomes have also produced mixed results. For example, the DAPA-MI trial reported improvements in cardiometabolic markers but no effect on cardiovascular death or hospitalization¹⁹. Likewise, the EMPACT-MI trial showed no reduction in first HF hospitalization or all-cause mortality²⁰. A secondary analysis of the same study which focused on HF outcomes revealed, however, a reduction in the risk of HF in patients with left ventricular dysfunction or congestion in response to empagliflozin²¹. Notably, CA125—which has shown promising potential as a predictor of clinical outcomes in both HF and AMI^2,11—remained unchanged in our study, aligning with the neutral findings of DAPA-MI and EMPACT-MI trials, and further contributing to the uncertainty surrounding the benefit of SGLT2i in post-AMI patients without preexisting heart failure or diabetes.

Although aligning with the neutral findings in some endpoints of the DAPA-MI and EMPACT-MI trials, the lack of effect of empagliflozin on CA125 in our study should be interpreted with caution for several reasons. First, this study represents a secondary analysis, as the primary endpoint of the EMMY study was the effect on NT-proBNP. Second, the observation period was 26 weeks only. Third, CA125 appears to be relatively insensitive to SGLT2i, as suggested by previous heart failure studies, which showed that changes in CA125 in response to SGLT2i were generally small, transient, or limited to specific subgroups. For example, in their analysis of the EMPEROR (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure)-Reduced and Preserved trials, Ferreira et al. found that empagliflozin therapy in patients with chronic HF had a small but statistically significant effect only at 12 weeks, with no notable changes observed at baseline or after 52 weeks of follow-up²². In a similar vein, Docherty et al. in their analysis of the DAPA-HF (Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure) trial observed a statistically significant reduction in CA125 in response to dapagliflozin only in patients with a history of hospitalization for HF, while this reduction was not observed in patients without a history of hospitalization or in the overall study cohort²³. It is important to note that our study investigated the effects of empagliflozin on CA125 in post-AMI patients regardless of heart failure status. Further research is needed to assess its impact specifically in the subgroup of patients with acute heart failure following AMI.

In our study, CA125 after AMI followed a distinct pattern, showing minimal, non-significant fluctuations at week 6 before dropping below baseline at week 26, whereas NT-proBNP and LVEF demonstrated steady improvement over time, beginning at week 6. In order to interpret the observed differences in trajectories between CA125 and other readouts of HF, it is important to consider that the HF readouts reflect different time windows. While echocardiogram parameters provide a real-time evaluation of cardiac function, the half-life of NT-proBNP is approximately 70 min²⁴. In contrast, the half-life of CA125 ranges between 6.3 and 14.3 days²⁵. The prolonged half-life of CA125 can explain its distinctive trajectory in comparison to other HF readouts observed in our study. It is worth noting that, although the temporal changes in CA125 were statistically significant, the mean change over time was minimal (< 1 U/mL). This lack of clear temporal variation, in contrast to the pronounced decline in NT-proBNP, underscores the limited dynamic range of CA125 and questions its utility as a follow-up biomarker in cohorts with relatively preserved myocardial function, such as the EMMY population.

The positive moderate association between log-transformed CA125 and log-transformed NT-proBNP in our study is consistent with previous observations^7,26. The relationship between CA125 and NT-proBNP is not strong in the majority of studies² indicating that the two markers are related but not necessarily interchangeable. Additionally, we observed a negative association between log-transformed CA125 and LVEF, as well as significant positive associations with LVEDD, LVEDV, LVESD, LVESV, and the E/e′ ratio, supporting the potential role of CA125 as a marker of filling pressure and congestion. The relationship between CA125 and left ventricular parameters as observed in echocardiography is not consistent across the literature³. Yet, our findings align with Yilmaz et al. (2011) and Rong et al. (2015), who reported a correlation between CA125 with LVEF and LVESD as well as elevated CA125 levels in the presence of left ventricular dilatation^26,27.

In our study, empagliflozin significantly altered the association between log-transformed CA 125 and log-transformed NT-proBNP, as well as the E/e′ ratio at 6 weeks, suggesting that empagliflozin may modulate the interplay between heart failure biomarkers over time, even without changing CA125 directly. This effect of empagliflozin underscores the impact of therapeutic intervention on the relationship between CA125 and other HF readouts—an aspect often overlooked when evaluating CA125 as a biomarker for HF.

Despite its correlation with other HF biomarkers and the growing evidence supporting CA125 as a prognostic and risk stratification marker in HF, no standardized cutoff has been established for diagnosis or follow-up. The commonly cited threshold of > 35 U/mL—originally derived from oncology—has shown prognostic value in HF^2,28, yet several studies suggest lower cutoffs. For example, levels < 23 U/mL have been linked to fewer 1-month death/HF readmission after admission with acute HF²⁹. A lower cutoff level (20.8 U/mL) was associated with increased all-cause mortality after the diagnosis of chronic HF in a retrospective study with a median follow-up period of 22.7 months³⁰. Following AMI, even lower thresholds have been proposed: Xu et al. identified 13.2 U/mL as a predictor of acute HF after STEMI⁹, and Falcão et al. reported 11.48 U/mL as predictive of all-cause mortality in STEMI patients undergoing coronary angioplasty¹¹. In our study, the median CA125 was 12 U/mL and the 75th percentile was 16 U/mL—values overlapping with those in Falcão et al. study—yet only three deaths occurred (less than 1%), compared to 18% in their cohort. Although differences in patient selection and assay platforms (Roche in Falcão et al. study vs. Abbott in our study) may partially explain this discrepancy, the substantial variation in CA125 cut-off values—and their overlap with current reference ranges—highlights the challenges of applying CA125 as a cardiac marker in clinical practice. Large, multicenter studies are still needed to define standardized thresholds across assay platforms and patient populations.

Limitations

This study has several limitations. First, the follow-up period of 26 weeks was relatively short. Second, the analysis focused on biomarkers and echocardiographic outcomes but did not include a clinical HF score. Third, patients in the EMMY trial were not stratified based on their HF status. Fourth, the study did not include patients with estimated glomerular filtration rate (eGFR) < 45 mL/min per 1.73 m².

Conclusions

Empagliflozin showed no direct benefit on CA125 levels in this secondary analysis of the EMMY trial. CA125 levels at 6 weeks did not differ significantly from those measured shortly after AMI but declined below baseline by 26 weeks; however, the overall temporal changes were modest in our cohort. CA125 was significantly associated with NT-proBNP and echocardiographic parameters of heart failure. Among these, the associations with E/e′ and NT-proBNP were modified by empagliflozin, suggesting a selective interaction with markers of diastolic function and congestion.

Methods

In this work, we conducted a secondary analysis of samples from the EMMY trial. As previously described^16,31, the EMMY trial was a multicenter, randomized, double-blind study that evaluated the effect of empagliflozin (10 mg once daily for 26 weeks) in patients with and without type 2 diabetes after confirmed AMI. Blood samples and echocardiograms were collected from all study patients at baseline (within 72 h after AMI), at 6 weeks, and at 26 weeks.

The main criteria for study enrollment were confirmed AMI, age between 18 and 80 years, eGFR > 45 mL/min per 1.73 m², hemodynamic stability with a blood pressure > 110/70 mmHg, and initiation of empagliflozin within a maximum of 72 h following AMI. Exclusion criteria were: any other form of diabetes mellitus other than type 2 diabetes mellitus, history of diabetic ketoacidosis, blood pH < 7.32, allergy to SGLT2i, hemodynamic instability, episodes of hypoglycemia in the last six months preceding the study, females of childbearing potential, patients with acute urinary or genital infections, and individuals who had previously been treated with any SGLT2i³¹.

CA125 was quantified from frozen serum samples using a chemiluminescent microparticle immunoassay (Alinity i CA125 II, Abbott GmbH, Vienna, Austria; manufactured by Fujirebio Diagnostics, Pennsylvania, USA for Abbott) at the Clinical Institute of Medical and Chemical Laboratory Diagnostics, Medical University of Graz. Measurements were performed on the Alinity i automated immunoassay analyzer (Abbott). According to the manufacturer’s specifications, the assay has a linear quantification range from 1.1 to 1000 U/mL, with a within-laboratory coefficient of variation (%CV) of 4.5% at a serum level of 5.4 U/mL and 2.4% at a serum level of 891.1 U/mL. Prior to measurement, samples underwent one freeze-thaw cycle. CA125 is reported to remain stable through up to eight freeze-thaw cycles³². All samples were obtained from the same blood draw used for NT-proBNP measurement, and echocardiography was performed on the same day, ensuring that all biomarkers reflect the same time point.

Outcomes of the study

In this secondary analysis, the main outcome of interest was the differential time course of CA125 levels between treatment groups across baseline, 6 weeks, and 26 weeks, modeled using R-LMEM.

Additional outcomes were the associations between CA125 and established HF markers: NT-proBNP and echocardiographic parameters including LVEF, LVEDD, LVEDV, LVESD, LVESV, and the E/e′ ratio.

Statistical analysis

Baseline measurements were presented as mean ± standard deviation (SD) or median with interquartile range (IQR) for continuous variables, and as frequencies with percentages (%) for categorical variables.

An R-LMEM was fitted to evaluate the dynamics of log-transformed CA125 following AMI in relation to treatment. Fixed effects included treatment group (empagliflozin vs. placebo), visit time points (weeks 6 and 26), baseline log-transformed CA125, along with interaction terms for treatment × time and baseline CA125 × time. The model was adjusted for age, sex, body mass index (BMI), smoking status, diabetes status, AMI type (ST-elevation myocardial infarction [STEMI] vs. non–ST-elevation myocardial infarction [non-STEMI]), and eGFR. Furthermore, pairwise comparisons with Holm’s adjustment were conducted to evaluate the effect of empagliflozin on log-transformed CA125 within each visit. To explore whether treatment effects varied by baseline LVEF, an additional model was fitted to include the interactions between treatment and baseline LVEF.

To explore the association between CA125 and classical HF markers, as well as the effect of empagliflozin on this relationship, a set of R-LMEMs was fitted using classical markers as outcome variables, including: log-transformed NT-proBNP, LVEF, LVEDD, LVEDV, LVESD, LVESV, and the E/e′ ratio. The explanatory fixed effects variables were: log-transformed CA125, treatment (empagliflozin vs. placebo), visit time points (6 weeks and 26 weeks), and the interaction between treatment, log-transformed CA125, and time points. The models were adjusted for age, sex, body mass index (BMI), smoking status, diabetes status, AMI type, and eGFR. The fixed explanatory variables were: log-transformed CA125, treatment group (empagliflozin vs. placebo), visit time points (weeks 6 and 26), and the three-way interaction between treatment, log-transformed CA125, and time.

Model fits were assessed using residual plots and robustness weights. Smoothed Huber Rho functions were used to reduce the influence of outliers. The statistical significance of interaction effects was determined at p ≤ 0.05 and adjusted for multiple comparisons over time using Holm’s method. All statistical analyses were performed using R (version 4.4.2).

Author contributions

HS and DvL are the principal investigators of the EMMY trial. DvL, HS, and MH planned and designed the current study. AMH measured CA125 and drafted the initial manuscript. FA performed the statistical analysis and prepared the figures and tables. NJT coordinated the EMMY trial and collected the clinical data. All authors reviewed the manuscript and approved the final version.

Funding

The EMMY study was funded by an unrestricted grant from Boehringer Ingelheim (no. 1245.151). HS’s research is supported by the Austrian Science Fund (FWF grant KLI-1076, PIN8074224) and the Horizon Europe project PoCCardio (grant number 101095432). DvL’s research is supported by the Austrian Science Fund (FWF grant KLIF-1155-B).

Data availability

Data are available upon reasonable request to Harald Sourij.

Declarations

Competing interests

HS is on the advisory board and speaker’s bureau of Boehringer Ingelheim, Novo Nordisk, Sanofi-Aventis, Amgen, AstraZeneca, Bayer, Eli Lilly, Kapsch, MSD, and Daiichi Sankyo. DvL is on the advisory board and speaker’s bureau of Boehringer Ingelheim, Novartis, Sanova, Sanofi, Orion, AstraZeneca, Bayer Recardio, Vaxxinity and Daiichi Sankyo. All other authors report no conflicts of interest related to this study.

Ethics approval and consent to participate

The EMMY trial was approved by the Ethics Committee of the Medical University of Graz, Austria (EK 29–179 ex16/17, EudraCT 2016-004591-22) and registered at ClinicalTrials.gov (NCT03087773). The trial methodology was in accordance with the 1964 Declaration of Helsinki and in compliance with the Good Clinical Practice Guidelines (ICH GCP E6). Written informed consent was obtained from all participants prior to inclusion in the study.