Combined prediction of anthracycline-induced cardiotoxicity in patients with breast cancer using neutrophil-to-lymphocyte ratio and left atrial reservoir strain: A prospective observational study

Abstract

Objective

Anthracycline-related cardiotoxicity may involve systemic inflammatory responses, and the neutrophil-to-lymphocyte ratio may change during chemotherapy as a marker for inflammatory and immune status. Considering this, we designed the present study to assess the predictive value of neutrophil-to-lymphocyte ratio combined with left atrial reservoir longitudinal strain for subclinical cancer therapy–related cardiac dysfunction during anthracycline chemotherapy in patients with breast cancer.

Methods

This prospective observational study enrolled 85 patients with breast cancer undergoing anthracycline chemotherapy. Neutrophil-to-lymphocyte ratio and left atrial reservoir longitudinal strain were measured before chemotherapy (T0), after the second cycle (T2), and after the fourth cycle (T4). Subclinical cancer therapy–related cardiac dysfunction was defined as a normal left ventricular ejection fraction with a ≥15% decrease in left ventricular global longitudinal strain compared with that at baseline. The predictive efficacy of the relative decrease in left atrial reservoir longitudinal strain, T2 neutrophil-to-lymphocyte ratio, and their combined model for subclinical cancer therapy–related cardiac dysfunction at T4 was evaluated using receiver operating characteristic curves.

Results

We found that 29.41% of the patients developed subclinical cancer therapy–related cardiac dysfunction. Multivariate analysis showed that both T2 neutrophil-to-lymphocyte ratio (odds ratio = 1.611, p = 0.002) and relative decrease in left atrial reservoir longitudinal strain (odds ratio = 1.426, p < 0.001) were independent predictors. Receiver operating characteristic analysis revealed an area under the curve of 0.896 for the combined model.

Conclusions

This study provides preliminary evidence for an association of T2 neutrophil-to-lymphocyte ratio and relative decrease in left atrial reservoir longitudinal strain with subclinical cancer therapy–related cardiac dysfunction during anthracycline chemotherapy in patients with breast cancer. In this cohort, their combined use suggested good predictive value.

Introduction

Breast cancer (BC) is one of the most common malignant tumors among women worldwide. ¹ With continuous advancements in early screening and treatment methods, the survival rate of BC patients has significantly improved. ² However, chemotherapy-induced cardiotoxicity, particularly cancer therapy–related cardiac dysfunction (CTRCD), remains a significant clinical challenge during the treatment process.^3,4 Anthracycline drugs, commonly used as first-line chemotherapy agents for BC, have markedly improved patient survival rates; however, they are also associated with a high risk of cardiotoxicity, especially CTRCD. ⁵

Currently, left ventricular ejection fraction (LVEF) is the most commonly used clinical indicator of monitoring cardiac function. However, LVEF changes are typically delayed, making it inadequate for the early detection of subclinical cardiac damage. ⁶ Therefore, a more sensitive and earlier monitoring tool is required to enable timely intervention in asymptomatic patients with cardiac damage.

In recent years, left atrial reservoir longitudinal strain (LASr) has emerged as a novel cardiac function assessment tool, capable of detecting subtle changes in cardiac function earlier than conventional measures. It is gradually becoming an important tool for early diagnosis of heart disease.^7,8 Meanwhile, the neutrophil-to-lymphocyte ratio (NLR), a simple and cost-effective inflammatory marker, has been shown to be closely associated with the prognosis of various solid tumors and chemotherapy toxicity. However, its application in the early detection of CTRCD has not been sufficiently explored.^9,10

Thus, this study aimed to assess the predictive capability of NLR combined with LASr for subclinical CTRCD during anthracycline chemotherapy in patients with BC. Through multivariate logistic regression and receiver operating characteristic (ROC) curve analyses, this study explored the value of their combined model in the early identification of CTRCD, providing a new and effective tool for cardiac function monitoring in patients with BC.

Materials and methods

Study participants

Study population

In total, 85 consecutive patients with BC who met the eligibility criteria and underwent anthracycline chemotherapy at the Department of Breast Surgery, First Affiliated Hospital of Jinzhou Medical University, between January 2024 and December 2025, were included in this study. All patients were administered a standardized chemotherapy regimen based on anthracycline drugs, specifically the doxorubicin plus cyclophosphamide (AC) protocol. Doxorubicin (60 mg/m²) and cyclophosphamide (600 mg/m²) were administered via intravenous infusion on day 1. The treatment cycle lasted for 21 days, and patients were administered at least 4 consecutive cycles.

AC–docetaxel (T)/paclitaxel (P)

This regimen included the AC protocol as described above for the first four cycles. After completing these cycles, a sequential taxane regimen was introduced with paclitaxel (80 mg/m²), administered weekly via intravenous infusion for 12 cycles. Some patients, based on their clinical condition, received 75–100 mg/m² docetaxel every 3 weeks. The specific drug dosages and administration frequency were individually adjusted by the attending physician based on the patient’s body surface area and tolerance.

Inclusion criteria

Patients were eligible if they met all the following criteria: (a) no previous history of malignancies other than BC; (b) no prior history of cardiac disease; and (c) no clinically significant abnormalities on baseline evaluations prior to chemotherapy, including electrocardiography, transthoracic echocardiography (TTE), complete blood count, and hepatic and renal function tests. In addition, only patients with normal baseline white blood cell count (WBC), neutrophil count (NEUT), and lymphocyte count (LYM) before chemotherapy were included.

Exclusion criteria

Patients were excluded if any of the following conditions were present: (a) pre-existing cardiac disease, including congenital heart disease, significant arrhythmias, valvular heart disease, cardiomyopathy, and prior myocardial infarction; (b) major cardiovascular risk factors (e.g. hypertension, diabetes mellitus, dyslipidemia, obesity, and smoking history) or prior exposure to radiotherapy and/or chemotherapy; (c) suboptimal echocardiographic image quality precluding reliable measurements due to factors such as postoperative scarring, obesity, and emphysema; and (d) poor treatment adherence during chemotherapy, changes in the chemotherapy regimen, failure to complete the planned cycles, or occurrence of severe non-cardiac adverse events.

Ethics approval

This study was approved by the Ethics Committee of the First Affiliated Hospital of Jinzhou Medical University, Jinzhou, Liaoning, China (Approval Number: KYLL-202536, approved on 22 May 2025). Written informed consent was obtained from all the participants. This study was conducted in accordance with the Declaration of Helsinki of 1975, as revised in 2024. All patient details were deidentified prior to analysis.

Study design

This was a prospective observational study. Blood samples were collected from all patients before chemotherapy (T0), at the second chemotherapy cycle (T2), and at the fourth chemotherapy cycle (T4). At each time point, patients also underwent TTE. According to the European Society of Cardiology (ESC) cardio-oncology guidelines, subclinical CTRCD was defined as a preserved LVEF with a relative reduction in left ventricular global longitudinal strain (LVGLS) of ≥15% compared with that at baseline (T0). ¹¹ Patients were classified into a CTRCD group and a non-CTRCD group based on the occurrence of subclinical CTRCD. Differences in relevant parameters between groups were compared, and the predictive value of each parameter for subclinical CTRCD was further evaluated.

Data collection

Clinical data

Baseline demographic and hemodynamic variables were collected, including age (years), height (cm), weight (kg), systolic blood pressure (SBP; mmHg), diastolic blood pressure (DBP; mmHg), and resting heart rate (HR; beats per minute (bpm)).

Laboratory measurements

Peripheral venous blood samples were collected at T0, T2, and T4 for hematological analysis. Blood samples were collected into ethylenediaminetetraacetic acid (EDTA)–anticoagulated tubes and analyzed promptly after collection according to standard laboratory procedures. WBC, NEUT, LYM, and platelet count (PLT) were measured using an automated hematology analyzer (Sysmex XN-2800, Sysmex Corporation). NLR was calculated as NEUT/LYM. Cardiac troponins and natriuretic peptides (B-type natriuretic peptide (BNP)/N-terminal pro–B-type natriuretic peptide (NT-proBNP)) were not routinely measured in this study and were therefore not included in the present analysis.

TTE

All echocardiographic examinations were performed using a Philips EPIQ 7C color Doppler ultrasound system by the same attending physician with 5 years of experience, and image analysis and quality control were conducted by a deputy chief physician specializing in echocardiography. Standard imaging views were acquired. The left atrial anteroposterior diameter (LAD) was measured in the parasternal long-axis view. Left ventricular end-diastolic volume (LVEDV), left ventricular end-systolic volume (LVESV), left ventricular fractional shortening (LVFS), LVEF, and left atrial volume index (LAVI) were quantified using the biplane Simpson method. Pulsed-wave Doppler (PW) was used to assess early (E-wave) and late (A-wave) transmitral inflow velocities and to calculate the ratio of early to late transmitral inflow velocities (E/A). Tissue Doppler imaging (TDI) was used to obtain e′, the ratio of early transmitral inflow velocity to early diastolic mitral annular velocity (E/e′), and Tei index. For speckle-tracking analysis, dynamic two-dimensional cine loops of the apical four-chamber, three-chamber, and two-chamber views were acquired over four consecutive cardiac cycles. Using the AutoStrain LV mode, the endocardial and epicardial borders of the left ventricle were automatically delineated and manually adjusted when necessary to ensure accurate border tracking; the software then calculated LVGLS. Left atrial strain was assessed using AutoStrain LA, with automatic delineation of the left atrial endocardium and end-diastole as the zero-strain reference, yielding LASr, left atrial conduit longitudinal strain (LAScd), and left atrial contractile longitudinal strain (LASct). In addition, apical four-chamber cine loops over four consecutive cardiac cycles were analyzed additionally using the 2DQ module to obtain maximum left atrial volume (LAVmax), minimum left atrial volume (LAVmin), and left atrial volume before atrial contraction (LAVpre). Left atrial total ejection fraction (LATEF), left atrial passive ejection fraction (LAPEF), and left atrial active ejection fraction (LAAEF) were calculated as follows:

$$\begin{array}{l}\text{LATEF}=\left(\text{LAVmax}-\text{LAVmin}\right)/\text{LAVmax}\\ \text{LAPEF}=\left(\text{LAVmax}-\text{LAVpre}\right)/\text{LAVmax}\\ \text{LAAEF}=\left(\text{LAVpre}-\text{LAVmin}\right)/\text{LAVpre}\end{array}$$

Statistical analyses

Statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) software (version 27.0, IBM). Continuous variables with a normal distribution were compared between groups using the unpaired Student’s t-test (independent-samples t-test) and are presented as mean ± SD. Non-normally distributed variables were compared using the Mann–Whitney U test and are presented as median (interquartile range (IQR)) values. For repeated measurements across chemotherapy cycles, repeated-measures analysis of variance (ANOVA) was applied when normality assumptions were satisfied; the Friedman test was used for nonparametric comparisons.

ROC curve analysis was performed to evaluate the predictive performance of T2 NLR, relative change in left atrial reservoir longitudinal strain at T2 compared with that at T0 (ΔLASr), and their combined model for subclinical CTRCD. The area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were calculated, and the optimal cutoff values were determined using the Youden index. A binary logistic regression model was constructed to assess the independent predictive value of T2 NLR and ΔLASr for subclinical CTRCD. Subclinical CTRCD was treated as the dependent variable (0 = no cardiotoxicity and 1 = cardiotoxicity), and T2 NLR and ΔLASr were included as independent variables. Regression results are reported as regression coefficients (β), odds ratios (ORs), 95% confidence intervals (CIs), and p-values. A two-sided p-value <0.05 was considered statistically significant.

Data visualization

To illustrate temporal trends, line plots were used to depict normally distributed variables across chemotherapy cycles, whereas box-and-whisker plots were used to present the distribution and changes in non-normally distributed variables at different time points. The combined use of these plots facilitates a comprehensive assessment of dynamic changes and supports clinical interpretation. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ¹²

Results

Baseline characteristics

At T0, a total of 85 patients with BC were enrolled, including 25 patients in the CTRCD group and 60 in the non-CTRCD group. No significant differences were observed in the clinical characteristics between the two groups, including age, height, weight, SBP, DBP, and resting HR (all p > 0.05).

Regarding laboratory parameters, the median NLR was 2.25 (1.56–2.87) in the CTRCD group and 2.17 (1.51–2.83) in the non-CTRCD group, with no statistically significant difference (p = 0.828). Other hematological indices, including WBC, NEUT, LYM, and PLT, were also comparable between the two groups (all p > 0.05).

Similarly, conventional echocardiographic parameters (LAD, LAVI, LVEDV, LVESV, LVEF, LVFS, E, A, E/A, E/e′, Tei index, LATEF, LAPEF, and LAAEF) and strain measurements (LVGLS, LASr, LAScd, and LASct) did not differ significantly between the two groups (all p > 0.05). These findings indicate that baseline cardiac function and systemic inflammatory status were comparable between the groups prior to chemotherapy, providing a robust foundation for subsequent analyses. The detailed results are presented in Table 1.

Table 1.

Comparison of the baseline (T0) clinical characteristics, laboratory parameters, and echocardiographic indices between the CTRCD and non-CTRCD groups (n = 85).

Variable	Overall (n = 85)	CTRCD (n = 25)	Non-CTRCD (n = 60)	p-value
Age (years)	56.36 ± 9.37	57.96 ± 9.71	55.70 ± 9.21	0.314
Height (cm)	160.34 ± 4.12	160.36 ± 4.60	160.33 ± 3.95	0.979
Weight (kg)	63.42 ± 10.17	60.16 ± 12.57	64.77 ± 8.76	0.057
SBP (mmHg)	124.42 ± 11.87	126.20 ± 13.06	123.68 ± 11.38	0.377
DBP (mmHg)	80.55 ± 5.79	81.08 ± 5.59	80.33 ± 5.90	0.591
HR (bpm)	84.95 ± 6.97	83.28 ± 8.10	85.65 ± 6.39	0.154
WBC (×10⁹/L)	6.26 (5.21–7.26)	5.78 (4.97–6.79)	6.43 (5.28–7.67)	0.081
NEUT (×10⁹/L)	3.83 (2.78–5.21)	3.71 (2.67–5.40)	3.85 (2.78–5.17)	0.710
LYM (×10⁹/L)	1.78 (1.50–2.10)	1.70 (1.45–1.95)	1.82 (1.57–2.28)	0.088
PLT (×10⁹/L)	255.0 (209.0–290.2)	270.0 (189.80–320.8)	254.5 (215.95–285.3)	0.703
NLR	2.25 (1.56–2.87)	2.34 (1.66–3.02)	2.17 (1.51–2.83)	0.828
LVEDV (mL)	88.83 ± 9.44	88.37 ± 9.56	89.02 ± 9.46	0.773
LVESV (mL)	33.54 ± 3.39	33.63 ± 3.18	33.50 ± 3.50	0.877
LVEF (%)	62.68 ± 3.48	61.94 ± 3.79	62.98 ± 3.32	0.210
LVFS (%)	33.79 ± 2.83	32.94 ± 2.76	34.15 ± 2.80	0.073
E/A	1.29 ± 0.33	1.28 ± 0.34	1.29 ± 0.33	0.916
E/e′	9.28 ± 2.01	9.22 ± 1.98	9.30 ± 2.04	0.868
E	89.27 ± 13.87	87.04 ± 12.29	90.20 ± 14.51	0.342
A	71.99 ± 14.21	70.73 ± 14.03	72.52 ± 14.37	0.599
Tei index	0.38 ± 0.04	0.38 ± 0.03	0.38 ± 0.04	0.994
LAD (mm)	32.10 ± 2.37	32.06 ± 2.12	32.12 ± 2.49	0.916
LAVI (mL/m²)	25.00 ± 4.70	24.35 ± 4.43	25.28 ± 4.83	0.407
LATEF (%)	59.95 ± 6.76	60.63 ± 7.28	59.67 ± 6.58	0.555
LAPEF (%)	36.59 ± 4.69	36.31 ± 3.79	36.72 ± 5.05	0.720
LAAEF (%)	37.90 ± 6.30	36.98 ± 6.04	38.29 ± 6.41	0.388
LASr (%)	39.71 ± 5.31	39.96 ± 5.89	39.61 ± 5.10	0.785
LAScd (%)	−23.50 ± 5.28	−23.73 ± 4.56	−23.40 ± 5.59	0.790
LASct (%)	−16.35 ± 3.85	−15.92 ± 4.31	−16.54 ± 3.67	0.500
LVGLS (%)	−21.83 ± 1.99	−21.38 ± 2.18	−22.02 ± 1.89	0.178

Normally distributed variables are presented as mean ± SD, whereas non-normally distributed variables are presented as median (interquartile range (IQR)).

p-values were calculated using the independent-samples t-test or Mann–Whitney U test, as appropriate. p-values reflect comparisons between the CTRCD and non-CTRCD groups; p < 0.05 indicates statistical significance.

CTRCD: cancer therapy–related cardiac dysfunction; SBP: systolic blood pressure; DBP: diastolic blood pressure; HR: heart rate; WBC: white blood cell count; NEUT: neutrophil count; LYM: lymphocyte count; PLT: platelet count; NLR: neutrophil-to-lymphocyte ratio; LVEDV: left ventricular end-diastolic volume; LVESV: left ventricular end-systolic volume; LVEF: left ventricular ejection fraction; LVFS: left ventricular fractional shortening; E/A: ratio of early to late transmitral inflow velocities; E: early diastolic transmitral inflow velocity; A: late diastolic transmitral inflow velocity; E/e′: ratio of early transmitral inflow velocity to early diastolic mitral annular velocity; LAD: left atrial anteroposterior diameter; LAVI: left atrial volume index; LATEF: left atrial total ejection fraction; LAPEF: left atrial passive ejection fraction; LAAEF: left atrial active ejection fraction; LASr: left atrial reservoir longitudinal strain; LAScd: left atrial conduit longitudinal strain; LASct: left atrial contractile longitudinal strain; LVGLS: left ventricular global longitudinal strain; bpm: beats per minute.

Temporal trends of key parameters

Regarding laboratory indices, the NLR increased progressively with advancing chemotherapy cycles, with median values of 2.25 (1.56–2.87) at T0, 2.71 (1.67–4.46) at T2, and 3.60 (1.97–5.29) at T4 (p < 0.001; Figure 1). In addition, both WBC and LYM decreased significantly during follow-up. WBC declined from 6.26 (5.21–7.26) × 10⁹/L to 4.89 (3.47–6.28) × 10⁹/L (p < 0.001; Figure 1), and LYM decreased from 1.78 (1.50–2.10) × 10⁹/L to 1.28 (0.72–1.63) × 10⁹/L (p < 0.001; Figure 1).

Figure 1.

Changes in the neutrophil-to-lymphocyte ratio (NLR), white blood cell count (WBC), and lymphocyte count (LYM) across chemotherapy cycles. Box-and-whisker plots show the distributions of NLR, WBC, and LYM at baseline (T0), after the second chemotherapy cycle (T2), and after the fourth chemotherapy cycle (T4). The central line indicates the median, the box indicates the interquartile range, the whiskers indicate the range within 1.5 interquartile ranges, and the dots indicate outliers.

With respect to echocardiographic parameters, LVEDV, LVESV, LVEF, and E/A showed no significant changes across the three time points (all p > 0.05), whereas the Tei index increased significantly at T4 compared with those at T0 and T2 (p < 0.001). Regarding left atrial function, both LASr and LAScd decreased significantly at T2 and T4 (both p < 0.001; Figure 2), and LVGLS demonstrated a significant decline during follow-up (p < 0.001; Figure 2). Specifically, LASr decreased from 39.71% ± 5.31% at T0 to 36.81% ± 5.16% at T4, and LAScd declined from −23.50% ± 5.28% at T0 to −21.50% ± 5.53% at T4; both changes were statistically significant (p < 0.001).

Figure 2.

Changes in left atrial reservoir longitudinal strain (LASr), left atrial conduit longitudinal strain (LAScd), and left ventricular global longitudinal strain (LVGLS) across chemotherapy cycles. Line plots show the temporal changes in LASr, LAScd, and LVGLS at baseline (T0), after the second chemotherapy cycle (T2), and after the fourth chemotherapy cycle (T4).

Collectively, LASr, LAScd, NLR, LYM, WBC, and LVGLS exhibited significant dynamic changes throughout the chemotherapy treatment period, suggesting that these parameters hold clinical utility for early monitoring of anthracycline-associated cardiotoxicity and systemic inflammatory responses.

Between-group comparisons (CTRCD vs. Non-CTRCD)

At T2, significant between-group differences were observed in several parameters. NEUT was significantly higher in the CTRCD group than in the non-CTRCD group (p = 0.020). NLR was also markedly elevated (p < 0.001), whereas LYM was significantly lower in the CTRCD group (p = 0.037). These findings suggest a more pronounced systemic inflammatory/immune response in patients who developed CTRCD during chemotherapy. In addition, LASr and LAScd were significantly reduced in the CTRCD group (p = 0.043 and 0.039, respectively), indicating impaired left atrial function. LVGLS was also significantly lower in the CTRCD group (p < 0.001), suggesting more substantial myocardial functional deterioration and higher risk of cardiotoxicity.

At T4, significant differences between the CTRCD and non-CTRCD groups persisted for several variables. NLR remained significantly higher in the CTRCD group (p = 0.041), whereas the differences in LYM and NEUT were no longer statistically significant (p = 0.780 and 0.195, respectively). This pattern may indicate that, although systemic inflammation persisted, immune-cell count changes became less distinct over time, reflecting ongoing treatment effects and adaptive immune modulation during chemotherapy. Echocardiography revealed that the LASr and LAScd remained significantly lower in the CTRCD group (p = 0.037 and 0.030, respectively), consistent with further left atrial functional impairment. LVGLS continued to be significantly reduced in the CTRCD group (p < 0.001), further supporting a progressive decline in myocardial function among patients with CTRCD.

Overall, conventional echocardiographic indices, such as LVEF, E/A, and E/e′, did not differ significantly between groups at either T2 or T4 (all p > 0.05). In contrast, NLR, LVGLS, LASr, and LAScd showed significant between-group differences during chemotherapy. These results suggest that traditional measures possess insufficient sensitivity in the early detection of CTRCD-related changes, whereas inflammatory and strain-based parameters capture earlier associations between systemic inflammation and cardiac functional impairment, providing greater sensitivity for early risk identification. Detailed results are presented in Table 2.

Table 2.

Comparison of the clinical characteristics, laboratory parameters, and echocardiographic indices between the CTRCD and non-CTRCD groups at T2 and T4 (n = 85).

Variable	Timepoint	Overall (n=85)	CTRCD (n=25)	Non-CTRCD (n=60)	p-value
Weight (kg)	T2	63.35 ± 10.16	60.10 ± 12.49	64.71 ± 8.78	0.056
	T4	63.35 ± 10.12	60.04 ± 12.41	64.73 ± 8.75	0.051
SBP (mmHg)	T2	122.51 ± 8.85	122.64 ± 9.51	122.45 ± 8.65	0.929
	T4	122.45 ± 8.27	121.76 ± 7.83	122.73 ± 8.50	0.624
DBP (mmHg)	T2	79.84 ± 5.12	79.64 ± 5.29	79.92 ± 5.09	0.822
	T4	79.93 ± 5.28	79.60 ± 5.31	80.07 ± 5.30	0.713
HR (bpm)	T2	85.02 ± 6.87	85.44 ± 8.21	84.85 ± 6.30	0.721
	T4	84.92 ± 7.07	83.96 ± 8.30	85.32 ± 6.53	0.424
WBC (×10⁹/L)	T2	5.29 (4.29–6.48)	4.97 (4.20–6.56)	5.32 (4.28–6.48)	0.919
	T4	4.89 (3.47–6.28)	4.89 (3.13–5.78)	4.87 (3.63–6.45)	0.469
NEUT (×10⁹/L)	T2	3.69 (2.35–7.26)	5.70 (3.22–7.94)	3.21 (2.09–6.06)	0.020
	T4	3.59 (2.02–6.49)	3.87 (2.45–6.53)	3.45 (1.54–6.51)	0.195
LYM (×10⁹/L)	T2	1.51 (1.13–1.90)	1.30 (1.22–1.46)	1.76 (0.98–1.98)	0.037
	T4	1.28 (0.72–1.63)	1.28 (0.87–1.68)	1.26 (0.60–1.62)	0.780
PLT (×10⁹/L)	T2	250.0 (203.35–305.0)	248.9 (206.9–301.35)	250.45 (197.43–309.75)	0.988
	T4	223.0 (177.7–300.25)	222.0 (167.95–291.55)	224.0 (188.58–306.60)	0.449
NLR	T2	2.71 (1.67–4.46)	3.92 (2.70–6.56)	2.14 (1.45–3.56)	<0.001
	T4	3.60 (1.97–5.29)	4.63 (2.76–5.47)	2.92 (1.87–4.82)	0.041
LVEDV (mL)	T2	88.81 ± 9.50	88.28 ± 9.73	89.03 ± 9.48	0.740
	T4	88.97 ± 9.52	88.72 ± 9.56	89.08 ± 9.58	0.874
LVESV (mL)	T2	33.73 ± 5.53	33.84 ± 5.27	33.69 ± 5.67	0.910
	T4	33.37 ± 3.69	33.33 ± 3.33	33.39 ± 3.85	0.940
LVEF (%)	T2	62.47 ± 3.52	62.08 ± 3.41	62.63 ± 3.57	0.509
	T4	62.07 ± 3.90	60.96 ± 3.65	62.53 ± 3.93	0.091
LVFS (%)	T2	33.42 ± 2.91	32.97 ± 2.64	33.62 ± 3.02	0.357
	T4	33.81 ± 3.42	32.96 ± 3.04	34.17 ± 3.53	0.138
E/A	T2	1.28 ± 0.34	1.26 ± 0.33	1.28 ± 0.35	0.748
	T4	1.28 ± 0.34	1.27 ± 0.35	1.29 ± 0.34	0.837
E/e′	T2	8.96 ± 1.98	8.80 ± 2.17	9.03 ± 1.91	0.642
	T4	9.04 ± 1.99	8.66 ± 2.17	9.19 ± 1.90	0.263
E	T2	89.01 ± 14.58	86.71 ± 12.56	89.97 ± 15.34	0.350
	T4	89.36 ± 14.59	87.32 ± 13.52	90.21 ± 15.05	0.409
A	T2	72.11 ± 13.87	70.80 ± 13.28	72.66 ± 14.19	0.575
	T4	72.47 ± 14.44	71.64 ± 13.78	72.82 ± 14.81	0.734
Tei index	T2	0.38 ± 0.04	0.39 ± 0.04	0.38 ± 0.03	0.100
	T4	0.42 ± 0.03	0.42 ± 0.03	0.41 ± 0.03	0.763
LAD (mm)	T2	32.88 ± 2.37	33.24 ± 2.19	32.73 ± 2.44	0.367
	T4	32.71 ± 2.29	32.94 ± 2.41	32.62 ± 2.25	0.547
LAVI (mL/m²)	T2	25.33 ± 4.78	23.78 ± 4.22	25.98 ± 4.88	0.053
	T4	25.70 ± 4.75	24.17 ± 4.36	26.33 ± 4.80	0.055
LATEF (%)	T2	59.66 ± 6.92	60.33 ± 7.07	59.38 ± 6.89	0.568
	T4	59.49 ± 7.89	60.16 ± 7.55	59.02 ± 8.04	0.400
LAPEF (%)	T2	35.52 ± 4.47	36.33 ± 4.07	35.19 ± 4.61	0.286
	T4	35.54 ± 4.54	34.51 ± 3.78	35.97 ± 4.79	0.177
LAAEF (%)	T2	37.68 ± 7.40	37.76 ± 6.84	38.07 ± 7.64	0.460
	T4	37.38 ± 8.12	36.96 ± 7.89	37.56 ± 8.27	0.756
LASr (%)	T2	37.79 ± 4.45	36.28 ± 4.36	38.42 ± 4.37	0.043
	T4	36.81 ± 5.16	35.00 ± 4.72	37.56 ± 5.18	0.037
LAScd (%)	T2	−22.29 ± 4.74	−19.59 ± 4.85	−22.29 ± 5.64	0.039
	T4	−21.50 ± 5.53	−19.59 ± 4.86	−22.30 ± 5.64	0.030
LASct (%)	T2	−15.45 ± 4.11	−15.78 ± 3.25	−15.31 ± 4.43	0.638
	T4	−15.21 ± 4.10	−15.52 ± 3.25	−15.08 ± 4.42	0.646
LVGLS (%)	T2	−20.16 ± 2.09	−18.81 ± 1.83	−20.73 ± 1.94	<0.001
	T4	−18.89 ± 2.05	−17.10 ± 1.75	−19.63 ± 1.69	<0.001

Normally distributed variables are presented as mean ± SD, and non-normally distributed variables are presented as median (interquartile range, IQR).

p-values were calculated using the independent-samples t-test or the Mann–Whitney U test, as appropriate. A two-sided p < 0.05 was considered statistically significant.

CTRCD: cancer therapy–related cardiac dysfunction; SBP: systolic blood pressure; DBP: diastolic blood pressure; HR: heart rate; WBC: white blood cell count; NEUT: neutrophil count; LYM: lymphocyte count; PLT: platelet count; NLR: neutrophil-to-lymphocyte ratio; LVEDV: left ventricular end-diastolic volume; LVESV: left ventricular end-systolic volume; LVEF: left ventricular ejection fraction; LVFS: left ventricular fractional shortening; E/A: ratio of early to late transmitral inflow velocities; E: early diastolic transmitral inflow velocity; A: late diastolic transmitral inflow velocity; E/e′: ratio of early transmitral inflow velocity to early diastolic mitral annular velocity; LAD: left atrial anteroposterior diameter; LAVI: left atrial volume index; LATEF: left atrial total ejection fraction; LAPEF: left atrial passive ejection fraction; LAAEF: left atrial active ejection fraction; LASr: left atrial reservoir longitudinal strain; LAScd: left atrial conduit longitudinal strain; LASct: left atrial contractile longitudinal strain; LVGLS: left ventricular global longitudinal strain; bpm: beats per minute.

Independent predictors: multivariable logistic regression analysis

Multivariable logistic regression was performed to assess the independent predictive value of T2 NLR, ΔLASr, and relative decrease in left atrial conduit longitudinal strain at T2 compared with T0 (ΔLAScd) for subclinical CTRCD. The analysis showed that T2 NLR and ΔLASr but not ΔLAScd were significant predictors. Specifically, T2 NLR was independently associated with subclinical CTRCD (OR = 1.611, 95% CI: 1.193–2.179; p = 0.002), indicating that each 1-unit increase in the NLR corresponded to an approximately 61.1% increase in the odds of subclinical CTRCD. The ΔLASr was also an independent predictor (OR = 1.426, 95% CI: 1.184–1.718; p < 0.001), suggesting that each 1% decrease in left atrial reservoir strain was associated with an approximately 42.6% increase in the odds of subclinical CTRCD. In contrast, the ΔLAScd did not reach statistical significance (OR = 4.903, 95% CI: 0.076–315.241; p = 0.454). Although the point estimate was large, the wide CI and non-significant p-value indicate substantial uncertainty and no evidence of an independent association in this model.

Overall, these findings suggest that changes in the T2 NLR and ΔLASr serve as early predictors of anthracycline-associated cardiotoxicity, whereas the potential role of ΔLAScd requires further confirmation in larger studies. The detailed results are presented in Table 3.

Table 3.

Multivariable logistic regression analysis of T2 NLR, ΔLASr, and ΔLAScd for predicting the occurrence of subclinical CTRCD.

Variable	β	OR	95% CI	p-value
T2 NLR	0.477	1.611	1.193–2.179	0.002
ΔLASr	0.355	1.426	1.184–1.718	<0.001
ΔLAScd	1.590	4.903	0.076–315.241	0.454

Subclinical CTRCD was defined as a relative reduction of ≥15% in the LVGLS compared with that at baseline.

A two-sided p < 0.05 was considered statistically significant.

β: regression coefficient; OR: odds ratio; CI: 95% confidence interval; NLR: neutrophil-to-lymphocyte ratio; ΔLASr: relative decrease in left atrial reservoir longitudinal strain at T2 compared with T0; ΔLAScd: relative decrease in left atrial conduit longitudinal strain at T2 compared with T0; LVGLS: left ventricular global longitudinal strain; CTRCD: cancer therapy–related cardiac dysfunction.

ROC curve analysis

To evaluate the predictive performance of T2 NLR and ΔLASr for subclinical CTRCD, ROC curves were generated for the T2 NLR, ΔLASr, and their combined model. The AUC for T2 NLR was 0.751, with an optimal cutoff value of 2.87, yielding a sensitivity of 76.0% and specificity of 63.3%. The AUC for ΔLASr was 0.856, with an optimal cutoff value of 0.07, corresponding to a sensitivity of 88.0% and specificity of 70.0%. The combined model (T2 NLR + ΔLASr) demonstrated superior discrimination, with an AUC of 0.896, sensitivity of 80.0%, and specificity of 85.0%, using an optimal predicted probability cutoff of 0.30. Detailed results are presented in Table 4 and illustrated in Figure 3.

Table 4.

Predictive performance of T2 NLR, ΔLASr, and their combined model for subclinical CTRCD.

Model	AUC	Optimal cutoff	Sensitivity (%)	Specificity (%)
T2 NLR	0.751	2.87	76.0	63.3
ΔLASr	0.856	0.07	88.0	70.0
T2 NLR + ΔLASr	0.896	0.30	80.0	85.0

For the combined model, the optimal cutoff of 0.30 refers to the predicted probability cutoff value derived from the logistic regression model.

A two-sided p < 0.05 was considered statistically significant.

AUC: area under the curve; NLR: neutrophil-to-lymphocyte ratio; ΔLASr: relative decrease in left atrial reservoir longitudinal strain at T2 compared with T0; CTRCD: cancer therapy–related cardiac dysfunction.

Figure 3.

ROC curves for the relative change in left atrial reservoir longitudinal strain (ΔLASr), T2 NLR, and the combined model. ROC curves show the predictive performance of ΔLASr at T2 compared with that at T0, neutrophil-to-lymphocyte ratio after the second chemotherapy cycle (T2 NLR), and their combined model for subclinical cancer therapy–related cardiac dysfunction (CTRCD). The corresponding areas under the curve (AUCs) are shown in the figure. ROC: receiver operating characteristic.

Overall, both T2 NLR and ΔLASr showed good early predictive ability for subclinical CTRCD, with ΔLASr achieving particularly high AUC and sensitivity. Importantly, integrating the T2 NLR with ΔLASr further improved predictive performance, especially in balancing sensitivity and specificity, as reflected by an AUC of 0.896. These findings suggest that combining an inflammatory marker with a strain-based functional parameter may enhance early risk stratification for anthracycline-associated cardiotoxicity in patients with BC, thereby supporting individualized monitoring and timely preventive interventions in clinical practice.

Discussion

In this study, we evaluated the early predictive value of T2 NLR and ΔLASr for the occurrence of subclinical CTRCD. Our findings demonstrated that both T2 NLR and ΔLASr were significant predictors of subclinical CTRCD, and that the combined model provided improved discriminatory performance. Specifically, T2 NLR and ΔLASr achieved acceptable sensitivity and specificity during the early phase of chemotherapy, with ΔLASr showing particularly strong performance in terms of AUC and sensitivity. Notably, the baseline NLR did not differ significantly between the CTRCD and non-CTRCD groups (2.25 vs 2.17, p = 0.828), suggesting that baseline inflammatory status was comparable between groups and that the predictive value of NLR in our cohort was dynamic, emerging during chemotherapy rather than at baseline.

Our results are consistent with previous evidence suggesting that inflammatory activity (e.g. the NLR) and cardiac functional indices (e.g. LASr) hold clinical utility for predicting cardiotoxicity. As a readily available inflammatory marker, the NLR reflects changes in systemic immune status, and previous studies have demonstrated that inflammatory responses are closely associated with cancer therapy–related cardiotoxicity.^13,14 During anthracycline chemotherapy, immune activation may exacerbate myocardial injury.^15–17 Accordingly, the increased NLR during chemotherapy compared with that at baseline may reflect treatment-related immune activation, which may help identify patients at increased risk of cardiotoxicity.^18–20 In our cohort, T2 NLR showed good performance for early CTRCD prediction (AUC = 0.751). Similarly, Baruch et al. have reported that in patients with BC, increases in the NLR during chemotherapy are strongly associated with a higher risk of cardiotoxicity. ¹⁰ In addition, the observed declines in WBC and LYM during chemotherapy may be consistent with the myelosuppressive effects of anthracycline-based treatment. However, bone marrow suppression, neutropenia, and lymphopenia were not predefined clinical endpoints and were not systematically graded in the present study.

ΔLASr, a measure of left atrial reservoir strain, can capture subtle alterations in cardiac function at an earlier stage. Previous studies have demonstrated that the left atrium plays a pivotal role in adapting to changes in cardiac loading conditions and that impairments in left atrial function may occur before detectable changes in the LVEF.^21,22 Accordingly, ΔLASr has been shown to be highly sensitive to early myocardial functional deterioration and performs well in the early identification of CTRCD.^8,23 In our study, ΔLASr achieved an AUC of 0.856 with a sensitivity of 88.0%.

Although ΔLAScd did not demonstrate a statistically significant independent predictive value in our study, it has been reported in the literature as a potential marker for cardiac functional impairment.^24,25 For example, based on their systematic review, Goyal et al. have reported that anthracycline chemotherapy significantly reduces left atrial strain parameters, including LAScd, suggesting that this index reflects subclinical myocardial dysfunction. ²³ Likewise, the review by Rusali et al. summarized multiple studies highlighting the potential role of LAScd in the early identification of cardiotoxicity. ²² In our cohort, ΔLAScd was not a significant early predictor of cardiotoxicity (p = 0.454), which may be attributable, at least in part, to the relatively small sample size; its predictive value may become more apparent in larger cohorts. Therefore, future studies integrating electrophysiological and/or advanced imaging markers are warranted to further clarify the potential role of ΔLAScd in CTRCD prediction.

In addition to risk prediction, recent evidence suggests that cardioprotective strategies are also relevant in anthracycline-treated patients. A recent systematic review and meta-analysis has reported that sodium–glucose cotransporter 2 inhibitors (SGLT2i) are associated with lower risks of all-cause mortality and heart failure hospitalization in patients with cancer and diabetes receiving antineoplastic therapy, with potentially favorable findings also observed in anthracycline-treated subgroups. ²⁶ However, the currently available evidence is derived mainly from observational studies and largely involves patients with concomitant diabetes; therefore, its applicability to the broader oncology population remains uncertain. In this context, early risk markers such as T2 NLR and ΔLASr may help identify patients who warrant closer surveillance and may be candidates for future evaluation of cardioprotective strategies, although this hypothesis still requires prospective validation.

In addition to risk prediction, recent literature has highlighted the importance of structured cardio-oncology management and preventive strategies in patients at increased risk of cardiotoxicity. ²⁷ A recent review has emphasized baseline risk stratification, surveillance, preventive strategies, and multidisciplinary management in patients with cancer at high and very-high risk of cardiotoxicity. In addition, recent evidence from inflammatory cardiac disease has suggested that the NLR may have broader prognostic relevance, although such evidence is not specific to cardio-oncology or anthracycline-treated populations. ²⁸ Therefore, our findings on dynamic markers such as T2 NLR and ΔLASr should be interpreted within a broader framework of risk stratification and future cardioprotective evaluation.

Despite the valuable clinical evidence provided by this study, certain limitations should be acknowledged. First, this was a single-center study with a relatively small sample size, and may therefore include selection bias, which could limit the generalizability of the findings. Multicenter, large-scale prospective studies are needed to validate our conclusions. Second, conventional cardiac biomarkers, including cardiac troponins and natriuretic peptides (BNP/NT-proBNP), were not routinely measured and therefore could not be incorporated into the analysis. As these biomarkers are important components of cardio-oncology risk stratification, their absence may have limited the comprehensiveness of risk assessment and the overall performance of the prediction model. Future studies should integrate circulating cardiac biomarkers with inflammatory and echocardiographic parameters to develop more robust multimarker prediction models. Third, although only patients with normal baseline WBC, NEUT, and LYM were included, autoimmune diseases, recent infections, inflammatory disorders, and hematologic disorders were not separately categorized in the exclusion criteria, which may have resulted in residual confounding in the interpretation of the NLR. Fourth, we did not integrate the Heart Failure Association–International Cardio-Oncology Society (HFA-ICOS) baseline cardiovascular toxicity risk assessment tool into the present study. As HFA-ICOS is an important framework for baseline risk stratification in ESC cardio-oncology guidelines, the absence of its integration may limit the alignment of our findings with guideline-based risk assessment. Future studies should evaluate whether dynamic markers such as T2 NLR and ΔLASr provide incremental predictive value beyond established baseline risk models such as HFA-ICOS. Fifth, we did not perform subgroup analyses according to BC molecular subtype. Because molecular subtype may influence treatment patterns, inflammatory status, and cardiotoxicity risk, its potential impact on NLR and subclinical CTRCD should be explored in future studies. Sixth, the follow-up period was relatively short, precluding the assessment of long-term cardiac outcomes after completion of chemotherapy. Longer-term follow-up is therefore necessary to evaluate clinical endpoints such as heart failure and other adverse cardiovascular events.

Future studies should further validate the utility of T2 NLR and ΔLASr across different cancer types and investigate their performance under various treatment modalities (e.g. chemotherapy, targeted therapy, and immunotherapy). Moreover, integrating imaging parameters, circulating biomarkers, and functional cardiac assessments into multivariable models may further improve the accuracy of early cardiotoxicity prediction. Finally, large-scale prospective clinical trials with longer follow-up are warranted to clarify the clinical applicability of these markers for cardiotoxicity surveillance and risk stratification.

Conclusion

This study provides preliminary evidence for an association of T2 NLR and ΔLASr with subclinical CTRCD during anthracycline chemotherapy in patients with BC. In this cohort, their combined use suggested potentially good predictive value.

Data availability statement

The data presented in this study are available on request from the corresponding author due to restrictions (e.g., privacy, legal, or ethical reasons).

Declaration of conflicting interests

The authors declare no conflicts of interest.

Institutional review board statement

This study was approved by the Ethics Committee of the First Affiliated Hospital of Jinzhou Medical University (Approval No. KYLL-202536). All procedures were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment.

Informed consent statement

Informed consent was obtained from all participants involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper.

Supplementary materials

No supplementary materials are provided for this manuscript, as all relevant figures, tables, and data have been included within the main text.

ORCID iD

Zhining Liu https://orcid.org/0009-0000-3073-9162