Dynamic cardiac changes in low cardiovascular risk patients with triple negative breast cancer treated with chemo-immunotherapy

Abstract

Background

Neoadjuvant chemotherapy-immunotherapy is the new standard of care for high-risk early-stage triple negative breast cancer (TNBC). As anthracyclines, pembrolizumab, and radiotherapy may each contribute to an increased risk of cardiovascular events, real-world assessment of early cardiovascular changes is of clinical interest.

Methods

Retrospective analysis of 85 women with early-stage TNBC treated with chemotherapy-pembrolizumab between 2018 and 2023 and had ≥ 1 transthoracic echocardiogram (TTE) available. Grade ≥ 2 cardiac common terminology criteria for adverse events (CTCAE) cumulative incidence estimates and Fine-Gray regressions (accounting for non-cardiac death as a competing risk) were calculated. Electrocardiogram (ECG) and TTE parameters during/following systemic therapy (vs. baseline) were compared.

Results

The median follow-up from immunotherapy start was 18.7 months [interquartile range (IQR) 13.6–39.1]. The median age was 50 years (IQR 38–61), 19% had hypertension, most (82%) with no detectable coronary artery calcium (CAC = 0), and 0% known cardiovascular disease. 9/85 (11%) experienced a grade ≥2 cardiac event with a median onset of 7.3 months (IQR 4.0–8.0) and a one-year cumulative incidence of 9.6%. Most (7/9) were grade 2 (n = 5 ejection fraction [EF] decline, n = 1 heart failure, n = 1 pericarditis); 2/9 were grade ≥ 3 (myocarditis, urgent percutaneous coronary intervention); all occurred among those receiving carboplatin, paclitaxel, doxorubicin, and cyclophosphamide-based therapy. Adjusting for age and CAC, mean left anterior descending coronary artery radiation dose was associated with an increased risk of cardiac events (sub-distribution hazard ratio 1.16/Gy, 95% confidence interval 1.01–1.35; p = 0.041). QTc prolongation ≥450ms was more common during treatment vs. baseline (39% vs. 15%; p = 0.025). On assessment for recovery, early grade 2 EF decline recovered in 3/5 patients (2/5 with absence of follow-up). In those with baseline and post-treatment TTE, 5/20 (25%) developed new moderate diastolic dysfunction, that persisted in a later TTE in 2/5 patients, downgraded to mild in 1/5, and not reevaluated by TTE in 2/5.

Conclusion

Early cardiovascular toxicity was observed during multi-modality TNBC treatment, even in young patients with low cardiovascular risk profiles, highlighting the importance of diligent surveillance. Longer follow-up and further studies are warranted, given the degree of recovery and later effects of these treatments may not yet be fully observed.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40959-025-00361-2.

Background

In women, breast cancer is the most common malignancy and the leading cause of cancer death worldwide [1]. Triple negative breast cancer (TNBC) is an aggressive subtype, accounting for 10–15% of all breast cancer cases and is most commonly diagnosed in women less than 50 years of age, carrying a higher rate of distant recurrence and greater risk of mortality [2]. Neoadjuvant chemoimmunotherapy per KEYNOTE-522 with an immune checkpoint inhibitor (ICI; pembrolizumab) together with carboplatin and anthracycline-containing chemotherapy is the new standard of care for high-risk early-stage TNBC, associated with a higher rate of pathologic complete response (pCR) and a significant improvement in event-free survival (EFS) [3, 4].

Both anthracyclines and pembrolizumab can cause adverse cardiovascular (CV) events, and combination therapy may further increase real-world CV risk. Anthracyclines carry an important risk of early, dose-dependent, and irreversible left ventricular dysfunction and heart failure [5]. ICI therapy has been associated with a 1–2% incidence of adverse cardiac events, for which myocarditis carriers a high risk of mortality (30–50%) [6, 7, 8, 9, 10, 11]. More recent studies have associated ICI therapy with accelerated atherosclerosis, atherosclerotic CV events [12], and conduction/interval prolongation abnormalities [13]. Furthermore, radiation therapy (RT) triggers an inflammatory cascade that accelerates the natural history of atherosclerosis [14]. In breast cancer survivors, left anterior descending (LAD) coronary artery radiation exposure is correlated with coronary artery calcium (CAC) score [15] and the need for mid-LAD revascularization [16]. In patients requiring revascularization following RT, nearly 30% present within the first 2-years and the median onset of significant LAD stenosis occurs less than 4 years from RT—demonstrating a small, but high-risk subgroup of patients experiencing significant, early CV events [16, 17]. Importantly, the correlation between LAD RT dose and cardiac events persists in the modern treatment era with advanced technology [18], especially as ongoing trials do not typically include dose limits to the LAD. Together, these observations highlight the need to better understand the cumulative associated cardiac risks during and after intensified therapy for TNBC.

The present study examines the cumulative risk of early grade ≥ 2 cardiac common terminology criteria for adverse events (CTCAE v5.0) in patients with TNBC receiving chemoimmunotherapy with or without RT. Additional goals were to identify factors associated with increased risk of grade ≥ 2 cardiac CTCAE and to characterize dynamic, short-term electrocardiogram (ECG) and transthoracic echocardiogram (TTE) changes during and after therapy.

Methods

Study population

The study cohort consisted of women aged ≥18 with newly diagnosed invasive TNBC treated with ICI therapy (pembrolizumab) between February 2018 and August 2023 at Cedars-Sinai Medical Center. We used Deep 6 AI (Deep 6 AI Inc., Pasadena, CA), a natural language processing and machine learning platform, to query our institution’s electronic medical record for patient records with structured data elements including “triple negative breast cancer” or “estrogen receptor (ER)/progesterone receptor (PR)/ human epidermal growth factor receptor 2 (HER2) negative status”, “pembrolizumab”, and “echocardiogram” (Supplemental Fig. 1). Patients with at least one TTE were included. Patients were excluded if they had metastatic or recurrent disease. The present study was approved by the Cedars-Sinai Medical Center Institutional Review Board with a waiver of informed consent due to minimal risk nature (STUDY00003061).

Baseline comorbidities and cancer information

Data on demographic, comorbidities, body mass index (BMI), smoking status, CV medications, laboratory values, breast cancer laterality, cancer stage, and cancer treatment details were retrospectively reviewed and collected manually from the electronic medical record. CV comorbidities included hypertension, hyperlipidemia, diabetes mellitus, coronary artery disease, ischemic heart failure, coronary heart disease equivalents (abdominal aortic aneurysm or peripheral artery disease), arrhythmia, and valvular disease. Any prior cancer diagnosed before the current TNBC was recorded. Cancer stage was based on the 8th edition of the American Joint Committee on Cancer staging of breast cancer.

Coronary calcification quantification

Coronary artery calcium (CAC) was manually measured on non-ECG gated, non-contrast-enhanced RT planning computed tomography (CT) scans using syngo.via (Siemens Healthineers, Malvern, PA) by a single investigator (KDS) with adjudication by a radiation oncologist experienced in cardiac anatomy (KMA). Helical RT planning CT scans were obtained with or without deep inspiration breath hold (DIBH), with slice thickness 2.5–5.0 mm (General Electric Medical Systems, Milwaukee, WI). If the patient did not undergo RT, the most recent diagnostic non-contrast enhanced chest CT or attenuation CT scan from positron emission tomography (PET)/CT studies with slice distances ≤ 5.0 mm acquired within six months before the initiation of chemoimmunotherapy was used to obtain a baseline CAC score. CAC was quantified using the Agatston method [19], with individual coronary vessel scores summed into a total CAC score and grouped into no atherosclerosis (0), minimal (1–10), mild (11–100), moderate (101–400), and high (> 400) risk.

Radiotherapy and dosimetry data

RT was planned with 3D conformal RT (3D-CRT) or intensity-modulated RT (IMRT) techniques using Varian Eclipse (Varian Medical Systems, Palo Alto, California). Dose fractionations were typically 50 Gy in 25 fractions or 42.6 Gy in 16 fractions, with or without a photon or electron tumor bed boost. The heart and the LAD coronary artery were delineated manually by the same radiation oncologist (JPN) and adjudicated by a radiation oncologist experienced in cardiac anatomy (KMA) according to a published cardiac contouring atlas [20]. The RT dose was converted to the equal dose in 2 Gy fractions (EQD2) and mean LAD and whole heart doses obtained.

Cardiac CTCAE, electrocardiogram, and echocardiogram endpoints

The primary outcome was grade ≥2 cardiac CTCAE (version 5.0) [21]. An in-depth manual medical record review was performed. Any cardiac event was examined, including CV death, myocardial infarction, coronary revascularization, unstable angina, heart failure/cardiomyopathy, ejection fraction (EF) decline, valvular disease, dysrhythmias/conduction abnormalities, and pericardial disease (e.g., pericarditis, pericardial effusion). Twelve-lead ECG and TTE data obtained within 6 months prior to pembrolizumab administration, and at any time after pembrolizumab was started, were extracted from patients’ records. The following automated ECG measurements were extracted from reports: heart rate, P-wave amplitude, QT interval, RR interval, corrected QT interval (QTc), QRS duration, PR interval, and QRS axis. ST-T changes were analyzed according to the AHA/ACCF/HRS (American Heart Association/American College of Cardiology Foundation/Heart Rhythm Society) recommendations [22]. The Cornell index and the Sokolow-Lyon index were evaluated as electrocardiographic LVH criteria [23, 24]. Any QTc prolongation, calculated by Bazett formula [25], was defined as ≥450 ms per CTCAE v5.0. TTEs were executed by dedicated sonographers. Evaluation of global function (left ventricular ejection fraction [LVEF]), structure (maximal wall thickness [MWT]), diastolic function (assessed and graded per the American Society of Echocardiography [ASE]/European Association of Cardiovascular Imaging (EACVI) guidelines) [26], left atrial volume index (LAVi), contractile function (global longitudinal strain [GLS]), pulmonary pressures (pulmonary artery systolic pressure [PASP]) and heart valves was performed. In patients with multiple ECGs and TTEs, the most abnormal (ECG: based on QTc interval and/or new arrythmia; TTE: based on EF and/or the presence of new/worsening diastolic dysfunction) and most recent studies during, and after pembrolizumab therapy were used for analysis. The index date was the start of pembrolizumab. All the patients were followed until loss to follow-up, death, or the end of January 2024, whichever came first.

Statistical analysis

Follow-up was calculated based on the start of ICI therapy using the reverse Kaplan Meier method [27]. Descriptive data are presented as number of patients (%) for categorical variables and mean (± standard deviation [SD]) or median (interquartile range [IQR]) for continuous variables. Cumulative incidence estimates of grade ≥ 2 and grade ≥ 3 cardiac events were estimated using non-CV death as a competing risk. Fine-Gray regression models were performed (accounting for non-cardiac death as a competing risk) [28], and results presented as sub-distribution hazard ratio (sHR) with 95% confidence interval (CI). Multivariable models included covariates with p ≤ 0.05 on univariable analysis and clinically pertinent variables. Comparisons between during and/or post-treatment ECG or TTE parameters versus baseline were calculated by a paired t-test or signed rank test for continuous variables, and McNemar’s test for categorical variables. Two-sided p-value ≤ 0.050 was considered statistically significant. All analyses were performed using StataSE, version 17.0 (StataCorp LLC) statistical software.

Results

Patient characteristics

The study cohort consisted of 85 patients (CONSORT diagram, Supplemental Fig. 1). The median follow-up from the start of ICI therapy was 18.7 months [interquartile range (IQR) 13.6–39.1 months]. Baseline patient characteristics are summarized in Table 1. The median age at TNBC diagnosis was 50 years (IQR 38–61), the median BMI was 24.0 (IQR 21.3–27.4), 31.8% were current or former smokers, 18.8% had hypertension, and 16.5% had hyperlipidemia. None had pre-existing coronary heart disease. The majority (81.9%) had baseline CAC of 0, while only 1.4% had CAC > 400. Most (96.5%) received neoadjuvant chemotherapy-pembrolizumab, with an anthracycline-based regimen in 78.8% of cases, though 3.5% received only adjuvant chemotherapy-pembrolizumab. Mastectomy was performed in 41 patients (48.2%). 67 patients (78.8%) received RT (61.2% left side; 47.5% with regional nodes (including internal mammary chain); 81.5% with boost; 69.5% standard fractionation; 81.4% with deep inspiration breath hold).

Table 1.

Clinical and treatment characteristics

Characteristics	No. (%)
Age, median (IQR, years)	50 (38–61)
BMI (kg/m2), median (IQR)	24.0 (21.3–27.4)
Tobacco
Never	58 (68.2%)
Former	23 (27.1%)
Current	4 (4.7%)
Medical history
Hypertension	16 (18.8%)
Hyperlipidemia	14 (16.5%)
Diabetes	2 (2.4%)
Heart disease	0 (0%)
DVT	3 (3.5%)
Prior cancer	6 (7.1%)
CV medications
ACE inhibitor	1 (1.2%)
Angiotensin receptor blocker	4 (4.7%)
Beta-blocker	6 (7.1%)
Calcium Channel blocker	6 (7.1%)
Statin	13 (15.3%)
Oral antidiabetic agent	2 (2.4%)
Insulin	1 (1.2%)
Anti-platelet	5 (5.9%)
CAC score†
No atherosclerosis: 0	59 (81.9%)
Minimal: 1–10	4 (5.6%)
Mild: 11–100	6 (8.3%)
Moderate: 101–400	2 (2.8%)
High: >400	1 (1.4%)
TNBC clinical stage (anatomic)
I	5 (5.9%)
II	62 (72.9%)
III	16 (18.8%)
IV	2 (2.4%)
TNBC post-neoadjuvant therapy pathological stage
No residual disease	52 (61.2%)
I	16 (18.8%)
II	8 (9.4%)
III	4 (4.7%)
Cancer treatments
Chemotherapy	85 (100%)
Doxorubicin	67 (78.8%)
Cyclophosphamide	69 (81.2%)
Carboplatin	71 (83.5%)
Taxane	84 (98.8%)
Capecitabine	20 (23.5%)
Pembrolizumab	85 (100%)
Mastectomy	41 (48.2%)
Radiotherapy	67 (78.8%)
Left side	41 (61.2%)
RNI (including IMN)‡	28 (47.5%)
Boost‡	53 (81.5%)
Standard fractionation‡	41 (69.5%)
DIBH‡	48 (81.4%)
Technique
3D-CRT	60 (92.3%)
IMRT	5 (7.7%)
Dose, median (IQR)‡
Heart mean, Gy	1.5 (1.0–3.0)
LAD mean, Gy	3.0 (0.6–5.6)

Values represent n (%) unless otherwise specified. Abbreviations: TNBC, triple negative breast cancer; IQR, interquartile range; BMI, body mass index; SD, standard deviation; CV, cardiovascular; ACE, angiotensin-converting enzyme; CAC, coronary artery calcium; DIBH, deep inspiratory breath-hold; Gy, Gray; LAD, left anterior descending coronary artery; RNI, regional nodal irradiation; IMN, internal mammary node; 3D-CRT, 3-dimensional conformal radiotherapy; IMRT, intensity modulated radiotherapy

^†Among n = 72 with baseline CAC available

^‡Based on patients with available radiotherapy information

The median mean heart and LAD coronary artery doses were 1.5 Gy (IQR 1.0–3.0) and 3.0 Gy (IQR 0.6–5.6), respectively. The median mean heart dose (MHD) was significantly higher in left- vs. right-sided breast cancer (2.6 Gy vs. 1.3 Gy, respectively; p = 0.008). Similarly, the mean LAD dose was also significantly higher in left- vs. right-sided cancer (median 5.5 Gy vs. 0.9 Gy, respectively; p < 0.001).

Grade ≥ 2 cardiac events

Following start of ICI therapy, 9 of 85 patients (10.6%) experienced a CTCAE grade ≥2 cardiac event with a median time to event of 7.3 months (IQR 4.0–8.0) and a one-year cumulative incidence of 9.6% (95% confidence interval [CI] 4.5–17.1%; Fig. 1). Most (7/9) were grade 2 (n = 5 EF decline, n = 1 heart failure, n = 1 pericarditis); 2/9 were grade ≥ 3 (myocarditis, cardiac chest pain requiring urgent percutaneous coronary intervention) (See Supplemental Table 1 for cardiac event details and patient risk factors). The myocarditis immune-related adverse event (irAE) occurred two months after initiation of ICI therapy and was delayed in diagnosis, in part due to transition of care from general cardiology to cardio-oncology. The myocarditis irAE occurred during the early time frame of this study, when awareness of ICI-related myocarditis amongst general cardiologists was less prevalent. Ultimately, the patient was treated with an angiotensin-converting-enzyme (ACE) inhibitor and beta blocker with resolution of symptoms (no glucocorticoids). The urgent percutaneous coronary intervention occurred in a 68-year-old woman with pre-existing hyperlipidemia one month after initiation of neoadjuvant ICI therapy, who required balloon angioplasty and placement of two stents in the LAD, who was later treated with adjuvant left-sided whole breast radiotherapy. Notably, all cardiac events (9/9) occurred in those receiving carboplatin, paclitaxel, doxorubicin, and cyclophosphamide chemotherapy. Among those receiving radiotherapy and dose volume histogram information available (n = 53), there were five cardiac events (9.4%). Adjusting for age and CAC score, mean LAD dose remained associated with an increased risk of cardiac events [sub-distribution hazard ratio (sHR) 1.16/Gy, 95% CI 1.01–1.35; p = 0.041] (Table 2). Among the 18 patients not treated with radiotherapy, there were four cardiac events (22.2%), a trend toward lower age versus those treated with radiotherapy (43 vs. 52 years, p = 0.084), and only one patient with CAC > 0. No other baseline cardiovascular or treatment factor was significantly associated with cardiac events on univariable analysis (p > 0.05) and was included in multivariable analysis.

Fig. 1.

Cumulative incidence of common terminology criteria for adverse events grade ≥ 2 cardiac events in patients with triple negative breast cancer treated with chemoimmunotherapy

Table 2.

Competing risk regression model for CTCAE grade ≥ 2 cardiac events

Variable	Univariable		Multivariable
	HR (95% CI)	P value	HR (95% CI)	P value
Age	1.05 (1.01–1.10)	0.024	1.07 (0.99–1.15)	0.051
BMI	1.09 (0.98–1.20)	0.11
Tobacco
Never	1.0 (Reference)
Ever	0.98 (0.26–3.73)	0.98
Hypertension	0.54 (0.07–4.01)	0.55
Hyperlipidemia	1.46 (0.32–6.72)	0.62
CV medications
Any antihypertensive	0.54 (0.07–4.01)	0.55
Statin	1.65 (0.36–7.54)	0.52
Anti-platelet	1.67 (0.23–12.16)	0.61
CAC score
CAC = 0	1.0 (Reference)		1.0 (Reference)
CAC > 0	2.11 (0.40-11.03)	0.38	1.43 (0.11–18.13)	0.78
Chemotherapy
Doxorubicin	†	-
Cyclophosphamide	†	-
Carboplatin	†	-
Taxane	†	-
Capecitabine	0.39 (0.05–3.22)	0.38
Mastectomy	0.86 (0.24–3.16)	0.82
RT dose
Mean heart dose (per Gy)	1.16 (0.93–1.44)	0.18
Mean LAD dose (per Gy)	1.16 (1.03–1.31)	0.015	1.16 (1.01–1.35)	0.041

Abbreviations: HR, hazard ratio; CI, confidence interval; BMI, body mass index; CV, cardiovascular; CAC, coronary artery calcium; RT, radiation therapy; LAD, left anterior descending coronary artery. †All failure events occurred in these groups, infinitely large hazard ratio, omitted from model

Dynamic electrocardiogram changes

There were 41 patients with ECGs obtained within 6 months prior to the initiation of pembrolizumab, 37 patients with ECGs during treatment, and 41 patients with ECGs after completion of treatment. The median time to latest post-treatment ECG was 14 months (IQR, 6–27 months). On ECG evaluation, there was increased frequency of QTc prolongation ≥450 ms on ECG obtained during treatment vs. baseline (38.9% vs. 14.6%; p = 0.025). The average heart rate was also significantly higher from 81±17 beats/min during pembrolizumab-based combination therapy vs. 72±12 beats/min at baseline (p = 0.036). Both observed overall differences resolved by the latest post-treatment ECG (Table 3). No additional statistically significant changes in conduction, rhythm, or other ECG parameters were observed in the overall cohort. Individually, among those that experienced grade ≥ 2 cardiac events, five patients additionally developed QTc prolongation during/after treatment (one resolved 3-years later, four did not have follow-up study to evaluate for resolution), one patient developed a type I AV block and LVH after treatment (no follow-up study), and one patient developed left posterior fascicular block during treatment (resolved 3-months later) and a right bundle branch block (not evaluated for resolution) (Supplemental Table 1).

Table 3.

Electrocardiogram characteristics before, during, and after pembrolizumab-based combination therapy

ECG characteristics	Baseline (n = 41)	†During treatment (n = 37)	‡p-value (vs. baseline)	†Post-treatment (n = 41)	‡p-value (vs. baseline)
HR (beats/min), mean (+/-SD)	72 (+/- 12)	81 (+/- 17)	0.036	76 (+/-17)	0.91
PR interval, median (IQR)	147 (+/-23)	150 (+/-22)	0.91	146 (22.6)	0.25
QTc length, median (IQR)	428 (+/-23)	441 (+/-23)	0.060	430 (+/-25)	0.46
QTc prolongation ≥ 450ms	6 (14.6%)	14 (38.9%)	0.025	10 (24.4%)	0.26
Conduction disorder	1 (2.4%)	2 (5.4%)	> 0.99	3 (7.3%)	0.25
PVC/PAC	0	2 (5.4%)	> 0.99	3 (7.3%)	0.50
Normal sinus rhythm	32 (80.0%)	25 (67.6%)	0.41	28 (68.3%)	0.48
Sinus tachycardia	1 (2.5%)	7 (18.9%)	0.16	3 (7.3%)	> 0.99
Sinus bradycardia	3 (7.5%)	3 (8.1%)	0.16	5 (12.2%)	0.69
Atrial fibrillation/flutter	0	0	-	0	-
LVH	0	0	-	2 (4.9%)	> 0.99
Low QRS voltage	2 (4.9%)	5 (13.5%)	0.50	4 (9.8%)	> 0.99
Abnormal P wave	0	0	-	0	-
Pathologic Q wave	0	0	-	0	-

The most abnormal data obtained during treatment, and the most recent data obtained after treatment, are shown

Data are presented as number of patients (column %), mean (± SD), or median (IQR, interquartile range)

P-value is calculated by a paired t-test or signed rank test for continuous variables, and McNemar’s test for categorical variables

^†Some variables may have smaller sample due to availability of reported parameter

^‡Statistical analysis was done in patients with both baseline and post-treatment electrocardiogram

Abbreviations: ECG, electrocardiogram; HR, heart rate; SD, standard deviation; IQR, interquartile range; QTc, corrected QT interval; PVC, premature ventricular contractions; PAC, premature atrial contractions; LVH, left ventricular hypertrophy

Dynamic echocardiographic changes

Comparably, there were 47 patients with TTEs obtained prior to the initiation of pembrolizumab, 47 patients with TTEs during treatment, and 30 patients with TTEs after completion of treatment. The median time to latest post-treatment TTE was 23 months (IQR 12–33 months). After initiation of pembrolizumab, five patients developed a grade 2 EF decrease, defined as 40–50% resting EF or 10–19% drop from baseline (Supplemental Table 1). Of these five patients, three ultimately had recovery in EF, though 2/5 did not have further studies to evaluate recovery by the time of data collection. In addition, among the twenty individual patients with a TTE obtained both at baseline and after initiation of pembrolizumab, 5/20 (25%) developed new moderate diastolic dysfunction, with a trend towards statistical significance (p = 0.103). No significant differences were observed in other echocardiographic parameters (Table 4). Notably, among the five patients who developed moderate diastolic dysfunction, all had CAC 0, only two had any baseline CV risk factors (age > 65 years, one of whom had hypertension and hyperlipidemia), and by the latest post-treatment TTE—two still had moderate diastolic dysfunction, one had down-graded to mild, but two did not have any post-treatment TTE performed after abnormality was noted during treatment.

Table 4.

Echocardiogram characteristics before, during, and after pembrolizumab-based combination therapy

TTE characteristics	Baseline (n = 47)	†During treatment (n = 47)	‡p-value (vs. baseline)	†Post-treatment (n = 30)	‡p-value (vs. baseline)
E/E’, median (IQR)	6.3 (5.1–7.7)	6.8 (6.0-8.3)	0.09	6.7 (5.2–7.8)	0.82
MWT (cm), mean (+/-SD)	0.9 (0.2)	0.9 (0.2)	0.17	0.9 (0.2)	0.82
LVEF (%), median (IQR)	64 (60–66)	63 (58–66)	0.10	63 (57–65)	0.78
LVEF ≤ 50%	3 (5.9%)	3 (6.4%)	> 0.99	2 (6.7%)	> 0.99
Diastolic dysfunction
Mild	13 (26.0%)	18 (43.9%)	> 0.99	11 (44.0%)	0.50
Moderate	1 (2.0%)	4 (9.8%)	> 0.99	3 (12.0%)	0.25
PASP (mmHg), mean (+/-SD)	19.6 (5.4)	20.2 (8.0)	0.70	20.2 (5.1)	0.73
LAVI, median (IQR)	17.7 (14.7–24.3)	19.8 (15.6–24.4)	0.85	24.4 (15.7–26.1)	0.17
GLS (%), mean (+/-SD)	-21.4 (2.3)	-19.2 (3.2)	0.17	-21.3 (3.2)	0.16
Valvular disease moderate	1 (2.1%)	3 (6.4%)	0.25	2 (6.7%)	> 0.99

The most abnormal data obtained during treatment, and the most recent data obtained after treatment, are shown

Data are presented as number of patients (column %), mean (± SD), or median (IQR, interquartile range)

P-value is calculated by a paired t-test or signed rank test for continuous variables, and McNemar’s test for categorical variables

^†Some variables may have smaller sample due to availability of reported parameter

^‡Statistical analysis was done in patients with both baseline and post-treatment echocardiogram

Abbreviations: TTE, transthoracic echocardiogram; IQR, interquartile range; MWT, maximal wall thickness; SD, standard deviation; LVEF, left ventricular ejection fraction; PASP, pulmonary artery systolic pressure; LAVI, left atrial volume index; GLS, global longitudinal strain

Individually, among those that experienced grade ≥ 2 cardiac events, four patients additionally experienced mild diastolic dysfunction during/after treatment, (one resolved three months later, one resolved two years later, and two did not have follow-up studies), and one patient developed a grade 1 pericardial effusion after (without follow-up).

Discussion

CV morbidity is an important endpoint for survivors of cancer, because of its impact on quality of life and future oncologic treatment possibilities. In this retrospective study of TNBC patients treated with modern, intensified, multi-modality, chemotherapy-pembrolizumab based treatment, we observed a 10% 1-year cumulative incidence of grade ≥2 cardiac events. Among those treated with RT and dose information available, mean LAD dose was significantly associated with the increased risk of grade ≥ 2 cardiac events, accounting for baseline CV risk. Importantly, these cardiac changes were noted in a cohort of younger TNBC patients (median age of 50) with relatively low CV risk profiles and low CAC burden (88% with CAC < 10). Together, these findings highlight the importance of diligent CV surveillance. Longer follow-up and further studies are warranted, given the degree of recovery, potential cardiac remodeling changes, and/or later effects of treatments may not yet be fully observed.

ICI therapy has been associated with electrocardiogram changes, including heart block, low-voltage, pathological Q wave, and QTc prolongation, though this has predominantly been reported in the context of ICI-related myocarditis [29, 30, 31]. This study observed that QTc prolongation ≥ 450 ms was more common during treatment compared to baseline, and with recovery post-treatment, though the extent to which this is due to anti-emetics versus ICI therapy is not discernable in this cohort. QT interval prolongation is associated with a potential increase in the risk of ventricular arrhythmia, specifically torsades de pointes (TdP) [32]. While the risk of TdP is generally more significant with a QTc > 500 ms, grade 2 changes should still be monitored and attention should be given to additive QTc prolonging drugs (i.e., anti-emetics, antibiotics) as well as correction of underlying electrolyte abnormalities. Though not incorporated in current cardio-oncology consensus guidelines for monitoring [33], if these findings were validated in other cohorts, it could warrant consideration for inclusion. Furthermore, as prolonged ICI therapy is known to accelerate atherosclerosis [12], it is unclear how these risks may be impacted by natural or treatment-induced menopause and development of additional metabolic risk factors.

Our analysis of dynamic TTE changes revealed early grade 2 LVEF deterioration in five patients, with subsequent recovery in three patients, and absence of follow-up study in the other two. Notably, the 2022 European cardio-oncology consensus guidelines recommend close surveillance of anthracycline-treated patients with cardiac biomarkers and TTE, including a TTE every few cycles—even in low-risk patients [33], though these recommendations have not been fully implemented in routine clinical practice. In addition, among patients with both baseline and during/post treatment TTE, 25% (5/20) developed moderate diastolic dysfunction, that persisted in a later study in two patients, was downgraded to mild in one patient, and was not reevaluated in two patients. None of these patients had CAC identified on baseline CT. Importantly, diastolic dysfunction has not been routinely assessed on TTEs in prior studies and requires longer-term follow-up to determine if associated with heart failure with reduced or preserved ejection fraction (HFrEF, HFpEF). Further, diastolic dysfunction is an important predictor of all-cause mortality in large epidemiologic studies [34], often precedes the development of systolic dysfunction [35], and has been shown to play an essential role in the pathophysiology of other cardiac diseases [36]. Though notably, diastolic dysfunction is also relatively common and may occur due to factors like elevated blood pressure, and while it may be linked to a small risk of systolic dysfunction [37]—it may not necessarily portend an increased risk of reduced ejection fraction associated with treatment. Together, these findings support monitoring diastolic and other echocardiographic changes that might identify patients at risk for developing overt left ventricular systolic dysfunction.

Prior studies have demonstrated an additive or synergistic effect of breast cancer therapies on CV risk. In a study by Kim et al., the investigators showed that left-side irradiation was associated with increased risk of CV events only in patients who also received a cumulative doxorubicin-equivalent dose ≥250 mg/m² [38]. Another study examining 55 patients with non-metastatic TNBC found a 13% rate of grade 3 myocarditis after neoadjuvant chemo-immunotherapy, but no additional myocarditis risk in the setting of adjuvant RT [39]. In the current study, we observed a 1% rate of grade 2 myocarditis, though non-myocarditis events were common (10%), and all occurred among those receiving carboplatin, paclitaxel, doxorubicin, and cyclophosphamide chemotherapy. Alternative anthracycline-free chemotherapy in the preoperative setting has been recently tested in early-stage TNBC. The phase 2 NeoPACT trial evaluated preoperative docetaxel with carboplatin for 6 cycles in addition to pembrolizumab and yielded a 58% pCR rate, as well as a 98% 3-year EFS rate for those achieving pCR, and 68% for those not achieving pCR, mirroring the EFS results from the KEYNOTE-522 trial [40]. The phase 3 OptimICE-PCR trial is examining omission of adjuvant pembrolizumab in TNBC patients with pCR after neoadjuvant chemo-immunotherapy [41]. Another consideration of emerging interest is the contribution of genetic underpinnings associated with increased cancer therapy associated cardiac toxicity. A study by Garcia-Pavia et al. reported that 7% of patients with anthracycline-induced cardiac dysfunction carried a likely pathologic variant within known cardiomyopathy genes (vs. <1% identified in matched control cohort) [42], suggestive that a certain subset of patients may be at even greater risk of toxicity. Together, these findings underscore the potential cumulative CV risk of multimodal cancer therapy, the need for close CV surveillance, and to optimize CV risk mitigation strategies. Improved adoption of existing cardio-oncology consensus guidelines-based risk assessment and monitoring [33] is warranted. Future, larger studies are needed to identify subgroup(s) that may be particularly vulnerable to combination chemotherapy-immunotherapy, to investigate novel blood-based and imaging biomarkers, and who may warrant closer surveillance with periodic ECGs and echocardiograms.

Given initial studies establishing the link between cardiac radiation dose and coronary heart disease in breast cancer survivors [43], significant efforts have been made to reduce heart dose exposure. Such strategies include deep inspiration breath hold (DIBH) technique to optimize distance between the heart and target [44], prone positioning (though results for cardiac dose sparing are mixed) [45, 46], target volume reduction (i.e., partial breast irradiation) [47, 48], and/or advanced planning techniques such as intensity-modulated RT or proton RT [49, 50, 51]. While heart doses have significantly reduced in the modern breast RT treatment era [52], recent data continue to support a correlation between LAD dose and adverse cardiac events [18]. Similarly, we observed an association between mean LAD dose and CTCAE grade ≥2 cardiac events (but not MHD), consistent with data supporting LAD as a better predictor (vs. MHD) of coronary events and MHD being a poor surrogate for LAD dose exposure during breast cancer RT [18, 53, 54]. Therefore, continued efforts are needed to optimize cardiac radiation sparing techniques in clinical practice to transform LAD dose into a modifiable cardiac risk factor, particularly given the potential interaction between RT and systemic therapies such as ICIs. It should additionally be noted that heart/LAD radiation dose is not incorporated in modern risk prediction models for cardio-toxicity [55] and further work is needed to account for cardiac radiation exposure in these models.

Limitations of our study include the retrospective nature with a relatively small sample size and limited follow-up that may be insufficient for drawing significant conclusions. Further, the shorter-term follow up may not fully capture the degree of recovery and/or cardiac remodeling or later effects of therapies. Specifically, radiation-induced heart disease can have a latency of several years, and while early events (within 2–5 years) occur [16, 17]—this shorter timeframe study may under-capture the extent of these risks. For instance, the patient who required acute coronary intervention following initiation of ICI was subsequently treated with left-breast radiotherapy, for which the added risk multi-modal therapy is likely insufficient to evaluate in this timeframe. In addition, we did not analyze cardiac events from patients who did not have echocardiograms available, which might have introduced a selection bias. Notably, only 47 patients had baseline echocardiographic data available, which limits our understanding of the extent to which observed ejection fraction changes were new or pre-existing, particularly given the known cardiotoxic potential of the commonly used chemotherapies. Lastly, the observed grade ≥ 2 cardiac event rate of 10% may be an under- or over-estimate given the incomplete nature of the echocardiogram and electrocardiogram surveillance.

Conclusion

Early CV toxicity related to multimodal treatment of TNBC was observed, even in young patients with low CV risk profiles. This highlights the importance of early CV surveillance, including improved cardiac monitoring during treatment, and continued efforts at cardiac toxicity risk mitigation strategies for breast cancer survivors. Longer follow-up and further studies are warranted, given degree of recovery and later effects of treatment may not yet be fully accounted for.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Conception: JPN, KMA, APN, RHM; methodology: JPN, KMA, APN, RHM; data acquisition: JPN, KDS, OP, AS, PB; data analysis: JPN, KMA, KDS; writing - original draft preparation: JPN, KMA; writing - review and editing: KMA, APN, RHM, MRK, JKJ, SLS, ACK, CR; supervision: KMA. All authors read and approved the final manuscript.

Funding

Not applicable.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study obtained ethical approval from the Cedars-Sinai Medical Center Institutional Review Board with a waiver of informed consent due to minimal risk nature (STUDY00003061).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.