Background

Cardiovascular diseases (CVD) in cancer patients are driven by treatment-related risk, which spans therapeutic classes and shares damage mechanisms leading to combined toxicity [1,2,3]. Cardio-oncology guidelines endorse surveillance with blood-based biomarkers (troponins and natriuretic peptides) and cardiac imaging [4]. Troponins are cardiac-specific, but not disease-specific, and natriuretic peptides are associated with heart failure (HF) [5]. Oxidative stress and inflammation induce C-reactive protein secretion, which is associated with poor outcomes in decompensated HF [5]. Association between radiotherapy (RT) and these biomarkers is inconclusive, therefore specific recommendations for radiation-induced cardiac damage are absent [4].

Our primary objectives were to evaluate: (a) the effect of RT on biomarkers reflective of myocardial injury/inflammation [high-sensitivity cardiac troponin I (hsTnI), N-terminal pro brain natriuretic peptide (NT-proBNP) and oxidative stress [high-sensitivity C-reactive protein (hsCRP)], and; (b) the association between mean heart dose (Dmean) and biomarkers.

Methods

Study population

This prospective, single-institution, IRB-approved (Protocol ID: Res/SCM/52/2022/40; IRB Approval ID: IRB-BHR/75/2022) observational study was conducted between June 2022 and July 2023. Women with left-sided breast cancer were eligible. Inclusion criteria were: (a) greater than 18 years with pathological stage I-III disease after either breast conservation surgery or mastectomy; (b) without coronary heart disease (CHD) or CHD risk equivalent; (c) requiring adjuvant RT (42.5 Gy/16Fx, 5 days/week) to whole breast or chest wall with/without elective regional nodal irradiation (sequential lumpectomy boost permitted; 10 Gy in 4Fx), and; (d) normal 2D transthoracic echocardiography prior to starting RT. Chemotherapy (neoadjuvant/adjuvant)(with trastuzumab for Her2-expressing tumors) was permitted. Adjuvant RT was delivered 3–4 weeks after completion of the preceding surgery or chemotherapy.

Study procedures

After obtaining informed consent, demographic, clinical, CVD risk factors, and treatment details were collected. Blood samples were collected pre-RT, immediately after RT completion (median: 0 days, Range: 0–18 days), and three months after RT completion (median: 92 days, IQR: 90–99 days). Since only 30% of patients return for a 3-month review post-RT (internal audit), 3-month biomarker measurement was optional. After accruing 52 patients, the data monitoring committee excluded hsTnI from the panel of tests, as 40 patients had values below the detection limit (0.012 ng/ml) at pre- and post-RT time points.

All patients were assessed for treatment in deep inspiration breath hold (DIBH)(RPM system, Varian Medical Systems, USA) and received RT via Field-in-Field Intensity-Modulated Radiotherapy Technique (FiF IMRT) or Volumetric Modulated Arc Therapy on a 6MV LINAC with daily kV-MV verification (Clinac 2100c, Varian Medical Systems, USA) [6]. Contouring for primary, nodal regions and organs-at-risk was performed as per RTOG 0413 protocol, RTOG consensus recommendations, and RTOG 1005 protocol (NCT01349322), respectively [7, 8]. Mean heart dose (Dmean) was recorded for the whole heart contour using a calculation grid size of 2.5 mm with AAA v15.6 algorithm (Varian Medical Systems, USA).

Biomarker measurements

All assays were performed on the Vitros 5600 platform (QuidelOrtho, USA). The hsCRP assay had a detection limit of 0.26 mg/L, coefficient of variation (CV) of < 8.3% at the 99th percentile with a reference limit of < 5.0 mg/L. The NT-proBNP assay had a measurement range of 11.1–35,000 pg/mL, CV of 11% at the 99th percentile with a reference limit of 125 pg/mL and 450 pg/mL for patients less than or greater than 75 years, respectively. The hsTnI assay had a measurement range of 0.012–80.0 ng/ml, CV of < 8.0% at the 99th percentile with a reference limit of 0.034 ng/mL. Whenever an elevated biomarker was detected, a cardiologist obtained a cardiac history and performed a cardiac examination.

Statistical analysis

The sample size for a relative effect size of 30% with 80% power and alpha = 0.05 with a two-tailed, paired-sample design assuming a normal distribution was 94. Another 15% were added for dropout, resulting in a sample size of 108. Baseline characteristics were reported as median with interquartile range (IQR)(continuous variables) or frequencies and percentages (categorical variables). All biomarkers had non-normal distribution; therefore, the Wilcoxon matched-pairs signed-rank test was used to compare pre- and post-RT levels with tied pairs handled by Pratt’s method [9]. The Friedman test was used to compare biomarkers at pre-, post-RT, and 3-months post-RT. Paired comparisons between time points utilized Dunn’s post-test [10]. Spearman rank correlation was used to assess correlation between biomarkers.

Multivariable quantile regression (QR) was used to test association of biomarkers with predictors without considering interactions [11]. Compared to ordinary least-squares (OLS) regression, QR demonstrates robust performance in non-normal distributions and in the presence of outliers. The Markov chain marginal bootstrap determined the standard error [12]. QR results were compared to OLS regression after log-transformation of biomarkers (Supplemental Materials).

This report complies with the ‘Strengthening the Reporting of Observational Studies in Epidemiology cohort’ guidelines (Supplemental Materials). Statistical analyses were performed using Prism v10 (DotMatics, USA) and R v4.2.3 (R Foundation for Statistical Computing, Austria). Statistical significance was set at p < 0.05 (two-sided).

Results

One hundred five patients completed the study and were analyzed for pre-/post-RT comparison, while 63 patients completed the 3-months post-RT evaluation (Fig. 1). Because the data monitoring committee recommended stopping hsTnI testing, 86 patients were analyzable for pre-/post-RT comparison, and 38 were analyzable at all three time points. The patients’ baseline characteristics are presented in Table 1. There was no correlation between markers (Supplemental materials). None of the patients developed a Major Adverse Cardiac Event (MACE) with a median follow-up of 15.0 months (IQR: 12.9–17.3) after the completion of RT. In addition, all patients with elevated biomarkers were asymptomatic on specialist examination, required no additional investigations and completed treatment without interruptions (Supplemental Materials).

Fig. 1
figure 1

Study flow diagram

Full size image
Table 1 Baseline Characteristics (n = 105) of the study population
Full size table

Biomarker comparison: Pre- vs. Post-RT (Fig. 2)

Fig. 2
figure 2

Wilcoxon signed-rank test, comparing biomarkers at pre- and post-RT time points

Full size image

The median (IQR) values of hsTnI, hsCRP, and NT-proBNP pre-RT were 0.012 ng/mL (0.012–0.012), 3.1 mg/L (1.5-6.0) and 59 pg/mL (26–116), respectively. The post-RT biomarkers were 0.012 ng/mL (0.012–0.012), 2.8 mg/L (1.6–6.8), and 45 pg/mL (24–101), respectively. The pre-/post-RT comparisons were not significantly different.

Biomarker comparison: Pre-, Post- and 3 months after RT (Fig. 3)

Fig. 3
figure 3

Friedman test, comparing biomarkers at pre-, post-, and 3-months post-RT time points

Full size image

The median (IQR) values of hsTnI, hsCRP, and NT-proBNP pre-RT were 0.012 ng/mL (0.012–0.012), 3.1 mg/L (1.4–6.1) and 56 pg/mL (29–125), respectively. The post-RT levels were 0.012 ng/mL (0.012–0.012), 2.4 mg/L (1.5–6.8), and 50 pg/mL (28–143), respectively. 3-months post-RT, the levels were 0.012 ng/mL (0.012–0.012), 2.7 mg/L (1.1-6.0), and 60 pg/mL (31–151), respectively. Comparisons between pre-, post-, and 3-months post-RT values were not significantly different.

Multivariable median quantile regression: pre- and Post-RT (Table 2)

Table 2 Results of multi-variable median quantile regression
Full size table

Multivariable median quantile regression demonstrated no influence of heart Dmean on any post-RT biomarker. No treatment-related factors (anthracycline or anti-Her2 exposure) influenced pre- or post-RT biomarkers. BMI and age were related to hsCRP and NT-proBNP at pre- and post-RT measurements, respectively. hsTnI analysis was not clinically meaningful as the variables had an estimate of < 0.0001 (Supplemental Materials), because the measurements were below the detection threshold. Finally, the results of QR were comparable to OLS regression (Supplemental materials).

Discussion

Identifying measurable markers associated with cardiac radiation exposure may permit MACE risk stratification of patients during follow-up and provide an objective measure to stratify risks with emerging RT techniques. Despite two decades of research, a reliable association has not been established (Table 3) [13,14,15,16,17,18,19,20,21,22,23,24,25,26]. Therefore, this prospective study of predefined, serial time-point biomarker measurements was deliberately designed in a homogeneous low-risk population of patients with left-sided breast cancer who received standardized modern RT to reduce variability, thereby isolating and enhancing the probability of detecting any RT effect. The biomarkers were chosen based on their role in monitoring cardiotoxic therapies and availability in community practice, in contrast to candidate research markers, which require specialized/centralized testing laboratories [25, 26].

Table 3 Overview of the literature on the association between cardiac biomarkers and cardiac radiation exposure. (Note: The criteria for inclusion were: (a) patients with breast cancer had to be treated with radiotherapy, and; (b) any cardiac biomarker was tested to determine its association with radiotherapy. Studies which measured biomarker change in relation to chemotherapy alone or were performed in animals, were excluded.)
Full size table

Early investigations reported elevation in troponins and natriuretic peptides with RT but were limited by retrospective design or indirect estimates of irradiated heart volume [13, 14]. Subsequent analyses were limited by modest sample sizes and statistical power, precluding conclusive association between RT and biomarkers [15, 17,18,19, 26]. In contrast, this study was adequately powered to detect a 30% change in biomarkers and used rigorous QR and OLS regression strategies with consistent results.

The achieved median heart Dmean (2.87 Gy) demonstrates that exposure can be minimized in the real world, and the threshold for measurable cardiac damage using these biomarkers is clearly above this dose. This is supported by the fact that the majority of reported heart Dmean of individual studies was around 3 Gy (Table 3), without clear association with absolute cardiac biomarker change. Contemporaneous cardiac damage detection will require ultrasensitive biomarkers or long-term studies of cardiac outcomes to establish the magnitude of effect (NCT04361240; NCT04790266; NCT03297346). However, it must be emphasized that the reducing cardiac exposure is adequately achieved with FiF IMRT, which was developed two decades ago and remains the benchmark for comparing newer, more complicated RT techniques [27].

The sample size was designed to be adequate to detect changes in biomarkers, but we acknowledge that it is modest for detecting subtler temporal trends. We intend to follow this cohort biennially for 15 years to record MACE, and perform a post-hoc analysis in the future, if statistically appropriate. We also did not measure these biomarkers pre-chemotherapy because we intended to establish association with RT [4]. Since heart Dmean is the most validated dosimetric parameter for cardiac outcomes, we chose not to investigate association with other sub-structures or parameters thereof, to avoid creating a multiple testing problem [28]. Enrolling right-sided breast cancer patients as controls was considered. But since their risk of developing RT-induced CVD is extremely low, repeated blood investigations were deemed unwarranted by our IRB. Our study population was intentionally composed of patients with low cardiac risk, to isolate the effect of radiotherapy on cardiac markers. It is plausible that patients at higher cardiac risk could demonstrate a more pronounced change in cardiac markers in response to radiotherapy, and could be an avenue for future research.

In conclusion, the lack of correlation between these biomarkers and cardiac radiation exposure will aid in narrowing the scope of future research. These results and prior reports clearly argue against their routine use to detect radiotherapy-induced cardiac injury with modern RT techniques.