Dosimetric Predictors of Symptomatic Cardiac Events After Conventional-Dose Chemoradiation Therapy for Inoperable NSCLC

Abstract

Introduction:

We hypothesized that higher cardiac doses correlates with clinically significant cardiotoxicity after standard-dose chemoradiation therapy (CRT) (~60 Gy) for inoperable NSCLC.

Methods:

We retrospectively reviewed the records of 140 patients with inoperable NSCLC treated with concurrent CRT from 2007 to 2015. Extracted data included baseline cardiac status, dosimetric parameters to the whole heart (WH) and cardiac substructures, and the development of post-CRT symptomatic cardiac events (acute coronary syndrome [ACS], arrhythmia, pericardial effusion, pericarditis, and congestive heart failure [CHF]). Competing risks analysis was used to estimate time to cardiac events.

Results:

Median follow-up was 47.4 months. Median radiation therapy dose was 61.2 Gy (interquartile range, 60 to 66 Gy). Forty patients (28.6%) developed 47 symptomatic cardiac events at a median of 15.3 months to first event. On multivariate analysis, higher WH doses and baseline cardiac status were associated with an increased risk of symptomatic cardiac events. The 4-year cumulative incidence of symptomatic cardiac events was 48.6% versus 18.5% for mean WH dose ≥ 20 Gy versus < 20 Gy, respectively (p = 0.0002). Doses to the WH, ventricles, and left anterior descending artery were associated with ACS/CHF, whereas doses to the WH and atria were not associated with Journal of Thoracic Oncology Vol. 13 No. 10: 1508–1518 supraventricular arrhythmias. Symptomatic cardiac events (p = 0.0001) were independently associated with death.

Conclusions:

Incidental cardiac irradiation was associated with subsequent symptomatic cardiac events, particularly ACS/CHF, and symptomatic cardiac events were associated with inferior survival. These results support the minimization of cardiac doses among patients with inoperable NSCLC receiving standard-dose CRT.

Introduction

Accumulating evidence suggests that radiation therapy (RT) potentially contributes to cardiotoxicity in patients with NSCLC.^1–5 In the Radiation Therapy Oncology Group (RTOG) 0617 study, patients treated on the high-dose, 74 Gy, chemoradiation therapy (CRT) arm experienced inferior overall survival (OS) compared to patients treated on the standard-dose, 60 Gy, CRT arm.⁶ Increasing heart volume receiving ≥ 5 Gy (V5) and heart volume receiving ≥ 30 Gy (V30) were independently associated with inferior OS after 2 years of follow-up, suggesting a possible association between inferior OS in the dose escalation arm and adverse cardiac sequelae of RT. Subsequent studies proposed a direct relationship between cardiac doses and symptomatic cardiac events in the setting of dose-escalated RT (median prescribed doses, 70 to 74 Gy) for locally advanced NSCLC.^3,4 One might expect the impact of RT-associated cardiotoxicity in NSCLC to increase as therapeutic advances improve longevity.^7–9

Although current evidence indicates the significance of cardiotoxicity after dose-escalated RT, the link between cardiac doses, clinically significant cardiotoxicity, and OS remains incompletely defined among patients treated with conventional, standard-dose RT (~60 Gy). Prior studies in this setting focused primarily on the association between cardiac dosimetry and OS, reaching conflicting conclusions.^10–13 Furthermore, the data appear limited about the significance of specific types of cardiotoxicities and the relationship with doses to cardiac subvolumes.¹⁴ We thus assessed these relationships within a modern cohort of inoperable NSCLC patients treated with curative standard-dose RT (median prescribed dose, 61.2 Gy) and concurrent chemotherapy.

Materials and Methods

Patient Population

We performed an IRB-approved review of the Rutgers Cancer Institute of New Jersey’s database to identify patients treated with curative intent CRT for unresectable stage II-III NSCLC or stage IV oligometastatic NSCLC between January 2007 and August 2015. We included patients with stage IV oligometastatic disease, defined as those with a solitary extrathoracic metastasis who received definitive CRT to the primary tumor and metastatic focus, based on comparable long-term survival rates to patients with locally advanced NSCLC in our and others’ experiences.^15,16 We identified 165 patients for this retrospective review. We included patients who subsequently received thoracic re-irradiation (e.g., for NSCLC recurrence) but censored them at the time of their re-irradiation. We excluded patients who did not receive concurrent chemotherapy (typically because of poor performance status) (n = 14) or complete the planned course of RT (because of either an inability to tolerate treatment or death) (n = 6). We also excluded patients who previously received thoracic RT (e.g., for a prior malignancy) (n = 5). This left 140 patients for the study. All patients underwent full staging work-up including positron-emission tomography/computed tomography (CT) scans. We used the American Joint Committee on Cancer seventh-edition criteria for staging classification.¹⁷

Treatment

RT was delivered with three-dimensional conformal radiotherapy (3DCRT), intensity-modulated radiation therapy (IMRT), or a mix of 3DCRT and IMRT to a typical dose of 60 to 66 Gy in 1.8 or 2 Gy per fraction. RT dose constraints were as follows: maximum spinal cord dose ≤ 50 Gy, mean lung dose < 20 Gy, lung V5 < 60% to 70%, lung volume receiving ≥ 20 Gy (V20) < 37%, and heart volume receiving ≥ 40 Gy (V40) < 100%. Following the publication of Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) in 2010, our institution attempted to limit the mean whole heart (WH) dose to ≤ 26 Gy (based on suggested limits to the pericardium).¹⁸ For reference, the National Comprehensive Cancer Network (NCCN) at the time recommended a mean WH dose ≤ 35 Gy.

The typical chemotherapy regimen consisted of intravenous infusional drug delivery with either weekly paclitaxel (45 mg/m²) and carboplatin (area under the curve = 2) or cisplatin (50 mg/m² on days 1, 8, 29, and 36) and etoposide (50 mg/m² on days 1–5 and 29–33).

Follow-up and Evaluation of Cardiotoxicity

Patients were typically seen by the radiation and medical oncologists every 1 to 3 months for the first year post-CRT, every 3 to 6 months for the following 2 years, and every 6 to 12 months thereafter. Follow-up chest CT scans were generally obtained 6 to 8 weeks after the completion of CRT, and then every 3 to 4 months for the first year, every 6 months for the following 2 years, and yearly thereafter.

We noted post-CRT symptomatic cardiac events via patient chart review. All available records were reviewed, including medical and radiation oncology follow-up notes, inpatient notes and discharge summaries, follow-up imaging (including CT, positron-emission tomography/CT, and echocardiogram reports), and outpatient cardiology notes. We defined five classes of symptomatic cardiac events, adapted from Wang et al.,³ as follows:

1. Acute coronary syndrome (ACS): myocardial infarction or unstable angina.

2. Significant arrhythmia: new onset arrhythmia requiring either medical or procedural intervention.

3. Symptomatic pericardial effusion: non-malignant effusions presenting with shortness of breath, confirmed on echocardiogram as hemodynamically significant, and/or requiring procedural intervention.

4. Pericarditis: radiographic-, echocardiographic-, or electrocardiogram-confirmed pericardial inflammation along with a presentation with shortness of breath or chest pain.

5. Congestive heart failure (CHF): new diagnosis of echocardiogram and cardiologist-confirmed heart failure or hospitalization for existing heart failure unassociated with other defined events.

Patients experiencing multiple events within a given class (e.g., ≥2 hospitalizations for CHF) were only counted for the earliest event experienced. However, we considered multiple events across different event classes for an individual patient.

For analysis of distinct event types, we grouped the five cardiac event classes defined above into three separate categories based on pathophysiology and number of events:

1. ACS/CHF events: ventricular/ischemic pathologies.

2. Supraventricular arrhythmic events: atrial pathologies.

3. Pericardial effusion/pericarditis events: pericardial pathologies.

We assessed baseline cardiac status by recording the presence of pre-treatment coronary artery disease (CAD), recording the presence of any pre-treatment cardiac disease (including CAD, significant arrhythmia, symptomatic pericardial disease, CHF unassociated with CAD, and symptomatic valvular disease), and calculating pre-treatment WHO/International Society of Hypertension (ISH) risk score. The WHO/ISH score, which is based on age, sex, blood pressure, smoking status, presence of diabetes mellitus, and epidemiological subregion, estimates the 10-year risk of a fatal or nonfatal major cardiovascular event in five strata (< 10%, 10% to < 20%, 20% to < 30%, 30% to < 40%, and ≥ 40%).¹⁹

Dosimetric Analysis

Treatment planning images were segmented (by one investigator, N.Y.R.) to define the WH, left ventricle (LV), right ventricle (RV), left atrium (LA), right atrium (RA), and left anterior descending artery (LAD), per previously published methods.²⁰ Image segmentations were reviewed for accuracy and consistency by a second investigator (S.K.J.). Based on the delivered RT plan, we extracted the mean dose, V5, V30, and volume receiving ≥ 50 Gy (V50) for each structure. We chose these metrics based on RTOG 0617.⁶

Study Endpoints

The primary endpoint was time to first symptomatic cardiac event. Secondary endpoints included time to first ACS/CHF event, supraventricular arrhythmic event, and pericardial event, and OS. We measured all endpoints from the end of RT to the event of interest.

Statistical Methods

We calculated median follow-up with the reverse Kaplan-Meier method. We estimated the rate of symptomatic cardiac events and distinct event types using the cumulative incidence method. Death in the absence of the event of interest was a competing event. We modeled the cumulative incidence function of each cardiac endpoint using Fine-Gray competing risks regression. We assessed associations between covariates of interest and each cardiac endpoint through univariate analysis. Dosimetric parameters were treated as continuous covariates and included depending on the endpoint: WH dosimetry for first symptomatic cardiac event; WH, LV, RV, and LAD dosimetry for first CHF/ACS event; and WH, left atrium (LA), and RA dosimetry for first supraventricular arrhythmic event. We found too few pericardial events for a separate time-to-event analysis. We considered clinically relevant covariates significant at the p = 0.15 level in the univariate setting for inclusion in the multivariate models. Because of the multicollinearity between different dosimetric parameters for a given structure, we only included mean dose in the multivariate models. Similarly, because different measures of baseline cardiac status are correlated, we included pre-treatment CAD, any pre-treatment cardiac disease, or pre-treatment WHO/ISH score, but not more than one such covariate, in any given multivariate model. We reported subdistribution hazard ratios (HRs) and 95% confidence intervals (CIs).

In addition to analyzing mean WH dose as a continuous covariate in the regression models for time to first symptomatic cardiac event, we stratified patients by mean WH dose in two ways, using: (1) The Contal and O’Quigley method to identify, in increments of 0.5 Gy, the cut-point that maximized the absolute value of the log rank statistic²¹; and (2) Previously suggested cut-points of < 10 Gy, 10 to 20 Gy, and ≥ 20 Gy, for comparisons of cumulative incidence functions via Gray’s test.³

We assessed associations between covariates and OS through univariate and multivariate Cox regression. Disease progression and symptomatic cardiac events were treated as time-dependent covariates for OS analysis. All hypothesis tests were two-sided and p < 0.05 was considered statistically significant. When analyzing associations between specific event types and doses to different cardiac subvolumes, we adjusted for multiple comparisons by applying the Bonferroni correction (p < 0.05 / number of cardiac structures tested). We performed analyses with SAS version 9.4 (Cary, North Carolina).

Results

Baseline characteristics are shown in Table 1. Median prescribed RT dose was 61.2 Gy (interquartile range: 60 to 66 Gy; range, 50.4 to 70.2 Gy). Baseline CAD was present in 30 patients (21.4%), and any baseline cardiac disease was present in 43 patients (30.7%). Baseline WHO/ISH 10-year risk was <10% for 70 patients (50%), 10% to 20% for 49 patients (35%), and ≥ 20% for 21 patients (15%). Most patients (63.6%) received concurrent carboplatin/paclitaxel. Dosimetric parameters to the WH and substructures are summarized in Supplementary Table 1. There was no association between mean WH dose and treatment year (Spearman’s ρ = 0.076, p = 0.37).

Table 1.

Baseline Characteristics

Characteristic	n (%)
Age (y)
Median	64
Range	40–85
Sex
Male	77 (55)
Female	63 (45)
ECOG PS
0	93 (66)
1	33 (24)
2	11 (8)
3	3 (2)
Baseline CAD
Yesa	30 (21)
No	110 (79)
Any baseline cardiac disease
Yesb	43 (31)
CAD	30
Significant arrhythmia	20
Symptomatic pericardial disease	1
CHF unassociated with CAD	2
Symptomatic valvular disease	3
No	97 (69)
Baseline WHO/ISH 10-year riskc
<10%	70 (50)
10% to 20%	49 (35)
≥20%	21 (15)
Clinical stage
IIA	8 (6)
IIB	7 (5)
IIIA	41 (29)
IIIB	69 (49)
IV (oligometastatic)	15 (11)
Histology
Adenocarcinoma	72 (51)
Squamous cell carcinoma	54 (39)
Poorly differentiated	14 (10)
Primary tumor laterality
Right	92 (66)
Left	48 (34)
GTV (cm3)
Median	126.9
Range	3.9–803.5
Prescribed dose (Gy)
Median	61.2
IQR	60–66
Range	50.4–70.2
RT Technique
3DCRT only	90 (64)
3DCRT/IMRT	23 (16)
IMRT only	27 (19)
Chemotherapy schedule
Induction	38 (27)
Concurrent	140 (100)
Consolidation	33 (24)
Concurrent chemotherapy agent(s)
Carboplatin + paclitaxel	89 (64)
Characteristic	n (%)
Cisplatin + etoposide	29 (21)
Other doublet	9(6)
Single agent	13 (9)

^aOf the 30 patients with CAD documented at baseline, 22 had myocardial ischemia requiring revascularization, 5 had myocardial ischemia confirmed on stress test and/or catheterization but not requiring revascularization, and 3 had CAD not otherwise specified.

^bOf the 43 patients with any cardiac disease documented at baseline, 13 had more than one type of cardiac disease: 10 had CAD and arrhythmia, 1 had CAD and valvular disease, 1 had arrhythmia and pericardial disease, and 1 had arrhythmia and valvular disease.

^cFor 6 patients, pre-treatment blood pressure was not available, so the average of the first two on-treatment blood pressure readings was used for calculation of the WHO/ISH 10-year risk.

ECOG PS, Eastern Cooperative Oncology Group performance status; CAD, coronary artery disease; CHF, congestive heart failure; ISH, International Society of Hypertension; GTV, gross tumor volume; IQR, interquartile range; RT, radiation therapy; 3DCRT, three-dimensional conformal radiotherapy; IMRT, intensity-modulated radiation therapy.

Symptomatic Cardiac Events

Median follow-up was 47.4 months (interquartile range: 32.2 to 74.0 months; range, 0.8 to 106.1 months). Forty patients (28.6%) developed 47 symptomatic cardiac events at a median of 15.3 months to first event post-CRT (range, 1 day to 75.3 months). Event details are summarized in Table 2 and Supplementary Table 2. The 1- and 4-year cumulative incidences of symptomatic cardiac events were 11.8% and 28.5% among all patients (Fig. 1A), 7.5% and 23.4% among patients without baseline CAD (Supplementary Fig. 1A), and 33.3% and 51.9% among patients with baseline CAD, respectively.

Table 2.

Symptomatic Cardiac Events Post-Chemoradiation Therapy^a

Event Class (n)	Description (n)	Intervention (n)
Acute coronary syndrome (8)	Fatal MI (1) Non-fatal MI (4) Unstable angina (3)	Stenting (3) Medical management (4)
New-onset arrhythmia (22)	Atrial fibrillation/atrial flutter (20) Multifocal atrial tachycardia (1) Ventricular tachycardia (1)	Ablation after unsuccessful medical management (3) Medical management ± attempted cardioversion (19)
Pericardial effusion (4)	Effusion with tamponade (2) Symptomatic effusion (2)	Pericardial window (2) Pericardiocentesis (1) Conservative management (1)
Pericarditis (1)	Constrictive pericarditis (1)	Conservative management (1)
Congestive heart failure (12)	New-onset, reduced EF (5) New-onset, preserved EF (2) Hospitalization for exacerbation, reduced EF (5)	Medical management (12)

^aSeven patients experienced two cardiac events within different classes, including congestive heart failure and supraventricular arrhythmia (n = 4), congestive heart failure and ventricular tachycardia (n = 1), and acute coronary syndrome and supraventricular arrhythmia (n = 2).

MI, myocardial infarction; EF, ejection fraction.

Figure 1.

(A) Cumulative incidence of first symptomatic cardiac event among all patients (n = 140), stratified by (B) mean whole heart dose ≥ 20 Gy versus < 20 Gy, and (C) mean whole heart dose ≥ 20 Gy, 10 to 20 Gy, and < 10 Gy. In (B) and (C), the curve for the ≥ 20 Gy group ends at 5.3 years because there were no more patients at risk.

On univariate analysis, all WH dosimetric parameters were associated with an increased risk of symptomatic cardiac events, as were baseline CAD and WHO/ISH stratum ≥ 20% versus < 10% (Table 3). Left-sided tumors (p = 0.088) and any baseline cardiac disease (p = 0.094) trended toward significance. In the multivariate models, baseline cardiac status (either CAD or WHO/ISH stratum ≥ 20%) and mean WH dose (as the representative cardiac dosimetric parameter) were the only factors associated with an increased risk of symptomatic cardiac events.

Table 3.

Regression Analysis for Time to First Symptomatic Cardiac Event

	Univariate Analysis			Multivariate Analysis
				Model 1a			Model 2b
	HR	95% CI	p Value	aHR	95% CI	p Value	aHR	95% CI	p Value
General
Age (y)	1.03	0.99–1.06	0.12	1.010	0.979–1.041	0.54
Gender (male vs female)	0.93	0.50–1.70	0.80
ECOG PS (≥1 vs 0)	0.82	0.42–1.60	0.57
Tumor laterality (left vs right)	1.71	0.92–3.16	0.088	1.28	0.66–2.50	0.46	1.28	0.65–2.54	0.48
GTV (cm3)	0.998	0.996–1.000	0.13
Prescribed dose (Gy)	0.978	0.914–1.045	0.51
IMRT only (vs all else)	0.57	0.23–1.39	0.21
IMRTonly + IMRT/3DCRT (vs 3DCRT only)	1.00	0.52–1.91	0.99
Concurrent chemo (carbo/taxol vs all else)	0.73	0.39–1.37	0.33
Induction chemo	0.89	0.46–1.73	0.74
Consolidation chemo	0.53	0.23–1.25	0.15
Baseline cardiac status
CAD	2.46	1.27–4.79	0.0080	3.54	1.79–6.99	0.0003
Any cardiac disease	1.73	0.91–3.26	0.094
WHO/ISH stratum (ref: <10%)
10% to 20%	1.09	0.52–2.30	0.81				0.96	0.46–2.02	0.92
≥20%	2.52	1.24–5.11	0.011				3.32	1.72–6.42	0.0004
Whole heart
Mean dose (Gy)	1.058	1.034–1.082	< 0.0001	1.065	1.038–1.093	< 0.0001	1.059	1.032–1.086	< 0.0001
V5 (%)	1.021	1.010–1.033	0.0001
V30 (%)	1.031	1.019–1.042	< 0.0001
V50 (%)	1.035	1.020–1.050	< 0.0001

^aModel used CAD as the representative metric for baseline cardiac status.

^bModel used WHO/ISH stratum as the representative metric for baseline cardiac status.

HR, subdistribution hazard ratio; CI, confidence interval; aHR, adjusted subdistribution hazard ratio; ECOG PS, Eastern Cooperative Oncology Group performance status; AC, adenocarcinoma; GTV, gross tumor volume; IMRT, intensity-modulated radiation therapy; 3DCRT, three-dimensional conformal radiotherapy; chemo, chemotherapy; carbo/taxol, carboplatin/paclitaxel; CAD, coronary artery disease; ISH, International Society of Hypertension; Vx, volume of structure receiving ≥ x Gy.

When stratified by mean WH dose ≥ 20 Gy versus < 20 Gy (identified by the Contal and O’Quigley method), the cumulative incidence curves for symptomatic cardiac events were significantly different (p = 0.0003) (Fig. 1B). The 1- and 4-year cumulative incidences were 23.4% and 48.6% for ≥ 20 Gy versus 7.7% and 18.5% for < 20 Gy. When stratified by mean WH dose ≥ 20 Gy, 10 to 20 Gy and < 10 Gy, the cumulative incidence curves were again significantly different (p = 0.0009) (Fig. 1C). The ≥ 20-Gy group had an increased risk of symptomatic cardiac events compared to both the 10- to 20-Gy (HR = 2.29, 95% CI: 1.18–4.46, p = 0.015) and <10 Gy (HR = 4.95, 95% CI: 1.76–13.9, p = 0.0024) groups, whereas the risk did not differ significantly between the 10- to 20-Gy and <10-Gy groups (HR = 2.16, 95% CI: 0.74–6.33, p = 0.16). Similar results were observed among patients without baseline CAD (Supplementary Fig. 1B and C) and a similar trend noted among those without any baseline cardiac disease (Supplementary Fig. 2).

ACS/CHF Events

Twenty patients experienced an ACS/CHF event, including myocardial infarction (n = 5), unstable angina (n = 3), new-onset CHF (n = 7), and hospitalization for CHF exacerbation (n = 5). The 1- and 4-year cumulative incidences were 5.8% and 16.0% (Supplementary Fig. 3), respectively. On univariate analysis, multiple dosimetric parameters (mean dose, V5, and V30) to the WH, LV, RV, and LAD associated with ACS/CHF events, as did baseline CAD, any baseline cardiac disease, and baseline WHO/ISH risk (Table 4). On multivariate analysis, mean dose to each structure retained significance when paired with baseline CAD and after adjusting for multiple comparisons (p < 0.013). Although not part of the original hypothesis, we found weak associations with mean LA dose (p = 0.024) and mean RA dose (p = 0.11), which were not significant after adjusting for multiple comparisons (p > 0.0083).

Table 4.

Regression Analysis for Time to First ACS/CHF Event (n = 20) and Supraventricular Arrhythmic Event (n = 21)

	Univariate Analysis			Multivariate Analysis
	HR	95% CI	p Value	aHR	95% CI	p Value
ACS/CHF event
Baseline cardiac status
CAD	3.48	1.44–8.38	0.0056	4.34	1.79–10.5	0.0012a
Any cardiac disease	3.08	1.29–7.35	0.011
WHO/ISH stratum (ref: <10%)
10% to 20%	3.16	1.10–9.10	0.033
≥20%	3.67	1.08–12.4	0.037
Whole heart
Mean dose (Gy)	1.051	1.020–1.083	0.0012	1.067	1.031–1.105	0.0003b
V5 (%)	1.021	1.007–1.035	0.0037
V30 (%)	1.028	1.011–1.044	0.0008
V50 (%)	1.024	1.001–1.047	0.039
Left ventricle
Mean dose (Gy)	1.035	1.010–1.061	0.0069	1.044	1.017–1.071	0.0013b
V5 (%)	1.013	1.002–1.024	0.023
V30 (%)	1.018	1.005–1.032	0.0078
V50 (%)	1.023	1.003–1.044	0.028
Right ventricle
Mean dose (Gy)	1.045	1.018–1.072	0.0009	1.057	1.026–1.090	0.0003b
V5 (%)	1.019	1.007–1.031	0.0024
V30 (%)	1.022	1.008–1.036	0.0016
V50 (%)	1.021	1.001–1.041	0.037
Left anterior descending artery
Mean dose (Gy)	1.041	1.015–1.068	0.0021	1.042	1.018–1.066	0.0005b
V5 (%)	1.016	1.005–1.028	0.0051
V30 (%)	1.020	1.007–1.034	0.0030
V50 (%)	1.017	0.996–1.039	0.11
Supraventricular arrhythmic event
Baseline cardiac status
CAD	1.53	0.59–3.98	0.38
Any cardiac disease	1.15	0.46–2.87	0.76
WHO/ISH stratum (ref: <10%)
10% to 20%	0.65	0.20–2.11	0.47
≥20%	3.35	1.37–8.17	0.0081
Whole heart
Mean dose (Gy)	1.028	0.990–1.067	0.15
V5 (%)	1.010	0.995–1.025	0.21
V30 (%)	1.014	0.994–1.035	0.16
V50 (%)	1.022	0.998–1.048	0.074
Left atrium
Mean dose (Gy)	1.015	0.987–1.044	0.29
V5 (%)	1.009	0.992–1.026	0.32
V30 (%)	1.008	0.993–1.023	0.31
V50 (%)	1.010	0.993–1.026	0.25
Right atrium
Mean dose (Gy)	1.007	0.987–1.028	0.51
V5 (%)	1.007	0.996–1.018	0.22
V30 (%)	1.002	0.990–1.014	0.72
V50 (%)	0.995	0.980–1.010	0.52

^aPaired with left ventricle mean dose.

^bPaired with coronary artery disease.

ACS, acute coronary syndrome; CHF, congestive heart failure; HR, subdistribution hazard ratio; CI, confidence interval; aHR, adjusted subdistribution hazard ratio; CAD, coronary artery disease; ISH, International Society of Hypertension; Vx, volume of structure receiving ≥ x Gy.

Supraventricular Arrhythmic Events

Twenty-one patients experienced a new onset supraventricular arrhythmic event, including atrial fibrillation/atrial flutter (n = 20) and multifocal atrial tachycardia (n = 1). The 1- and 4-year cumulative incidences were 6.0% and 13.9%, respectively. On univariate analysis, doses to the WH, LA, and RA were not associated with supraventricular arrhythmic events (Table 4). Although not part of the original hypothesis, we also found no associations with mean LV dose (p = 0.18) or mean RV dose (p = 0.19).

Pericardial Events

Five patients experienced a pericardial event, including effusion with tamponade (n = 2), symptomatic effusion (n = 2), and constrictive pericarditis (n = 1). The 1- and 4-year cumulative incidences were 1.5% and 4.2%, respectively. The low number of events precluded a reliable regression analysis. Nonetheless, the median mean WH dose was higher for these five patients than for the entire cohort (30.9 Gy versus 15.8 Gy).

Overall Survival

Median OS for the cohort was 24.1 months (95% CI: 15.0–34.9 months), and the actuarial 5-year OS was 28.7% (95% CI: 19.8–38.1%). Death was documented in 90 patients (64.3%) and disease progression in 97 patients (69.3%). On multivariate analysis, symptomatic cardiac events (adjusted HR = 2.62, 95% CI: 1.60–4.29, p = 0.0001), when treated as a time-dependent covariate, predicted for worse OS after accounting for baseline WHO/ISH risk and disease progression (Table 5). There were no direct associations between WH doses and OS.

Table 5.

Overall Survival Analysis

	Univariate Analysis			Multivariate Analysisa
	HR	95% CI	p Value	aHR	95% CI	p Value
General
Age (y)	1.045	1.022–1.068	0.0001
Sex (male vs. female)	1.31	0.86–2.01	0.21
ECOG PS (≥1 vs. 0)	1.26	0.82–1.96	0.30
Histology (non-AC vs. AC)	2.65	1.71–4.12	<0.0001	2.33	1.48–3.68	0.0003
Tumor laterality (left vs. right)	1.11	0.72–1.71	0.64
GTV (cm3)	1.002	1.000–1.003	0.014	1.001	1.000–1.003	0.061
Prescribed dose (Gy)	1.039	0.988–1.092	0.14
Concurrent chemo (carbo/taxol vs. all else)	0.84	0.55–1.28	0.41
Baseline cardiac status
CAD	1.87	1.16–3.00	0.0099
Any cardiac disease	1.78	1.15–2.74	0.0095
WHO/ISH stratum (ref: <10%)
10% to 20%	2.20	1.38–3.50	0.0010	1.99	1.22–3.23	0.0057
≥20%	2.00	1.12–3.57	0.019	1.98	1.08–3.62	0.026
Whole heart
Mean dose (Gy)	1.010	0.990–1.031	0.33
V5 (%)	1.003	0.995–1.012	0.43
V30 (%)	1.005	0.994–1.016	0.40
V50 (%)	1.008	0.994–1.024	0.27
Time-dependent
Disease progression	7.64	4.38–13.3	<0.0001	6.99	3.92–12.5	<0.0001
Symptomatic cardiac event	4.59	2.90–7.27	<0.0001	2.62	1.60–4.29	0.0001

^aModel used WHO/ISH stratum as the representative metric for baseline cardiac status.

HR, hazard ratio; aHR, adjusted hazard ratio; CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; AC, adenocarcinoma; GTV, gross tumor volume; chemo, chemotherapy; carbo/taxol, carboplatin/paclitaxel; CAD, coronary artery disease; ISH, International Society of Hypertension; Vx, volume of structure receiving ≥ x Gy.

Discussion

Within a modern cohort of patients with inoperable NSCLC, we note three main findings: (1) Symptomatic cardiac events appeared relatively common after standard-dose concurrent CRT, occurring in 28.6% of patients; (2) Cardiac doses were associated with an increased risk of symptomatic cardiac events, particularly ACS/CHF; and (3) These events were associated with a 2.6-fold increased hazard for death. Findings (2) and (3) persisted after controlling for baseline cardiac risk. These data add to the growing literature of cardiotoxicity after CRT for NSCLC, suggesting that such relationships can be significant even when using conventional RT doses.

Traditionally, most have considered RT-associated cardiotoxicity a late manifestation, occurring more than 10 years after treatment and therefore a concern for breast cancer and Hodgkin lymphoma survivors who face more favorable prognoses.^22–26 However, Marks et al.²⁷ observed new LV perfusion defects as early as 6 months post-RT in 27% of patients with breast cancer, and in 42% of patients by 2 years post-RT. Additionally, Darby et al.²⁸ found that the greatest percentage increase in the rate of major coronary artery events per Gy of heart irradiation occurred within the first four years after RT for breast cancer. These observations challenge the long-held belief that clinical RT-associated cardiac events represent a delayed toxicity. Patients with inoperable NSCLC usually receive higher cardiac doses than patients with breast cancer, and are often chronic smokers with significant cardiac risk factors, the combination of which likely accentuates RT-induced cardiac damage. The time course of RT-associated cardiotoxicity appears considerably shorter than previously recognized, as noted in this cohort of patients with NSCLC.

Although prior studies have shown a cardiac dose-toxicity relationship among patients with inoperable NSCLC receiving dose-escalated RT (median prescribed doses, 70 to 74 Gy^3,4,14,29), the evidence remains limited and conflicting among patients treated with more conventional RT doses (median prescribed doses, 60 to 66 Gy).^{10,14,30–32} Speirs et al.¹⁰ found that higher cardiac doses were associated with worse OS and grade ≥1 cardiotoxicities. However, they found no association between cardiotoxicities and OS, perhaps because the most common cardiotoxicity was pericardial effusions incidentally noted on follow-up imaging. Other studies found weak or no associations between cardiac doses and cardiac events.^30–32 The median follow-up for our cohort was 47.4 months, which we believe led to a more accurate reflection of cardiotoxicity. Pertinent studies on RT-associated cardiotoxicity for inoperable NSCLC are summarized in Supplementary Table 3.

Because different types of cardiac events reflect distinct pathophysiologic backgrounds, we also looked at the relative frequency of specific event types and their association with cardiac subvolume dosimetry. ACS/CHF events and supraventricular arrhythmic events appeared most common (4-year rates, 16% and 13.9%, respectively) whereas pericardial events appeared less frequent (4-year rate, 4.2%). ACS/CHF events strongly correlated with higher WH, LV, RV, and LAD doses, conceivably because these represent ventricular/ischemic pathologies. While RT potentially leads to accelerated late atherosclerosis, it potentially also leads to early microvascular changes which may manifest as early cardiac perfusion defects.^27,33–37 Moreover, prior studies found that higher ventricular doses associated with increased troponin levels³⁸, increased B-type natriuretic peptide levels³⁹, and impaired LV ejection fraction.⁴⁰ Although we found weak associations between ACS/CHF events and mean LA and RA dose, these associations might be expected given the multicollinearity between ventricular and atrial doses (Spearman’s ρ = 0.75, p < 0.0001 between mean LV dose and mean LA dose), and were not significant after adjusting for multiple comparisons.

We found no significant association between doses to the WH, LA, or RA and supraventricular arrhythmic events. Our study may lack adequate power to detect an association. Additionally, because arrhythmias are relatively common in the general population and may arise in the setting of noncardiac illness, the arrhythmia endpoint remains subject to additional confounders. Wang et al.¹⁴ also found arrhythmic events as the least specific in their dose-toxicity analysis.

The low rate of symptomatic pericardial events precluded an in-depth analysis. Whereas asymptomatic pericardial effusions occur in up to 40.3% of patients after RT, symptomatic pericardial effusions occur only in 0.7% to 6.3%.^4,14,41 Nevertheless, prior work found associations between cardiac doses and mostly asymptomatic pericardial or pleural effusions.^41,42

A few actionable steps may potentially limit cardiotoxicity after RT for inoperable NSCLC. First, on the basis of a high cardiac event rate among patients receiving mean WH dose ≥20 Gy (4-year rate, 48.6% in this study and 41% in Wang et al.³), a reasonable constraint for routine planning might be to keep mean WH dose <20 Gy, below the current recommendation of ≤26 Gy by the NCCN.⁴³ We note, however, that this may not necessarily represent an “optimal” threshold, as the risk of cardiotoxicity generally increased with increasing mean WH dose. Therefore, we advocate for the minimization of WH doses (e.g., mean, V5, V30, and V50) as much as possible, without compromising tumor coverage, and while respecting doses to other normal tissues. Most patients in our study (64%) received 3DCRT and not IMRT or proton therapy; the latter two techniques afford increased flexibility in steering/reducing cardiac, and other normal tissue, doses.^44,45 Second, because ACS/CHF events were relatively common and strongly associated with doses to the WH, LV, RV, and LAD, we believe that patients at higher risk of these types of events (e.g., those with baseline CAD) may benefit from more careful sparing of these cardiac structures. Third, we believe that pre- and post-CRT cardiac testing may better risk stratify patients and allow for timely initiation of cardioprotective agents to hopefully minimize RT-associated cardiotoxicity.

Our study shows several limitations. First, we assessed cardiotoxicity retrospectively, possibly underestimating the true frequency of events. However, the median follow-up of 47.4 months was relatively high, and we found similar event rates to Wang et al.³ and Dess et al.,⁴ studies which included patients on prospective clinical trials. The follow-up time, while relatively high for patients with NSCLC, is still limited. Cardiotoxicities seen in the first few years after RT may differ from those seen decades after RT. Second, because patients usually did not undergo pre-CRT cardiac testing, baseline cardiac morbidity was possibly underestimated. Nevertheless, we assessed baseline cardiac risk by recording pre-treatment CAD, any pre-treatment cardiac disease, and pre-treatment WHO/ISH 10-year risk stratum, which represent reasonable methods used by others.^3,28 Third, our ability to associate specific event types with doses to certain cardiac subvolumes was limited by the multicollinearity of subvolume dosimetry, incomplete follow-up for all patients which precluded area under the curve analyses, and the fact that 7 of 40 patients experienced multiple types of cardiac events. Fourth, because all patients received concurrent chemotherapy, with 85% receiving either carboplatin/paclitaxel or cisplatin/etoposide, we cannot exclude the possibility that chemotherapy agents contributed to cardiotoxicity. In a recent multicenter phase III trial, patients with stage III NSCLC randomized to receive carboplatin/paclitaxel experienced increased pulmonary toxicity and inferior 3-year OS compared to those randomized to receive cisplatin/etoposide.⁴⁶ However, in our cohort we found no association between carboplatin/paclitaxel use and either cardiotoxicity or OS. Fifth, we excluded from our analysis patients who did not receive concurrent chemotherapy or complete RT which likely led to a slightly more favorable median OS of 24.1 months. For such patients who face poorer prognoses, the risk of cardiotoxicity is likely more difficult to measure.

In conclusion, incidental cardiac irradiation appeared to increase the risk of clinically significant cardiotoxicity, particularly ACS and CHF, among patients with inoperable NSCLC receiving conventional RT doses. While additional study is warranted to better establish the dose/volume/outcome relationship for cardiotoxicity, we advocate a reasonable constraint for the present as a mean WH dose < 20 Gy.