Cardiac substructure-guided radiotherapy planning for esophageal cancer: a novel strategy for cardiotoxicity reduction

Abstract

Background

A novel method was proposed to protect the sensitive cardiac substructures in esophageal cancer patients and reduce the risk of cardiotoxicity.

Methods

A retrospective study was conducted on 20 patients with esophageal cancer. For each patient, a conventional clinical plan (c-IMRT) and a plan to protect sensitive cardiac substructures (p-IMRT) were designed. Dosimetric parameters of the two plans were calculated. The tumor control probability (TCP) of the target and the normal tissue complication probability (NTCP) of normal tissue were compared between two plans. Paired sample t test and Wilcoxon signed-rank test were used in to assess significant differences (p < 0.05) between data set.

Results

In the p-IMRT plans, the dose covering 2% of planning target volume (PTV) (D2) and homogeneity index (HI) of PTV were significantly higher than those in c-IMRT plans, D98 of PTV was significantly lower than that in c-IMRT plans. No significant differences in other indicators. The percentages of total lung volume receiving ≥ 5 Gy and 20 Gy (V5 and V20), as well as the maximum dose (Dmax) to the spinal cord, were comparable between the p-IMRT and c-IMRT plans. Compared with the c-IMRT plans, the p-IMRT plans significantly reduced the dose of six sensitive cardiac substructures, include left ventricle (LV), pulmonary artery (PA), left main artery (LMA), left anterior descending artery (LAD), left circumflex (LCX) and right coronary artery (RCA). Mean dose (Dmean) of LMA, LAD, LCX and RCA decreased by 23.41%, 45.45%, 27.84% and 28.18%, respectively. Dmax of LCX and RCA decreased by 12.42%, 27.32%, 22.03% and 24.62%. Notably, the NTCP of the heart in the p-IMRT plans (0.10 ± 0.11‰) was significantly lower than that in c-IMRT plans (0.13 ± 0.09‰).

Conclusions

The research shows that the novel method for protecting sensitive cardiac substructures can effectively reduce the dose of cardiac substructures without significantly increasing the dose of other organs at risk (OARs). By delineating sensitive cardiac substructures, it could significantly decrease the NTCP of the heart while increasing TCP of the target, thereby improving the quality of clinical plans.

Introduction

Esophageal cancer remains one of the most aggressive and lethal malignancies of the digestive tract globally [1]. Recent cancer statistics indicate that it constitutes approximately 3.2% of all new cancer diagnoses worldwide (ranking seventh) and is responsible for about 5.3% of total cancer-related deaths (ranking sixth) [2]. At present, radiotherapy is one of the most important treatment methods for patients with esophageal cancer [3, 4].

In general, cardiotoxicity is considered to be the main late side effect of radiotherapy [5, 6]. Evidence from treatments for Hodgkin’s lymphoma and breast cancer has established a clear link between thoracic irradiation and an elevated risk of cardiac injury, manifesting as conditions, such as ischemia, pericardial effusion, coronary artery disease, myocardial fibrosis, congestive heart failure, and myocardial infarction [7–13].

In esophageal cancer radiotherapy, the proximity of the planning target volume (PTV) to the heart inevitably results in considerable cardiac dose exposure. With the improvement of survival rate among esophageal cancer patients, the risk of radiotherapy-related cardiotoxicity has also increased. Therefore, cardiotoxicity in esophageal cancer radiotherapy has garnered growing attention.

In the current design of radiotherapy planning, planners and radiation oncologists assess cardiac risk through indicators of whole heart, such as mean heart dose (MHD) [14–16]. Some researchers have studied the relationship between the whole heart dose and overall survival (OS). The results showed that the volume of heart receiving dose ≥ 5 Gy (V5), V10 and V20 can be used as independent prognostic factors for OS [17].

However, the heart is an anatomically complex organ comprising distinct substructures, such as ventricle, coronary artery, left anterior descending artery (LAD) and other substructures. Whole-heart dose parameters, derived from averaging across the entire organ, fail to capture the specific irradiation of these individual components. Studies [18–23] have indicated that different cardiac substructures exhibit varying radiosensitivity, prompting researchers to focus on the radiation dose of cardiac substructures.

Atkins [24] et al. investigated the relationship between the radiation dose to cardiac substructures such as LAD and mortality in patients with non-small cell lung cancer (NSCLC). The results showed that the volume of LAD receiving ≥15 Gy (V15) was an independent factor related to mortality. The findings of Jang et al. [25] showed that higher radiation dose to the left venture (LV) may increase the risk of acute coronary syndrome event in patients with stage III NSCLC. In the context of esophageal cancer, Zhang et al. [26] analyzed the impact of cardiac substructure dosimetry on prognosis following intensity modulated radiation therapy (IMRT). The results demonstrated that both the mean and maximum doses to the right coronary artery (RCA) and LAD significantly influences OS.

The above research findings provide a theoretical basis for reducing the dose to sensitive cardiac substructure during the planning process. Therefore, we developed a novel planning strategy designed to selectively spare these sensitive cardiac substructures. It is expected that, without significantly increasing the dose to other organs at risk (OARs), the dose to cardiac substructures can be reduced, thereby lowering radiotherapy-induced cardiotoxic side effects and achieving the goal of further protecting the heart.

In this study, we delineated heart, left atrium (LA), right atrium (RA), LV, right ventricle (RV), left main artery (LMA), LAD, left circumflex (LCX), RCA, superior vena cava (SVC), inferior vena cava (IVC), ascending aorta (AA), descending aorta (DA), pulmonary artery (PA) and pulmonary vein (PV). Based on prior findings in the literature [24–26], LV, PA, LMA, LAD, LCX and RCA were selected as the sensitive cardiac substructures requiring special protection.

Two plans were designed for each esophageal cancer patient, namely, clinical IMRT plan (c-IMRT) and protect the sensitive cardiac substructure IMRT plan (p-IMRT). We compared the dosimetric parameters between the two plans and calculated the tumor control probability (TCP) for the target volume and the normal tissue complication probability (NTCP) for OARs. This evaluation aimed to determine whether the observed dosimetric improvements could translate into potential biological benefits. The findings of this study may provide a new idea for reducing cardiotoxicity and improving the prognosis of patients.

Materials and methods

Patients characteristic and image acquisition

This retrospective study included 20 patients with esophageal cancer who received radiotherapy at our department from July 2019 to September 2021. The median age of the patients was 72 (range 57–83), of which 18 were males and 2 were females. Among the 20 patients, 19 patients were squamous cell carcinoma and 1 was small cell carcinoma. Two patients underwent surgery, while the remaining 18 did not. All patients were in the supine position, fixed with thermoplastic mask with their arms placed alongside the body. Siemens SOMATOM Definition AS CT scanner system (Siemens Healthcare, Erlangen, Germany) was used to perform CT scan. The scanning range is from the upper edge of the second cervical vertebra (skull base) to the lower edge of the second lumbar vertebra, with a slice thickness of 5 mm. The scanned images were transmitted to Philips Pinnacle 9.10 3D treatment planning system (Philips health, Fitchburg, WI, USA).

Delineation of target and organs at risk

The target, spinal cord, total lung and heart of each patient were delineated by an experienced radiation oncology on Philips Pinnacle 9.10 3D treatment planning system (TPS) (Philips health, Fitchburg, WI, USA). For patients who did not undergo surgery, the gross tumor volume (GTV) was delineated on CT images with reference to case reports, esophagography and esophagoscopy reports. Clinical target volume (CTV) was defined as GTV expanded by 3 cm in craniocaudal direction and 0.6–0.7 cm radially, with large blood vessels and vertebral bodies entirely excluded from CTV. Taking into account the positioning errors, respiratory movement and other factors, the PTV was generated by expanding the CTV isotropically by 0.5–0.6 cm. For postoperative patients, preoperative CT images, esophagography, esophagoscopy report, surgical records and postoperative pathology was used to determine the location of the primary tumor. CTV was defined as the esophageal lymphatic drainage region. With consideration of positioning error and respiratory movement, PTV was generated by isotropically expanding the CTV by 0.6–1.0 cm. Total lung was defined as the sum of left and right lungs minus GTV. For heart, the upper boundary was started at the level of the lower edge of the pulmonary artery, crosses the midline and extends downward to the apex of the heart.

The delineation of cardiac substructures was performed in MIM Maestro Version 7.1.4 (MIM Software Inc, Cleveland, OD). First, the radiation oncologist delineated LA, RA, LV, RV, LMA, LAD, LCX, RCA, SVC, IVC, AA, DA, PA and PV of a patient. The cardiac substructures of the remaining 19 patients were automatically generated through the rigid and deformable registration function of the MIM software. Then, the radiation oncologist performed slice by slice examination, and finally completed the delineation of the cardiac substructures of 20 patients.

Treatment planning generation

Two plans were generated for each patient, c-IMRT and p-IMRT. All plans were created using the auto-planning module of pinnacle 9.10 TPS. The direct machine parameter optimization (DMPO) algorithm is applied with the tissue heterogeneity taken into account. The dose calculation was performed with a grid resolution of 3 mm. The prescription dose for the PTV was 50.4 Gy, delivered in 1.8 Gy per fractions, 5 days per week, using 6 MV photon beams generated by a Varian Edge accelerator. For fair comparison, each plan was normalized, such that a minimum of 95% of the PTV received 100% of the prescribed dose. In addition, the following dose constraints were applied according to the protocol of our department: V5 of total lung ≤ 50%, V20 of total lung ≤ 25%, mean dose of total lung (MLD) ≤ 15 Gy; maximum dose of spinal cord (Dmax) ≤ 50 Gy; V30 of heart ≤ 40%, V40 of heart ≤ 30%, and mean dose of heart (MHD) ≤ 26 Gy.

The beam angles of c-IMRT plan were determined based on factors, such as the shape, size, location and other factors of the tumor. The beam angles of p-IMRT plan were consistent with those of c-IMRT plan. If the dose constraints could not be met, the beam angles were slightly adjusted according to the specific clinical situation. In c-IMRT plan, the optimization objectives of whole heart were set. In p-IMRT plans, on the basis of the whole heart optimization, additional dose constraints were applied to sensitive cardiac substructures, including LV, PA, LMA, LAD, LCX and RCA. All other optimization settings remained the same.

Evaluation of treatment planning

In this paper, the dosimetric differences between p-IMRT plan and c-IMRT plan were compared. The dose–volume histogram was used to evaluate the dose of target and OARs. The mean dose (Dmean), the dose to 2% of PTV (D2), D98, conformal index (CI) and heterogeneous index (HI) were used for target evaluation. The evaluation parameters of OARs included V5, V20 and MLD of total lung, Dmax of spinal cord, MHD, V30 and V40 of heart, Dmax, Dmean and partial volume dose of 14 cardiac substructures. In addition, the monitor units (MU) were obtained for each plan.

Radiobiological modeling calculation

TCP of target and NTCP of total lung, heart and spinal cord were calculated to evaluate the biological differences caused by dose differences between treatment plans. TCP was calculated based on the EUD model [27] and using a MATLAB program, as shown in the following formula:

$$TCP=\frac{1}{1+{\left(\frac{TC{D}_{50}}{\mathit{EUD}}\right)}^{4{\gamma }_{50}}}$$ 1

$$EUD={\left[\sum _{i=1}\left(v_i{D}_i^{\alpha }\right)\right]}^{\frac{1}{\alpha }}$$ 2

where TCD₅₀ (tumor dose controlling 50% tumor) = 51.24 Gy and γ₅₀ (expected TCP change due to 1% dose change of TCD₅₀) = 0.83, α = 0.30. D_i is the uniform dose irradiated to the fractional volume V_i [28].

NTCP was calculated based on Lyman Kutcher Burman (LKB) model [29] using MATLAB program. The formula is as follows:

$$NTCP=\frac{1}{\sqrt{2\pi }}{\int }_{-\infty }^t{e}^{-\frac{x^2}{2}dx}$$ 3

$$t=\frac{D_{\mathit{eff}}-T{D}_{50}}{mT{D}_{50}}$$ 4

$$D_{\mathit{eff}}={\left(\sum _i{v}_i{D}_i^{1/n}\right)}^n$$ 5

where for pneumonia, the parameters published by Semenko et al. [30] were adopted, TD₅₀ = 29.9 Gy, n = 1, M = 0.41. For pericarditis, the parameters published by Martel et al. [31] were adopted, TD₅₀ = 50.6 Gy, n = 0.64, M = 0.13. For myelitis/necrosis, the parameters published by Luo [32] were adopted, with TD50 = 66.5 Gy, n = 0.05, and M = 0.175.

Per-voxel EQD2 conversion was performed before computing TCP and NTCP [33]. The formula is as follows:

$$EQD2=Di\ast \frac{{\text{d}}_{\text{i}}+\alpha /\beta }{2+\alpha /\beta }$$ 6

where di is the fractional dose. α/β is a tissue-specific LQ parameter of the exposed organ. In this study, α/β values of 3 Gy, 2 Gy, and 3 Gy were used for the lung, spinal cord, and heart, respectively.

Statistical analysis

SPSS 20.0 (IBM Corp., Armonk, NY, USA) statistical software was used for data analysis. To determine the statistical significance between groups, paired t test was used for parameters with normal distribution, and Wilcoxon signed-rank test was used for parameters without normal distribution. p < 0.05 was statistically significant.

Results

Case example

In this study, a total of forty plans were generated for 20 patients, all of which were considered clinically acceptable by radiation oncologist. To meet clinical requirements, the beam angles of the p-IMRT plans for 5 out of the 20 patients were slightly different from those of the c-IMRT plans. Figures 1 and 2 show a comparison of dose distribution and dose–volume histogram between two plans for the same patient.

Fig. 1.

Dose distribution of two plans for one patient

Fig. 2.

Dose–volume histogram of two plans for one patient

Dosimetric evaluations for PTV

Table 1 presents the data of PTV dosimetric parameters in the two plans. In the p-IMRT plan, D2 (54.08 ± 0.41 Gy) and HI (0.09 ± 0.01) were slightly higher than those in the c-IMRT plan, with statistically significant differences (p < 0.05). MU in p-IMRT plan was also significantly higher than that in c-IMRT plan (p < 0.05). D98 in p-IMRT plan was significantly lower than that in c-IMRT plan (p < 0.05). Dmean and CI were equivalent between p-IMRT plan and c-IMRT plan. No significant differences were observed between the two plans, with p values of 0.09 and 0.09.

Table 1.

PTV dosimetric comparison between p-IMRT and c-IMRT plans

	p-IMRT	c-IMRT	p value
Dmean (Gy)	52.35 ± 0.22	52.27 ± 0.20	0.09
D2 (Gy)	54.08 ± 0.41	53.86 ± 0.37	0.01
D98 (Gy)	49.55 ± 0.21	49.79 ± 0.24	< 0.001
CI	0.76 ± 0.06	0.78 ± 0.07	0.09
HI	0.09 ± 0.01	0.08 ± 0.01	< 0.001
MU	505.85 ± 69.40	445.70 ± 75.56	< 0.001

Figure 3 provides a more intuitive representation of the differences in PTV metrics between p-IMRT plan and c-IMRT plan, with the displayed box plots showing results.

Fig. 3.

Box plots for dosimetric dmetrics of PTV used for comparing p-IMRT and c-IMRT

Dosimetric evaluations for OARs

The dosimetric parameters of total lung and spinal cord in the two plans were summarized in Table 2. As shown in Table 2, there were no significant differences in V5, V20 of total lung and Dmax of spinal cord between p-IMRT plan and c-IMRT plan (p > 0.05). Although there was statistical difference in MLD, the absolute value difference was relatively small.

Table 2.

Total lung and spinal cord dosimetric comparison between p-IMRT and c-IMRT

	p-IMRT	c-IMRT	p value
Total Lung
MLD (Gy)	9.05 ± 2.36	8.92 ± 2.31	0.04
V5 (%)	41.36 ± 9.92	41.41 ± 10.13	0.68
V20 (%)	16.80 ± 5.89	16.79 ± 5.41	0.97
Spinal Cord
Dmax (Gy)	42.06 ± 1.81	42.27 ± 1.74	0.50

Bold means significant difference

Figure 4 provides the differences in total lung and spinal cord between p-IMRT plan and c-IMRT plan, with the displayed box plots showing results.

Fig. 4.

Box plots for dosimetric metrics for total lung and spinal cord in p-IMRT and c-IMRT

The dosimetric comparison results of heart and cardiac substructures are shown in Table 3. The results showed that except for the V40 of heart, Dmax of LA, PA, IVC, DA and Dmean of RA, SVC, IVC, DA, the dosimetric parameters of other heart and cardiac substructures in p-IMRT plan were lower than those in c-IMRT plan. MHD of heart, Dmean of LA, LV, PA, PV, LMA, LAD, LCX and RCA, V35 of LA, V5, V35 and V13 of LV, V10 of RV, V35 of PA, V15 and V30 of LAD, V15 of LCX, Dmax of LMA, LAD, LCX and RCA in p-IMRT plan were significantly lower than those in c-IMRT plan (p < 0.05). Compared with c-IMRT plan, in p-IMRT plan, Dmean of LMA decreased by 23.41% and Dmax decreased by 12.42%, Dmean of LAD decreased by 45.45% and Dmax decreased by 27.32%, Dmean of LCX decreased by 27.84% and Dmax decreased by 22.03%, Dmean of RCA decreased by 28.18% and Dmax decreased by 24.62%, dee Table 3 for details.

Table 3.

Heart and cardiac substructures dosimetric comparison between p-IMRT and c-IMRT

	p-IMRT	c-IMRT	p value
Heart
MHD (Gy)	22.00 ± 3.65	23.35 ± 3.38	< 0.001
V30 (%)	29.45 ± 6.06	30.57 ± 4.95	0.13
V40 (%)	16.96 ± 4.19	14.99 ± 4.63	0.02
LA
Dmean (Gy)	40.73 ± 4.97	41.71 ± 4.59	< 0.001
Dmax (Gy)	54.49 ± 0.68	54.06 ± 0.65	0.03
V35 (%)	73.17 ± 12.84	75.95 ± 12.17	< 0.001
LV
Dmean (Gy)	13.58 ± 7.33	16.19 ± 6.84	< 0.001
Dmax (Gy)	47.48 ± 8.36	48.26 ± 6.49	0.94
V5 (%)	72.46 ± 30.61	81.69 ± 23.82	< 0.001
V15 (%)	35.18 ± 27.42	48.37 ± 26.42	< 0.001
V23 (%)	19.37 ± 16.78	24.53 ± 17.06	< 0.001
RA
Dmean (Gy)	26.52 ± 8.24	24.87 ± 6.96	0.02
Dmax (Gy)	50.75 ± 4.53	49.64 ± 5.46	0.01
RV
Dmean (Gy)	17.30 ± 7.75	17.81 ± 7.12	0.20
Dmax (Gy)	43.03 ± 12.64	43.07 ± 11.02	0.68
V10 (%)	71.57 ± 28.30	76.16 ± 28.33	0.01
PA
Dmean (Gy)	25.64 ± 8.27	28.98 ± 8.75	< 0.001
Dmax (Gy)	52.37 ± 6.74	52.23 ± 6.65	0.34
V35 (%)	31.65 ± 14.74	38.17 ± 17.14	< 0.001
PV
Dmean (Gy)	30.10 ± 9.29	32.39 ± 8.46	< 0.001
Dmax (Gy)	52.50 ± 3.68	52.71 ± 2.02	0.84
SVC
Dmean (Gy)	31.26 ± 14.48	30.13 ± 13.88	0.01
Dmax (Gy)	46.85 ± 11.18	46.96 ± 10.98	0.94
IVC
Dmean (Gy)	24.54 ± 16.93	23.85 ± 16.63	0.04
Dmax (Gy)	41.10 ± 20.13	41.00 ± 19.78	0.54
AA
Dmean (Gy)	30.05 ± 10.66	30.52 ± 10.03	0.09
Dmax (Gy)	50.42 ± 6.79	50.63 ± 4.77	0.33
DA
Dmean (Gy)	39.02 ± 5.98	38.92 ± 5.84	0.44
Dmax (Gy)	54.72 ± 0.59	54.38 ± 0.55	0.44
LMA
Dmean (Gy)	24.28 ± 10.27	31.70 ± 8.59	< 0.001
Dmax (Gy)	33.71 ± 10.42	38.49 ± 8.01	< 0.001
LAD
Dmean (Gy)	12.05 ± 6.66	22.09 ± 7.91	< 0.001
Dmax (Gy)	21.95 ± 10.74	30.20 ± 10.31	< 0.001
V15 (%)	36.50 ± 36.77	78.61 ± 35.50	< 0.001
V30 (%)	1.63 ± 3.60	15.79 ± 25.50	0.01
LCX
Dmean (Gy)	17.13 ± 11.79	23.74 ± 11.23	< 0.001
Dmax (Gy)	25.76 ± 14.21	33.04 ± 11.44	< 0.001
V15 (%)	47.46 ± 41.92	76.74 ± 34.44	< 0.001
RCA
Dmean (Gy)	14.68 ± 7.60	20.44 ± 6.05	< 0.001
Dmax (Gy)	21.0 ± 10.63	27.86 ± 8.64	< 0.001

Figure 5 illustrates the dosimetric differences in the heart and substructure between the two plans.

Fig. 5.

Box plots for dosimetric metrics of heart and substructure used for comparing p-IMRT and c-IMRT

TCP of target and NTCP of total lung, heart and spinal cord are summarized in Table 4. TCP in p-IMRT plan was significantly higher than that in c-IMRT plan (p < 0.05). NTCP of heart in p-IMRT plan was significantly lower than that in c-IMRT plan (p < 0.05). No statistical differences were observed in other parameters (p > 0.05).

Table 4.

TCP and NTCP comparison between p-IMRT and c-IMRT

	p-IMRT	c-IMRT	p value
TCP (%)	51.76 ± 0.36	51.60 ± 0.31	0.02
NTCP (‰)
Total Lung	47.22 ± 16.13	45.94 ± 15.35	0.03
Heart	0.10 ± 0.11	0.13 ± 0.09	0.02
Spinal Cord	3.69 ± 1.90	3.97 ± 1.98	0.27

Discussion

In this study, we aimed to propose a novel method to protect sensitive cardiac substructures in patients with esophageal cancer. We explored the dosimetric and biological improvements achieved with this method compared with conventional radiotherapy plans for esophageal cancer. The results showed that compared with conventional radiotherapy plans, our novel method significantly reduced the dose to sensitive cardiac substructures while maintaining comparable doses to other OARs, and such dosimetric reduction may translate into biological benefits. Dmean of LV, PA, LMA, LAD, LCX and RCA decreased by 16.12%, 11.53%, 23.41%, 45.45%, 27.84% and 28.18%, respectively, and NTCP of heart decreased by 23.08%. These results provide supporting data for the new method proposed in this study to protect sensitive cardiac substructures in patients with esophageal cancer.

Radiation-induced heart disease is one of the major complications of radiotherapy for esophageal cancer [34]. Studies have reported that patients who receive radiotherapy are 1.62 times more likely to die from heart disease than those who do not [35]. With advancements in radiotherapy technology, the survival rate of esophageal cancer patients has been significantly improved. Consequently, there is growing research focus on the cardiotoxicity associated with esophageal cancer radiotherapy [36, 37]. Due to varying radiosensitivity among different cardiac substructures, accumulating evidence suggests that specific cardiac substructures such as the LV, PA, LMA, LAD, LCX, and RCA may better predict patient prognosis. To date, most research on cardiac substructure protection has focused on Hodgkin's disease, breast cancer, and lung cancer, while there is relatively little research in esophageal cancer. Compared with Hodgkin’s lymphoma and breast cancer, esophageal cancer radiotherapy typically involves a higher prescription dose; moreover, the proximity of the target volume to the heart results in greater cardiac radiation exposure. Therefore, investigating strategies to protect cardiac substructures in esophageal cancer holds considerable clinical significance. Based on previous studies, this study proposed a novel method to protect LV, PA, LMA, LAD, LCX and RCA, aiming to reduce radiation-induced heart disease and improve the prognosis of patients.

In conventional radiotherapy planning for esophageal cancer, planners set dose constraints for the whole heart. During the evaluation of radiotherapy plans, both planners and radiation oncologists assess cardiotoxicity using whole-heart dosimetric parameters, such as MHD, V30 and V40 of heart. Cardiac substructures are generally not considered during OARs contouring, plan design, or plan evaluation. Compared with conventional radiotherapy plans for esophageal cancer patients, the key innovation of our proposed method for protecting sensitive cardiac substructures lies in the incorporation of specific optimization objectives for these substructures during the planning process. This method effectively achieved dose reduction to sensitive cardiac substructures in a simple way without increasing the risk of other OARs. Among all the patients included in this study, 75% (15 out of 20) could achieve the goal of reducing the dose to sensitive cardiac substructures and improving cardiotoxicity our method without changing the beam angles. Thus, this method can be applied in clinic practice conveniently without significantly increasing the difficulty for planners in design plan.

All 40 plans of 20 patients in this study met the clinical requirements. Table 1 shows that, compared with c-IMRT plans, CI of PTV was equivalent in the p-IMRT plans, while the HI was significantly reduced. In Table 4, TCP of PTV in p-IMRT plans was significantly higher than that in c-IMRT plan. This improvement in TCP may be attributed to the significantly increased D2 and Dmean to the PTV in the p-IMRT plans, which concurrently led to a reduction in dose uniformity. Mu in p-IMRT plan was significantly higher than that in c-IMRT plan. As the target of esophageal cancer patients is located close to heart, to better protect sensitive cardiac substructures, it may increase the complexity of the plan to a certain extent, resulting in the increase of MU. However, the absolute difference between p-IMRT plan and c-IMRT plan was small, and the impact of increased MU on treatment implementation efficiency, patient comfort, and plan robustness are clinically acceptable.

As indicated by Stam et al., [18] reducing cardiac dose at the expense of increased dose to the total lung may not be an optimal strategy in esophageal cancer radiotherapy. Therefore, in this study, we evaluated the dosimetric parameters of total lung. It can be discovered from Table 2 that compared with c-IMRT plans, p-IMRT plans did not increase the dose to total lung and other OARs, confirming that the novel method proposed in this study can significantly reduce the dose of sensitive cardiac substructures without increasing the dose of other OARs.

The results in Table 3 present the dose difference of each cardiac substructure between p-IMRT and c-IMRT plans. It can be found that in addition to the six sensitive cardiac substructures we focused on protecting, including LV, PA, LMA, LAD, LCX and RCA, other substructures such as RV and PV also received a certain degree of protection. This finding is similar to that of Turtle et al. [15] they conducted research on cardioprotective radiotherapy in patients with locally advanced non-small cell lung cancer and found that purposeful reduction of cardiac dose could reduce the interlocking of cardiac substructures, such as RA and LV. In p-IMRT plans, the Dmax of LA, RA, PA, IVC and DA were slightly higher than those in c-IMRT plans, but the increased absolute doses were very small, 0.43 Gy, 1.11 Gy, 0.14 Gy, 0.1 Gy and 0.34 Gy, respectively. Moreover, Dmax represents a point dose, its impact on patient treatment and related side effects is likely negligible.

Table 4 summarizes the differences in TCP and NTCP between p-IMRT and c-IMRT plans. TCP and NTCP of heart in p-IMRT plans were significantly better than those in c-IMRT plan (p < 0.05). But the absolute differences appear marginal. Whether these differences may translate into meaningful clinical benefit need to be further investigation through prospective study. The heart NTCP in this study was calculated based on the dosimetry results of the whole heart rather than on the dose of cardiac substructure. The formula for calculating NTCP is relatively simple and may not fully reveal the biological differences between the two plans. However, it is the most effective tool we currently have. Future studies calculating NTCP based on cardiac substructure doses could allow for a more comprehensive analysis of the biological effects of substructure protection. However, as substructure-specific NTCP parameters (TD50, m, n) have not yet been established, the whole-heart model was employed in this study. Exploring the NTCP of cardiac substructures is one of the future research directions. We can collect prospective data for NTCP modeling, and obtain TD50, m, and n required for calculating the NTCP of cardiac substructures through fitting, so as to derive the NTCP of cardiac substructures more accurately.

There are also some limitations in this study. First, this study is a retrospective study, and a prospective study may better verify our results. Second, all esophageal cancer patients included in this study were treated with prescription dose of 50.4 Gy. The extent of cardiac substructure protection may vary with different dose regimens. Third, the use of other new radiotherapy techniques, such as proton radiotherapy, may enable the novel method for protecting cardiac substructures proposed in this study to achieve clinical benefits.

Conclusion

This study explored a novel approach to reduce cardiotoxicity in patients with esophageal cancer by decreasing the dose to sensitive cardiac substructures. Compared with the conventional radiotherapy plans, the proposed method achieved a reduction in the Dmean to these substructures ranging from 11.53% to 45.45%. In addition, the NTCP of heart decreased by 23.08%. These findings suggest a promising strategy for enhancing cardiac protection in esophageal cancer radiotherapy and provide theoretical support for future prospective clinical trials.

Abbreviations

MHD

Mean heart dose

OS

Overall survival

NSCLC

Non-small cell lung cancer

IMRT

Intensity modulated radiation therapy

OAR

Organ at risk

LA

Left atrium

RA

Right atrium

LV

Left ventricle

RV

Right ventricle

LMA

Left main artery

LAD

Left anterior descending branch

LCX

Left circumflex

RCA

Right coronary artery

SVC

Superior vena cava

IVC

Inferior vena cava

AA

Ascending aorta

DA

Descending aorta

PA

Pulmonary artery

PV

Pulmonary vein

TCP

Tumor control probability

NTCP

Normal tissue complication probability

GTV

Gross tumor volume

CTV

Clinical target volume

PTV

Planning target volume

DMPO

Direct machine parameter optimization

Dmean

Mean dose

D2

Dose received by the hottest 2%

D98

Dose received by the coolest 98%

CI

Conformity Index

HI

Homogeneity Index

V5

Volume receiving at least 5 Gy

V20

Volume receiving at least 20 Gy

MLD

Mean lung dose

MU

Machine unit

Dmax

Maximum dose

Author contributions

Authors Yan Shao and Zhangru Yang contributed equally to the manuscript. Yan Shao: Data Collection and Analysis, Interpretation of Results, Writing and Revision of the Manuscript. Zhangru Yang: Delineation Contours, Interpretation of Results, Writing and Revision of the manuscript. Aihui Feng: Interpretation of Results, Review and Editing. Hengle Gu, Hao Wang, Hua Chen, Yanhua Duan, Ying Huang and Zhenjiong Shen Review and Editing. Zhiyong Xu: Conception and Design of the Study, Review and Editing, Approval of the Final Version.

Funding

National Natural Science Foundation of China (Grant No. 82303947, 82403785, 81803044),. Natural Science Foundation of Shanghai (Grant No. 24ZR1464200).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study is a retrospective study. When the study began, all selected patients signed informed consents and completed radiotherapy. Ethical standards and patients’ confidentiality were ensured and in line with regulations of the local institutional review board and data safety laws. This study was approved by the Ethics Committee of Shanghai Chest Hospital (the committee's reference Number: KS24052).

Consent for publication

All authors of the manuscript have read and agreed to its content and are accountable for all aspects of the accuracy and integrity of the manuscript.

Competing interest

The authors declare no competing interests.