Introduction

Breast cancer is the most common malignant tumor globally, particularly among women1. Adjuvant radiotherapy plays a critical role in the comprehensive treatment of breast cancer by reducing local recurrence and disease-specific mortality2. However, for left-sided breast cancer patients, the proximity of vital organs such as the heart and lungs to the treatment area increases the risk of radiation-induced heart disease (RIHD) and radiation pneumonitis (RP). Therefore, optimizing dose distribution and protecting critical organs are essential to improving long-term patient outcomes.

Traditional radiotherapy typically employs the free-breathing (FB) technique, where respiratory motion may enlarge the target volume, thereby increasing radiation exposure to normal tissues. To address this issue, the deep inspiration breath-hold (DIBH) technique was introduced. Initially described in breast cancer radiotherapy in 20013, DIBH has gained widespread adoption only in recent years4. By performing radiation delivery during deep inspiration, DIBH increases lung volume, displaces the heart away from the radiation field, and reduces tumor movement, thereby improving both organ sparing and dosimetric precision5,6.

This study compares the dosimetric parameters of clinical target volume (CTV), heart, lungs, and other organs at risk (OARs) under DIBH and FB conditions in patients undergoing postoperative radiotherapy for left-sided breast cancer. It aims to determine the mode that optimally balances target coverage and OAR sparing, providing a basis for precision radiotherapy.

Materials and methods

Patient selection

A total of 94 patients with left-sided breast cancer who underwent modified radical mastectomy or breast-conserving surgery between February 2023 and March 2024 were included. Inclusion criteria were: (1) females aged 18 to 75 years; (2) pathological confirmation of left-sided breast cancer; (3) planned postoperative chest radiotherapy; (4) no prior breast reconstruction; and (5) adequate performance status (Karnofsky Performance Status ≥ 70; ECOG-PS 0–1). Exclusion criteria included: (1) inability to cooperate with respiratory control due to fear or respiratory gating intolerance; (2) limited arm mobility; (3) DIBH duration < 30 s; and (4) comorbid respiratory conditions such as asthma or bronchitis.

All patients practiced DIBH under professional guidance prior to CT simulation to ensure consistent breath-hold amplitude and duration. The study was approved by the hospital’s ethics committee (approval number S2024-758-01), and all participants provided informed consent.

Methods

CT simulation

CT simulation was performed after repeated DIBH training to ensure breath-hold durations exceeded 30 s. Patients were immobilized using breast boards with both arms raised above their heads. Lead markers were placed on the affected breast for reference, and a Siemens SOMATOM Definition AS 64-slice CT scanner was used for imaging. Scans covered the region from the mandible’s lower edge to 10 cm below the inframammary fold (approximately the T10 level) with a 5-mm slice thickness. Both FB and DIBH CT datasets were acquired for all patients, FB CT scan was firstly performed.

Target volume delineation

CT images were transferred to the Eclipse 13.1 (Varian) treatment planning system. Target volumes and OARs were contoured by the same attending physician following Radiation Therapy Oncology Group (RTOG 1304) guidelines7, with subsequent review by a senior physician to minimize interobserver variability. They were blinded to the breathing technique. The CTV included the chest wall, whole breast, or involved lymph nodes, as appropriate.

Treatment planning

Radiotherapy plans were designed under both DIBH and FB conditions using either intensity-modulated radiotherapy (IMRT) or volumetric-modulated arc therapy (VMAT). Treatment areas were determined by N-stage. For N0 patients, only the chest wall or breast was irradiated; for N1 patients, the chest wall/breast and internal mammary nodes were included; and for N2–3 patients, the supraclavicular region was additionally covered. Prescription doses were selected based on clinical indications: 36.5 Gy (10 fractions), 42.5 Gy (16 fractions), or 50 Gy (25 fractions).

Dosimetric analysis

Dosimetric parameters for CTV and OARs were extracted from dose-volume histograms (DVHs) under both breathing modes. Parameters included maximum dose (Dmax), mean dose (Dmean), minimum dose (Dmin), D98, D95, D90, D50, D10, D5, D2, homogeneity index (HI), conformity index (CI)8, and coverage for CTV. OARs dosimetric parameters include: (1) heart: Dmax, Dmean, Dmin, D0.35, D5 and D15; (2) left lung: Dmax, Dmean and Dmin; (3) right lung: Dmax, Dmean and Dmin; (4) whole lung: Dmax and Dmean; (5) esophagus and spinal cord: Dmax and Dmean. D98, D95, D90, D50, D10, D5 and D2 represented the irradiation doses received by 98%, 95%, 90%, 50%, 10%, 5% and 2% of the target volume, respectively. All dose parameters were converted to equivalent doses of 2 Gy fractions (EQD2) to facilitate comparison across different fractionation regimens. EQD2 = nd [(d + α/β) ÷ (2 + α/β)], where n represented the number of fractions, d represented the dose per fraction (Gy), right breast, and lung α/β were taken as 4, heart and esophagus α/β were taken as 3, and spinal cord α/β was taken as 2 (citations for α/β values selection as shown in Table S1). The dose parameters of CTV need to consider repopulation effects and were converted according to the method proposed by Dale et al.9.

Statistical analysis

Statistical analysis was performed using SPSS 21.0. Quantitative data were expressed as mean ± standard deviation. Paired t-tests were used for normally distributed data, while Wilcoxon signed-rank tests were applied for non-normally distributed data. Independent sample t-test was used for non-paired data that conformed to normal distribution, and Mann-Whitney U test was used for data that did not conform to normal distribution. The false discovery rate (FDR) corrected P values (term Q values) were calculated in addition to raw P values to adjust for multiple comparisons. Statistical significance was set at Q < 0.05. The normality test adopted the Shapiro-Wilk test, and P > 0.05 was considered to conform to a normal distribution. As an exploratory analysis, receiver operating characteristic (ROC) analysis was performed to find out the threshold value of the lung volume expansion for cardiac sparing which may help in selecting the patient for DIBH.

Results

Patients characteristics

The median age of patients was 47.5 years (range: 29–71). Among the cohort, 42 patients received IMRT and 52 received VMAT. Sixty-two patients received 36.5 Gy (10 fractions), 20 received 42.5 Gy (16 fractions), and 12 received 50 Gy (25 fractions). The T and N stages were shown in Table 1.

Table 1 Patients characteristic.
Full size table

CTV dosimetric parameters

DIBH significantly improved the CI of the CTV (Q < 0.05) but had no significant impact on HI. D95, D90, and coverage were significantly reduced under DIBH (Q < 0.05). Other parameters, including Dmax and Dmean, showed no significant differences between the two-breathing mode, as shown in Table 2. The D90 of CTV under two breathing modes showed good consistency, with an average decrease of 0.097 GyEQD2 in DIBH compared to FB, as shown in Fig. 1. In addition, there were significant differences in the distribution of CI and Coverage, as shown in Fig. 2.

Table 2 Comparison of CTV dosimetric parameters under two breathing modes (\(\bar{x}\)±s).
Full size table
Fig. 1
Fig. 1
Full size image

Bland-Altman plot of the D90 of CTV under two breathing modes. There was a good consistency for D90 between DIBH and FB, with an average decrease of 0.097 GyEQD2. SD: standard deviation.

Fig. 2
Fig. 2
Full size image

Violin plots of CI and Coverage of CTV under two breathing modes. Compared with FB, DIBH has a higher median CI and a lower median coverage.

Cardiac dose sparing

DIBH significantly reduced cardiac Dmax, Dmean, Dmin, D0.35, D5, and D15 compared to FB (Q < 0.05), with the most substantial reduction observed in D0.35 (20.739 GyEQD2). The mean heart dose under DIBH was 2.089 GyEQD2, a reduction of 2.631 GyEQD2 compared to FB. The differences were statistically significant (Q < 0.05), as shown in Table 3.

Table 3 Comparison of cardiac dosimetric parameters under two breathing modes (\(\bar{x}\)±s).
Full size table

Lung dose sparing

DIBH significantly reduced Dmax and Dmean for both the left and right lungs (Q < 0.05). The volumes of the left and right lungs during DIBH were significantly larger than that during FB (Q < 0.05), as shown in Table 4.

Table 4 Comparison of lung dosimetric parameters under two breathing modes (\(\bar{x}\)±s).
Full size table

Esophagus and spinal cord

Compared to FB, DIBH significantly reduced the maximum and mean doses to the esophagus and spinal cord (Q < 0.05), as shown in Table 5.

Table 5 Comparison of dosimetric parameters of other oars under two breathing modes (\(\bar{x}\)±s).
Full size table

The relationship between the dose of oars and lung volume

A weak positive correlation was observed between the increase in left lung volume during DIBH and the decrease in the mean dose to OARs. Specifically, the correlation coefficients were as follows: for the heart, R = 0.37 (P < 0.05); for the left lung, R = 0.39 (P < 0.05), as illustrated in Fig. 3. Furthermore, ROC curve analysis indicated that the increase in left lung volume possesses a certain predictive capability regarding the reduction of the Mean Heart Dose (MHD) by 2.631 GyEQD2. The area under the curve (AUC) of the ROC analysis was determined to be 0.714, with a threshold value of 807.1 cc for the increase in left lung volume, as shown in Fig. 4.

Fig. 3
Fig. 3
Full size image

Distribution diagram of ∆left lung volume and ∆mean OARs dose. Left lung extension volume in DIBH was weakly positively correlated with cardiac and left lung dose reduction, for the heart, R = 0.37 (P < 0.05); for the left lung, R = 0.39 (P < 0.05).

Fig. 4
Fig. 4
Full size image

ROC curve for ∆left lung volume predicting > 2.631 GyEQD2 reduction of MHD (corresponding to 22.1% reduction of the relative risk of acute cardiac events) with a cut-off value of 807.1 cc. AUC: area under the curve.

Discussion

Breast cancer is one of the most prevalent tumors globally, affecting a large number of women each year. The treatment of breast cancer typically encompasses surgical resection, radiotherapy, and adjuvant pharmacotherapy. Postoperative radiotherapy plays a pivotal role in the management of breast cancer10; however, due to the proximity of the left breast to critical structures such as the heart and lungs, conventional radiation therapy may subject these organs to harmful radiation doses, resulting in adverse effects including cardiac and pulmonary injuries. Thus, minimizing damage to normal tissues while ensuring the efficacy of the treatment has become a crucial consideration in postoperative radiotherapy.

DIBH represents a respiratory control strategy utilized in the field of radiotherapy. By guiding the patient to hold their breath during deep inhalation, DIBH leverages the chest’s expansion to displace the heart and lungs away from the breast tissue, consequently reducing the radiation doses received by these vital organs11,12,13 and enhancing the accuracy of dose distribution14. Numerous studies have demonstrated that DIBH can markedly decrease doses to the heart, left anterior descending artery (LAD), and ipsilateral lung in the context of left breast cancer radiotherapy15. Nevertheless, there remains a paucity of research addressing the dosimetric implications of DIBH for the target area in postoperative radiotherapy, and its efficacy in improving target area dose remains a topic of debate. This study aims to compare dosimetric parameters of the CTV, heart, lung, and other organs at risk under both DIBH and FB modes, thereby elucidating the advantages of employing DIBH in postoperative radiotherapy for left breast cancer.

Our findings revealed no statistically significant differences in the Dmax and Dmean of CTV between DIBH and FB modes, aligning with the results reported by Tang et al.16. Gaal et al.17 and Yeh et al.8 observed a reduction in HI under the DIBH mode, with Yeh’s study reporting significant results; conversely, Gong et al.18 identified an upward trend in HI. Our results suggest that DIBH technology does not markedly enhance HI. Similarly, the CI findings in Yeh’s study mirrored our own8, while Gong’s study indicated a negligible effect of DIBH on CI18. A higher CI value denotes greater conformity of the radiation therapy plan to the target volume. Additionally, our study documented significant reductions in D90, D95, and coverage; however, no marked differences were found in Dmin, D98, D50, D10, D5, and D2. Therefore, an acceptable radiation dose to the CTV is maintained through the DIBH technology.

Exposure of the heart to radiation elevates the risk of adverse cardiac events post-breast cancer treatment. Darby et al.19 demonstrated that for every 1 Gy increment in mean heart dose throughout the follow-up period, the incidence of acute cardiac events (ACEs) rises by 7.4%. During the first nine years following radiation exposure, the relative increase was approximately 16%, corroborated by findings from Van Den Bogaard et al.20. Moreover, Darby et al. noted a robust correlation (correlation coefficient of 0.98) between mean heart dose and biological effective dose (EQD2), with an increase of 8.4% in the probability of ACEs for every 1 GyEQD2 increment in mean heart dose. In our research, DIBH mode resulted in significant reductions in maximum, minimum, and mean heart doses, alongside D0.35, D5, and D15. A multitude of studies have substantiated that DIBH effectively minimizes mean heart dose21,22,23. Our findings indicated that the mean heart dose in DIBH was 2.089 GyEQD2, reflecting a 2.631 GyEQD2 decrease compared to FB, translating to a relative risk reduction of approximately 22.1% for ACEs post-radiation therapy. Furthermore, we observed a notable decrease in maximum heart dose due to DIBH, consistent with the work of Ferdinand et al.24 and Ferini et al.25. Rudat et al.26 established that the relative increase in left lung volume associated with DIBH was a good predictor for the 0.56 Gy decrease in mean heart dose ([AUC] = 0.89, n = 26), demonstrating significant correlation (R = 0.65, P = 0.0003). Similarly, Gaál et al.16 reported a significant relationship (R = 0.40, P < 0.001) between increases in left lung volume and reductions in the mean heart dose, closely mirroring our findings (R = 0.37, P < 0.05).

In clinical settings, the mean lung dose serves as a commonly utilized correlate for radiation pneumonitis. Some studies indicate a significant association between dual lung’s V5 and radiation-induced lung injury27. Currently, reports on the dose impact of DIBH on the ipsilateral lung exhibit inconsistency, with certain studies asserting that it fails to reduce radiation dose28. In contrast, our investigation revealed significant reductions in Dmax and Dmean for both left and right lungs under DIBH. Research conducted by Yamauchi et al.29 demonstrated significant decreases in Dmean for both lungs, including a marked reduction in V5 for the left lung. Ferdinand et al.24 documented a significant decrease in mean dose for the left lung, although the whole lung’s mean dose did not exhibit significant decline. Song et al.30 identified substantive changes in Dmean for both lungs, as well as V5, V10, and V20 for the left lung, with findings consistent with those of Chen et al.31, who observed significant changes in V30 for the left lung. Our results align with these findings, underscoring the capability of DIBH to diminish the radiation dose to the ipsilateral lung. Oechsner32 explored the difference of V20 for the left lung in three directions and Mankinen et al.33 found that DIBH could reduce normal tissue complication probability (NTCP) for clinical pneumonitis. These are very valuable, and our future research will study these indicators. Furthermore, we identified that DIBH significantly reduced both maximum and mean doses delivered to the esophagus and spinal cord. Dumane et al.34 found the maximum dose to the esophagus under FB was higher than under DIBH, while other studies reported that DIBH can decrease both maximum and mean doses to the spinal cord18,35. Although these findings did not achieve statistical significance, the observed trends were congruent with our results.

The influence of DIBH on the radiotherapy target dose remains a subject of ongoing debate. Nonetheless, the outcomes regarding its effects on organs at risk appear more consistent. The DIBH technique effectively reduces doses administered to at-risk organs, particularly the heart and left lung. This investigation distinguished itself by conducting a self-controlled experiment based on a larger dataset, enhancing the diversity of samples while mitigating individual differences’ impact on the results. Additionally, employing paired tests for dose parameters facilitates the detection of significant differences with improved precision.

This study has the following limitations: This study is a single-center study, and the selected sample cannot represent all patients. Clinical indicators such as body mass index (BMI), age, and previous heart disease were not analyzed in our study. In addition, we did not report any clinical outcome and evaluate the advanced toxicity of radiotherapy. After that, we will further evaluate the efficacy and safety, and conduct long-term follow-up of adverse reactions and survival results. Through ROC analysis, left lung volume expansion may be a parameter that predicts the reduction of MHD, but a further verification is required.

Conclusion

Although DIBH has a marginal reduction in target coverage, it offers superior cardiac and pulmonary protection in left-sided breast cancer radiotherapy compared to FB, especially with a mean cardiac dose reduction of 2.631 GyEQD2. Its integration into clinical practice should be prioritized for suitable patients to enhance therapeutic outcomes and minimize toxicity.