Dosimetric and LET_d comparison of horizontal port and gantry carbon-ion radiotherapy for early-stage non–small cell lung cancer: an exploratory study

Article information

Radiat Oncol J. 2026;.roj.2026.00080

Publication date (electronic) : 2026 August 24

doi : https://doi.org/10.3857/roj.2026.00080

Seung-Hyeop Baek^1,*, Keith M. Furutani^2,*, Seyjoon Park¹, Justin Chunjoo Park^2,3, Min Cheol Han³, Changhwan Kim³, Jin Sung Kim³, Chris J Beltran², Kyung Hwan Kim^,3, Chae-Seon Hong^,3

¹Department of Radiation Oncology, Yonsei Cancer Center, Seoul, Republic of Korea

²Department of Radiation Oncology, Mayo Clinic Florida, Jacksonville, FL, USA

³Department of Radiation Oncology, Yonsei Cancer Center, Heavy Ion Therapy Research Institute, Yonsei University College of Medicine, Seoul, Republic of Korea

Correspondence: Chae-Seon Hong Department of Radiation Oncology, Yonsei Cancer Center, Heavy Ion Therapy Research Institute, Yonsei University College of Medicine, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 03722, Korea E-mail: CS.HONG@yuhs.ac

Kyung Hwan Kim Department of Radiation Oncology, Yonsei Cancer Center, Heavy Ion Therapy Research Institute, Yonsei University College of Medicine, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 03722, Korea E-mail: KYUNGHKIM@yuhs.ac

*These authors contributed equally to this work.

Received 2026 January 15; Revised 2026 July 3; Accepted 2026 July 6.

Abstract

Introduction

Radiotherapy, along with chemotherapy and surgery, is one of the principal treatment modalities for cancer. Since the discovery of X-rays in 1895, radiotherapy has undergone continuous advancement for over a century [1]. During the early to mid-20th century, kilovoltage energies were commonly employed; however, subsequent technological progress led to the adoption of megavoltage energies as the clinical standard. This transition enabled the development of a range of megavoltage-based treatment techniques, including three-dimensional conformal radiotherapy, intensity-modulated radiotherapy (IMRT), volumetric modulated arc therapy, and stereotactic body radiotherapy. Among these, IMRT is particularly well-suited for cases involving complex anatomy or tumors located adjacent to critical organs at risk (OARs), and it is therefore widely used in the treatment of head and neck, breast, prostate, and lung cancers [2-6].

In 1946, Wilson first proposed that the Bragg peak characteristic of protons could be exploited for cancer treatment [7]. In 1954, clinical studies were initiated at the University of California, Berkeley [8]. In these early investigations, proton beams were not applied to malignant tumors; instead, irradiation of the pituitary gland was performed to treat disorders associated with abnormal hormone production [9,10]. At present, proton therapy is commonly used in the management of various malignancies, including head and neck, prostate, lung, and liver cancers [11-15].

Carbon ions, similar to protons, can deliver highly focused radiation to the target by exploiting the Bragg peak, and they exhibit less lateral scattering than protons [16]. In addition, at the Bragg peak, carbon ions demonstrate a higher relative biological effectiveness (RBE), approximately 2–3 times that of protons, along with a higher linear energy transfer (LET). Owing to these advantages, carbon ion radiotherapy (CIRT) has emerged as one of the most promising radiotherapy modalities worldwide, and it is particularly attractive for treating radioresistant malignancies such as sarcomas, melanomas, and certain adenocarcinomas [17-19]. The National Institute of Radiological Sciences in Japan, which initiated carbon ion therapy in 1994, is among the few institutions with over 30 years of experience in CIRT, and it consistently reports clinical outcomes [20-22]. Additional CIRT facilities have now been established in Gunma, Osaka, and Yamagata in Japan [23-25], as well as in Heidelberg, Germany [26]. In South Korea, the Yonsei Cancer Center began CIRT in 2023, becoming the first facility in the country to offer this treatment [27]. In addition, the Mayo Clinic Florida is currently constructing the first CIRT facility in the United States [28].

CIRT equipment can be broadly categorized into fixed-beam port and gantry systems, each with distinct advantages and limitations. Fixed-beam port systems use horizontal and/or vertical beam ports and are associated with lower installation costs. In contrast, gantry systems enable irradiation from patient-specific optimal beam angles, providing greater flexibility in treatment delivery; however, they require higher installation costs and involve more complex mechanical structures. Accordingly, gantry systems are currently available in only four of the 17 CIRT facilities worldwide, and in only five of the 53 treatment rooms [29].

In lung cancer treatment, CIRT can reduce the volume of low-dose irradiation to normal lung tissue owing to its small lateral penumbra compared with other treatment modalities. In studies of early-stage non–small cell lung cancer (NSCLC), Miyamoto et al. [30,31] reported that CIRT achieved outcomes comparable to those of surgery for stage I disease. Demizu et al. [32] further demonstrated that CIRT provides a superior dose distribution compared with stereotactic body radiotherapy, thereby offering a safer option for patients with impaired lung function. Similarly, Okano et al. [33] reported excellent local control with CIRT and no significant adverse effects in patients with early-stage lung cancer and interstitial lung disease. For locally advanced NSCLC, Hayashi et al. [34] suggested that CIRT may represent an effective treatment alternative for patients who are unable to undergo surgery or chemotherapy, with acceptable toxicity profiles.

In CIRT, beam angle selection is an important consideration in treatment planning because it affects target dose conformity, normal tissue sparing, and the spatial distribution of the high-LET region near the distal end of the beam. Although gantry systems offer greater flexibility, fixed-beam port systems remain more widely used due to cost and facility-related limitations. Several studies have therefore compared the feasibility of fixed-beam port systems with that of gantry systems. Koom et al. [35] showed that, in pancreatic cancer, a gantry system achieved equivalent target coverage while reducing dose to surrounding organs compared with a fixed-beam port system. Miyasaka et al. [36] compared a fixed-beam port system with couch roll to a gantry system in liver cancer and reported no significant differences in doses to OARs, except for the clinical target volume (CTV) and gross tumor volume. Yagi et al. [37] evaluated nine cancer sites, including lung cancer, and found that a horizontal port (HP) system with couch roll was effective for centrally located tumors, whereas its effectiveness varied according to tumor location. These findings suggest that the feasibility of HP-based CIRT for lung cancer remains uncertain because of tumor location, thoracic anatomy, and tissue heterogeneity. Moreover, although dose-averaged LET (LET_d) is increasingly recognized as an important metric in CIRT, LET_d-based comparisons between HP and gantry systems for lung cancer remain limited. Therefore, this exploratory study evaluated the feasibility of HP-based CIRT for early-stage NSCLC by comparing RBE-weighted dose and LET_d metrics with those of gantry-based plans.

Materials and Methods

1. Patient selection and image acquisition

Five patients who underwent CIRT for lung cancer between June 2024 and January 2025 were included in this study, and their clinical characteristics are summarized in Table 1. All patients received respiratory training before undergoing a four-dimensional computed tomography (4D CT) scan. The scan was then acquired under free-breathing conditions using a 4D CT scanner (Aquilion ONE Vision, Canon Medical Systems, Otawara, Japan), with respiratory signals recorded using an Anzai respiratory monitoring system (AZ-733V, Anzai Medical, Tokyo, Japan).

Table 1.

Patient characteristics

2. Treatment planning

The CTV was delineated by a radiation oncologist in each phase of the 4D CT images. Both gantry and HP plans were generated using RayStation 2024A (RaySearch Laboratories, Stockholm, Sweden) with the pencil-beam dose-calculation algorithm. RBE-weighted doses were calculated using the microdosimetric kinetic model [38]. The prescribed dose was 60 Gy (RBE) to the CTV, delivered in four fractions. Dose constraints for the CTV and OARs were as follows: CTV V95% (volume receiving at least 95% of the prescribed dose) > 99.9%; ipsilateral lung V5 Gy (RBE) (volume receiving at least 5 Gy [RBE]) < 55% and V20 Gy (RBE) < 25%; spinal cord Dmax (maximum dose within the organ) < 15 Gy (RBE) and V16 Gy (RBE) < 0.2 cm³; bronchus V26 Gy (RBE) < 0.2 cm³; trachea V30 Gy (RBE) < 0.2 cm³; and esophagus V20 Gy (RBE) < 0.2 cm³. Treatment planning was performed using a single-field uniform dose optimization approach, with each fraction delivering 15 Gy (RBE). Fig. 1 shows the couch coordinate system and the setup configurations for the gantry and HP plans. As this was a retrospective study, the HP plans were generated in the gantry system using the available computed tomography (CT) images. The intended HP configuration consisted of horizontal beam directions, corresponding to gantry angles of 90° or 270°, with a ±15° couch roll and a ±30° couch yaw. Since the required couch roll exceeded the allowable range in the treatment planning system, it was represented by modifying the gantry angle. Accordingly, HP configurations with gantry angles of 90° and 270° combined with a ±15° couch roll were implemented using gantry angles of 75°/105° and 255°/285°, respectively (Table 2). For four-dimensional robust optimization, setup and range uncertainties were set to 3 mm and 3.5%, respectively. The gating phase was selected from phases in which the displacement of the CTV isocenter is within 2 mm relative to the end-of-exhale phase (Table 3).

Fig. 1.

(A) Definition of couch rotation directions (yaw, pitch, and roll). Yaw is the rotation around the posterior–anterior axis, pitch is the rotation around the right–left axis, and roll is the rotation around the inferior–superior axis. For a head-first supine patient, positive angles are defined as clockwise rotations when viewed from below, right, and feet for yaw, pitch, and roll, respectively. (B, C) The beam configurations for patient A in the treatment planning system, corresponding to the gantry system (B) and the horizontal port (HP) system emulated using the gantry system (C), respectively. In this figure, G, Y, and R denote the gantry angle, couch yaw, and couch roll, respectively. Parentheses (≈) represent the intended beam configurations for the HP system.

Table 2.

Gantry and couch angle information for each patient

Table 3.

Treatment planning and computation settings

Results

1. Dosimetric evaluation

Fig. 2 shows the mean DVHs for the CTV and OARs of the five patients in the gantry and HP plans. The evaluated DVH parameters, including the CI and HI, are summarized as mean ± standard deviation in Table 4, and the individual patient values are provided in Supplementary Table S1. CTV V95% was 100% in all plans, satisfying the institutional dose constraints. The ipsilateral lung V5 Gy (RBE) was 12.4% ± 2.77% for the gantry plans and 13.54% ± 1.19% for the HP plans, while the corresponding V20 Gy (RBE) values were 5.77% ± 0.75% and 6.15% ± 0.30%, respectively. The maximum dose to the spinal cord was 1.07 ± 1.92 Gy (RBE) for the gantry plans and 1.87 ± 1.51 Gy (RBE) for the HP plans. The V16 Gy (RBE) of the spinal cord, V30 Gy (RBE) of the trachea, and V20 Gy (RBE) of the esophagus were 0.0 cm³ in all cases. The bronchus V26 Gy (RBE) was 0.09 cm³ only in the HP plan of patient C and 0.0 cm³ in all other cases. The CI was 0.34 ± 0.07 for the gantry plans and 0.32 ± 0.07 for the HP plans, while the HI was 0.02 ± 0.01 and 0.02 ± 0.00, respectively.

Fig. 2.

Mean dose–volume histograms for the clinical target volume (CTV) and organs at risk in the gantry and horizontal port (HP) plans for the five patients. RBE, relative biological effectiveness.

Table 4.

Dose-volume histogram parameters of the gantry and HP plans

2. LET_d evaluation

Mean LET_d–volume histograms showed that the minimum LET_d within the CTV was higher in the gantry plan than in the HP plans (Fig. 3). Mean V40 keV/μm and LET_d98% values for the CTV are presented in Table 5, with values for each patient in Supplementary Table S2. V40 keV/μm for the CTV was 100% in all plans. Mean LET_d98% of the CTV was 65.27 keV/μm for the gantry plans and 57.07 keV/μm for the HP plans, with the gantry plans showing a higher value by 8.2 keV/μm.

Fig. 3.

Mean dose-averaged linear energy transfer (LET_d)–volume histograms for the gantry and horizontal port (HP) plans in five patients. The red vertical line indicates the LET_d threshold of 36.4 keV/μm reported by Li et al. [42] as a reference value related to local recurrence in non–small cell lung cancer. CTV, clinical target volume.

Table 5.

LET_d-volume histogram parameters of the gantry and HP plans for five patients

3. Beam range in water

Fig. 4 presents the beam energy and beam range in water for the five patients. In all patients, the beam energy and beam range were lower and shorter, respectively, under the gantry plan than under the HP plan.

Fig. 4.

Box plots of beam energy and corresponding range in water for the gantry and horizontal port (HP) plans in each patient. The left y-axis represents beam energy (MeV/u), and the right y-axis represents range in water (g/cm²).

Discussion and Conclusion

This study evaluated the clinical feasibility of using the HP system in CIRT for five patients with NSCLC. In the dosimetric evaluation, the CTV coverage achieved with the HP plan was comparable to that of the gantry plan, and doses to OARs satisfied institutional constraints. One limitation of the HP system is the restricted beam angle selection, which may lead to increased OAR doses in certain scenarios; however, in this study, adequate separation between the target and adjacent OARs allowed all constraints to be met. Consistent with these findings, Yagi et al. [37] reported no significant differences between gantry and HP plans. In lung tissue, the relatively low stopping power ratio can result in dose extension beyond the distal edge of the target. As illustrated in Fig. 5, the low-dose region extended predominantly in the anterior–posterior direction in the gantry plan, whereas extension occurred mainly in the superior–inferior direction in the HP plan.

Fig. 5.

Representative dose distributions of the gantry and horizontal port plans for patient A. RBE, relative biological effectiveness.

For LET_d evaluation, both the gantry and HP plans achieved a V40 keV/µm of 100%, with mean LET_d98% values of 65.27 keV/µm and 57.07 keV/µm, respectively. Li et al. [42] reported that, among 62 patients with locally advanced NSCLC, 16 experienced local recurrence. In that study, the mean LET_d98% was 37.5 keV/µm in the local control group and 32.0 keV/µm in the local recurrence group, while the corresponding mean V40 keV/µm values were 89.2% and 75.4%. Based on these results, Li et al. [42] proposed that the LET_d98% of the internal gross tumor volume should exceed 36.4 keV/μm and that V40 keV/µm should exceed 88%. In the present study, both planning approaches satisfied these LET_d criteria. For the ipsilateral lung, LET_d-based evaluation metrics have not yet been established. Although relatively high LET_d values were observed in the ipsilateral lung, these regions were confined to low-dose areas, and their clinical relevance is therefore likely to be limited.

Beam ranges in water differed between the gantry and HP plans due to differences in beam angle and tissue heterogeneity along the beam paths. The beam ranges in the gantry plan were shorter than those in the HP plan for all patients. Although the differences in beam range may be associated with the LET_d differences observed within the CTV between the two plans, LET_d is influenced by multiple factors, such as beam arrangement, tissue heterogeneity, beam modulation, and optimization strategy. Therefore, these factors should be considered collectively when evaluating LET_d.

This study has several limitations. First, the cohort was small, comprising only five patients; therefore, statistical significance was not evaluated, which limits the generalizability of the findings. Second, implementation of the HP system requires CT scans to be acquired with couch roll. However, additional CT scans with couch roll were not acquired because this was a retrospective study. Therefore, the HP configuration was emulated in the gantry system by adjusting the gantry angle instead of applying couch roll. Third, the potential uncertainties associated with couch roll were not evaluated in this study. When couch roll is performed, internal organs and soft tissues may shift or deform due to gravitational effects [43]. Such anatomical changes may affect the beam range [44]; however, these uncertainties were not considered in this study. Finally, the pencil beam algorithm used in this study may introduce dose calculation uncertainties in heterogeneous regions, such as the lung, compared with Monte Carlo calculations [45-47].

In conclusion, within this small exploratory feasibility study, the HP plan showed preliminary dosimetric findings that were generally comparable to those of the gantry plan. However, given the small cohort, these findings should be interpreted with caution and not be overgeneralized. When non-planning factors such as setup stability, organ motion, and reproducibility are considered, gantry-based delivery may offer greater clinical reliability. In addition, HP-based treatment would require two CT simulations and deformable image registration, whereas the gantry system eliminates these steps, potentially reducing clinical workload and improving overall treatment efficiency.

Supplementary Materials

Article information Continued