Analysis of the Association Between Low Dose Bone, Lung, and Heart Irradiation and Survival Rates for Patients Who Received High-dose Proton Beam Therapy With Concurrent Chemotherapy for Stage III Non-small Cell Lung Cancer

Masatoshi Nakamura University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Hitoshi Ishikawa University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Kayoko Ohnishi (  ohnishi@pmrc.tsukuba.ac.jp ) University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei https://orcid.org/00000002-6497-5099 Yutarou Mori University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Keiichiro Baba University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Kensuke Nakazawa University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Toshihiro Shiozawa University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Ikuo Sekine University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Kazushi Maruo University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Toshiyuki Okumura University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei Hideyuki Sakurai University of Tsukuba Faculty of Medicine: Tsukuba Daigaku Igaku Iryokei


Background
The standard treatment for patients with unresectable and locally advanced stage III non-small cell lung cancer (NSCLC) is chemoradiotherapy (CRT) [1,2]. Recently, the PACIFIC study revealed that durvalumab after CRT for NSCLC improved treatment outcomes of patients with stage III NSCLC [3], and immunooncology is becoming more popular in clinical practice. With the advent of immune checkpoint inhibitors (ICI), immuno-oncology is garnering attention in the eld of radiation oncology. Furthermore, lymphocytes, especially T-cell lymphocytes, are known to play an important role in the cancer immune system [4,5]. Some studies have reported that, in the treatment for various cancers, survival rates are associated with lymphopenia and neutrophil lymphocyte ratio (NLR) as representative markers [6][7][8][9][10].
In the eld of radiotherapy (RT), many previous studies have attested to the important roles of radiationinduced immune response in the success of cancer treatment and there has been renewed attention to this subject since the introduction of ICIs [11][12][13]. It is well known that lymphocytes are radiosensitive cells and that RT-induced lymphopenia results from a decrease of circulating lymphocytes in the lung and heart and depletion of progenitor cells in the bone marrow and spleen [14][15][16][17]. However, the lung, heart, and lymphoid organs like bone marrow are exposed to unnecessary radiation doses during thoracic RT. These days, the ability to calculate dose volume histograms (DVHs) of the targets and organs at risk (OARs) enables the examination the effects of irradiation on various OARs [18]. However, despite advances in RT and the adoption, in many facilities, of intensity modulated radiotherapy (IMRT), which enables a more intensive high-dose irradiation of the target than three-dimensional conformal RT, concerns remain over lower doses of radiation to the normal tissues becoming more diffuse [19].
Among the more recent developments in RT, proton beam therapy (PBT) is recognized for its unique ability to deliver high-dose irradiation to the target while reducing unnecessary irradiation of healthy tissues, because a spread-out Bragg peak of protons can be created to match the depth and thickness of the target [20,21]. Therefore, compared to X-ray RT, PBT can potentially yield better clinical outcomes and is also considered to have the advantage of minimizing RT-induced lymphopenia. Regarding the use of PBT in NSCLC, a randomized trial has been conducted for locally advanced NSCLC [22]. As for the effect of irradiation on the lung, which is regarded as a blood pool, it has been reported that lymphopenia is associated with lung V5, and that lower lymphocyte nadirs during RT were correlated with worse overall survival (OS) [23]. However, no report has investigated the effects of irradiation of the bone marrow on lymphopenia and patient prognosis in RT or PBT combined with chemotherapy for stage III NSCLC.
Therefore, the purpose of the current study was to analyze the clinical outcomes of patients with stage III NSCLC who received de nitive PBT with concurrent chemotherapy and examine the associations with survival rates and doses to normal tissues including lung, heart, and bone marrow.

Patient population
Data of 41 patients with stage III unresectable locally advanced NSCLC who received de nitive PBT at 74 GyE with concurrent chemotherapy between November 2007 and December 2017 at our institution were retrospectively reviewed. Patient characteristics are shown in Table 1. There were 31 men and 10 women, and the median age was 62 years (range = 42-79 years). According to the 7th version of the Union for International Cancer Control TNM classi cation, the clinical stage was IIIA in 12 patients and IIIB in 29, and histopathological examination revealed squamous cell carcinoma in 11, adenocarcinoma in 24, and NSCLC in 6. Proton beam therapy For treatment planning, chest computed tomography (CT) images were taken at 2.5-mm or 5.0-mm intervals with the patients in a body cast in the treatment position (Engineering System Co., Matsumoto, Japan) using a respiratory-gated system during the end-expiratory phase. Passive-scattering PBT plans were constructed, and dose calculations were performed using the pencil beam method for PBT (Proton Treatment Planning Software version 1.7 or 2, Hitachi Inc., Ibaraki, Japan). Proton beams of 155 to 250 MeV were used in the treatment plans. The treatment planning system automatically estimated the conditions required for beam delivery, which included a ridge lter, range shifter, collimator, and bolus. The beam delivery system created a homogenous dose distribution at the prescription dose using the spread-out Bragg peak.
In general, initial clinical target volume (CTV1) encompassed the primary tumor, the positive lymph nodes, and hilar and mediastinal lymph nodes as prophylactic areas where clinically positive lymph nodes existed. Clinically positive lymph nodes were de ned as nodes measuring ≥ 1 cm (as visualized on CT) or as positron emission tomography (PET)-positive lymph nodes. Second CTV (CTV2) encompassed the primary tumor and the positive lymph nodes, and the third CTV (CTV3) included only the primary tumor.
The planning target volume (PTV) encompassed the CTV with a 7-to 10-mm margin in all directions and an additional 5-mm margin in the caudal direction to compensate for the respiratory motion. After delivering a dose of 40 GyE in 20 fractions to the PTV1, 66 GyE in 33 fractions was delivered to the PTV2, followed by a total boost of 74 GyE in 37 fractions to the PTV3. In general, two to three ports in the optimal direction were used to meet the following dose constraints: the percentage of the lung volume receiving a dose of ≥ 20 GyE (V20) ≤ 35%, maximum dose to the spinal cord < 46 GyE biologically equivalent dose in 2 GyE per fraction (EQD2), maximum dose to the esophagus < 70 GyE (EQD2), and maximum dose to the bronchus < 70 GyE (EQD2).

Dosimetry analysis and evaluation of blood cell counts
Dosimetry parameters of patients were obtained from available DVHs of the bone, lung, and heart. In the present study, the vertebrae from Th1 to Th10, the bilateral rst to seventh ribs, and the whole sternum were contoured on planning chest CT as bone for DVH analysis. The organ contoured as bone included all irradiated bones in every patient.
During PBT with concurrent chemotherapy, complete blood count (CBC) was performed at least once a week. When grade 3 or severe cytopenia according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 occurred, CBC was performed at least twice a week until cytopenia improved to grade ≤ 2. For evaluation of maximum and minimum absolute lymphocyte counts (ALCmax and ALCmin), and maximum NLR (NLRmax), results of CBCs performed from the rst to the last day of PBT were used, whereas CBCs within 2 to 3 days after administration of steroids used as antiemetic drugs were excluded.

Follow-up and statistical analysis
The patients were followed up with a physical examination, chest radiography, blood test, CT or PET/CT, and magnetic resonance imaging every 2-3 months during the rst year and at 3-to 6-month intervals thereafter. Local progression at the primary site was de ned as an increase in tumor size, signi cant positive accumulation on PET/CT, or histological diagnosis. Regional recurrence was de ned as regrowth or new lymphadenopathy in the hilar, mediastinum, or supraclavicular lesion. Distant metastasis was de ned as failure at any other site. Adverse events were assessed according to the CTCAE version 4.0.
The follow-up interval was de ned from the rst day of PBT to the date of death or the last follow-up. The OS, progression-free survival (PFS), distant metastasis-free survival (DMFS), local progression-free (LPF), and regional control (RC) rates were calculated from the rst day of PBT to the date of that event or the last follow-up using the Kaplan-Meier method. Signi cant differences between survival curves were assessed using the generalized Wilcoxon test and Cox proportional hazard model. A p value < 0.05 was considered signi cant. SAS version 9.4 (SAS Institute, Cary, NC, USA) was used for the statistical analyses.

Survival and locoregional control
Because the correlation coe cients of lung parameters were similar, lung V5 was used as an evaluation value for lung dose based on the radiosensitivity of lymphocytes, in accordance with previous reports [23,24]. Heart doses did not correlate with ALCmax, ALCmin, and NLRmax. Heart V5 was used as the evaluation value in accordance with a previous report [19]. Scatter plots of the bone, lung, and heart V5 versus these three indicators of lymphopenia are shown in Fig. 2.  (Fig. 3). Table 3 shows the patient characteristics of the high and low ALCmin groups. There were more women than men in the high ALCmin groups, but sex (male vs. female) was not a factor associated with OS (p = 0.962), PFS (p = 0.855), and DMFS (p = 0.507) in this study population.

Discussion
We conducted this study hypothesizing that, in patients with stage III NSCLC undergoing PBT, irradiation of bone tissues might cause lymphopenia owing to depletion of progenitor cells, which would, in turn, reduce antitumor immunity and thereby in uence survival. Our ndings revealed that bone V5 and lung V5 correlated with lymphopenia and that lymphopenia was signi cantly associated with survival rates. In this study, however, there was no signi cant impact of bone V5 on survival rates, but lung V5 affected to DMFS on the multivariate analysis.
In RT with concurrent chemotherapy for NSCLC, it has been pointed out that low-dose irradiation of the thoracic vertebral body is associated with grade ≥ 3 leukopenia, which can result in poor survival and control rates owing to incomplete chemotherapy or treatment that could not be performed as planned [25]. Some studies have also reported that doses to the bone were related to the survival rate in CRT for other primary tumors or with radiation alone [26,27]. In the present study, the lymphocyte count tended to decrease with increasing bone irradiation doses, but lung V5 rather than bone V5 strongly correlated with lymphopenia. Conceivably, lung V5 irradiation doses increased relative to the bone V5 irradiation dose, and it is possible that lymphopenia, caused by the increase in doses to the bone, may be a spurious correlation (Fig. 4). However, there was no signi cant difference in the survival rate related to irradiation doses to the bone. In the present study, the thoracic vertebrae (Th1 to Th10), the sternum, and the rst to seventh ribs were contoured as bones. Hayman et al. reported that the relative contribution of the thoracic vertebra, sternum, and ribs/clavicle to the active proliferating bone marrow was approximately 20%, 3%, and 9%, respectively [28]. Thus, one reason that bone irradiation may not have a large effect on the survival rate could be that myelosuppression is mainly caused by the concurrent use of chemotherapy.
In the current study, we examined not only ALC but also NLR as lymphocyte-related factors, as there are various reports describing how these factors relate to the prognosis of surgery and systemic treatment [9,10,29]. It has been reported that in ammatory cytokines are involved in cancer progression and associated with chemotherapy [30,31]. Furthermore, high NLR levels are in ammatory markers that are one of the poor prognostic factors for programmed cell death receptor-1 (PD-1) inhibitor treatment in patients with lung cancer [30,31]. Likewise, high NLR was a predictive factor for lower PFS and DMFS on multivariable analysis in the present study. Therefore, because the standard treatment for locally advanced lung cancer is CRT and immune checkpoint inhibitor therapy, it is important to reduce lymphopenia during RT.
Radiation pneumonitis (RP) is a potentially life-threatening adverse event in chest RT, and lung V20 is frequently used as an index of RP [32][33][34][35]. The National Comprehensive Cancer Network Guidelines have described lung V5 as an RP risk factor; however, owing to the increased use of IMRT and the results of the RTOG 0617 study, the description of lung V5 has been removed from the guidelines regarding dose constraints in lung cancer treatment [35,36]. In the present study, lung V5 strongly correlated with lymphopenia and have a signi cant impact on DMFS. This suggests that while lung V5 does not often reduce the survival rate due to RP, it may inhibit a patient's anticancer immunity in association with lymphopenia. In the PACFIC study, it is suggested that distant metastases might be reduced by maintaining anti-tumor immunity, in which lymphocytes play an important role [3]. Thus, careful attention should be paid not only to lung V20 but also to lung V5 in the treatment planning for NSCLC.
Nowadays, irradiation doses to the heart are known to be important in NSCLC patients treated with CRT.
In the RTOG 0617 study, heart V5 was identi ed as a prognostic factor for OS [19]. In the present study, heart V5 was signi cantly associated with OS in univariable analysis. Unlike IMRT, PBT has the advantage of concentrating high doses of irradiation on the CTV while avoiding low doses to the lung and heart [20]. Therefore, in the treatment of lung cancer, PBT is considered to be more useful than IMRT because it can lower lung V20 and heart V5 exposure while suppressing the increase of lung V5. In chest irradiation with concurrent chemotherapy for esophageal cancer, PBT has advantages over IMRT in terms of lymphopenia and survival rate, and a prospective study is being conducted [12,37,38]. In lung cancer, a randomized control trial comparing passive-scattering PBT and IMRT for locally advanced NSCLC did not prove the superiority of PBT because, despite PBT, DVHs of the lung and heart remained extremely high [39]. We speculate that in this randomized control trial, technical de ciencies in the delivery of the PBT probably affected the results because RP and local failure rates at 12 months for patients enrolled before versus after the trial midpoint were 31.0% and 13.1%, respectively (p = 0.027). Therefore, the results of the NRG 1308 trial, which compares PBT and IMRT and is focused on low doses to at-risk organs and lymphopenia, are highly anticipated to shed light on this important issue [22].

Conclusions
This analysis showed that lymphopenia was associated with a lower irradiation dose to the lung as well as bone in CRT using proton beams for patients with stage III unresectable locally advanced NSCLC. Furthermore, patients with severe lymphopenia during the course of CRT had poor survival rates.
Although lung doses were associated with DMFS, bone doses were not associated with both OS and DMFS.
Taken together, our ndings indicate lung doses are more important than bone doses in CRT for stage III NSCLC and add weight to the argument that PBT has advantages over photon therapy because it is not only capable of delivering high-dose irradiation to lesions but also highly effective for reducing doses to surrounding healthy organs. Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare no competing interests. Kaplan-Meier estimate of survival, local-progression, and regional control rates. (a) Overall, progressionfree, and distant metastasis-free survival curves for the patients in this study. Straight, dashed, and dotted lines indicate overall, progression-free, and distant metastasis-free survival, respectively. (b) Local progression-free and regional control rates. Straight and dotted lines indicate local progression-free and regional control rates, respectively.