Background
Lung cancer is one of the most common cancers worldwide, and non-small cell lung cancer (NSCLC) comprises 85% of lung cancer cases [
1]. Radiotherapy is the primary method of treating patients with medically inoperable or unresectable locally advanced NSCLC (LA-NSCLC) [
2‐
4]; addition of concurrent chemotherapy to radiotherapy (CRT), which is standard care for unresectable LA-NSCLC, achieves a median OS of 17–27 months [
5‐
8]. Nevertheless, some patients suffer early recurrence and severe adverse events, such as radiation-induced pulmonary and esophageal injury, as well as massive hemoptysis.
Optimizing the radiation plan and minimizing radiation-induced toxicities requires limiting the dose to the organs at risk (OAR). Dose-volume parameters of lung, such as V20 and mean lung dose (MLD), are known to play an important role in predicting severe radiation pneumonitis and early deaths [
9]. The pulmonary artery (PA) is the most commonly tumor-involved thoracic great vessel [
10,
11]. Patients who had PA invasion by tumor were usually considered to have unresectable disease, though vascular reconstruction has been applied in selected patients. Our previous study of 100 patients with inoperable NSCLC in the United State first categorized the grade of PA invasion by contrast-enhance Computed Tomography (CT) or Positron Emission Tomography (PET) scan and indicated that only 2 out of 4 patients with grade 5 PA invasion experienced massive bleeding caused death, and overall grades of PA invasion were not associated with overall survival in patients with inoperable NSCLC. However, in patients with LA-NSCLC treated with concurrent chemoradiation, the proportion of PA volume that received more than 45–60 Gy was associated with shorter OS [
12].
The heart is known to be an important OAR for patients receiving thoracic radiotherapy. Postoperative radiation therapy for lung cancer was associated with increased mortality from heart disease [
13]. In the RTOG 0617 trial, there were less than 5% cardiac adverse events (≥ grade 3) in either the high dose or low dose arm, and multivariate analysis showed that the higher proportions of heart volume receiving ≥ 5 Gy (H V5) or ≥ 30 Gy (H V30) were significantly correlated with shorter survival [
7]. However, Tucker et al. [
14] did not confirm the result of that trial regarding the impact of heart dose volume on OS.
In our previous study [
12], the heart was not evaluated in the univariate or multivariate analysis. In addition, data from another independent research center is needed to confirm these findings regarding the PA. Hence, we performed a retrospective analysis to re-evaluate the results of the previous study and verify the impact of dosimetric parameters of the PA and heart on the survival of patients with NSCLC treated with definitive conventional fractionated radiotherapy (CFRT) with or without chemotherapy and to determine whether PA should be considered an OAR and what PA dose/volume limit should recommended for patients receiving definitive CFRT.
Discussion
The present study demonstrated that the score of KPS, CCI, stages of T and N, PA invasion grade, as well as percentage of PA volumes that received 40–55 Gy were independent prognostic factors of OS for patients with NSCLC treated with definitive CFRT. PA invasion grade was significantly associated with OS, though it was not associated with bleeding during or after radiotherapy. In addition, patients with PA V40 > 80%, V45 > 68%, V50 > 45% or V55 > 32% had a significantly shorter OS. The results further confirmed our previous conclusions [
12] and suggested that the PA should be regarded as an OAR during conventionally fractionated thoracic radiotherapy or chemoradiotherapy, and that the grade of PA invasion and PA dosimetric parameters V40-55 could be used to predict survival for those patients with NSCLC who received definitive CFRT.
Radiation induced lung injury, esophagitis, and heart injuries are the most frequent radiation related toxicities and have been widely investigated recently. The RTOG 0617 trial [
7] revealed that high dose radiotherapy given to patients with LA-NSCLC failed to improve local control and prolong OS compared with standard dose radiotherapy. The possible damage to the local immune microenvironment of the field receiving high dose radiation, and the radiation related toxicities might be the major reason why patients with high dose radiotherapy had shorter survival. In this trial, H V5 and H V30, but not radiation modalities were correlated with OS [
7]; The present study also evaluated the relationship between OS and heart dosimetric parameters including maximum or mean dose to the heart, or the percentage of heart volume that received a specific dose (H V5-70). However, no significant associations were found between them, although the radiation related heart injury was not evaluated herein. The reason for this finding might be attributed to relatively strict radiation dose constraints for heart that were used in our study. Our results were in accordance with Tucker et al’s study [
14], which found that heart dosimetric parameters did not significantly affect survival.
A great vessel is deemed as a dose-limiting organ when lesions are given hypofractionated or stereotactic body radiotherapy (SBRT) [
17]. Some researchers thought there were greater risks of severe hemoptysis and early mortality among patients who had severe PA invasion during or after radiotherapy. A previous study stated that vessel rupture may result from the penetration of the tumor into the vessel wall itself [
18]. However, a later study [
19] showed that this explanation might be implausible because penetration of a tumor into the wall of an elastic artery is so unusual except in the following circumstances: the vessel itself (e.g., atherosclerosis), had exposed vasa vasorum during surgery and damage to the adventitia caused by ulcer, fistulas or infection around artery that were independent of radiotherapy [
19]. Another study showed that toxicity to the aorta after reirradiation was relatively rare, even when maximum composite doses to the aorta exceeded 100 Gy [
20].
Recently, our study [
12] first proposed grading criteria for PA invasion from grade 0 to grade 5, describing the distance of tumor to the PA, degree of circumferential involvement of the PA, and status of vessel wall damage. In that study, we found patients with PA invasion of grade 0–1 had a longer median OS than those with grade 2–5 (33.4 months vs. 18.3 months,
P = 0.242), and two of the 4 patients with grade 5 PA invasion died suddenly from massive hemorrhage at 3 and 4.5 months after completion of radiotherapy. To facilitate its clinical use, we simplified the grading system to grade 0, 1, 2, 3 for no, minimal, moderate, and extensive invasion, respectively. In this group of patients, we found a significant relationship between grade of PA invasion and survival. Two patients with extensive PA invasion died of massive bleeding after radiotherapy. We believe that a higher grade of PA invasion indicates worse clinicopathologic behavior and prognosis, though few bleeding events were observed during or after radiotherapy in both studies. In the future, reliable imaging or pathological assessment of great vessel rupture needs to be developed to confirm this assumption.
The present study also found that the percentages of PA volume that received 40–55 Gy were significantly associated with OS regardless of target volume, KPS, CCI and stage. Especially, patients with PA V45 > 68%, and those with PA V55 > 32% had a significantly shorter OS. Interestingly, the results were highly consistent with our previous study in which high dose volume of PA V45-60 was predictive of shorter OS, and the cut-off values of PA V45 and V60 were 70% and 37%, respectively. Both studies suggest that PA V45 (68-70%) is a better cut-off value predictive of shorter OS in patients with LA-NSCLC treated with CFRT. Our findings differed from those using SBRT, in which maximum dose exceeding 50 Gy to the PA was related with massive hemoptysis and caused death [
21]; in our study, when CFRT was administrated, the maximum dose to the PA was not related with OS.
In the present study, we found that GTV, CTV, PTV and dose coverage of target volumes were not associated with OS, which seemed contradictory to the effect of the factors T or N stage on OS, because a higher T or N stage often results in a larger target volume. However, we did not see any correlation between T or N stage and the target volumes GTV, CTV or PTV. Correlation analysis revealed only weak correlations between target volumes and PA V40-V55, thus ruling out an influence of target volume on PA volume dose. Certainly, our study may be too small to distinguish the independent or synthetic effect of heart and PA dosimetric parameters on survival. Future prospective studies with larger numbers of patients are needed to pursue the combined effects of heart and PA dosimetry. In addition, in our study only two patients with extensive PA invasion died of fatal hemoptysis after radiotherapy during follow-up. Hence, it appears that in this cohort, these rare outcomes, such as fatal hemoptysis or massive bleeding, are not good end points guide recommendations for dosimetric constraints for this OAR. PA grading variation from the start of radiotherapy to death may be an alternative end point. However, this is not evaluated in the present study because the accurate PA grading variation largely depends on reliable imaging or pathological analysis of abnormalities, such as ulcer, fistula, and stricture. On the other hand, we believe that the PA is an OAR similar to the heart, and that higher dose-volume irradiation to the PA might also cause serious consequences such as pericarditis, arrhythmia, and fatal hemoptysis. Future studies are needed to validate these postulations.