Addressing the challenge of proper delineation of lymph node stations in modern radiotherapy for lung cancer.

We greatly appreciate the recent publication of a Japan Lung Cancer Society and Japanese Society for Radiation Oncology consensus-based computed tomographic atlas for defining regional lymph node stations (LNSs) for radiotherapy for lung cancer. This atlas is based on the International Association for the Study of Lung Cancer (IASLC) consensus lymph node map, which resolved differences between the Japanese and American classification and was incorporated into the seventh edition of the TNM staging system. The authors must be commended for providing a very clear, comprehensive presentation of the borders of the mediastinal and hilar LNSs, using a multidisciplinary approach. The Atlas created by Itazawa et al. is the second published consensus-based atlas for defining the LNSs for lung cancer radiotherapy according to the IASLC-endorsed LNS map. The differences between the two published atlases of LNS delineation in radiation oncology are discussed.

We should welcome all attempts to standardize radiotherapy procedures. The Japan Lung Cancer Society and the Japanese Society for Radiation Oncology recently published a consensus-based computed tomographic atlas for defining regional lymph node stations (LNSs) for radiotherapy for lung cancer, and it is of great interest to the radiation oncology community [1]. This atlas is based on the International Association for the Study of Lung Cancer (IASLC) consensus lymph node map, which resolved differences between the Japanese [2] and American classifications [3] and was incorporated into the seventh edition of the TNM staging system [4]. The authors must be commended, not only for providing a very clear, comprehensive presentation of the borders of the mediastinal and hilar LNSs, but especially for their establishment of the principles of contouring by a multidisciplinary team containing thoracic surgeons and radiologists in addition to the radiation oncologists. It is of crucial importance for radiation oncologists such as ourselves to have the benefit of insights from the other oncological specialties, because radiotherapy procedures should be based on clinical, pathological, radiological and surgical data.

Considerable technological advances in lung cancer radiotherapy have been made during the last two decades, and it is now possible to accurately delineate and treat moving targets in the thorax; the clinical application of these advances has been guided by the recommendations of the expert groups [5]. As the current guidelines do not recommend the use of elective nodal irradiation (ENI), one may think that a computed tomographic atlas for defining the regional LNSs for radiotherapy for lung cancer is of lesser importance for radiation oncologists. However, some forms of ENI are still in use, i.e. in postoperative radiotherapy, or in the case of delineation of a whole LNS as a margin for subclinical invasion around a pathological lymph node.

To facilitate the delineation of the LNSs in a reproducible manner, CT-based atlases for radiotherapy have been published. In lung cancer radiotherapy, the first such an attempt was based on the Mountain and Dressler LNS classification, and a consensus-based atlas known as the Michigan Atlas was published in 2005 [6]. Their recommendations were shown to be reproducible (assuming adequate training has been provided) and useful for ENI portals design [7, 8]. A new classification of the mediastinal and hilar LNSs for lung cancer, endorsed by the IASLC and adopted by the 7th edition of the TNM [4], superseded the recommendations for outlining the mediastinal and hilar LNSs provided by the Michigan Atlas. The Atlas created by Itazawa et al. [1] is a second attempt (after the publication by Lynch et al. [9]) to provide a consensus-based atlas for defining the LNSs for lung cancer radiotherapy according to a new LNS map. There are some substantial differences in the recommended LNS borders, not only between the atlas based on the Mountain and Dressler classification [6] and those based on the IASLC-endorsed new LNS map, but also between both new propositions [1, 9]. The major differences in the definition of the borders of LNSs are summarized in Table 1.

Table 1.

Major differences in the borders of the lymph node station as proposed by the consensus-based atlases for the purposes of radiotherapy

LNS number and border	Atlas Michigan [6]	Atlas by Lynch et al. [9]	Atlas by Itazawa et al. [1]
1R/L; cranial border	upper limit of the sternal notch	lower margin of the cricoid cartilage
1R/L; caudal border2R/L; cranial border	upper rim of brachiocephalic vein where it ascends to left, crossing in front of trachea at its midline	1R/L caudal border; level of the clavicles bilaterally and the upper border of the manubrium in the midline2R/L cranial border; apex of the lung and pleural space; in the midline, upper border of the manubrium	apex of the lung
2R; caudal border4R; cranial border	superior border of the aortic arch	intersection of the caudal margin of the left brachiocephalic vein with the midline of the trachea
2R and 4R; left border2L and 4L; right border	midline of the trachea	left border of the trachea
4R; caudal border10R; cranial border	right upper lobe bronchus level or where the right pulmonary artery crosses the midline of the mediastinum	lower margin of the azygos arch
4L; caudal border10L; cranial border	left upper lobe bronchus	upper margin of the left main pulmonary artery
LNS 5; caudal border	the lowest image in which the right pulmonary artery is maximally visualized	level of the carina	upper margin of the left main pulmonary artery
LNS 5 and 3A separation	at the midline of ascending aorta	at the line running horizontally from the anterior border of the ascending aorta	no contact with 3A, anterior border is an imaginary line that is tangential to the posterior aspect of the ascending aorta and is in contact with LNS 6 at this level
LNS 6: caudal border	lower border of the aortic arch		level of the carina
10R/L	location at the level of lobar bronchi; external to mainstem bronchi	location immediately adjacent to the mainstem bronchi and hilar vessels, including the proximal portions of the pulmonary veins and main pulmonary artery; internal to lobar bronchi.in the Atlas by Itazawa et al. [1] additional distinction of 10*L, as vascular part of 10L

LNS = lymph node station, R = right, L = left.

Standardization of radiotherapy procedures through use of the delineation atlas would help radiation oncologists to provide consistently high-quality treatment. On the other hand, it has been shown that strict following of the Michigan Atlas would lead to the construction of larger ENI volumes than when using conventional field settings [8]. Radiation oncologists should be aware that, in some clinical scenarios, strict adherence to these guidelines may lead to unnecessary toxicity. In addition, the discrepancies in defining the LNS borders between the existing Atlases may lead to further confusion and potential errors. In the Atlas by Itazawa et al. [1]—in line with the IASLC definition—the caudal border of LNS 5 is at the level of the upper margin of the left pulmonary artery, whereas in the remaining atlases [6, 9] LNS 5 was located laterally to the ascending aorta and left pulmonary artery, extending to the level of the carina [9] or to the lowest image when the right pulmonary artery is maximally visualized [6]. This area is defined by Itazawa et al. [1] as 10*L—left hilar nodes adjacent to the proximal portions of the pulmonary veins, left pulmonary artery and the hilar vessels. The division of LNS 10 to the three subgroups (10R, L and *L) was made for the purpose of contouring in lung cancer radiotherapy, but the overall boundaries of LNS 10 as established by Itazawa et al. [1] are in full concordance with the IASLC definition. Despite that, when contouring LNS 10L for radiotherapy purposes, the inclusion of the nodes located laterally to the left pulmonary artery may lead to excessive toxicity. If we do not treat LNS 5, should we still include in LNS 10 those nodes adjacent to the hilar vessels, which in the remaining Atlases [6, 9] are defined as part of LNS 5? And a question remains: what is the concordance between surgeons in regard to a definition of LNS 5 and 10L at this level. This is important, because such discrepancies could affect treatment decisions and analysis of the outcome due to their impact on staging. The need to include this part of the hilar region into 10L probably differs in each individual case. Radiation oncologists need more pathological/surgical data on the frequency of involvement of this part of LNS 10L to establish strong recommendations. We hope that the planned guidelines for contouring an elective Clinical Target Volume (CTV) for lung cancer will address such questions.