Journal of Medical Imaging and Radiation Oncology

Transcription

1 Journal of Medical Imaging and Radiation Oncology 63 (2019) RADIATION ONCOLOGY REVIEW ARTICLE Contouring consensus guidelines in breast cancer radiotherapy: Comparison and systematic review of patterns of failure Harriet E Gee, 1,2,3 Lauren Moses, 1* Kirsty Stuart, 1,2,3 Najmun Nahar, 1 Ken Tiver, 1,4 Tim Wang, 1,2,3 Rachel Ward 1 and Verity Ahern 1,2,3 1 Crown Princess Mary Cancer Centre, Westmead Hospital, Sydney, New South Wales, Australia 2 Sydney Medical School, C24 Westmead Hospital, The University of Sydney, Sydney, New South Wales, Australia 3 Westmead Breast Cancer Institute, Westmead Hospital, Sydney, New South Wales, Australia 4 Nepean Cancer Care Centre, Nepean Hospital, New South Wales, Australia HE Gee MBBS, DPhil, FRANZCR; L Moses MBBS; K Stuart MBBS, FRANZCR, PhD; N Nahar MBBS, FRANZCR; K Tiver MBBS, FRANZCR; T Wang MBBS, MPH, FRANZCR; R Ward BMRS(RT); V Ahern MBBS, FRANZCR. Correspondence Dr Lauren Moses, Radiation Oncology, St George Hospital Cancer Care Centre, 1 Short Street, Kogorah, NSW 2217, Australia. lauren.moses@health.nsw.gov.au Conflict of interest: None. *Present address: Radiation Oncology, St George Hospital Cancer Care Centre, Kogorah, NSW, Australia Submitted 3 March 2018; accepted 16 August doi: / Summary Adequate coverage of sites harbouring potential microscopic disease is paramount, where the clinical decision has been made to include regional lymph node radiotherapy for patients with breast cancer. This must be achieved in balance with minimising dose to normal tissues. Several international consensus guidelines detailing clinical target volumes (CTVs) are available, but there is currently no agreement as to which is most appropriate for a given clinical situation. Contouring guidelines are beneficial for routine practice and essential for clinical trial quality assurance. The aims of this study were as follows: to provide a single point of comparison of four commonly used contouring guidelines, including one used in a current Trans-Tasman Radiation Oncology Group trial; and to undertake a systematic review of existing studies which map sites of breast cancer recurrence against contouring guidelines. Two international consensus guidelines (European Society for Radiotherapy and Oncology, and Radiation Therapy Oncology Group) were compared with two clinical trial guidelines (TROG PET LABRADOR and the Proton/Photon trial NCT RADCOMP). Comprehensive literature search for patterns of failure studies was undertaken using Embase and Pubmed. We detail the small but significant differences between the breast consensus guidelines, particularly the supraclavicular (SCF) and internal mammary chain CTVs. Seven series were found mapping recurrence patterns. These results are discussed in the context of the contouring guidelines. Several studies found the SCF CTV is the area at greatest risk of geographical miss. This review will facilitate further discussion about guideline selection and modification, particularly for future clinical trials in Australia and New Zealand. Key words: breast cancer; contouring guidelines; radiotherapy. Journal of Medical Imaging and Radiation Oncology Introduction Randomised clinical trials have shown that the addition of regional nodal irradiation (RNI) to whole breast adjuvant radiotherapy (RT) following breast-conserving surgery in stage I III breast cancer improves disease-free survival 1,2 with a marginal effect on overall survival. Post-mastectomy RT for patients with involved axillary nodes (particularly >3) has been shown in meta-analysis to reduce loco-regional recurrence (LRR), overall recurrence and improve overall survival. 3 The indications for RNI are an ongoing subject of expert discussion, which must be personalised to the individual patient risk assessment. In the situation where the clinical decision has been made to include RNI for patients at high risk of LRR, adequate coverage of sites at potential for harbouring microscopic disease is paramount. RT for breast cancer has become more sophisticated over the last decades, with the adoption of 3D and 4D planning, more complex beam arrangements such as field-in-field, and 102

2 Breast contour guidelines and failure patterns complex modulated fields intensity-modulated RT (IMRT) or volumetric arc therapy (VMAT) and protons. 4 These techniques can reduce dose heterogeneity and increase normal tissue sparing. 5,6 Centres across Australia and New Zealand (ANZ) are increasingly contouring breast/chest wall and regional node clinical target volumes (CTVs) for treatment planning purposes. 7 However, there is significant intra-observer variability in contouring for breast cancer target volumes, 8 which can affect dosimetry and dose to normal tissues. 9 Contouring guidelines can reduce intra-observer variability in breast cancer. 10 In the setting of pancreatic and advanced head and neck cancer, contouring guidelines have been shown to increase consistency of treatment and enable comparison across centres and heterogeneous clinical cohorts. 11,12 Furthermore, adherence to contouring guidelines is important for quality assurance in clinical trials. 11,12 Several breast cancer consensus guidelines are published. The Royal Australian and New Zealand College of Radiologists (RANZCR) Faculty of Radiation Oncology 13 recommends the guidelines of the European Society for Radiotherapy and Oncology (ESTRO) 14,15 and the breast cancer contouring atlas of the North American-based Radiation Therapy Oncology Group (RTOG). 16 Two recent trial guidelines are also worthy of comparison. Following concerns about LRR at the edge or just outside these guidelines, a consortium developed a trial version of RTOG guidelines, here called RADCOMP, for a Phase III clinical trial comparing the use of photons and protons to treat breast cancer where RNI is required. 17 The Trans-Tasman Radiation Oncology Group (TROG) Trial 12.02: PET Scans for Locally Advanced Breast Cancer and Diagnostic MRI to Determine the Extent of Operation and Radiotherapy (PET-LABRABOR) examines the use of PET Scans and MRI to determine extent of operation and RT for locally advanced breast cancer. This protocol uses a modified version based on methodology described by Dijkema et al. 18 and Kirova et al. 19 and closely follows the TROG STARS study. 20 Though variations between anatomical boundaries suggested by different breast contouring atlases for RNI are relatively minor, these could have significant clinical implications for patients. There is a greater risk of recurrence if microscopic disease is inadequately treated, but the risk of acute and late toxicity increases if excessive normal tissue is included in the field. 14 Several international studies have been published recently reporting failure patterns of the commonly utilised breast contouring guidelines These are timely to inform debate. There is currently no clear consensus in ANZ institutions on the most efficacious breast contouring atlas and the most appropriate contouring atlas to use for future TROG clinical trials has not been determined. Across ANZ institutions, patient selection for RNI remains an area of wide variation in practice among Radiation Oncologists, with 65.9% of respondents in a recent survey reporting they would sometimes treat according the MA.20 trial criteria. 2,7 A range of selection criteria including tumour size, nodal burden, and inflammatory breast cancer were used to individualise treatment. The majority of clinicians use ESTRO consensus guidelines. 2,7 Appropriately, many clinicians make individual adjustments to guidelines based on clinical judgement. The rationale for this review was to provide a single point of comparison of four current contouring guidelines, including that used in a recent TROG trial. Next, a comprehensive literature review was undertaken of existing studies mapping the sites of recurrence data against these contouring guidelines to facilitate further discussion about guideline selection and modification, and to inform whether there is a need for ANZ breast consensus guidelines. Methods The guidelines for analysis were selected due to their use in ANZ. First, the ESTRO guideline is based on international expert consensus, 14 and was updated following feedback in The guideline is based on discussion around multiple previous guidelines, 10,18,19,24,25 and is vessel-based. 26 The breast cancer contouring atlas of the North American-based RTOG is also based on multicentre expert consensus. 16 Both are very frequently used worldwide. Two recent contouring guidelines developed for clinical trials were also compared. The first was developed for the PET-LABRADOR study. 27 This was chosen as a relatively comprehensive protocol, intended for patients requiring RNI due to high risk of LRR, but has been found to be complex to follow in practice. The second was the atlas for ClinicalTrials.gov Identifier: NCT a Pragmatic Phase III Randomized Trial of Proton vs. Photon Therapy for Patients with Non-Metastatic Breast Cancer: A Radiotherapy Comparative Effectiveness (RADCOMP) Consortium Trial. 17,28 Institutional ethics approval was obtained to contour model CTVs on the CT simulation images from two cases. The cases were selected because they had been previously used to credential the centre for a TROG trial. The objective of our systematic review was to evaluate available evidence pertaining to the patterns of locoregional failure of published breast contouring guidelines. Studies analysing the coverage of regional lymph node recurrence or metastases in relation to the anatomical borders stipulated by various breast contouring guidelines were selected for review. Studies were selected according to the eligibility criteria outlined below: Eligibility criteria Study characteristics Participants: Women of any age with a diagnosis of breast cancer with involvement of regional lymph nodes, 103

3 HE Gee et al. regardless of disease stage. Women with or without a history of previous treatment including surgery, chemotherapy or radiotherapy to the breast, chest wall or regional lymph nodes were eligible for inclusion. Intervention: Transfer of the locations of metastatic lymph nodes onto representative CT images to allow for coverage analysis against the selected breast contouring guideline(s). Outcome: Determination of the distribution of geographical miss of lymph node metastases by the selected guideline(s). Comparison: Nil. Study designs: Single case reports, conference abstracts, letters, comments and editorials were excluded from review. Report characteristics The search was limited to studies published from 2006 onwards, in accordance with the emergence of the first breast contouring guidelines. Non-English language literature was not included. Unpublished material was eligible for inclusion. Study selection Studies were identified by searching electronic databases, evaluating the reference lists of retrieved articles and liaising with experts in the field. Searches were performed using PubMed and Embase (2006 present) with the following search terms: breast cancer, computed tomography, radiation*, target, miss, map and field. The most recent search was undertaken on 18 January Our search strategy was peer reviewed by our departments clinical trials coordinator. Eligibility assessment was performed independently by two reviewers, with any disputes resolved by discussion and consensus. The search of electronic databases yielded 464 citations, adjusted to 402 following the removal of duplicates (Appendix S1). Two articles recommended by experts in the field overlapped with the search results, while a further three articles were classified as editorials and excluded from review. Title and abstracts were reviewed in the first instance, with 114 publications excluded based on their classification as conference abstracts, editorials or single case reports. In the case of one relevant conference abstract, the author was contacted to confirm a fulltext publication was not available. A further 254 articles did not meet the inclusion criteria. The review of the full-text of the remaining 34 articles was undertaken. Twenty-four articles failed to satisfy the inclusion criteria. A further three articles examined patterns of loco-regional failure without analysis against published contouring guidelines and were therefore excluded from review. A total of seven articles were included for review. The selected studies were all classified as case series (Level IV evidence). The study characteristics are summarised in Table 1. Results The anatomical boundaries of the CTVs covered by the contouring guidelines are presented in Table 2. These include the anatomical boundaries of the supraclavicular fossa (SCF), infraclavicular fossa (ICF), posterior neck, axillary levels I, II and III, interpectoral nodes (IP), internal mammary chain (IMC), and CTV breast and chest wall (CW). For consistency, in this review, single naming conventions have been adopted although there are minor variations between the guidelines (discussed below). Using these guidelines, the CTVs described were contoured on two patients: one who had been treated with breast-conserving surgery and one treated with mastectomy. Representative images from these contours are shown in Figures 1 and 2. Full contoured cases are available as Figures S1 and S2. The guidelines lacked consensus for normal tissue contouring, and these organs were therefore not compared in this review. There were a number of key differences between consensus guidelines. Differences between guidelines: nomenclature The ESTRO guidelines refer to the SCF as lymph node level 4, as their definition of the supraclavicular CTV for breast cancer is not the same as it is, for example, for head and neck cancer, and also to reflect the continuum of the lymphatic drainage pattern of breast cancer. PET LABRADOR protocol divides the SCF into a medial and lateral component; however, since these are contiguous with each other, we treated them as a single SCF for the purpose of this review. The infraclavicular fossa (ICF) was only separately identified in the PET LABRADOR protocol. The ICF would be almost entirely included in RTOG or RADCOMP guidelines if axillary level II/III were contoured, but in the ESTRO guideline, the extent of ICF coverage depends both on patient anatomy, and the inclusion of axillary level II/III. The posterior neck volume described in RADCOMP is not included in either ESTRO or RTOG guidelines, but is mostly covered in PET LABRADOR (as part of the SCF volume). The IP (Rotter s) nodes are not separately described in RTOG or RADCOMP but CTVs covering the CW/breast and axillary volumes would cover these nodes. Differences between guidelines: CTV coverage The boundaries of the described CTVs are different between the guidelines. The most obvious difference is the extent of the SCF volumes (Figs 1a c and 2a,b and d). Superiorly, ESTRO guideline limits the SCF to the whole of the subclavian vein and artery, as compared 104

4 Breast contour guidelines and failure patterns Table 1. Study Details: Patient Populations and Co-interventions Publication Year of recruitment Institutions Contouring guidelines evaluated Patients Patients with distant metastases (%) Total events Regions evaluated Loco-regional metastases at primary diagnosis (Patients) Loco-regional recurrence lesions post therapy (Patients) Loco-regional recurrence: Initial therapy (patients) Surgery with BCS or mastectomy Adjuvant radiotherapy (RNI) Chemotherapy Endocrine treatment Brown et al RTOG (52) 161 SCF 117 (44) 44 (18) 18 8 (4) 17 9 Jing et al RTOG Madu (63.6) 524 SCF N/S (17) N/S (38) 38 3 (N/S) 34 N/S et al. Atean et al. Dijkema et al. Reed et al Varied (87) 52 SCF N/S (6) N/S (26) N/S N/S (3) 26 N/S Jethwa et al RTOG (23) 130 IMN 115 (60) 15 (7) N/S N/S (N/S) N/S N/S Gentile et al RTOG 30 3 (10) N/S Axilla N/S (30) n/a n/a n/a n/a n/a Chang et al ESTRO (36.4) 235 Multiple n/a 235 (129) (49) 99 (Adj.) 67 RTOG Chang et al ESTRO RTOG (10) 337 Multiple n/a 337 (234) (130) 66 (Neo.) 161 (Adj.) 102 Adj., adjuvant; BCS, breast-conserving surgery; IMN, internal mammary nodes; n/a, not applicable; N/S, not specified; Neo., neoadjuvant; RNI, regional nodal irradiation; SCF, supraclavicular fossa. 105

5 HE Gee et al. Table 2. Comparison of anatomical boundaries of contouring guidelines ESTRO RTOG RTOG-RADCOMP TROG PET LABRADOR Supraclavicular Fossa Cranial Whole of subclavian vein and artery (or 1 2 cm sup to involved LN) Caudal Subclavian vein with a 5 mm margin (connecting to IMN) Medial Jugular vein without margin; exclude thyroid and common carotid artery Lateral Anterior scalene mm and connects to medial border level 3 Caudal to cricoid cartilage Below the cricoid Inferior aspect of cricoid cartilage Junction of brachiocephalic-axillary veins/caudal edge clavicle head (superior border of breast/cw field) Connect to IMN (include subclavian vein) Most superior CT section of jugulo-subclavian junction, or inferior most slice of external jugular vein (runs in close proximity to carotid artery and both included) Exclude thyroid and trachea Medial edge carotid artery M: medial edge of internal carotid artery and the internal jugular vein; L: medial SC LN Cranial: lat edge SCM; caudal: junction first rib-clavicle Lateral edge SCM, clavicle and connecting to axilla level III M: lateral border of SCM and scalenus anterior L: not lateral to lateral most aspect of clavicle Anterior SCM, posterior edge clavicle SCM Dorsal surface of SCM, clavicle or strap M: posterior surface SCM; L: clavicle or skin mm and adipose tissue directly deep to skin Posterior Pleura Anterior aspect scalene mm Scalenus (ant and med), levator scapulae M: scalenus muscle, posterior to the jugulocarotid sheath; L: anterior surface of the omohyoid, levator scapulae and scalenus medius muscles. The post border should extend posteriorly to a muscle, in a line that follows the space between the clavicle and levator scapulae ESTRO RTOG RTOG-RADCOMP TROG (PET LABRADOR) Infraclavicular fossa Cranial Partially included in axillary level II/III volume Included in axillary level II/III volumes Included in axillary level/iii volumes Lateral and medial SC LN where space exists between pec major and clavicle Caudal Inferior most slice of coracoid process Medial Skin and origin of pectoralis major on clavicle Lateral Medial border of the coracoid process, and pectoralis minor Anterior Pectoralis major (and skin inferomedially) Posterior Clavicle and subclavius muscle Posterior neck Cranial Below the cricoid Included in supraclavicular fossa volumes Caudal Obliteration of fat space Medial Longus coli Lateral Platysma, level 2/3 or scapula Anterior Supraclavicular volume Posterior Trapezius Axilla level III Cranial Cranial extent of the subclavian artery (i.e. 5 mm cranial of subclavian vein) Caudal 5 mm caudal to subclavian vein (or top ALND) Axillary vessels cross-medial edge of pectoralis minor Pectoralis minor insert on cricoid At or below pectoralis minor insertion on coracoid Obliteration of fat space between pectoralis major and CW Inferior to infraclavicular volume Inferior CT slice of the axillary vessels 106

6 Breast contour guidelines and failure patterns Table 2. (continued) ESTRO RTOG RTOG-RADCOMP TROG (PET LABRADOR) Medial Junction of subclavian and internal jugular veins > level 4 Thoracic inlet Obliteration of fat space and supraclavicular volume Clavicle, rib, and the lateral aspect of the jugulo-subclavian junction Lateral Medial side of the minor pectoral muscle Medial border of pectoralis minor Level 2/medial border pectoralis minor Medial border pectoralis minor Anterior Pectoralis major Posterior surface pectoralis major Posterior pectoralis major Posterior surface pectoralis major Posterior Up to 5 mm dorsal of subclavian vein or to costae and intercostal muscles Axilla level II Cranial Cranial extent of the axillary artery (i.e. 5 mm cranial of axillary vein) Caudal Caudal border of pectoralis minor mm, (or top of ALND) Ribs and intercostal mm CW Ribs and axillary vessels Axillary vessels cross medial edge of pectoralis minor m Axillary vessels cross lateral edge of pectoralis minor m At or below pectoralis minor insertion on coracoid Obliteration of fat space between pectoralis major and pectoralis minor or CW (may include pectoralis minor mm for ease of contouring) Medial Medial border pectoralis minor Medial border pectoralis minor Medial border pectoralis minor/level III axilla Superior slice of the axillary vessels posterior or deep to pectoralis minor Last slice where pectoralis minor has not attached to CW Medial border pectoralis minor/level III axilla Lateral Lateral border pectoralis minor Lateral border pectoralis minor Lateral border pectoralis minor/level I Lateral border pectoralis minor/level I axilla Anterior Pectoralis minor Anterior surface pectoralis minor Posterior pectoralis major Posterior pectoralis major Posterior Up to 5 mm dorsal of axillary vein or to costae and Ribs and intercostal mm Chest wall 5 mm margin on posterior border of axillary intercostal muscles vessels; or CW (no margin) Axilla level I Cranial Medial: 5 mm cranial to the axillary vein Lateral: max up to 1 cm below the edge of the humeral head, 5 mm around the axillary vein Caudal To the level of rib 4 and 5, taking also into account the visible effects of the sentinel lymph node biopsy Medial Axilla level II, the interpectoral level and the thoracic wall Lateral Cranially up to an imaginary line between the major pectoral and deltoid muscles, and further caudal up to a line between the major pectoral and latissimus dorsi muscles Axillary vessels cross lateral edge of pectoralis minor m Anterior Pectoralis major and minor mm Plane defined by anterior surface of pectoralis major and lat dorsi Posterior Cranially up to the thoracodorsal vessels, and more caudally up to an imaginary line between the anterior edge of the latissimus dorsi muscle and the intercostal muscles Intrapectoral nodes Cranial Includes the cranial extent of the axillary artery (i.e. 5 mm cranial of axillary vein) Caudal Caudal border of pectoralis minor Axillary vv at lateral edge pectoralis minor & below humeral head Pectoralis major insert onto ribs Pectoralis major inserts into ribs (around fourth and fifth rib) Tendon of latissimus dorsi, generally not above superior aspect of coracoid process Superior to visible breast tissue, wherever vessels can be seen laterally in the space between the lateral aspect of pectoralis major and the teres muscles Lateral border pectoralis minor Lateral border pectoralis minor/level II axilla Pectoralis mm, CW, and the level II axilla Medial border of latissimus dorsi Latissimus dorsi, at line connecting latissimus dorsi and deltoid or up to skin Posterior pectoralis major or skin Skin Anterior surface subscapularis muscle Anterior border subscapularis and latissimus dorsi Teres major, subscapularis and skin Post border of axillary vessels, subscapularis and teres mm Included in CW/breast volumes Included in CW/axillary volumes Any space between pectoralis minor and major mm 107

7 HE Gee et al. Table 2. (continued) ESTRO RTOG RTOG-RADCOMP TROG (PET LABRADOR) Medial Medial edge pectoralis minor Lateral Lateral edge pectoralis minor Anterior Pectoralis major Posterior Pectoralis minor Internal mammary chain Cranial Caudal limit of SCF (CTVn_L4) Superior aspect medial first rib Inferior supraclav volume or caudal to head of clavicle Caudal Cranial side of fourth rib (fifth rib in select cases) Cranial aspect of fourth rib Cranial border fourth rib Medial 5 mm margin on internal mammary vein and artery Encompass internal mammary/thoracic Sternum vessels Internal mammary artery and vein are used as surrogates Lateral 5 mm margin on internal mammary vein and artery Include any visible fat Anterior Anterior limit of vascular area Posterior chest wall Posterior Pleura To pleura (include fat but no lung) CTV breast post conservation Cranial Upper border of palpable/visible breast tissue; Clinical reference + second rib insertion All soft tissue of breast encompasses breast maximally up to the inferior edge of the outline markers. May include skin if tumour sternoclavicular joint extends to/includes area Caudal Most caudal slice with visible breast Clinical reference + loss of CT apparent breast Medial Lateral to the medial perforating mammary vessels; maximally to the edge of the sternal bone Lateral Lateral breast fold; anterior to the lateral thoracic artery Sternal rib junction Clinical reference and midaxillary line, typically excludes latissimus dorsi mm Anterior 5 mm under skin surface Skin Posterior Pectoralis major or costal/intercostal mm where no Excludes pectoralis mm, CW mm, ribs Down to pectoralis fascia, may include if muscle tumour extends to/includes fascia CTV CW post-mastectomy Cranial Guided by palpable/visible signs; if appropriate guided by the contralateral breast; maximally up to the inferior edge of the sternoclavicular joint Caudal Guided by palpable/visible signs; if appropriate guided by the contralateral breast Medial Guided by palpable/visible signs; if appropriate guided by the contralateral breast. Lateral Guided by palpable/visible signs; if appropriate guided by the contralateral breast. Usually anterior to the midaxillary line Caudal border clavicle head Interior to head of clavicle As per Whole Breast_CTV_pre and _post (preliminary simulation and MRI/PET) Clinical reference + loss of CT apparent contralateral breast Sternal rib junction To sternum Clinical reference and midaxillary line, typically excludes latissimus dorsi mm 1-2 cm below breast tissue (or prior breast tissue) To midaxillary line (or if not including level 1&2, may extend more posterior) Anterior 5 mm under skin surface Skin Skin Skin Posterior Pectoralis major or costal/intercostal mm where no Includes pectoralis mm, CW mm, ribs Up to but not including ribs Extends to pectoralis fascia, may include if muscle tumour extends to/through ALND, axillary lymph node dissection; CW, chest wall; IMN, internal mammary nodes; L, lateral SCF volumes; LN, lymph node; M, medial; Mm, muscle. 108

8 Breast contour guidelines and failure patterns (a) (b) (c) (d) Fig. 1. Representative image from right-sided case (breast conservation surgery). Blue = ESTRO; Red = RADCOMP; Yellow = PET LABRADOR; Green = ESTRO humoral head PRV. (a) SCF superior, lateral and medial borders; (b) Axilla levels I, II and III and medial SCF; (c) SCF posterior extent; (d) IMC superior/inferior boundaries. (a) (b) (c) (d) Fig. 2. Representative image from left-sided case (mastectomy). Blue = ESTRO; Red = RADCOMP; Yellow = PET LABRADOR; Green = ESTRO humoral head PRV. (a) SCF superior, lateral and medial borders; (b) Axilla levels I, II and III and medial SCF; (c) IMC and chest wall; (d) IMC superior/inferior boundaries. with the other three guidelines which extend to the inferior aspect of cricoid cartilage. However, the ESTRO guidelines do acknowledge that in cases of locally advanced breast cancer (e.g. if there are pathological nodes in level III) it may be relevant to modify the cranial border to account for lymphatic spread beyond the first lymph node levels. In such cases, a mm margin is suggested to add to the pathological node to define the CTV. 14 Medially, the ESTRO guideline stays lateral of the internal carotid artery, while the other three guidelines include the artery. The posterior neck volume in RADCOMP effectively extends the SCF volume posteriorly to the levator scapulae to include the posterior triangle of the neck. This boundary is similar to the posterior aspect of the SCF volume in PET LABRADOR and is more posterior than RTOG (anterior aspect of the scalene muscle), or ESTRO border. The limits of the axillary volumes are more subtly different between the guidelines. While for RTOG, the inferior limit of level I is the pectoralis major insertion onto 109

9 HE Gee et al. ribs, ESTRO recommends the (typically more superior) 4th 5th rib. ESTRO guidelines also typically are more anterior than RTOG for level I (imaginary line between the anterior edge of the latissimus dorsi muscle and the intercostal muscles, vs. extending to subscapularis). The ESTRO and PET LABRADOR guidelines keep the superior border of level I axilla inferior to that of RTOG/RADCOMP, to avoid irradiating the humoral head (with a 1 cm margin for dose drop-off from field edge, Fig. 1a). The ESTRO volumes are also inferior to RTOG/RADCOMP in the superior extent of the axilla level III (with RTOG including up to the pectoralis minor insertion onto coracoid, versus 5 mm above the axillary vein for ESTRO). The extent of the IMC CTV also varies between the guidelines. While the ESTRO volumes are continuous with the SCF, RTOG and RADCOMP stop caudally at the first rib (Figs 1d and 2d). Inferiorly, the ESTRO guidelines permit using the cranial side of the fifth rib in select cases, while the other guidelines use the cranial side of the fourth rib. Furthermore, ESTRO recommends a 5 mm margin on the vessels, while the RTOG and PET LABRA- DOR recommend the vessel as surrogate (Fig. 2c). RAD- COMP follows ESTRO, by including the perivascular fat space. Finally, the chest wall muscles and ribs are included in the post-mastectomy CW CTV for RTOG but these volumes are less extensive for ESTRO or RADCOMP. The skin is included in RTOG and RADCOMP guidelines but ESTRO specifies 5 mm below skin surface. Patterns of failure studies A total of seven case series were reviewed. Most consisted of a mixture of patients with nodal disease at diagnosis (Brown et al. 29, Jethwa et al. 21 and Gentile et al. 30 ). Only small numbers of patients had a LRR after known RT (RNI or to SCF). Supraclavicular Reed et al. 31 analysed the anatomical distribution of supraclavicular metastases at presentation or recurrence. Three patients had a history of known supraclavicular RNI. Most patients (87%) had concurrent distant metastases (DM). The posterior region of the supraclavicular fossa was identified as a potential site of geographical miss, with metastatic supraclavicular nodes located posterior to the vertebral body transverse process in 16%, and 21% of nodes located in level V (posterior triangle). Brown et al. 29 mapped supraclavicular nodal metastases against RTOG target volumes. A little over twothirds (N = 117/161) of these were found at primary diagnosis, and about 50% (N = 32/62) of the patients had concurrent DM. Four patients had prior RNI to SCF CTV. Overall, the RTOG volume failed to cover 41% of all SCF nodal metastases. The most common site of nodes outside the RTOG CTV was at the level of the cricoid and thyroid cartilage, followed by the posterolateral SCF (Fig. 3c). The third location of potential geographical miss was the lateral low SCF, lateral to the RTOG volume. Jing et al. 32 examined CTV coverage of four contouring guidelines (RTOG, 16 Madu et al., 33 Atean et al., 10 Dijkema et al. 18 ) against supraclavicular metastases. A little over two-thirds (N = 38/55) of patients in the series had recurrent disease, (although no prior RT to the SCF), while one-third were at initial presentation, and the majority had concurrent DM. Overall, the RTOG atlas covered nearly two-thirds of involved supraclavicular nodes. The atlas by Dijkema et al. (on which PET LABRA- DOR was based) covered the majority of involved nodes in the medial and lateral supraclavicular regions. Again, the most prominent location of inadequate CTV coverage was posterolateral: 82% of SCF metastases were beyond the lateral edge of the sternocleidomastoid muscle, and 67% beyond the posterior edge of the anterior scalene muscle (Fig. 3b). Half of the patients had nodes located above the caudal edge of the cricoid bone. Internal mammary chain Jethwa et al. 21 delineated the locations of IMN metastases (the majority at primary diagnosis) against RTOG target volumes. Most patients had either N3 disease or DM. The area cranial to the first intercostal space and medial and lateral to the IMN vessels were found highest risk for geographical miss (in approximately 8% of cases). The majority of the nodal metastases were located within the first three intercostal spaces, but only half would have been appropriately covered by RTOG CTVs as they were several millimetres either medial or lateral to the vessels. Axilla Gentile et al. 30 evaluated 30 patients with clinically involved axillary lymph nodes on pretreatment CT against RTOG volumes. Tumour extension was evaluated relative to the axillary anatomical borders as determined by the RTOG consensus volumes. Tumour extension at level I was common at the cranial and anterior borders (>80% cases). The caudal border was the most prevalent site of tumour extension for levels II and III. Multiple CTV areas In the large, multi-institutional study by Chang et al., 22 recurrent nodal lesions were evaluated against ESTRO and RTOG CTVs. All patients experienced LRR (the majority of which were solitary), and had been treated initially with surgery followed by adjuvant therapy in the anti-her2 era. Some form of RNI had been given previously in 6.3% of patients with N0 disease, 48.1% of 110

10 Breast contour guidelines and failure patterns (a) A B C (b) (c) Fig. 3. Patterns of failure in the SCF. (a) Patterns of failure from a single institution (A = no adjuvant RT; B = adjuvant RT; C = adjuvant RT Breast only), previously published by Nielsen and Offersen, 2015; (b) Transferred enlarged nodes (lime green) from patients at either relapse or initial diagnosis, with normal anatomical structures. No previous radiation therapy to the SCV region, previously published by Jing et al., 2017; (c) Transferred enlarged nodes from patients at diagnosis. Cyan = solitary SCF/no DM. Magenta = solitary SCF + DM. Blue = multiple SCF, no DM. Red = multiple SCF/DM previously published by Brown et al., Reproduced with permission. patients with N1 disease and 98% of patients with N2 3 disease, although the specific inclusion of each nodal group treated was not individually described. Overall, the proportion of potential geographical miss outside the ESTRO, but inside the RTOG CTVs was 7.3%, and outside the RTOG CTV was 9%. The three regions deemed most vulnerable to potential geographical miss included the pectoralis muscle, the area cranial to the SCF and the area posterolateral to the SCF. The ESTRO breast/cw CTV did not encompass the observed pectoralis recurrence well, especially following mastectomy, although the RTOG CTV covered a little better. The authors commented that this may be related to thin chest walls in the patient cohort. In the SCF, the most prevalent site of regional recurrence, only a small percentage (~9%) was outside the ESTRO, inside the RTOG CTVs, cranial to the subclavian artery. However, a significant proportion of SCF recurrences (20%) were outside both the ESTRO and RTOG CTVs, posterolateral to the anterior scalene muscle. Previously, the same group published their single institution results, from patients treated in an earlier era. 23 The findings were similar to the multi-institutional study (and there may have been overlap in cases). Fourteen per cent of LRR were located outside the ESTRO, inside the RTOG CTVs. The SCF was again highlighted as a region susceptible to geographical miss. Here, 20% of recurrences were posterolateral to the ESTRO and RTOG CTVs, while a quarter were outside the ESTRO, but inside the RTOG CTV. Patients with locally advanced disease or DM were significantly more likely to experience recurrence outside the ESTRO-, inside the RTOG CTVs, or in the posterolateral SCF, when compared to patients with early stage disease. Discussion Differences in contouring nomenclature Many of the differences between the guidelines are likely to be clinically insignificant, especially nomenclature. For example, the interpectoral nodes are treated as a separate volume in ESTRO and PET LABRADOR but are included, along with the pectoralis minor muscle, as part of the axillary volumes in RTOG/RADCOMP. This may have little impact on planning where radiation fields are used to treat the CTVs as the muscles will be included in the fields; however, this could result in under dosage where volumetric planning using IMRT or particle therapy are used. Similarly, when the SCF is treated with 3D CRT 111

11 HE Gee et al. using a single anterior field off-cord, there is typically significant dose to the posterior neck region (in the order of 40 Gy), which may have been historically treating small volume microscopic disease. In contrast, when IMRT or protons are used, dose is much more conformal while this spares normal tissue and late toxicity, consideration needs to be given as to whether the posterior neck should be intentionally treated. Patterns of failure data and CTV coverage All seven studies identified as investigating this issue have been classified as case series (level IV evidence), acknowledging that there are significant limitations to producing the ideal study to assess the failure patterns of breast contouring guidelines. First, a long follow-up period is required to ascertain recurrence rates, particularly in the setting of early breast cancer and in an era of effective systemic therapy, 34,35 with corresponding reductions in LRR rates All studies were retrospective in nature, with potential bias from multiple sources, including incomplete patient information in clinical records. To circumvent these issues, several studies assessed patients with loco-regional metastases at diagnosis, with the authors acknowledging this was based on the assumption that sites of gross metastatic disease translate to regions at high risk of harbouring subclinical or micrometastatic disease in the setting of early stage disease. Small population sizes precluded them from drawing statistically significant conclusions. Apart from the editorial by Nielsen, 40 the studies do not specify the guidelines used at the time of treatment. Finally, there would be variability in their applications between institutions and individual clinicians depending on patient characteristics and treatment. Supraclavicular The most consistent finding was that the posterior neck region, posterolateral to RTOG and ESTRO volumes, was a site of metastasis or recurrence in three series 22,29,32 (Fig. 3b,c). The exact frequency and risk of this recurrence is difficult to assess, and reflects the composition of the case series, with higher rates of theoretical miss associated with higher nodal disease burden, stage IV disease, and possibly lower rates of current standard chemotherapy. In both papers by Chang et al., 22,23 which only examined recurrent lesions, the rate of SCF LN metastases outside the ESTRO/RTOG CTV was lower than in the other two series, which also included patients who presented with SCF metastases at diagnosis. 29,32 However, taken together, these studies suggest that the posterior volume, which extends beyond the scalene muscles, is at risk from metastatic breast cancer. This volume has been incorporated into the RADCOMP protocol; the posterior neck volume extends the SCF posteriorly to the levator scapulae. This is more posterior to the traditional RTOG SCF border of the anterior aspect of the scalene muscle. While the cranial border of the SCF was another prominent area of lymph nodes, with lesions noted superior to the cricoid cartilage in these three series, again, the percentage of lesions was much higher in patients with locally advanced or metastatic disease at diagnosis or recurrence (4% vs %). This suggests that for most patients treated with curative intent, the cricoid remains an appropriate superior border, especially given that disease above this may be considered systemic. Furthermore, it is expected that in the context of gross tumour, the extent of the CTV would vary that is the target [CTV] equally sufficient for cancer control in those with 1-3 axillary nodal metastases as in those with >10?. 41 Conflicting data exist on the difference between the ESTRO and RTOG superior SCF border (Fig. 3). On the one hand, one multi-institutional study found that 9% of supraclavicular lesions were superior to the ESTRO volume, all within 6 mm of the cranial border (subclavian artery). 22 In contrast, a large single institution database of 73 patients with 101 regional recurrences (patients treated with fields similar to the ESTRO guidelines), found no evidence for enlarging the SCF borders (or indeed, the other regional nodal CTVs). 40 Internal mammary chain Chang et al. 23 found the majority of IMC metastases were located in the first three intercostal spaces, whereas only a few were caudal to the fourth rib. This was similar to that of Jethwa et al., 21 and does not support routinely lowering the caudal margin of CTV. However, the use of a margin on the IMC vessels (rather than only the vessel itself) is supported by patterns of failure. Jethwa et al. 21 suggest that an expansion of 4 mm on the vessels would cover 90% of pathologically involved IMC nodes. The necessity of adequately covering the deep part of the IMC where it extends superiorly, as this can be challenging to plan without excessive dose to normal lung, has been discussed in the literature previously. The ESTRO guideline asks: whether this most cranial part of the [IMC] is clinically relevant, since the internal mammary veins drain into the brachiocephalic veins 1 2 cm caudal to the corresponding arteries...in the EORTC and DBCG IMN studies the [IMC] dorsal to the sternoclavicular joint was often not treated to the full dose. Based on data showing that the higher IMC are mostly involved in the context of lower IMC as well, Jethwa et al. comment that in women with known IMN involvement, cranial extension to the confluence of the IM vein with the brachiocephalic vein with or without caudal extension to the fourth or fifth interspace may be considered provided that normal tissue constraints are met. 112

12 Breast contour guidelines and failure patterns Axilla Chang et al. 23 found that most recurrences in axilla levels I and II were within the ESTRO CTV, with a low number of recurrences in level III. An unusual study design mapping pretreatment axillary lymph nodes detectable on CT onto simulation CTs (performed after neoadjuvant chemotherapy and dissection) found that most patients had an aspect of the presumed involved lymph node lying outside the anatomical borders of an axillary level. 30 The clinical significance of these gross nodes (compared with a more standard microscopic disease setting) is unclear. Chest wall An editorial fell outside our search criteria but analysed the patterns of recurrence following mastectomy for 278 patients across five series. 42 The most common site of recurrence was within skin and subcutaneous tissues anterior to the pectoralis musculature, with the second most common site being in pectoralis itself. The authors recommended revision in the RTOG guidelines to bring the chest wall CTV posterior border to the anterior rib surface, a change which has been incorporated into the RADCOMP guideline. Several studies, however, recommended maintaining a degree of personalisation to the posterior border, balancing the risk of a pectoralis recurrence against the increased toxicity to lung (and heart). 22 The ESTRO guidelines are predominantly intended for early breast cancer, and avoidance of enlargement of treated area compared with field-based planning was considered paramount, in order to spare late effects. In contrast, RADCOMP (which was designed for patients with stage II/III breast cancer) and PET LABRADOR were designed for more comprehensive coverage of patients potentially at greater risk of recurrence. The studies repeatedly demonstrate that the coverage of nodal disease would be lower in the setting of increased nodal stage disease burden and stage IV disease when guidelines are followed. Chang et al. 23 attempted to find clinicopathological factors associated with geographical miss, and while they were limited by small numbers, there appeared to be associations with locally advanced tumours, higher nodal burden, younger age, and unfavourable molecular subtype. Thus, more generous coverage may be recommended in patients with multiple LN, high risk or more locally advanced disease at presentation. While no guideline suggests a strict application of the approach, the patterns of failure data highlight the importance of clinical judgement and personalisation of therapy. This review has limited consideration of dosimetry and in-field recurrences. Chen et al. examined patterns of LRR in 21 patients who had received adjuvant RT, including half to the SCF. Only a small percentage were classified as in-field, while approximately one-third were marginal recurrences approximately equally distributed across breast, CW, and regional nodes. The out-of-field LRRs were found in IMC and SCF, areas not fully covered by the prescribed RT dose, or purposely excluded. Patterns of failure may well be different in the era of more effective systemic therapy, and a trend towards less extensive axillary surgery. For example, the difference between the Danish and other studies may reflect different institutional practices around the use of chemotherapy and the extent of RNI, with smaller fields acceptable in the context of highly effective systemic therapy. Additionally, there are different thresholds for the recommendation of RNI; the Danish RT protocol for many years has been full RNI even if only one axillary node was positive. DeSelm et al. 43 recently mapped patterns of failure after definitive therapy for breast cancer, in a large single institution cohort (published in abstract form). The axilla was the most common site for recurrence (40%), followed by SCF (31%) and IMN (29%). Nearly half had isolated LRR (no concurrent DM). Around 75% of the patients had not had prior nodal radiotherapy; prior RT reduced the rate but did not change the pattern of recurrence (DeSelm, Personal Communication). We await the full publication for further discussion of the high rate of axillary recurrence, which suggests axillary management may have been by sentinel lymph node biopsy only. Even in the presence of guidelines, in situations such as contouring a seroma cavity for boost, where there is significant anatomical uncertainty, there is still a significant degree of inter- observer variation. 44 Training is required to use guidelines correctly. We propose a patterns of practice study to understand how ANZ practitioners are applying guidelines to their patients as the first step in a Delphi consensus. The aim is to develop contouring guidelines for future TROG trials. Conclusions There are small but clinically significant differences between guidelines, reflecting their applicability to patients of various stages and risk groups. The most relevant consideration is that more generous coverage may be recommended for patients with multiple LN, high risk or more locally advanced disease at presentation. An ANZ consensus breast radiotherapy contouring guideline would inform clinical practice and facilitate the implementation and quality assurance of future TROG breast study protocols. Acknowledgements The authors acknowledge with gratitude the contribution of Tracy Pearl-Larson with ethics submission, Jacqueline Buck with the literature search and the staff and patients of the Crown Princess Mary Cancer Centre. 113

13 HE Gee et al. References 1. Poortmans PM, Collette S, Kirkove C et al. Internal mammary and medial supraclavicular irradiation in breast cancer. N Engl J Med 2015; 373: Whelan TJ, Olivotto IA, Parulekar WR et al. Regional nodal irradiation in early-stage breast cancer. N Engl J Med 2015; 373: EBCTCG, McGale P, Taylor C et al. Effect of radiotherapy after mastectomy and axillary surgery on 10-year recurrence and 20-year breast cancer mortality: meta-analysis of individual patient data for 8135 women in 22 randomised trials. Lancet 2014; 383: MacDonald SM, Jimenez R, Paetzold P et al. Proton radiotherapy for chest wall and regional lymphatic radiation; dose comparisons and treatment delivery. Radiat Oncol 2013; 8: Harsolia A, Kestin L, Grills I et al. Intensity-modulated radiotherapy results in significant decrease in clinical toxicities compared with conventional wedge-based breast radiotherapy. Int J Radiat Oncol Biol Phys 2007; 68: van Asselen B, Raaijmakers CP, Hofman P, Lagendijk JJ. An improved breast irradiation technique using three-dimensional geometrical information and intensity modulation. Radiother Oncol 2001; 58: Nguyen K, Mackenzie P, Allen A et al. Breast interest group faculty of radiation oncology: Australian and New Zealand patterns of practice survey on breast radiotherapy. J Med Imaging Radiat Oncol 2017; 61: Li XA, Tai A, Arthur DW et al. Variability of target and normal structure delineation for breast cancer radiotherapy: an RTOG Multi-Institutional and Multiobserver Study. Int J Radiat Oncol Biol Phys 2009; 73: Crook J, Milosevic M, Catton P et al. Interobserver variation in postimplant computed tomography contouring affects quality assessment of prostate brachytherapy. Brachytherapy 2002; 1: Atean I, Pointreau Y, Ouldamer L et al. A simplified CTbased definition of the supraclavicular and infraclavicular nodal volumes in breast cancer. Cancer Radiother 2013; 17: Abrams RA, Winter KA, Regine WF et al. Failure to adhere to protocol specified radiation therapy guidelines was associated with decreased survival in RTOG 9704 a phase III trial of adjuvant chemotherapy and chemoradiotherapy for patients with resected adenocarcinoma of the pancreas. Int J Radiat Oncol Biol Phys 2012; 82: Peters LJ, O Sullivan B, Giralt J et al. Critical impact of radiotherapy protocol compliance and quality in the treatment of advanced head and neck cancer: results from TROG J Clin Oncol 2010; 28: Faculty of Radiation Oncology. Quality Guidelines for Volume Delineation in Radiation Oncology, 2015, 23 October Available from URL: om/college/document-library/quality-guidelines-forvolume-delineation-in-radiation-oncology. 14. Offersen BV, Boersma LJ, Kirkove C et al. ESTRO consensus guideline on target volume delineation for elective radiation therapy of early stage breast cancer. Radiother Oncol 2015; 114: Offersen BV, Boersma LJ, Kirkove C et al. ESTRO consensus guideline on target volume delineation for elective radiation therapy of early stage breast cancer, version 1.1. Radiother Oncol 2016; 118: White JT, Arthur D, Buchholz T et al. Breast Cancer Atlas for Radiation Therapy Planning: Consensus Definitions, accessed 21 February Available from URL: aspx?fileticketzvzjfhpabipez 17. Bekelman J, Cahlon O, McDonald S. Pragmatic Randomized Trial of Proton vs. Photon Therapy for Patients With Non-Metastatic Breast Cancer: A Radiotherapy Comparative Effectiveness (RADCOMP) Consortium Trial [accessed 21 February 2017]. Available from URL: NCT Dijkema IM, Hofman P, Raaijmakers CP, Lagendijk JJ, Battermann JJ, Hillen B. Loco-regional conformal radiotherapy of the breast: delineation of the regional lymph node clinical target volumes in treatment position. Radiother Oncol 2004; 71: Kirova YM, Servois V, Campana F et al. CT-scan based localization of the internal mammary chain and supra clavicular nodes for breast cancer radiation therapy planning. Radiother Oncol 2006; 79: Graham P. A randomised comparison of anastrozole commenced before and continued during adjuvant radiotherapy for breast cancer versus anastrozole and subsequent anti-oestrogen therapy delayed until after radiotherapy. (STARS study of anastrozole and radiotherapy sequencing). TROG Research Group; Jethwa KR, Kahila MM, Hunt KN et al. Delineation of internal mammary nodal target volumes in breast cancer radiation therapy. Int J Radiat Oncol Biol Phys 2017; 97: Chang JS, Lee J, Chun M et al. Mapping patterns of locoregional recurrence following contemporary treatment with radiation therapy for breast cancer: a multi-institutional validation study of the ESTRO consensus guideline on clinical target volume. Radiother Oncol 2017 Oct 16. PubMed PMID: Chang JS, Byun HK, Kim JW et al. Three-dimensional analysis of patterns of locoregional recurrence after treatment in breast cancer patients: validation of the ESTRO consensus guideline on target volume. Radiother Oncol 2017; 122: Nielsen MH, Berg M, Pedersen AN et al. Delineation of target volumes and organs at risk in adjuvant radiotherapy of early breast cancer: national guidelines 114