Delineation of Neck Node Levels for Patients with Locally Advanced Supraglottic Cancer Receiving Radical IMRT: A Cross-Sectional Study in Mainland China

Yi Xu Chinese Academy of Medical Sciences and Peking Union Medical College Meilin He Chinese Academy of Medical Sciences and Peking Union Medical College Yang Liu Chinese Academy of Medical Sciences and Peking Union Medical College Jun-Lin Yi Chinese Academy of Medical Sciences and Peking Union Medical College Ye Zhang (  ivy2019xu@163.com ) Chinese Academy of Medical Sciences and Peking Union Medical College


Introduction
Laryngeal cancer (LC) is a challenging clinical problem with a relatively high incidence rate among head and neck cancers. An estimated 26.4 per 100,000 new diagnoses and 14.5 per 100,000 deaths of LC occurred in 2015 in China 1 . In the United States, newly diagnosed LC affected an estimated 12,410 people in 2019 2 . More attention is paid to maintaining quality of life without compromising survival when choosing the treatment modality because the larynx is a vital functional organ, used as speaking, breathing and swallowing. De nitive chemoradiation therapy as a larynx preservation strategy is recommended as the mainstay treatment for LC according to the NCCN guidelines. Patients who refuse surgery often prefer radiotherapy.
Precise delineation of the target volume is important for radical radiotherapy, especially in the era of highly conformal techniques, including intensity-modulated radiotherapy (IMRT). Compared to twodimensional and three-dimensional conformal radiotherapy, IMRT reduces long-term toxicities such as xerostomia and dysphagia; nevertheless, radiation-induced dysphagia resulting in feeding tube intubation or weight loss remains a serious issue 3 . Meanwhile, the selection of lymph node clinical target volumes (LN-CTVs) has not yet reached consensus. For example, international guidelines support irradiating bilateral levels II-IV in LC patients with clinically negative lymph nodes 4 ; however, the Danish Head and Neck Cancer Group excludes bilateral level IV from CTV 5 . The de nitions of high-risk (HR) and low-risk (LR) CTV also vary. European criteria de ne isotropic expansion of 1 cm from the edge of positive lymph node (LN+) as HR-CTV 4 , while North American criteria de ne node levels with LN + 6 . All anatomic subgroups of LC are always discussed together; however, supraglottic cancer has a much higher prevalence of LN metastases than does glottic cancer, especially in locally advanced stages 7 . In China, there is also no standard guideline for LN-CTV delineation for locally advanced supraglottic cancer (LA-SGC), and large differences exist among hospitals. Therefore, this cross-sectional study aimed to evaluate hospital variations in LN-CTV delineation for LA-SGC patients receiving radical IMRT in mainland China. The data regarding the selection and de nition of LN target volume were addressed. We also analysed risk factors that may in uence the determination. Finally, this work hopefully provides a basis for improving delineation consistency for LN-CTV.

Hospital selection
All provincial cancer hospitals, one randomly selected provincial general hospital and one randomly selected municipal general hospital were enrolled from each of 30 provinces (including 22 provinces, 4 province-level municipalities, and 4 autonomous regions) in mainland China. The Tibetan autonomous region was excluded for lack of radiotherapy equipment. Radiotherapy apparatus is scarce in Hainan, Qinghai and Ningxia provinces, and each had only tumour hospitals included. Two attending radiation oncologists in charge were considered for being recruited from the selected hospitals, with one mediumgrade professional title and one high-grade professional title. All doctors specialised in head and neck cancer. The informed consent was obtained from the all participants or their respective legally authorized persons.

Case 1
The patient was a 64-year-old woman who had been smoking for more than 40 years. Diagnosis: Welldifferentiated SGC with stage cT2N1; invading right ventricular band and aryepiglottic fold; an LN+ (the maximum short diameter was 2 cm, no invasion of capsule on MRI) in right level II.

Case 2
The patient was a 59-year-old man with hypertension for 5 years who had been smoking and drinking for 40 years. Diagnosis: SGC with stage cT4aN0 well differentiation; invading right ventricular band, vocal cords, paraglottic space, aryepiglottic fold, medial wall of pyriform sinus, belt-shaped muscle (limited to the right), and no positive LN on MRI.

Case 3
The patient was a 66-year-old man who had been smoking for 40 years and drinking for 20 years. Diagnosis: SGC with stage cT3N2b well differentiation; invading right ventricular band, paraglottic space, aryepiglottic fold, medial wall of pyriform sinus and xed vocal cords; two LNs+ (the largest short diameter of the largest LN + was 3 cm, with extracapsular spread and necrosis on MRI) in right level II.

Case 4
The patient was a 54-year-old man with hypertension for 30 years who had been smoking and drinking for 40 years. Diagnosis: SGC with stage cT4aN2c well differentiation; invading right ventricular band, vocal cords, aryepiglottic fold, medial wall of pyriform sinus, paraglottic space, arytenoid cartilage, thyroid cartilage plate and belt-shaped muscle; ve LNs+ (the largest short diameter of the largest LN + was 3 cm, without extracapsular spread and necrosis on MRI) in right levels II and III, two LNs+ (the largest short diameter was 2 cm, with extracapsular spread and necrosis on MRI) in left level II.
The neck node level de nition followed the international consensus guideline updated in 2013 9 . We asked whether each lymph node level should be irradiated or not, then whether it should it be included in LR-CTV (dose 44-50 Gy) or HR-CTV (dose 54-63 Gy).
Numerous studies have identi ed that risk factors, including subregion involvement, pathological differentiation and midline involvement, affect the risk of LN metastasis, which may in uence the choices of respondents [10][11] . Then, we attached these strings: subglottic region involvement to cases 2-4, poor differentiation and tumour crossing the midline to cases 1-4. Their in uence on the selection of lymph node levels was also estimated. Meanwhile, participants were asked "which factors will affect your determination of LN-CTV? For example, extracapsular spread, T stage and N stage, and so on." Statistical analysis Different selections of neck node level were presented in gures (0 = unirradiated, 1 = as HR region, 2 = as LR region) for each case. The frequencies (n, %) as descriptive statistics of agreement were calculated. The chi-square test was used to determine whether there were signi cant differences between cases when strings were added, and a p-value < 0.05 was considered signi cant. The percentage of participants agreeing or strongly agreeing that risk factors had an impact on LN-CTV delineation was calculated using 95% con dence intervals.

Results
A sample of 164 valid questionnaires from 82 hospitals were nally received, and the recovery rate was 100%. The distribution of respondents by different quali cations and working years is shown in Table 1. Of the participants, nearly half were chief physicians with over 10 years of working experience.

Risk factors
When the subglottic region was involved, 50.6%, 46.2% and 56.2% of respondents selected ipsilateral level VIb as HR/LR-CTV in cases 2 to 4, respectively, compared to 13.1%, 15%, 26.2%, respectively, when the subglottic region was not involved.

Discussion
Most respondents preferred to follow the North American criteria to delineate HR-CTV, while few physicians followed the European criteria. The lower adjacent level next to the LN + level was also included in HR-CTV. In four cases, bilateral levels II-IV were mainly discussed, and physicians tended to enhance treatment for stage N2 compared with stage N0-1. The rate of selecting levels Ib and V as CTV increased in patients with stage N2. Subglottic region involvement indicates irradiation of level VIb. Tumours crossing the midline, extracapsular spread, T stage and N stage were considered risk factors for in uencing nodal level determination.

cN0-1 stage
In the cN0-1 stage, ipsilateral levels II and III selected as HR-CTV and ipsilateral level IV as LR-CTV reached good agreement. The controversy turned on whether contralateral levels II and III should be included in HR-or LR-CTV and whether contralateral level IVb should be excluded from CTV. Ipsilateral levels II and III are considered HR-CTVs because they are the most frequently involved, even for patients with cN0 stage. According to previous studies, the rate of occult metastases exceeded 20% in SGC 7 12 13 .
Irradiating contralateral levels II and III also reached good agreement, regardless of whether the primary tumour was limited to one side or not, because SGC has a tendency to develop contralateral lymph node metastasis 14 . Because of the rare skip metastasis of nodal levels 13 15 , ipsilateral level IV was included in LR-CTV if ipsilateral level II was involved, which would be in HR-CTV if level III was involved.
We found that the T stage may affect the delineation of level IIb by comparing cases 1 and 2. Some studies showed that level IIb rarely developed occult metastases (1-6%) in the cN0 stage when level IIa and IIb were analysed based on the pathological test results after surgery 16 20 . In our series, the rates of including contralateral level II in HR-or LR-CTV were approximately equal, while in T4 stage, more respondents included contralateral levels II and III in HR-CTV other than LR-CTV (60.6% vs 36.6%; 56% vs 39.4%). According to this result, T stage has more in uence on contralateral neck metastases than ipsilateral single positive lymph nodes, which made more respondents choose contralateral levels II-III as HR-CTV in advanced T stage.
Some investigators suggested not irradiating bilateral level IV in the cN0 stage because occult metastasis rarely occurs 21 . Danish national guidelines for head and neck cancer radiotherapy also excluded them from CTV 24 . However, these previous studies were not much concerned about risk factors. Lim et al. observed that 33.3% of patients with LC (57.5% were SGC) in the T4N0 stage had ipsilateral level IV metastases, while 0%, 3.3%, and 5.9% had ipsilateral level IV metastases in the T1-3 stages, respectively. Our data also showed that more responses agreed to irradiate contralateral level IV in the T4N0 stage (69.4%) than in the T2N1 stage (53.7%). Perhaps bilateral level IV should not be omitted in the T4N0 stage, and further research is needed.
cN2 stage In the cN2 stage, the bilateral levels II-V other than HR-CTV regions were all included in LR-CTV, which was indicated in most respondents (75-92.5%). Compared to the cN0-1 stage, more respondents preferred to include ipsilateral level IV and contralateral levels II-III in HR-CTV. Levels Ib and Va/b were more likely included in CTV when there were multiple LNs + in the ipsilateral neck. Level VIb was selected as the CTV when the subglottic region was involved. For the question regarding which risk factors may in uence physicians in delineating CTV, most respondents took tumours crossing the midline, extracapsular spread, T stage and N stage into consideration. However, how these factors change radiation oncologists' decisions in speci c situations needs deeper investigation.
There are several limitations to our study. First, the disputes regarding which method of HR-CTV delineation would be better were not addressed. European criteria are complicated in clinical application compared to North American criteria. It is reasonable to apply the 1-cm expansion around LN + for HR-CTV because it would cover more than 99% of likely tumour extension 27 . The geometric margins resulted in more uniform and left less room for misinterpretation 28 , but it adds burden to physicians when formulating treatment plans, which may lead most hospitals in mainland China to use the North American criteria. Second, we only chose some risk factors associated with LN metastasis from published reports in which certain missing information is inevitable. We could also not analyse how the other risk factors mentioned in the questionnaire in uent physicians in speci c situations. In addition, HPV status, conditions of retropharyngeal lymph nodes and combination treatments (chemotherapy, targeted therapy or immunotherapy) are not present in designed cases, which may affect the nal results.

Conclusion
In mainland China, most physicians follow North America criteria of CTV delineation. For patients with LA-SGC, bilateral levels II-IV are generally irradiated. However, which nodal level should be included in HR-or LR-CTV showed poor agreement, especially for contralateral levels II-IV. Levels Ib, Va/b and VIb are selected as CTVs under some circumstances, such as multiple lymph node metastasis in the ipsilateral neck and subglottic region involvement. Other risk factors, such as tumour crossing the midline, extracapsular spread, T stage and N stage, also in uenced decisions regarding delineating LN-CTV, which needs further research to indicate