第一部分

目的：回顾分析本中心治疗头颈部神经内分泌肿瘤(HNNEN)的临床特点，治疗疗效、失败模式及预后因素。

方法：回顾性分析2000年11月至2021年11月我院收治的93例初诊无远地转移的头颈部神经内分泌肿瘤患者的临床资料。其中男性71例，女性22例，发病中位年龄为53岁。最常见的原发部位为：鼻腔(25例，26.9%)、口腔(23例，24.7%)、副鼻窦(19例，20.4%)。根据第七版AJCC头颈部肿瘤分期，共有Ⅰ-Ⅱ期33 例，Ⅲ-Ⅳ期59例，1例不明分期。治疗方案包括：化疗联合放疗(C+RT组)40例，手术后放疗(S+RT组)34 例，单独手术后挽救治疗(S+Sa组)19例。

结果：中位随访时间为64.5月(3-180月)， 总体患者的5年总生存率(OS)、无进展生存率(PFS)、局部无复发生存率(LRRFS)和无远地转移生存率(DMFS)分别为64.5%、51.6%、66.6%和 62.1%。根据不同治疗模式，对于Ⅰ-Ⅱ期，接受放疗(C+RT和S+RT组)和未接受放疗(S+Sa组)的患者的5年LRRFS分别为75.0%和12.7%(p=0.015)。而对于Ⅲ-Ⅳ期患者，接受和未接受放疗患者的5年LRRFS分别为77.8%和50.0%(p=0.006)；接受全身治疗(C+RT组)和未接受全身治疗(S+RT组和S+Sa组)的5年DMFS分别是71.2%和51.5%(p=0.075)。共44例(47.3%)患者出现治疗失败，远地转移是主要失败模式。单因素预后分析提示分化程度是影响OS和DMFS的预后因素，治疗模式是影响LRRFS的预后因素；多因素分析提示接受放疗是提高LRRFS的独立预后因素，全身治疗是提高DMFS的独立预后因素。

结论：放射治疗显著改善头颈部神经内分泌肿瘤的局部控制，并在HNNEN的治疗中发挥了重要作用。 HNNEN的最佳治疗方案仍然是局部治疗和全身治疗相结合。

第二部分  

目的：头颈部神经内分泌肿瘤(HNNEN)是一种罕见的恶性肿瘤，其颈部淋巴结引流规律尚未见报道，是否颈部预防照射(ENI)是目前争议的焦点之一。本研究旨在总结HNNEN颈部淋巴结引流规律，探讨颈部预防照射在无淋巴结转移(N0)的HNNEN的价值。

方法：回顾性分析2000年11月至2021年11月我院收治的93例初诊无远地转移的头颈部神经内分泌肿瘤患者的临床资料。其中38例为淋巴结阳性(N+)的患者，分析淋巴结转移的比例及引流规律。55例为N0患者，其中42例接受放射治疗，根据是否接受预防性颈部照射进行分组。

结果：头颈部神经内分泌肿瘤的颈部淋巴结转移发生率为40.86%，最常见的区域是Ⅱ区和Ⅲ区。42例接受放射治疗的N0患者中，28例接受颈部预防照射，随访中无颈部区域淋巴结复发；14例未接受颈部预防照射，其中3例(21.4%)出现颈部区域淋巴结复发(p=0.040)。有无颈部预防照射两组的总生存(OS)、无进展生存(PFS)、无远地转移生存(DMFS)，均未见统计学差异。接受颈部预防照射组有更低的区域失败，但是局部复发率和远地转移率两组之间没有差异。

结论：头颈部神经内分泌肿瘤的颈部淋巴结转移基本符合逐站转移规律。对于NO的病变，颈部预防照射能够显著提高区域控制率，但是不能提高总生存率。

第三部分  

目的：头颈部鳞癌(HNSCC)是全球第六大常见癌症，尽管治疗模式在更新，其治疗效果仍不能令人满意。本研究应用生物信息学方法，基于放疗敏感性及cGAS-STING通路相关基因进行头颈部鳞癌分型，以预测预后及免疫治疗敏感性。

方法：从GEO数据库下载GSE67614(共包含102例HNSCC样本，其中放疗后未复发68例，放疗后复发34例)和GSE65858数据集。从 UCSC-Xena数据库下载TCGA-HNSCC的mRNA表达谱和临床数据。分析差异表达基因(DEGs)。从Msigdb数据库中获得78个与cGAS-STING通路相关的基因。将DEGs与cGAS- STING通路相关基因相交，得到cGAS-STING相关的差异表达基因(CSDEGs)。应用生物信息学方法构建CSDEGs评分，验证CSDEGs评分对预后及免疫治疗敏感性的预测作用。

结果：根据9个CSDEGs，将HNSCC分为两个亚型，A型和B型。B型的肿瘤免疫相关的通路均得分更高，提示更好的预后和免疫治疗敏感性。我们分析了CSDEGs评分与免疫细胞浸润的相关性，发现CSDEGs评分高的患者免疫细胞浸润水平升高。存活的患者与已经去世的患者相比，CSDEGs评分更高。高CSDEGs评分组的免疫检查点表达升高，进一步支持了CSDEGs评分高的肿瘤患者可能对免疫治疗反应更好。

结论：CSDEGs评分可以较好的预测HNSCC的预后以及对免疫检查点抑制剂的敏感性，可辅助临床筛选免疫治疗的获益人群，为HNSCC患者带来获益。

Part 1

Background：Neuroendocrine neoplasm is a rare cancer of the head and neck. This study aimed to evaluate the clinical features, treatment outcomes and prognostic factors of neuroendocrine neoplasm of head and neck treated at a single institution.

Methods: Between Nov.2000 and Nov.2021, ninety-three patients diagnosed with neuroendocrine neoplasms of head and neck treated at our institution were reviewed retrospectively. The initial treatments included chemotherapy (induction, adjuvant, or concurrent) combined with radiotherapy in 40 patients (C+RT group), surgery followed by post-operative RT in 34 (S+RT group), and surgery plus salvage therapy in 19 patients (S+Sa group).

Results: The median follow-up time was 64.5 months. The 5-year overall survival rate (OS), progression-free survival rate (PFS), locoregional relapse-free survival free rate (LRRFS) and distant metastasis-free survival rate (DMFS) were 64.5%, 51.6%, 66.6%, and 62.1%, respectively. For stage I-II, the 5-year LRRFS for patients’treatment regimen with or without radiotherapy (C+RT and S+RT groups versus S+Sa group) were 75.0% versus 12.7% (p=0.015) while for stage III-IV, the 5-year LRRFS were 77.8% versus 50.0% (p=0.006). The 5-year DMFS for patients with or without systemic therapy (C+RT group versus S+RT or S+Sa) were 71.2% and 51.5% (p=0.075). 44 patients (47.3%) experienced treatment failure and distant metastasis was the main failure pattern.

Conclusions:Radiotherapy improved local-regional control and played an important role in the management of HNNENs. The optimal treatment regimen for HNNENs remains the combination of local and systemic treatments.

Part 2

Background: Neuroendocrine neoplasms of the head and neck (HNNEN) is a rare malignant neoplasm with great controversies on management of the neck. The aim of this study is to characterize the incidence and regional lymphatic spread patterns of N+ disease and assess the role of elective neck irradiation (ENI) in patients with N0 disease.

Methods: Between Nov.2000 and Nov.2021, ninety-three patients diagnosed with neuroendocrine neoplasms of head and neck treated at our institution were reviewed retrospectively. Among them, 38 patients were lymph node positive (N+). The proportion of lymph node metastasis and drainage pattern were analyzed. 55 patients were N0 patients, 42 of whom received radiotherapy, and were divided into groups according to whether they received elective neck irradiation.

Results: The incidence of cervical lymph node metastasis in head and neck neuroendocrine tumors is 40.86%, and the most common areas are areas II and III. Among 42 N0 patients who received radiotherapy, 28 received ENI, and there was no recurrence of regional lymph nodes in the neck，14 patients did not receive ENI, and 3 of them (21.4%) experienced recurrence of regional lymph nodes in the neck (p=0.040). There were no statistical differences in overall survival (OS), progression-free survival (PFS), and distant metastasis-free survival (DMFS) between the two groups with and without ENI.

Conclusions: Cervical lymph node metastasis of head and neck neuroendocrine tumors basically follows the station-by-station metastasis pattern. For N0 patients, ENI can significantly improve the regional control rate, but it cannot improve the overall survival rate.

Part 3

Background: Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer and the treatment effects are still unsatisfactory. This study intends to apply bioinformatics methods to predict prognosis and immunotherapy sensitivity of HNSCC based on radiotherapy sensitivity and cGAS-STING pathway related genes.

Methods: GSE67614 (containing a total of 102 HNSCC samples, including 68 cases without recurrence after radiotherapy and 34 cases with recurrence after radiotherapy) and GSE65858 data sets were download from the GEO database. The mRNA expression profile and clinical data of TCGA-HNSCC were download from the UCSC-Xena database. Differentially expressed genes (DEGs) were analysed. 78 genes related to the cGAS-STING pathway were obtained from the Msigdb database. The DEGs were intersected with cGAS-STING pathway-related genes to obtain cGAS-STING-related differentially expressed genes (CSDEGs). Bioinformatics methods were used to construct CSDEGs scores and verify the predictive effect of CSDEGs scores on prognosis and immunotherapy sensitivity.

Results: HNSCC is divided into two clusters (A and B), based on 9 CSDEGs. Cluster B tumors have higher scores on immune-related pathways, indicating better prognosis and sensitivity to immunotherapy. We analyzed the correlation between CSDEGs score and immune cell infiltration and found that patients with high CSDEGs scores had elevated levels of immune cell infiltration. Surviving patients had higher CSDEGs scores compared with deceased patients. The expression of immune checkpoints in the high CSDEGs score group was increased, further supporting that tumor patients with high CSDEGs scores may respond better to immunotherapy.

Conclusions: The CSDEGs score can well predict the sensitivity of HNSCC to immune checkpoint inhibitors, and can be used in clinical screening of people who will benefit from immunotherapy.