1.1部分

目的：对于术后病理分期 pT1N1M0(IB 期)的胃癌患者，术后辅助化疗是否都能生存获益，目前仍存在争议。我们希望通过回顾性分析为 pT1N1M0(IB 期)胃癌的化疗指征提供依据。

方法：我们收集了 2011 年 1 月-2017 年 12 月就诊于中国医学科学院肿瘤医院并行根治性手术治疗的所有早期胃癌患者的临床病理资料。根治术后经组织病理证实为pT1N1M0 期的胃癌中具有完整数据资料的共 179 例。我们对这部分病人进行回顾性分析，分类变量采用 Fisher 精确检验或 χ2 检验进行比较。采用 Kaplan-Meier 生存分析，并采用 log-rank 检验进行比较，以评估差异是否具有显著性。使用 Cox 风险比例模型进行单因素和多因素生存分析，P 值<0.05 认为有统计学意义。

结果：将符合纳入标准的 179 例 pT1N1M0 患者根据术后是否接受辅助化疗分为两组（接受化疗组和不接受化疗组），生存分析显示，接受化疗治疗组的总生存期(OS)和肿瘤特异性生存期(CSS)较不接受化疗组更长，具有显著差异(P <0.01)。 Cox回归分析发现对于 pT1N1M0 肿瘤患者，性别(男，危险比[HR] 2.470, 95%可信区间[CI] 1.294-4.718)，清扫淋巴结总数(EN)(EN≤15，[HR] 2.402；95% CI1.329-4.341)和是否辅助化疗(未化疗，HR 2.554；95% CI 1.393-4.681)是影响OS 的独立危险因素为，也是影响 CSS 的独立危险因素。我们利用多因素分析中的两个重要预后因素(性别和 EN)，分为四个亚组(男，EN≤15；男，EN>15；女，EN≤15；女，EN >15)。生存分析显示，四组间 OS 和 CSS 存在显著差异，但生存曲

线存在重叠(男，EN>15)和(女，EN≤15)预后相似。因此，我们将这两个亚组进行合并，风险级别分为：高风险(男，EN≤15)、中风险(男，EN>15 或女，EN≤15)和低风险(女性，EN>15)。新的分类似乎有一个更佳的预后分层。各组间 OS、CSS均有显著差异(P<0.05)。

结论：对于部分 pT1N1M0 胃癌患者，术后辅助有望带来额外的生存益处，尤其是男性患者，当 EN≤15 时可能更需要进行辅助化疗。

1．2部分

目的:对于术后病理分期为Ⅰ期(T1N0 和 T2N0)且淋巴结为阴性(N0)的胃癌患者，术 后的辅助化疗是否能给所有患者带来生存获益，目前仍存在争议。

方法:我们收集了 2011 年 1 月至 2017 年 12 月就诊于我院，行根治性手术切除且术后组织病理证实为 pT1N0 和 pT2N0 的所有胃癌患者，筛选出具有完整病例和随访 数据资料的 1601 例进行回顾性分析。

结果:我们对 1601 例，符合纳入标准的早期胃癌患者进行 Cox 回归分析发现，影响总生存期（OS）的独立危险因素包括：男性(危险比[HR] 1.454, 95%可信区间 [CI] 1.127-1.876)、年龄≥65 岁(HR: 1.367；95% CI 1.071-1.744)，T2 期(HR 1.283；95% CI 1.005-1.638)，肿瘤大小>3cm (HR 1.704；95% CI 1.346-2.158)，清扫淋巴结总数(EN)≤15 (HR 1.327；95% CI 1.058-1.664)和非印戒细胞癌的病 理类型(HR 1.639；95% CI 1.123-2.392)。进一步用 Cox 回归分析发现，影响肿瘤 特异性生存期（CSS）的独立危险因素包括 T2 期(HR 1.410；95% CI 1.026-1.936)， 肿瘤大小>3cm (HR 1.755；95% CI 1.285-2.397)，清扫淋巴结总数(EN)≤15 (HR 1.489；95% CI 1.101-2.015)。我们根据这两个影响预后的危险因素(肿瘤直径和 EN)将 pT2N0 患者分为四组，组 1(EN≤15, 肿瘤直径≤3 cm)，组 2（EN>15, 肿 瘤直径≤3 cm），组 3（EN≤15, 肿瘤直径>3 cm）和组 4（EN>15, 肿瘤大小>3 cm )， 分组分析发现只有在第 3 组(EN≤15，肿瘤直径>3 cm)患者 CSS 的延长可能受益于 辅助化疗(P=0.049)，并且在第 3 组内的非印戒细胞癌患者中，辅助化疗对延长 CSS 的作用可能更为显著(P=0.034)。

结论：并非所有的 pT2N0 胃癌患者都需要术后辅助化疗，但是部分高危的患者，如 EN≤15 且肿瘤直径>3cm 的非印戒细胞癌患者，术后辅助化疗可能对延长生存期更 有帮助。

第二部分

目的：对于有胰头侵犯的胃癌患者，有研究表明多器官联合切除(extended multi-organ resection, EMR)可提高患者的生存率。然而，也有研究显示联合脏器切 除会导致并发症发生率和死亡率升高。对于T4b胃癌的联合脏器切除治疗仍有争议。 本研究旨在探讨 pT4b 期胃癌合并胰头侵犯的合理手术方式。

方法：我们收集了 2006 年 1 月至 2016 年 12 月到中国医学医学院肿瘤医院就诊的 胃癌患者资料，其中有 143 例胃癌侵犯胰头患者行手术治疗。我们对 76 例接受胃 切除术联合胰十二指肠切除术组(GP 组)或单纯胃切除术组(GA 组)，且术后病理证 实为胃癌侵犯胰腺的患者进行回顾性分析，比较患者的临床病理特征、手术结果和 预后因素，其余 67 例患者行探查手术或姑息短路手术。

结果：纳入分析的 76 个患者中，24 例(16.8%)接受了根治性胃切除术+胰十二指肠切 除术，52（83.2%）例行单纯胃切除术，术中或术后病理显示胃癌侵犯胰腺。两组 比较，GP 组肿瘤直径较大(P< 0.001)，尤其是胰瘘的 发生率明显高于 GA 组。GP 组患者的 3 年总生存率显著长于 GA 组(47.6%，中位 30.3 个月 vs. 20.4%，中位 22.8 个月，P=0.010)。多因素分析显示新辅助化疗(危险 比[HR] 0.290；95%可信区间[CI] 0.103-0.821；P=0.020)，皮革胃(HR 2.614；95% CI 1.024-6.675；P=0.033)，手术切缘(HR 0.274；95% CI 0.102-0.738；P = 0.010)，N 分 期(HR 3.489；95% CI 1.334-9.120；P=0.011)和术后放化疗(HR 0.369；95% CI 0.163-0.836；P=0.017)是 pT4b 期胃癌侵犯胰头患者生存相关的独立危险因素。

结论：对于胃癌侵犯胰头的部分患者可以通过根治性联合脏器切除提高生存率。胰 头的侵犯并不是手术的绝对禁忌症，但是手术的并发症发生率增高。因此联合脏器 切除需要谨慎开展，手术的风险和获益需要更多研究去探索。

第三部分

目的：胃癌具有高度异质性，胃腺癌细胞与印戒细胞常以不同比例形式存在于同一个肿瘤。本研究试图根据印戒细胞的含量不同分组，比较各组间预后的差异，并进 一步分析不同含量的胃印戒细胞癌的肿瘤突变负荷的差异，基因变异特征与预后关系，从基因水平揭示预后差异的原因，探究不同印戒细胞含量的胃癌组织中多种肿 瘤浸润免疫细胞的密度和阳性率，试图寻找免疫浸润细胞种类和含量的差异在预测 预后方面的价值，为临床准确预测治疗反应和评估预后提供理论依据。

方法：共收集 27例含不同比例胃印戒细胞癌成分的胃癌患者临床病例信息，对术后标本蜡块切片，采用基因二代测序技术捕获 520 个与肿瘤发生发展的生物学原理 及个性化治疗方案高度相关的基因突变特征，采用多重免疫荧光的方法检测肿瘤实 质和间质中 CD3+T，CD8+T,CD56+NK，CD68+CD163-M1，CD68+CD163+M2 及 PD1 和 PDL1 阳性细胞密度和阳性率。根据印戒细胞含量不同分为高印戒细胞组 （H-SRC，印戒细胞含量>50%）和低印戒细胞组（L-SRC，印戒细胞含量<50%），对患者临床病理特征、基因突变、免疫状态进行回顾性统计分析。

结果：H-SRC 组的肿瘤直径稍大于 L-SRC 组 (5.54 cm vs. 4.06 cm, P = 0.043)，其他基本特征两组间无明显的差异。对两组进行生存分析显示，L-SRC 组的预后较差，其中Ⅱ、Ⅲ期患者中生存差异具有统计学意义（P=0.045）。两组的基因突变谱及免疫标记物表达情况均无明显差异。Cox 单因素生存分析显示：T 分期、淋巴结阳性率、神经侵犯及 KRAS、ARID1A、MET 、FANCA4 个基因突变与预后相关，且Cox 多因素分析中以上 7 个指标仍具有统计学意义，均为预后不佳的高危因素。

结论：较低印戒细胞含量的胃癌预后更差，且在Ⅱ、Ⅲ期患者中差异显著。T 分期，淋巴结阳性率，神经浸润，KRAS、ARID1A、MET 、FANCA 基因突变是预后不佳的高危因素。需要扩大样本量明确研究结论的准确性，也需要更深层次探索预后不佳的原因，为临床精准化、个体化治疗提供基础理论指导。

Part1.1

Purpose: Whether adjuvant chemotherapy (AC) has a survival beneft for all patients with pathological stage pT1N1M0 (Stage IB) gastric cancer (GC) remains controversial.

Methods: A total of 179 with surgically resected, histologically confrmed pT1N1M0 GC between January 2011 and December 2017 at the National Cancer Center, China, were retrospectively reviewed. The Fisher's exact test or χ2 test was used to compare the categorical variables. The cumulative survival rate was calculated by Kaplan-Meier method and compared by log-rank test to evaluate whether the difference was significant. Univariate and multivariate survival analyses were performed using Cox risk scale models. Single factor with statistical difference was subsequently included in multivariate analysis, and P < 0.05 was considered statistically significant.

Results: A total of 179 patients with pT1N1M0 were identifed. Survival analysis showed that both overall survival (OS) and cause-specifc survival (CSS) were significantly different between patients treated with and without AC (P < 0.01). Independent risk factors for reduced OS identified in the Cox regression analysis in patients with pT1N1M0 cancer were sex (male sex, hazard ratio [HR] 2.470, 95% confidence interval [CI] 1.294–4.718), examined lymph nodes (EN) (EN ≤ 15, HR 2.402; 95% CI 1.329– 4.341), and AC (treated without AC, HR 2.554; 95% CI 1.393–4.681), which were also independent risk factors for reduced CSS. We divided patients with pT1N1M0 into three risk categories (high, moderate, and low) according to two significant prognostic factors (sex and EN) and found that both OS and CSS were significantly different between the three risk groups (P < 0.05).

Conclusion: An additional survival benefit related to AC is expected for selected pT1N1M0 patients. Male patients with EN ≤ 15 may be particularly appropriate candidates for AC.

Part 1.2

Purpose: Whether adjuvant chemotherapy (AC) has a survival benefit for pathological stage Ⅰ (T1N0 and T2N0) gastric cancer (GC) patients with negative lymph node (N0) remains controversial.

Methods: Patients with surgically resected, histologically confirmed pT1N0 and pT2N0 GC between January 2011 and December 2017 at the National Cancer Center, China, were retrospectively reviewed.

Results: A total of 1601 patients who met the inclusion criteria were identified. Independent risk factors for reduced overall survival (OS) identified in the Cox regression analysis were male sex (hazard ratio [HR] 1.454, 95% confidence interval [CI] 1.127–1.876), age ≥ 65 years (HR 1.367; 95% CI 1.071–1.744 ), T2 stage (HR 1.283; 95% CI 1.005-1.638), tumor size > 3cm (HR 1.704; 95% CI 1.346-2.158), examined lymph nodes (EN) ≤ 15 (HR 1.327; 95% CI 1.058–1.664), and non-signet ring cell carcinoma (Non-SRCC) (HR 1.639; 95% CI 1.123–2.392). While only T2 stage (HR 1.410; 95% CI 1.026-1.936), tumor size > 3cm (HR 1.755; 95% CI 1.285-2.397), examined lymph nodes (EN) ≤ 15 (HR 1.489; 95% CI 1.101–2.015) were independent risk factors for cause-specific survival (CSS). We divided patients with pT2N0 into four sub-categories according to two significant prognostic factors (size and EN) and found that only patient in group 3 (EN ≤ 15, size >3 cm) with improved CSS benefit from AC (P = 0.049). More significant CSS benefit from AC was identified in Non-SRCC patients within group 3 (P = 0.034).

Conclusion: An additional survival benefit related to AC is expected for selected pT2N0 patients. Non-SRCC patients with EN ≤ 15 and tumor size > 3 cm may be particularly appropriate candidates for AC.

Part 2

Purpose: For advanced gastric cancer patients with pancreas head invasion, some studies suggested that extended multi-organ resections (EMR) improved the survival. However, other reports showed high rates of morbidity and mortality after EMR. EMR for T4b gastric cancer remains controversial.This study aims to evaluate the surgical approach for pT4b gastric cancer with pancreas head invasion.

Methods: A total of 143 gastric cancer patients with pancreas head invasion were surgically treated between 2006 and 2016 from China National Cancer Center. Of these patients, 76 confirmed by postoperative pathology were retrospectively analyzed. They were divided into the gastrectomy plus en-bloc pancreaticoduodenectomy group (GP group) and gastrectomy alone group (GA group) by comparing the clinicopathologic features, surgical outcomes, and prognostic factors of these patients.

Results: Of the 76 consecutive patients included, 24 patients (16.8%) underwent radical gastrectomy plus en-bloc pancreaticoduodenectomy（GP group) that had significantly larger lesions (P < 0.001), a higher incidence of advanced N stage (P = 0.030), and less neoadjuvant chemotherapy (P < 0.001) than the gastrectomy alone group (GA group).There was no significant difference in mortality between GP group and GA group (4.2% vs. 4.8%, P = 1.000), but the incidence of postoperative complications in GP group (62.5% vs. 11.5%, P < 0.001), especially the incidence of pancreatic fistula was significantly higher than that in GA group. The overall 3-year survival rate of the patients in the GP group was significantly longer than that in the GA group (47.6%, median 30.3 months vs 20.4 %, median 22.8 months, P = 0.010). Multivariate analysis identified neoadjuvant chemotherapy (hazard ratio [HR] 0.290; 95% confidence interval [CI] 0.103–0.821; P = 0.020), linitis plastic (HR 2.614; 95% CI 1.024–6.675; P = 0.033), Surgical margin (HR 0.274; 95% CI 0.102–0.738; P = 0.010), N stage (HR 3.489; 95% CI 1.334–9.120; P = 0.011), and postoperative chemoradiotherapy (HR 0.369; 95% CI 0.163–0.836; P = 0.017) as independent predictors of survival in patients with pT4b gastric cancer and pancreas head invasion.

Conclusion: For some patients with gastric cancer invading the head of pancreas, radical combined viscerectomy can improve the survival rate. Invasion of the head of the pancreas is not an absolute contraindication for surgery, but the incidence of surgical complications is increased. Therefore, combined viscerectomy needs to be carried out cautiously, and more studies are needed to explore the risks and benefits of surgery.

Part 3

Objective: Gastric cancer is highly heterogeneous. Gastric adenocarcinoma cells and sig-ring cells often exist in the same tumor in different proportions. This research attempts to different groups according to the content of signet ring cell, compare the prognosis of differences between groups, and further analysis of different levels of gastric signet ring cell carcinoma tumor mutation load difference, genetic variation characteristics and prognosis of relations, from the gene level to reveal the reason of the difference of prognosis, to explore the different contents of signet ring cell gastric cancer tissue in a wide variety of tumor infiltrating the density of immune cells and positive rate.

Methods: The clinical information of 27 gastric cancer patients with different proportions of signet-ring cell carcinoma was collected. The wax sections of postoperative specimens were collected and 520 genes highly related to the biological principle of tumor development and personalized treatment plan were captured by gene sequencing technology. Multiple immunofluorescence method was used to detect CD3+T, CD8+T,CD56+NK, CD68+CD163-M1, CD68+CD163+M2, PD1 and PDL1 positive cells density and positive rate in tumor parenchyma and stroma. According to different sig-ring cell content, the patients were divided into high sig-ring cell group (H-SRC, signet-ring cell content >50%) and low sig-ring cell group (L-SRC, signet-ring cell content <50%), and the clinicopathological characteristics, gene mutation and immune status of the patients were retrospectively analyzed.

Results: The tumor diameter of the H-SRC group was slightly larger than that of the L-SRC group (5.54 cm vs. 4.06 cm, P = 0.043), and there was no significant difference in other basic characteristics between the two groups. Survival analysis of the two groups showed that the prognosis of L-SRC group was poor, and the difference of survival in stage Ⅱ and Ⅲ patients was statistically significant (P = 0.045). There were no significant differences in gene mutation profile and expression of immune markers between the two groups. Cox univariate survival analysis showed that T stage, lymph node positive rate, nerve invasion and mutations of KRAS, ARID1A, MET and FANCA4 genes were correlated with prognosis, and the above 7 indicators were still statistically significant in Cox multivariate analysis, all of which were high risk factors for poor prognosis.

Conclusion: The prognosis of gastric cancer with low signet ring cell content is worse, and the difference is significant in stage Ⅱ and Ⅲ stage patients. T stage, lymph node positive rate, nerve invasion, KRAS, ARID1A, MET and FANCA gene mutations are high risk factors for poor prognosis. It is necessary to expand the sample size to clarify the accuracy of the study conclusions, and further explore the causes of poor prognosis, so as to provide basic theoretical guidance for clinical precision and individualized treatment.