第一部分 子宫内膜癌分子分型相关临床病理及分子特征研究

背景：

子宫内膜癌（Endometrial carcinoma，EC）是妇科常见的恶性肿瘤。随着第二代测序（The next generation sequencing，NGS）技术的发展，我们对EC的分子机制有了更深入的认识，精确的分子分型对患者的预后评估及辅助治疗选择十分重要。通过检测DNA聚合酶ε（DNA polymerase epsilon，POLE）基因突变状态、错配修复（Mismatch repair，MMR）蛋白表达/微卫星不稳定（Microsatellite Instability，MSI）状态及TP53基因突变/p53蛋白表达状态，可将EC分为四个分子亚型：POLE基因超突变（POLE ultramutated，POLE mut）型、错配修复缺陷（Mismatch repair deficiency，MMRd）型、p53异常（p53 abnormality，p53 abn）型及无特殊分子改变（Non-specific molecular profile，NSMP）型。目前，国内关于分子分型的临床应用尚处于起步阶段，检测方法亦有待规范。本研究基于对我们中心的EC病例进行分子分型相关检测，对所有病例进行分子分型，探讨分型过程中遇到的问题，并且探究各分子亚型的临床病理及预后特征，以期为EC分子分型的实践提供经验。

方法：

收集2012-2024年间在北京协和医院行EC分期手术的患者及在外院行EC分期手术的会诊患者共320例，收集他们的术后病理切片及福尔马林固定石蜡包埋（formalin-fixed and paraffin-embedded，FFPE）组织块。完善MMR蛋白（MLH1、PMS2、MSH2、MSH6）及p53蛋白免疫组织化学（Immunohistochemistry，IHC）染色，提取石蜡样本DNA行POLE基因和TP53基因突变及MSI状态检测。行EC分子分型，收集临床、病理及预后相关信息。最后，对各分子亚型EC的临床病理特征及预后作出分析。

结果：

320例EC患者中，POLE基因致病性突变共39例（12.2%）。5种“热点突变”占所有致病性突变病例的87.2%，其中频率最高的突变为P286R（46.2%）。

9例（2.8%）存在异质性MMR蛋白表达（Heterogeneous MMR expression，MMR-het）现象，涉及的分子机制包括MLH1启动子高甲基化、POLE基因致病性突变及MMR基因胚系/体系突变，其中多重分子特征（multiple-classifier）的病例占66.7%；7例MMR IHC与MSI NGS结果不一致，一致率为97.8%。其中2例错配修复正常（proficient mismatch repair，pMMR）/微卫星高度不稳定型（microsatellite instability-high，MSI-H）均为POLE mut型；41例p53 IHC和TP53结果不一致，分子分型为：MMRd型22例、POLE mut型14例、p53 abn型3例及NSMP型2例，其中multiple-classifier的病例占73.2%。

POLE mut型39例（12.2%）、MMRd型100例（31.3%）、p53 abn型42例（13.1%）和NSMP型139例（43.4%）。组间分析示四个亚组的发病年龄、FIGO分期、EEC分化程度及是否辅助治疗差异具有统计学意义（P＜0.05）。生存分析示POLE mut型预后最好，p53 abn型预后最差。

Multiple-classifier为40例（12.5%），包括22例MMRd + p53 abn，11例POLE mut + p53 abn，5例POLE mut + MMRd和2例POLE mut + MMRd + p53 abn。p53 IHC最常见的模式为亚克隆表达（subclonal expression，SE）（51.5%）；发生多个TP53突变（＞1个突变）的事件数明显多于单纯p53 abn型（P＜0.001）。

结论：

综上，本研究验证了EC各分子亚型独特的临床病理特征及预后意义。并且，本研究强调在分子分型过程中判读顺序的重要性、检测内容的全面性以及不同检测方法的必要性。一方面，multiple-classifier的病例并不罕见，仅检测一种分子指标不足以确保分子分型的准确性。另一方面，无论是MMR蛋白/MSI状态之间还是p53蛋白表达/TP53基因突变结果之间，均存在不一致性的情况，并且绝大多数为multiple-classifier的病例。

第二部分 子宫内膜癌错配修复蛋白异质性表达模式及分子机制研究

背景：

子宫内膜癌（Endometrial carcinoma，EC）中错配修复缺陷（Mismatch repair deficiency，dMMR）的发生率为20%-40%。dMMR检测被推荐用于所有新诊断的EC患者，它不仅有助于识别林奇综合征患者，并且可用于指导EC的分子分型、预后分析和治疗策略。免疫组织化学（Immunohistochemistry，IHC）常规用于检测实体瘤中的dMMR，其有着经济、快速且判读较为简便的优点，已在临床实践中广泛开展。但在MMR IHC的广泛应用过程中出现了多种表型并在解读上存在挑战，其中异质性MMR蛋白表达（Heterogeneous MMR expression，MMR-het）便是其中一种，其被定义为MMR蛋白染色完整的肿瘤区域与染色缺失的肿瘤区域同时存在且界限分明。不同的MMR-het模式和各自的分子机制可能会对后续的检测策略和免疫治疗具有意义，但是目前仅偶有报道但尚未被系统性地研究。本研究选取了具有MMR-het的EC作为研究对象，探索MMR-het的不同染色模式及其背后的分子机制，以期加深对这一特殊MMR IHC表型的认识及合理判读，并为其分子检测方法的选择及分子分型实践提供指导。

方法：

复阅2020年1月至2022年12月期间在北京协和医院病理科诊断为EC且有完整4个MMR蛋白染色患者的石蜡样本，筛选出具有MMR-het的样本25例。总结MMR-het的染色模式，并通过全面的基因组测序（Comprehensive genomic profiling，CGP）和MLH1启动子甲基化分析检测其分子改变和肿瘤突变负荷（Tumor mutation burden，TMB），分析并总结MMR-het的不同染色模式及其相关的分子机制。

结果：

根据染色丢失（完全丢失或异质性表达）的MMR蛋白的数量，MMR-het染色模式被分为四个亚组：“单丢失”（3例）、“MLH1/PMS2 双丢失”（14例）、“MSH2/MSH6 双丢失”（6例）和“三/四个丢失”（2例）。EC大多呈现“MLH1/PMS2双丢失”表型。

10例MMR-het 病例表现出组织学分化异质性，即MMR蛋白丢失通常局限于低分化或高分化的肿瘤区域。

所有“单丢失”病例均具有MMR基因体系突变和POLE核酸外切酶结构域突变（POLE exonuclease domain mutations，POLE-EDM）。“MLH1/PMS2 双丢失”肿瘤均存在MLH1启动子高甲基化（MLH1 promoter hypermethylation，MLH1-hm），且均无MMR胚系突变。在“MSH2/MSH6双丢失”亚组中，3例存在MSH2/MSH6胚系突变，另外2例存在多种MSH2/MSH6体系突变，有2例同时伴有POLE-EDM。“三/四个丢失”亚组均存在MLH1-hm，并且同时伴有MSH2/MSH6体系或胚系突变。

18例（18/23，78.3%）MMR-het EC表现为为TMB-H，所有POLE突变病例均表现出超高TMB（111.4至524.2 mut/Mb）。

结论：

我们的研究结果强调了EC中MMR-het的不同染色模式均存在MMR基因相关的分子机制。准确解释MMR-het染色模式有助于指导更有效的分子检测和临床治疗策略。

Abstract

Part One: Analysis of Clinicopathological and Molecular Characteristics Related to Molecular Classification of Endometrial Carcinomas

Background:

Endometrial Carcinoma (EC) is common malignant tumors of the female reproductive system. With the development of the next generation sequencing technology (NGS), we have a deeper understanding of the molecular mechanism of EC. Accurate molecular classification is very important for prognosis assessment and adjuvant treatment selection. EC can be divided into four molecular subtypes, namely POLE gene hypermutated (POLE mut), mismatch repair deficiency (MMRd), p53 abnormality (p53 abn) and non-specific molecular profile (NSMP), by detecting DNA polymerase epsilon (POLE) gene mutation status, mismatch repair (MMR) protein expression/Microsatellite Instability (MSI) status and TP53 gene mutation/p53 protein expression status. At present, the clinical application of EC molecular classification in China is still in its infancy, and detection methods still need to be standardized. The purpose of this study is to investigate the clinicopathological and prognostic features of EC molecular subtypes and explore the problems related to the molecular classification, and to provide practical experience for clinical practice.

Methods:

In this study, a total of 320 patients who underwent EC staging surgery at Peking Union Medical College Hospital (PUMCH) and consultation patients who underwent EC staging surgery in other hospitals from 2012 to 2024 were collected. Postoperative pathological sections and formalin-fixed and paraffin-embedded (FFPE) specimens of these patients were collected. The immunohistochemical (IHC) staining of MMR proteins (MLH1, PMS2, MSH2, MSH6) and p53 protein of each patient was completed, and DNA from paraffin samples was extracted for POLE gene and TP53 gene mutations and MSI status detection. Molecular classification was performed and the clinical pathological information and prognostic information of the patients was collected. Finally, the clinicopathological characteristics and prognosis of each molecular subtype of EC were analyzed.

Results：

(1) Among the 320 ECs in this study, a total of 39 cases (12.2%) of pathogenic mutations were detected, including five "hotspot mutations" (87.2%), among which the most frequent mutation was P286R (46.2%).

(2) Nine patients (2.8%) had completely heterogeneous expression of MMR proteins. The molecular mechanisms involved include MLH1 promoter hypermethylation, POLE gene pathogenic mutations, and MMR gene germline/somatic mutations. Multiple-classifier cases accounted for 66.7%; Seven patients had inconsistent MMR IHC and MSI NGS detection results, with a concordance rate of 97.8%. Among them, 2 cases of pMMR/MSI-H were all of POLE mut subtype. Forty-one patients had inconsistent p53 IHC and TP53 mutation detection results. The molecular classification was as follows: 22 cases of MMRd subtype, 14 cases of POLE mut subtype, 3 cases of p53 abn subtype, and 2 cases of NSMP subtype. Multiple-classifier cases accounted for 73.2% of this cases.

(3) The results of molecular classification were 39 cases (12.2%) of POLE mut subtype, 100 cases (31.3%) of MMRd subtype, 42 cases (13.1%) of p53 abn subtype and 139 cases (43.4%) of NSMP subtype. Between-group analysis showed statistically significant differences between the four subgroups in age at onset, FIGO stage, degree of EEC differentiation, and adjuvant therapy (P < 0.05). Survival analysis suggested that the POLE mut type had the best prognosis aand the p53 abn type had the worst prognosis.

(4) Multiple-classifier features were found in 40 cases (12.5%), including 22 cases of MMRd + p53 abn, 11 cases of POLE mut+p53 abn, 5 cases of POLE mut+MMRd, and 2 cases of POLE mut+MMRd+p53 abn. The most common expression pattern of p53 IHC was subclonal expression (51.5%); and the number of events in which multiple TP53 gene mutations (>1 mutation) occurred was significantly higher than that of the p53 abn subtypes ( P < 0.001).

Conclusions：

In summary, this study verified the unique clinicopathological characteristics and prognostic significance of each molecular subtype. Moreover, it is also emphasized the importance of interpretation sequence, comprehensiveness of detection content and the necessity of different detection methods in the process of molecular classification. On the one hand, multiple-classifier cases are not rare, and detecting only one indicator is not enough to ensure the accuracy of molecular classification. On the other hand, there are inconsistencies between MMR protein/MSI status or p53 protein expression/TP53 gene mutation results, and the vast majority are multiple-classifier cases.

Part Two: Heterogeneous Expression Patterns and Molecular Mechanisms of Mismatch Repair Proteins in Endometrial Carcinomas

Background: 

The incidence of mismatch repair deficiency (dMMR) in endometrial carcinoma (EC)  is 20%-40%. dMMR analysis is increasingly recommended for all newly diagnosed endometrial carcinomas, as it identifies Lynch syndrome patients, and is emerging as a prognostic classifier to guide molecular classification, prognostic analysis and adjuvant strategy. Immunohistochemistry (IHC) is routinely performed to detect dMMR in solid tumors and it has been widely adopted in clinical practice due to its advantages of being economical, fast and relatively easy to interpret. However, during the widespread application of MMR IHC, a variety of IHC phenotypes have emerged and face interpretation challenges. Heterogeneous MMR expression (MMR-het) is one of them, which is characterized by tumor areas of intact staining with adjacent sharply demarcated areas of abrupt loss of staining. However, different MMR-het patterns and respective genetic mechanisms, which may have significance of subsequent testing strategy and immunotherapy, have been reported occasionally but not systemically studied yet. Our study focused on dMMR ECs showing heterogeneous MMR expression, aimed to uncover the diversity of MMR-het patterns and molecular mechanisms of this rare yet distinct subset, which may guide MMR IHC interpretation and have important implications for molecular detection decision making. 

Methods: 

All paraffin-embedded specimens from patients with EC and with 4 MMR protein staining slides diagnosed by the pathology department of PUMCH from 2020.1 to 2022.12 were reviewed. Twenty-five samples with MMR-het were screened. The staining patterns of MMR-het in these samples were summarized. We further analyzed MMR genetic alternations and tumor mutational burdens (TMB) through comprehensive genomic profiling and MLH1 promoter hypermethylation analysis to summarize the molecular mechanisms related to staining patterns.

Results: 

According to deficient MMR proteins (either complete loss or heterogeneous loss), the MMR-het patterns were classified into four subgroups, namely “single-loss” (3 cases), “MLH1/PMS2 double-loss” (14 cases), “MSH2/MSH6 double-loss” (6 cases), and “triple/tetra-loss” (2 cases). Most ECs show the “MLH1/PMS2 double loss” phenotype. 

Ten MMR-het cases exhibited histological heterogeneity, in which MMR protein loss was generally confined to either poorly differentiated or well-differentiated tumor areas. 

All “single-loss” ECs had MMR somatic mutations and coexisting POLE exonuclease domain mutations (POLE-EDM). “MLH1/PMS2 double-loss” tumors unexceptionally harbored MLH1 promoter hypermethylation (MLH1-hm) without MMR germline mutations. In the “MSH2/MSH6 double-loss” subgroup, three cases had MSH2/MSH6 germline mutations, whilst another two cases had multiple MSH2/MSH6 somatic mutations. Additional POLE-EDM were identified in two cases. All two tumors in the “triple/tetra-loss” subgroup generally had MLH1 abnormalities, and coexistent somatic or germline mutations on MSH2/MSH6. 

Eighteen cases (18/23, 78.3%) were TMB-H, and all POLE-mutated cases exhibited ultra-high TMB (111.4 to 524.2 mut/Mb). 

Conclusions: 

Our findings highlighted the existence of MMR gene-related molecular mechanisms of different MMR-het staining patterns in endometrial carcinoma. Accurate interpretation of heterogeneous MMR protein staining patterns could help guide more effective molecular testing and clinical treatment strategy.