Endometrial cancer (EC) is the sixth most commonly diagnosed cancer in women in the world and the most common gynecological malignancy in developed countries, with approximately 417,000 new cases and 97,000 deaths globally in 2020 [1,2]. About 70% of ECs are confined to the uterus at diagnosis and have a favorable prognosis, whereas patients with metastatic and high-risk histologic subtype have poor prognosis [3]. Platinum-based chemotherapy is the standard first-line treatment for advanced and recurrent EC, and for platinum-insensitive ECs with mismatch repair proficiency (pMMR), the combination of pembrolizumab and lenvatinib can be used as a second-line treatment option. However, for patients with metastatic advanced disease, the 5-year overall survival rate is still only 20% [4]. In recent years, the emergence of immune checkpoint inhibitors (ICIs) based on programmed death-1/programmed death ligand-1 (PD-1/PD-L1) pathway has kindled the light of hope for patients with EC. Several recent studies have demonstrated unprecedented durable response of ICIs in patients with EC. The NRG-GY018 study showed that pembrolizumab combined with chemotherapy reduced the risk of death in both pMMR and mismatch repair deficiency (dMMR) patients [5]. Similar results were confirmed in the RUBY trial, where dostarlimab combined with chemotherapy improved survival in patients with both microsatellite instability high (MSI-H)/dMMR and microsatellite stability (MSS)/pMMR [6].

Nevertheless, the factors influencing the efficacy of immunotherapy have not been fully understood. Predictive immunotherapy biomarkers that have gained regulatory approval include PD-L1 expression, MSI-H/dMMR and TMB-H [[7], [8], [9]]. However, the response to ICIs is not completely consistent with any of these existing biomarkers, which may occur in low PD-L1 expression population, but lack in patients with MSI-H and TMB-H [10,11], making it often necessary to detect all three biomarkers simultaneously in clinical practice to avoid missing any patients who might benefit from immunotherapy. On the other hand, the detection methods of these biomarkers are diverse and the threshold evaluation criteria are multifarious, which has not yet formed a unified gold standard. Moreover, the difficulty of obtaining tissue samples and the high cost of large panel or whole-exome sequencing (WES) limit the extensive application of these biomarkers in treatment selection. With the continuous deepening of research, some novel immunotherapy biomarkers have been proposed, such as aneuploidy score [12], tumor microenvironment (TME) [13,14] and POLE/POLD1 gene exonuclease domain mutation [15,16], but their clinical application value is still relatively limited. Taken together, there is a clear need to identify novel biomarker of response to immune checkpoint inhibitor therapy, especially if such a biomarker can combine the features of multiple existing biomarkers simultaneously and can overcome all of the above deficiencies.

Traditionally, EC can be divided into two groups according to the factors such as relationship to estrogen, histopathological and epidemiological characteristics [17]. Among them, type I is hormone-dependent, and the pathological type is mainly endometrioid carcinoma, with a favorable prognosis; type II is primarily hormone-independent serous carcinoma with poor prognosis. However, this classification method is greatly influenced by tumor heterogeneity and pathologist's subjective factors, and its prognostic value and guidance for individual standardized treatment are suboptimal. Nowadays, molecular classification of EC has been widely studied and applied in combination with pathological typing to guide postoperative adjuvant therapy and prognosis prediction [18]. By integrating genomic sequencing, copy number variation, microsatellite instability result and immunohistochemical technique, the most commonly used TransPORTEC classification identified four classes of EC: ultramutated ECs with POLE somatic proofreading domain mutation, MSI-H or dMMR subset, copy number low or no-specific molecular profile (NSMP) type, and copy number high or p53 mutated type [19]. The POLE-hypermutant group possesses the best prognosis due to the high mutation rate and consequent enhanced immunogenicity that can increase the ability to stimulate antitumor immunity [20]. The copy number high EC has the worst prognosis, while the prognosis of MSI-H subgroup is moderate [21,22]. Unfortunately, this molecular classification alone cannot explain the differential response to systemic therapy, and a combined exploration of the TME is needed to better identify new therapeutic targets and improve prognosis [23].

FAT3 and LRP1B gene mutations are common in a variety of solid tumors. FAT3 is cadherin superfamily members involved in tumor suppression and planar cell polarity (PCP) and belongs to the upstream regulator of the Hippo signaling [24]. LRP1B encodes a low-density lipoprotein (LDL) family receptor with broad roles in normal cell function and development, and negatively regulates Wnt signaling [25]. However, there are still relatively few studies on these two tumor suppressor genes at present, and most of the several previous studies only confirmed the predictive value of single gene mutation of these two genes in terms of prognosis and induction of host immune activation [[26], [27], [28]]. Our study in non-small cell lung cancer (NSCLC) have illustrated that FAT3 and LRP1B co-mutation defines a unique subset and is associated with immunotherapy efficacy [29]. Given that the effects and roles of FAT3 and LRP1B in EC are still unclear, and their mutations are relatively common, we hypothesized that these two genes may also be worth further exploration in EC. In this study, we sought to speculate the prognosis and the predictive value for precision medicine treatment selection of FAT3 and LRP1B co-mutation in endometrial cancer by assessing the multiple immunotherapy biomarkers and tumor microenvironment characterization. A prognostic model was constructed and a unique subset with better prognosis was defined for co-mutation samples in MSI-H subtype.