The burden of esophageal squamous cell carcinoma (ESCC) in China is substantial, with 85% of the cancers being in the progressive stage. The treatment for advanced ESCC are extremely limited, and immunotherapy, represented by PD-1 inhibitors, has demonstrated a promising application potential. However, the effectiveness of PD-1 inhibitors varies significantly among patients with different types of ESCC, and currently, there is no effective method to predict the response to PD-1 inhibitors. In this study, investigators aim to construct a multimodal deep learning-based model to predict the level of immune infiltration and the efficacy of immunotherapy for ESCC, integrating both pathological image features and clinical information of patients with ESCC, thereby enhancing the level of individualized and precise treatment for ESCC.

Inclusion Criteria:

1. Availability of Hematoxylin and Eosin (H&E) stained images, molecular data obtained through RNA sequencing (RNA-Seq, DNA-seq), and comprehensive clinical information including patient age, gender, history of alcohol consumption, history of smoking, AJCC Tumor, Node, Metastasis Stage, specific location of oesophageal cancer occurrence, and history of reflux.
2. The sample collection is restricted to cancerous tissue, encompassing both primary tumor samples and those from metastatic sites.

Exclusion Criteria:

1. Patients diagnosed with adenosquamous carcinoma or presenting with a combination of other types of oesophageal cancers;
2. Cases involving combined adenocarcinoma affecting the gastroesophageal junction;
3. Individuals with high-grade tumors that have not penetrated the basement membrane, as confirmed by postoperative pathological examination;
4. Subjects in whom postoperative pathology confirms an absence of residual malignant tissue.

110 patients with ESCC in the Changhai hospital

A planned initiative that includes sharing DNA and RNA sequencing data, alongside the provision of calculating code.