Overview

This study aims to initially utilize machine learning on pan-cancer DNA methylation data from public databases to construct a DNA methylation classification model (PaCIFiC-CUP, pan-cancer integrated fingerprinting classifier of CUP) for diagnosing various types of cancer, particularly the primary site of cancer of unknown primary. The goal is to achieve diagnosis of cancer pathology type by analyzing the DNA methylation patterns of cancer specimens, thereby guiding subsequent precision treatment for cancer.

Eligibility

Inclusion Criteria:

1. The patient specimens were obtained from the Sun Yat-sen University Cancer Center and affiliated cooperating centers, with written consent from the patients authorizing the use of the specimens for research purposes.
2. Following standard assessments (medical history, physical examination, complete blood count, biochemistry, computed tomography scans of the neck, chest, abdomen, and pelvis, targeted evaluations of all symptomatic areas, pathology, and immunohistochemistry), the diagnosis was determined as a primary site unknown tumor (including adenocarcinoma, squamous cell carcinoma, undifferentiated carcinoma, neuroendocrine carcinoma, sarcoma, etc).
3. The diagnosis was confirmed at the participating institution and the patient had received systemic therapy.
4. Complete clinical, pathological, and follow-up data for the patients can be obtained.
5. ECOG performance status score: 0-2 points.

Exclusion Criteria:

1. Pregnant or lactating female patients.
2. Tumor tissue sample size is too small (tumor tissue accounts for <70% in the biopsy or slice tissue).
3. Organ transplant or history of non-autologous (allogeneic) bone marrow or stem cell transplantation.
4. History of previous tumors, with the current condition being a recurrent tumor.
5. Hematological malignancies (excluding lymphoma).
6. Other diseases that may severely impact patient survival, such as severe cardiovascular or cerebrovascular diseases, sepsis, severe trauma or burns, etc.