Proteomic Map of Cervical Cancer Reveals Subtypes and a New Drug Target

Cervical cancer has long been treated as a single disease with a few histological variations, but a large-scale proteomic study published in Genome Medicine is challenging that assumption. A team led by researchers at Zhejiang Cancer Hospital and the Hangzhou Institute of Medicine, Chinese Academy of Sciences, analyzed 404 cervical cancer tissue samples using deep mass spectrometry-based proteomics, producing one of the most detailed molecular portraits of the disease to date. Their findings, published on 5 October 2026, reveal that the major histological types of cervical cancer carry sharply distinct protein signatures, and that squamous cell carcinoma, the most common form, can be divided into three molecular subtypes with markedly different immune landscapes and patient outcomes. The work also pinpoints a protein called CCM2 as a functional driver of invasion in cervical adenocarcinoma, offering a candidate therapeutic target for a tumor type that has historically lagged behind in precision oncology.

The scale of the study is what sets it apart. The cohort comprised 298 squamous cell carcinomas, 74 adenocarcinomas, 22 adenosquamous carcinomas, and 10 neuroendocrine carcinomas, allowing the researchers to compare the four major histological categories within a single analytical framework. Cervical cancer is a leading malignancy in women worldwide, and while human papillomavirus vaccination and screening have reduced incidence in many countries, patients who do develop the disease face highly variable prognoses and treatment responses. That variability has been partly explained by histology: squamous cell carcinoma, adenocarcinoma, adenosquamous carcinoma, and neuroendocrine carcinoma differ in behavior and outcome. But the molecular underpinnings of these differences, particularly for the rare adenosquamous and neuroendocrine forms, have remained poorly characterized, and no dedicated molecular classification existed for squamous tumors.

To build their molecular map, the researchers used data-independent acquisition mass spectrometry, a technique that enables deep, quantitative profiling of thousands of proteins across large sample sets. They complemented the proteomic measurements with pathway enrichment analyses, tumor microenvironment deconvolution, survival analyses, and machine-learning approaches, including non-negative matrix factorization for subtype discovery and random forest models for classification. Quality control procedures and orthogonal statistical methods such as orthogonal partial least squares-discriminant analysis helped ensure that the differences they detected reflected genuine biology rather than batch effects or technical noise. Informed consent was obtained from all participants, and the study was approved by the Ethics Committee of Zhejiang Cancer Hospital under reference IRB-2024-464.

The first major finding concerns the divide between squamous cell carcinoma and adenocarcinoma, the two most common histotypes. Comparative proteomics revealed distinct signatures separating the two, with adenocarcinomas characterized by activation of epithelial-mesenchymal transition, the developmental program that allows epithelial cells to acquire migratory and invasive properties, and by heightened activity of glycoprotein metabolic pathways. Epithelial-mesenchymal transition is a well-known hallmark of aggressive carcinomas, associated with metastatic spread and therapy resistance, so its proteomic activation in adenocarcinoma fits the clinical observation that these tumors can behave more aggressively than their squamous counterparts. The glycoprotein metabolism finding suggests that adenocarcinoma cells rewire their protein glycosylation machinery, a change that could potentially be exploited for imaging, biomarker development, or targeted therapy.

Within the adenocarcinoma signature, one protein stood out: CCM2, or cerebral cavernous malformations 2, a gene best known for its role in vascular development and endothelial cell junctions. The team functionally validated CCM2 as a candidate mediator of adenocarcinoma cell invasion and migration, meaning that manipulating the protein in laboratory models altered the ability of cancer cells to move and invade. This is a critical step beyond correlation. Many proteomic studies identify proteins that are merely associated with aggressive disease; demonstrating a functional role elevates CCM2 to a plausible therapeutic target. Because CCM2 participates in cell-cell adhesion and cytoskeletal organization, its dysregulation could provide adenocarcinoma cells with the mechanical flexibility needed to break through tissue boundaries, a hypothesis the authors suggest warrants further investigation as a route to precision treatment for this histotype.

The second major contribution is a new molecular classification of squamous cell carcinoma, which the researchers named cervical squamous molecular subtypes, or CSMSs. Using unsupervised clustering of proteomic data, they identified three groups. CSMS1 is described as immunosuppressive, characterized by a tumor microenvironment that appears hostile to immune attack. CSMS2 is immune-enriched, suggesting tumors that harbor robust immune cell infiltration and may respond well to immunotherapy. CSMS3 is defined by oxidative stress response, indicating tumors that rely heavily on antioxidant defenses to survive the reactive oxygen species generated by their own metabolism and by treatment. Each subtype presumably reflects distinct dependencies that could be targeted with different therapeutic strategies, from immune checkpoint inhibitors for CSMS2 to redox-modulating agents for CSMS3.

Crucially, the classification was not merely descriptive. When the researchers applied it to an independent external cohort, the three subtypes significantly stratified patient survival, with a log-rank p value of 0.022, providing statistical confidence that the molecular grouping carries genuine prognostic weight beyond the discovery dataset. The team then went further, combining molecular subtypes with clinical characteristics in what they call a clinical-molecular stratification. This analysis showed that the CSMS1 subtype, the immunosuppressive group, had the poorest prognosis in specific patient subgroups: those older than 45, those with a body mass index below 25 kilograms per square meter, those with moderately differentiated tumors, and those with FIGO stage III or IV disease. Notably, the prognostic disadvantage of CSMS1 was especially pronounced among patients receiving non-surgical treatment, suggesting that molecular typing could help identify which patients with advanced disease need intensified or alternative therapeutic approaches.

The implications for clinical practice are potentially significant. Current treatment decisions in cervical cancer rest primarily on staging, histology, and a handful of clinical variables, but two patients with the same stage and histotype can experience dramatically different outcomes. By integrating the CSMS classification with conventional clinical parameters, oncologists could in principle identify high-risk patients earlier and tailor follow-up intensity and treatment selection accordingly. For patients with immune-enriched CSMS2 tumors, immunotherapy might be prioritized, while the immunosuppressive CSMS1 group might require strategies to overcome a hostile microenvironment, such as combination approaches or trials of agents that remodel the tumor immune landscape. The authors argue that this clinical plus molecular framework may improve the identification of high-risk patients and facilitate precision management of the disease.

The study also sheds light on the rare histotypes. With 22 adenosquamous and 10 neuroendocrine carcinoma cases, the dataset provides some of the first large-scale proteomic information on these uncommon tumors, which have been difficult to study because few institutions accumulate enough cases for meaningful molecular analysis. Neuroendocrine carcinoma of the cervix in particular is an aggressive entity often managed by extrapolation from small cell lung cancer, and molecular data confirming or refuting those parallels could refine treatment. While the numbers for these rare subtypes remain modest and the findings there are more exploratory, their inclusion reflects a growing recognition that cervical cancer is not one disease but a family of molecularly distinct malignancies united by a common viral origin.

Limitations and next steps remain. Proteomics captures the functional layer of biology, the proteins that actually execute cellular behavior, but translating subtype assignments into routine clinical tests will require simpler, more scalable assays, perhaps targeting a small panel of marker proteins rather than full proteomic profiling. The functional validation of CCM2, while compelling, now needs to be extended with target-development work to determine whether the protein can be pharmacologically modulated. Independent validation in additional cohorts and across diverse populations will also be essential before the CSMS system enters clinical use. Nevertheless, the study marks a milestone: by deconstructing cervical cancer heterogeneity at the protein level across more than 400 tumors, it converts a histologically defined disease into a molecularly stratified one, and in doing so hands researchers both a map and a set of destinations, from CCM2-directed therapies for adenocarcinoma to immune-informed treatment algorithms for squamous carcinoma, that could shape the next generation of precision care for patients worldwide.

Subject of Research: Proteomic characterization of cervical cancer histological and molecular heterogeneity

Article Title: Deconstructing cervical cancer heterogeneity: squamous cell carcinoma subtypes and CCM2 vulnerability in adenocarcinoma

Article References: Ye, H., Feng, Y., Wang, Z., Pang, G., Ma, L., Lv, X., Zhang, X., Tian, J., Tang, J., Zhu, W., Lou, H., & Zhang, X. (2026). Deconstructing cervical cancer heterogeneity: squamous cell carcinoma subtypes and CCM2 vulnerability in adenocarcinoma. Genome Medicine. https://doi.org/10.1186/s13073-026-01784-x

Image Credits: AI Generated

DOI: 10.1186/s13073-026-01784-x

Keywords: cervical cancer, proteomics, mass spectrometry, squamous cell carcinoma, adenocarcinoma, CCM2, molecular subtypes, tumor microenvironment, epithelial-mesenchymal transition, precision medicine, biomarkers, Genome Medicine