For most cancers, the path to understanding begins with a biopsy, a clinical trial, and a cohort of patients who happen to live near a major medical center. But for a cancer as rare as angiosarcoma, a malignant tumor of the blood vessel walls that strikes roughly one person per million each year, that path has never been viable. No single hospital sees enough cases. No traditional trial can enroll enough participants. For decades, angiosarcoma has remained one of oncology’s most poorly understood malignancies, its molecular drivers hidden behind the twin barriers of scarcity and heterogeneity.
A study published July 24 in Nature Communications has changed that. Led by researchers at the Broad Institute of MIT and Harvard, in collaboration with the Angiosarcoma Project, the largest comprehensive multiomics analysis of angiosarcoma ever conducted reveals not just the cancer’s molecular subtypes, but something arguably more important: that a patient-partnered research model, where patients drive recruitment, sample contribution, and study design, can succeed where conventional infrastructure has failed.
Two Cancers, One Name
The study, led by Hoyin Chu, Marissa Hollyer, Brittany Borden, and senior authors Corrie Painter and Saud AlDubayan, analyzed tumor transcriptomes, somatic mutations, and germline DNA from 254 angiosarcoma patients recruited entirely through the Angiosarcoma Project, a distributed research network that empowers patients to enroll from anywhere in the world. The result is a molecular atlas that definitively separates angiosarcoma into two biologically distinct diseases.
Subcutaneous angiosarcomas, tumors arising in the deeper soft tissues, show consistent upregulation of TGF-beta and receptor tyrosine kinase (RTK) signaling pathways, with recurrent driver mutations in the genes KDR (which encodes VEGFR2, a key blood vessel growth receptor), PLCG1, and POT1. These tumors carry molecular signatures that align more closely with other soft tissue sarcomas than with their superficially occurring counterparts.
Cutaneous angiosarcomas, those forming in the skin, tell a completely different story. They are driven by MYC-centered transcriptional programs, bear the telltale mutational fingerprints of ultraviolet light damage, and express high levels of immune checkpoint genes, including PD-L1 and CTLA-4. Their mutation landscape features recurring alterations in TP53, FLT4, and BRAF.
The two subtypes have distinct cell-of-origin signatures, distinct microenvironment compositions, and, crucially, distinct therapeutic vulnerabilities. The study suggests that subcutaneous angiosarcomas may be targetable through TGF-beta and RTK inhibitors, while the immune-enriched cutaneous subtype may respond to immune checkpoint blockade, a hypothesis that aligns with emerging clinical observations but has never before had molecular evidence of this scale to support it.
The POT1 Signal
The most striking finding to emerge from the germline analysis concerns POT1, a gene whose protein product protects chromosome ends from being recognized as DNA breaks. A growing body of evidence has linked POT1 loss-of-function variants to familial melanoma, glioma, and cardiac angiosarcoma. The new study puts the risk in stark numbers: carriers of germline POT1 pathogenic variants face a 92.7-fold increased risk of developing angiosarcoma compared with non-carriers.
Even more telling is the double-hit phenomenon the researchers identified. Patients who carried both a germline POT1 variant and a second, somatic POT1 mutation in their tumor developed disease decades earlier than those with only somatic alterations. The tumors from these patients also displayed a distinct molecular fingerprint: underexpression of TERT, the telomerase catalytic subunit, and overexpression of CHAMP1, a chromosome alignment protein. This combination suggests an alternative telomere maintenance mechanism that could itself be a therapeutic target.
The finding has immediate clinical relevance. The researchers propose that germline POT1 testing should be considered for patients with angiosarcoma, especially those diagnosed at unusually young ages or with a family history suggestive of cancer susceptibility.
Why the Model Matters
The biological discoveries are significant. But the study’s most important contribution may be methodological.
Angiosarcoma is so rare that the United States National Cancer Institute estimates only about 1,000 new cases per year nationwide. Traditional clinical trials, which require centralized enrollment, onsite tissue collection, and standardized protocols at academic medical centers, have never been able to assemble a cohort large enough to power meaningful multiomics analysis. Prior genomic studies of angiosarcoma typically analyzed fewer than 50 specimens, often pooled across multiple institutions with inconsistent collection methods.
The Angiosarcoma Project, part of the Count Me In network based at the Broad Institute, takes a fundamentally different approach. Patients enroll online, consent electronically, and mail in saliva samples for germline DNA analysis. Tumor tissue is retrieved from the hospital where the patient received care, regardless of that hospital’s affiliation with the research team. Clinical data are abstracted from medical records by a centralized team. The entire model is built around patient convenience and agency, not institutional convenience.
Of the 254 patients in the study, the vast majority would never have been within driving distance of a single angiosarcoma research center. Through the patient-partnered model, they became part of the largest molecular characterization of their disease ever performed.
A Template for Rare Disease Research
The implications extend well beyond angiosarcoma. There are roughly 7,000 known rare diseases, most of them cancers or cancer predisposition syndromes. The vast majority lack even the rudimentary molecular characterization that angiosarcoma has now received. For each of them, the central obstacle is the same: too few patients, too widely dispersed, for traditional research infrastructure to capture.
The Angiosarcoma Project demonstrates that direct-to-patient recruitment, paired with centralized multiomics analysis, can overcome that obstacle. When patients are treated as partners rather than subjects, and when the logistical barriers to participation are removed, the sample sizes needed for statistical power become achievable. The model produced a cohort 5 to 10 times larger than any previous angiosarcoma molecular study, and that scale is what made the subtype discovery possible.
The study was funded by the Department of Defense, the National Institutes of Health, the Edward P. Evans Foundation, the American Society of Clinical Oncology, the Rally Foundation for Childhood Cancer Research, and others. The authors reported that funders had no role in study design, data interpretation, or manuscript preparation.
What Comes Next
With the molecular architecture of angiosarcoma now mapped, the field faces a new set of challenges. The subtype-specific therapeutic hypotheses need clinical validation. The POT1 double-hit mechanism needs mechanistic dissection. And the patient-partnered model itself needs to be scaled to other rare cancers where it could yield similar breakthroughs.
The researchers have made all genomic data publicly available through the Angiosarcoma Project’s data portal, ensuring that the cohort will continue to yield insights long after the current study is published. For the patients who contributed their samples and their stories to the effort, that open-access commitment is perhaps the most fitting outcome of a study that began not with a hypothesis in a laboratory, but with a decision to let patients lead the way.
Reference: Chu, Hollyer, Borden et al. Patient-partnered multiomics reveals the molecular architecture of angiosarcoma. Nature Communications (2026). DOI: 10.1038/s41467-026-75810-2. Open Access (CC BY-NC-ND 4.0).