A Molecular Atlas of Endometrial Cancer, Built From 95 Tumors and Seven Layers of Data

The CPTAC discovery cohort profiles uterine cancer patients with unprecedented multi-omics depth, revealing how obesity, tumor grade, and genomic subtypes intersect in a disease that kills more than 13,000 American women each year.

Every year, roughly 66,000 women in the United States are diagnosed with cancer of the uterine lining — endometrial carcinoma — and the incidence is climbing, driven in part by rising obesity rates. Most cases are caught early, when abnormal bleeding sends a postmenopausal woman to her doctor, and the prognosis is generally favorable. But a stubborn minority of tumors are aggressive from the start: high-grade, prone to spread, resistant to standard treatment. Telling those tumors apart from their indolent counterparts — not just by how they look under a microscope, but by how their proteins behave, which genes are mutated, and which chemical switches are flipped — is the central promise of molecular oncology.

The Clinical Proteomic Tumor Analysis Consortium (CPTAC) set out to do exactly that. Its uterine cancer discovery cohort assembled approximately 95 treatment-naive patients, predominantly with endometrioid histology and spanning FIGO stages I through IV. Each tumor was interrogated with whole-exome sequencing, copy-number profiling, RNA sequencing, microRNA and circular RNA quantification, genome-wide methylation arrays, and deep quantitative proteomics — including phosphorylation and acetylation maps. The result is one of the most comprehensive molecular portraits of endometrial cancer ever compiled. What follows is a guided walk through that portrait, from the women themselves to the mutations etched into their genomes.

Who Is in the Cohort

The first question about any cancer cohort is whether it looks like the disease it claims to represent. Endometrial carcinoma is overwhelmingly a disease of postmenopausal women; the median age at diagnosis in national registries sits in the early sixties. The CPTAC cohort mirrors that pattern. The age distribution peaks squarely in the 55-to-70 range, consistent with epidemiological expectations. A modest left tail extends below 50 — clinically significant, because younger patients are disproportionately likely to carry Lynch syndrome, an inherited defect in DNA mismatch-repair genes such as MLH1, MSH6, and PMS2. Those patients warrant cross-referencing with the mismatch-repair immunohistochemistry data collected alongside the multi-omics measurements.

The Obesity Connection

Obesity is the single strongest modifiable risk factor for endometrial cancer. Excess adipose tissue drives higher circulating estrogen levels — the hormonal fuel for most endometrioid tumors — and activates insulin-signaling cascades that feed directly into cancer-promoting pathways like PI3K/AKT/mTOR. The BMI distribution in this cohort reflects that biology plainly: it is right-skewed, with a mode in the obese range (BMI ≥ 30) and a long tail stretching past 40 into morbid obesity. Those extreme-BMI patients are not just statistical curiosities. Their proteomic and phosphoproteomic profiles may carry distinct metabolic signatures — altered insulin-receptor signaling, shifted adipokine levels — that could confound or illuminate downstream analyses.

When BMI is plotted against age at diagnosis, a subtle but expected pattern emerges. Younger patients tend to carry higher BMIs, consistent with the model that obesity-driven estrogen excess triggers type I (endometrioid) disease earlier. Older patients, by contrast, span a wider BMI range, reflecting the admixture of lean women with aggressive serous-like (type II) tumors that arise through estrogen-independent mechanisms.

Stage, Grade, and Histology: The Clinical Tripod

Three variables dominate clinical decision-making in endometrial cancer: FIGO stage (how far the tumor has spread), tumor grade (how abnormal the cells look), and histological subtype (endometrioid versus serous versus rarer forms). In this cohort, early-stage disease predominates — most patients fall into FIGO stage I or II, diagnosed before the cancer has escaped the uterus. That skew is entirely expected; postmenopausal bleeding triggers early investigation, and most tumors are caught while still confined to the endometrium or superficial myometrium.

Histologically, endometrioid tumors form the large majority, with a smaller fraction of serous and mixed cases. This ratio sets a hard ceiling on statistical power: any comparison between subtypes will be constrained by the smaller serous group. Tumor grade shows a spread across G1, G2, and G3, with grade 3 tumors — the most aggressive, often correlating with copy-number-high molecular profiles and worse outcomes — present but less numerous than their lower-grade counterparts.

Three variables dominate clinical decision-making in endometrial cancer: FIGO stage, tumor grade, and histological subtype. In this cohort, early-stage disease predominates.

Disentangling Stage From Grade

Stage and grade are correlated — advanced-stage tumors tend to be high-grade, and vice versa — but the correlation is imperfect, and the discordant cases are among the most biologically interesting. A stage I, grade 3 tumor has aggressive cellular features but was caught before it spread; a stage III, grade 1 tumor looks deceptively bland under the microscope yet has already invaded beyond the uterus. These off-diagonal patients offer a rare opportunity to tease apart the molecular programs that drive aggressive morphology from those that drive actual metastatic behavior.

The cross-tabulation of stage and grade in this cohort reveals which cells in that matrix are populated and which are sparse. Well-populated diagonal cells confirm the expected clinical correlation. But sparse or empty off-diagonal cells — say, stage III with grade 1 — indicate natural combinations the cohort simply cannot interrogate with statistical confidence. Any researcher planning a grade-adjusted proteomic analysis needs to check these counts before proceeding.

Histology Shifts With Stage

One of the most important confounds in endometrial cancer research is the entanglement of histology and stage. Serous tumors are biologically aggressive and tend to present at advanced stages; endometrioid tumors dominate early-stage disease. If a researcher finds a protein that differs between stage I and stage III tumors, is that a stage effect or a histology effect? The answer often depends on how thoroughly the cohort's composition has been mapped.

Breaking down histological subtypes by pathological stage makes the confound visible. Endometrioid cases fill the early-stage bins almost exclusively. Serous and mixed histologies appear more frequently at higher stages. When one stage category is almost entirely a single subtype, any stage-associated molecular finding risks being a histology finding in disguise. The same logic applies to BMI distributions across stages: if early-stage patients are systematically heavier, metabolic proteomic signals attributed to stage may actually reflect BMI-driven biology.

The Genomic Landscape: Mutations That Define Subtypes

The oncoprint — a matrix of somatic mutations and copy-number alterations across the most recurrently altered genes — is where the molecular subtypes of endometrial cancer come into sharpest focus. Each column is a patient; each row is a gene, ranked by how often it is altered.

Two patterns emerge immediately. The first is the PTEN-mutant, PIK3CA-mutant, microsatellite-unstable block: columns dense with small colored tiles, indicating many mutations per patient. These are the hypermutated or ultramutated tumors, often driven by mismatch-repair deficiency (loss of MLH1, MSH6, or PMS2) or by mutations in the proofreading polymerase POLE. They correspond almost exclusively to endometrioid histology and tend to carry favorable prognoses.

The second pattern is the sparse column: few mutations, but prominent TP53 alterations and widespread copy-number changes. These are the copy-number-high, serous-like tumors — molecularly akin to high-grade serous ovarian cancer, with poor outcomes and limited therapeutic options. The mutual exclusivity of PTEN and TP53 alterations is not absolute, but it is striking, and it recapitulates one of the most robust findings in endometrial cancer genomics.

Between these poles sit the copy-number-low endometrioid tumors: moderate mutation burden, PTEN and PIK3CA alterations present, but no mismatch-repair deficiency. These patients lack a clear actionable target today, making them a priority for proteomic and phosphoproteomic discovery.

What emerges from this dashboard is not a single headline discovery but something more foundational: a carefully characterized starting point. The CPTAC endometrial cancer cohort is overwhelmingly early-stage, endometrioid, postmenopausal, and obese — a faithful mirror of the disease as it is most commonly encountered. Its molecular layers — mutations, copy-number shifts, protein abundances, phosphorylation events — are all anchored to patients whose clinical features have been mapped in granular detail. That anchoring matters, because the next step is to ask harder questions: Which phosphorylation events predict recurrence in the copy-number-low tumors that lack clear druggable targets? Do morbidly obese patients show metabolic proteomic signatures distinct enough to warrant tailored treatment? Can the handful of young, lean patients with mismatch-repair deficiency teach us something about immunotherapy response? The atlas is built. The interrogation is just beginning.