Endometrial biopsy detects endometrial cancer with roughly 90% to 95% sensitivity and 95% or higher specificity when sufficient tissue is obtained, but accuracy drops sharply when the sample is scant or the lesion is small and focal. Sensitivity is highest for endometrioid adenocarcinoma and lowest for certain high-grade and rare histologies.
The test's real-world performance depends less on the needle than on what the pathologist receives. An endometrial biopsy samples a small fraction of the uterine cavity, so the accuracy figures quoted in reviews describe an idealized adequate specimen. This explainer unpacks the mechanism, the measured ranges, the edge cases that defeat the test, and the practical thresholds clinicians use when a result does not fit the picture.
What the Procedure Actually Samples
Endometrial biopsy threads a thin catheter through the cervix and suctions or scrapes tissue from the endometrial lining. The device, often a Pipelle-style sampler, removes a strip of mucosa a few millimeters wide. That strip is fixed, sectioned, stained, and read under a microscope by a pathologist looking for architectural crowding, nuclear atypia, and loss of normal gland spacing.
The accuracy ceiling comes from geometry. The endometrial cavity has a surface area on the order of tens of square centimeters, and a biopsy captures a small sampling of it. A cancer that occupies a large, diffuse area is nearly impossible to miss. A cancer confined to a polyp or a small corner of the cavity can be entirely bypassed. This is why a negative biopsy is reassuring but not definitive, and why clinicians sometimes still proceed to further sampling when symptoms persist.
A pathologist reading the slide faces a second problem: distinguishing cancer from its mimics. Complex atypical hyperplasia, polypoid changes, and secretory-phase alterations can look architecturally busy. The diagnosis rests on nuclear features and stroma invasion patterns that require adequate, well-oriented tissue. A fragmented or blood-clotted specimen degrades that judgment.
Quantified Accuracy
Across systematic reviews of diagnostic accuracy, endometrial biopsy carries a pooled sensitivity near 90% and specificity above 95% for endometrial cancer in women with abnormal uterine bleeding. Many individual studies report sensitivity in the 90% to 95% range when inadequate samples are excluded. The distinction matters: excluding insufficient specimens inflates apparent accuracy, because the samples that fail are disproportionately the ones that miss disease.
The most important number is the inadequacy rate. Depending on operator experience, device, and patient anatomy, roughly 5% to 20% of office biopsies yield tissue insufficient for diagnosis. In those cases, the test has not ruled anything out; it has simply failed to answer. Many reported sensitivity figures treat these as a separate category rather than as false negatives, which flatters the headline number.
Comparative data help. A hysteroscopic-directed biopsy, in which the clinician visualizes the cavity and samples the abnormal area directly, generally achieves higher sensitivity than blind office sampling for focal lesions. Dilatation and curettage samples more of the cavity than a suction sampler but is less targeted than hysteroscopy. Each step up in invasiveness trades convenience for coverage.
One counterintuitive finding recurs in the literature: adding a second blind sampling pass improves yield far less than clinicians expect. The gain from repeating an office biopsy is modest compared with moving to a visualized, targeted sampling method. Repeating the same blind technique tends to reproduce the same blind spots.
Where the Numbers Break Down
Accuracy is not a property of the test alone; it is a property of the test applied to a particular patient. Several factors shift the true sensitivity well below the pooled average.
Postmenopausal atrophy: a thin, atrophic lining yields scant tissue, raising the inadequacy rate and the chance of a false-negative. Focal or polypoid disease: a lesion occupying a small fraction of the cavity can sit outside the sampler's path. High-grade and rare histologies: uterine serous carcinoma and carcinosarcoma can shed cells that are harder to characterize on a small strip, and some non-endometrioid tumors are patchy. Prior tamoxifen exposure: associated endometrial changes and polyps complicate both sampling and interpretation. Cervical stenosis: anatomy that blocks catheter passage produces a failed, not a negative, test.
The professional caveat that follows is blunt: a negative office biopsy in a woman with persistent postmenopausal bleeding does not close the case. Guidelines from major gynecologic societies treat ongoing bleeding after a benign biopsy as an indication for further evaluation, typically with imaging or hysteroscopy. The biopsy answered a question about the tissue it reached, not about the whole cavity.
Common Misconceptions
The most frequent error is treating a negative biopsy as a rule-out. A negative result lowers the probability of cancer substantially but does not eliminate it, especially when symptoms continue. The residual risk after a negative biopsy in a symptomatic postmenopausal woman is low but not zero, and it is concentrated in exactly the focal and atrophic cases described above.
A second misconception conflates biopsy with curettage or with hysteroscopy. These are different procedures with different coverage of the cavity and different accuracy profiles. A patient told she had "a biopsy" may assume the entire lining was examined; often only a small strip was sampled.
A third error is reading a diagnosis of hyperplasia as a benign dead end. Atypical hyperplasia is a precursor lesion with a meaningful progression risk to cancer, and its presence changes management. The biopsy's job is not only to find cancer but to stratify risk, and a report of atypia is a clinically actionable finding rather than a reassurance.
Finally, patients and clinicians sometimes overweight a single atypical or suspicious report. Interobserver agreement among pathologists on borderline endometrial proliferations is imperfect, and a second expert review is reasonable when a diagnosis is uncertain or management hinges on it.
When to Escalate Beyond Biopsy
The practical question is not whether biopsy is accurate in the abstract but when its result should be trusted and when it should be superseded. The decision rests on symptom pattern, imaging, and the adequacy of the specimen.
If the pathology report states the sample was insufficient or scant, the test is uninterpretable and further sampling is required. If the specimen was adequate but the patient has continued bleeding, further evaluation is warranted regardless of a benign reading. If transvaginal ultrasound shows a thickened or heterogeneous endometrium that does not fit a benign biopsy, the imaging and the tissue result must be reconciled rather than one simply overriding the other.
A comparison of sampling approaches clarifies the trade-offs:
| Method | Cavity coverage | Typical sensitivity context | Invasiveness |
|---|---|---|---|
| Office suction biopsy | Partial, blind | Pooled ~90% with adequate tissue | Low, office-based |
| Dilatation and curettage | Broader, still blind | Higher yield than suction alone | Operative setting |
| Hysteroscopic-directed biopsy | Targeted to visualized lesion | Highest for focal disease | Procedural |
| Transvaginal ultrasound | No tissue; risk stratification | Adjunct, not diagnostic alone | Noninvasive |
The table's message is that no single test dominates. Ultrasound triages, biopsy characterizes sampled tissue, and hysteroscopy resolves focal disease the blind sampler missed. Accuracy lives in the combination, not in any one instrument.
A representative clinical time course illustrates how results sequence:
| Stage | What happens | Decision point |
|---|---|---|
| Presentation | Postmenopausal bleeding or abnormal bleeding | Trigger evaluation |
| Imaging | Transvaginal ultrasound assesses lining thickness | Thickened or heterogeneous lining raises concern |
| Office biopsy | Suction sample obtained | Adequate tissue with diagnosis ends workup |
| Inadequate result | Report states insufficient specimen | Repeat with better method |
| Persistent symptoms | Bleeding continues despite benign result | Escalate to hysteroscopy |
| Definitive sampling | Targeted biopsy or curettage | Tissue diagnosis established |
| Management | Plan based on final histology | Treatment or surveillance |
The timeline's key point is that a "negative" result at the office-biopsy stage is provisional until symptoms and imaging agree.
Practical Thresholds and Actions
The actionable core of this topic is a small set of rules that convert accuracy data into decisions. First, confirm the specimen was adequate before interpreting a negative result; an inadequate sample is a non-answer. Second, treat persistent postmenopausal bleeding after a benign biopsy as an unresolved problem requiring further evaluation. Third, when imaging and biopsy disagree, resolve the disagreement rather than defaulting to the more reassuring result.
For clinicians, the practical step is to document specimen adequacy explicitly and to counsel patients that a negative office biopsy reduces but does not eliminate risk. For patients, the practical step is to report ongoing bleeding rather than assume a benign result is final. These are not technical nuances; they are the difference between a test used well and a number quoted out of context.
What the Accuracy Figures Cannot Tell You
Every pooled sensitivity estimate describes a population of specimens, not the slide in front of a particular pathologist. The figures assume adequate tissue, a representative sample, and a lesion within reach of the device. When those assumptions fail, the quoted 90% to 95% sensitivity does not apply to that patient's result.
The honest summary is that endometrial biopsy is a strong but imperfect first test. It reliably detects diffuse endometrioid cancer and reliably samples the tissue it reaches. It is weakest precisely where clinical vigilance matters most: thin atrophic linings, small focal lesions, and high-grade histologies that behave differently on a small strip. Accuracy is high enough to anchor a workup and too low to end one when symptoms persist.