The Fast-Growing Breast Lump That Is Not Carcinoma: Phyllodes Tumours
A breast lump appears, and then it grows. Not over years. Over weeks. It is smooth, it slides under the finger, and it looks exactly like the harmless lump most people have heard of. The core needle biopsy comes back benign, the word fibroadenoma lands like a release, and sometimes that release is premature. Healz was built to read the growth history and the biopsy wording as one question instead of two.
A lump that grows fast is a different question

In a 2024 series of 57 patients in F1000Research, a palpable mass was the presenting symptom in 91.2% of them. The band usually quoted is 35 to 55, which those authors take from older literature rather than from their own cohort; the peak is between 40 and 50 and the average age at diagnosis around 45 (Clinics, 2025). The series' own mean age was 38.8 years, and age did not separate its benign cases from its non-benign ones.
Fibroadenomas turn up in a younger group and grow more slowly, so a firm, mobile, painless mass enlarging quickly in a woman in her forties is the picture most often described. Speed is suggestive, not decisive: juvenile fibroadenoma also grows fast, in teenagers, and in that same series the time from noticing the mass to diagnosis did not differ between the benign and non-benign groups, a median of 4 months against 5, though tumour size did.
Ask Healz.
One chat instead of ten apps. 1M+ rare cases checked. The root cause, not the label. Private.
Imaging describes the mass, it does not classify it. On ultrasound it is usually a solid, hypoechoic lesion with partially indistinct or well-circumscribed margins and posterior acoustic enhancement (Clinics, 2025). Fluid-filled clefts in a predominantly solid mass are highly suggestive (F1000Research, 2024), and on MRI the T2 images often show slit-like spaces (Clinics, 2025). Larger size raises suspicion, and cystic areas are more common in malignant tumours (Clinics, 2025); in the 57-patient series only size separated the groups (F1000Research, 2024). A mass over 10 cm is commonly described as giant, though that is convention rather than a formal definition; over 10 cm is the T3 staging boundary and is variably reported as a recurrence predictor (Clinics, 2025). Our piece on what microcalcifications on a mammogram mean covers the other common reason a breast study gets re-read.
What a phyllodes tumour actually is
Phyllodes tumours are fibroepithelial lesions: they contain both an epithelial and a stromal component (Clinics, 2025). The same family holds the fibroadenoma. What makes it a tumour is the stroma: those cells are the neoplastic part, and the epithelium is carried along. Strip out the epithelial component and the lesion is classified as a sarcoma rather than a phyllodes tumour.
The name describes what a pathologist sees. Diagnosis rests on leaf-like fronds protruding into cystically dilated spaces, with stromal hypercellularity underneath (Histopathology, 2016). That architecture is also what gets misread, because a cellular fibroadenoma can produce something that looks close to it.
Hence "not carcinoma". Breast carcinoma arises from the epithelium lining the ducts and lobules, so the vocabulary of breast cancer care, hormone receptors, HER2, nodal staging, sits awkwardly on a stromal tumour. They are also rare, about 2% to 3% of fibroepithelial lesions in Western countries (Clinics, 2025).
Why the needle biopsy often cannot settle it
Core needle biopsy is the standard preoperative test, and on this question it is weaker than people assume. In the same 2025 review, the false-negative rate for core needle biopsy runs 25% to 30%, with an underestimation rate of 18.8% to 37.5%. The 2016 Histopathology consensus review puts it directly: separating a benign phyllodes tumour from a cellular fibroadenoma is especially problematic on core biopsies, and one study found only 60% concordance against the excision diagnosis.
The reason is structural, not sloppiness. A needle takes a few thin cores out of a heterogeneous mass, and several of the features that define and grade a phyllodes tumour belong to the whole specimen. The tumour border cannot be judged on a core at all: the core does not contain the edge. Stromal overgrowth and the highest mitotic count may sit where the needle never reached.
This is where an AI second opinion earns its keep, not by overruling the pathologist but by holding the report next to the history. Our guide on how to read a biopsy result walks through what the wording of a pathology report is and is not claiming.
The three grades and what the grading measures
Phyllodes tumours are classified as benign, borderline or malignant on five stromal and architectural features (Histopathology, 2016; Clinics, 2025). The epithelium does not decide the grade.
The common mitotic thresholds are 0 to 4 mitoses per 10 high-power fields for benign, 5 to 9 for borderline, and 10 or more for malignant (Clinics, 2025). The same scale is now also written per square millimetre: under 2.5 benign, 2.5 to under 5 borderline, 5 or more malignant (Histopathology, 2025). One caveat: a high-power field is not a fixed area across microscopes, so the same tumour can cross a threshold depending on the optics.
The malignant grade is a full house. The current WHO classification asks for all five features together, a permeative border, marked stromal cellularity, marked atypia, 10 or more mitoses per 10 high-power fields and stromal overgrowth, and grades a tumour missing any one of them as borderline. The exception is a malignant heterologous element, such as osteosarcoma, chondrosarcoma, rhabdomyosarcoma or a pleomorphic or myxoid liposarcoma, which makes the tumour malignant on its own. That bar is contested: Tan and colleagues, lead author of the 2016 consensus cited above, argued in 2025 that four of the five should be enough (Histopathology, 2025).
Stromal overgrowth sounds vague and is not. It means stroma without any epithelium filling at least one low-power field at four times magnification (Histopathology, 2016). It is absent in benign tumours and is one of the strongest pointers toward malignancy.
Most of these tumours are benign: that grade accounts for roughly 60% to 75% of cases (Annals of Breast Surgery, 2021). But grading is subjective and tumours commonly carry features from more than one category, so two expert pathologists can read the same slides and land on benign versus borderline.
Surgery, margins, and why the armpit is usually left alone
Surgical excision is the treatment, aiming for histologically clear margins, by wide local excision where the breast allows it and by mastectomy where it does not. What is genuinely unsettled is how wide. Most guidelines have advised at least 1 cm, and the 2025 Clinics review calls that figure debatable. The 2025 UK Association of Breast Surgery guideline grades it instead, calling a generic 10 mm margin for every phyllodes tumour overtreatment: complete excision with the capsule intact for benign, a 5 mm target for borderline with re-excision below 3 mm, and a 10 mm target for malignant with re-excision below 5 mm (British Journal of Surgery, 2025). A positive margin is the main predictor of recurrence while wider margins did not reduce it, and as of the version cited in a 2021 Annals of Breast Surgery review, the NCCN recommended wide excision for borderline and malignant tumours without requiring a mastectomy when 1 cm is unachievable. The pathology consensus review, finding no robust evidence for a specific width, suggests it may be pragmatic to count tumour on ink or within 1 mm as a positive margin.
The axilla surprises people who know breast cancer surgery. Routine axillary dissection is not recommended (Histopathology, 2016). Reported nodal metastasis rates range from under 1% to 3.8% (Clinics, 2025), because spread, when it happens, is haematogenous rather than lymphatic, going mainly to the lungs (66%), bones (28%) and brain (9%) (Clinics, 2025). Enlarged axillary nodes are reported in about 17% of patients, while histologically proven nodal metastases occur in about 1% (Annals of Surgical Treatment and Research, 2026); in malignant phyllodes tumours specifically, reactive lymphadenopathy is reported in 10% to 15% of cases (Clinics, 2025). Nodes are sampled when clinically suspicious, not as staging.
What follow-up is actually watching for
Follow-up after a phyllodes tumour is built around the breast, not the rest of the body. Reported local recurrence rates are 10% to 17% for benign, 14% to 25% for borderline and 23% to 30% for malignant tumours (Histopathology, 2016), though more recent series report lower figures, around 8% to 10%, 13% to 14% and 18% (Annals of Breast Surgery, 2021). Recurrence generally happens within the first two to three years (Clinics, 2025).
Radiation improves local control without a demonstrated survival benefit and is considered for borderline and malignant tumours, particularly where a recurrence would cause significant morbidity (Annals of Breast Surgery, 2021). The 2025 UK guideline is more specific: not recommended for benign tumours, not recommended in most borderline ones, considered for malignant tumours over 5 cm or multifocal disease. Adjuvant chemotherapy is not recommended in non-metastatic disease (British Journal of Surgery, 2025).
How Healz reads a growth history against a biopsy report
Two documents decide this case, and they are almost never read at the same time: the growth history, meaning the dates and the measurements of the lump, and the biopsy report with its exact wording. Separately, each looks reassuring. Together, a benign core on a mass that doubled in a season is the mismatch that moves a case.
Healz is equipped with root-cause technology, which means it does not stop at the first label. It asks what a rushed visit skips, cross-checks the case against 1M+ rare cases, and drills toward the true cause instead of accepting fibroadenoma because the core said so. Healz's memory pulls the medical facts out of every upload, dates each measurement and keeps the abnormal values up front, so a size from March sits beside a size from June. Frontier AI works the whole file, so the pathology report wording, the imaging and the timeline are read against each other rather than in sequence. It all sits in one chat, not ten apps and four portals, which is what a cancer ai should actually give you. You can bring a board-certified doctor into the same chat for a second opinion.
Frequently asked questions
- Can a phyllodes tumour come back after surgery?
Yes, and local recurrence is the main thing follow-up is looking for. Reported local recurrence rates are 10% to 17% for benign tumours, 14% to 25% for borderline and 23% to 30% for malignant (Histopathology, 2016); a 2021 Annals of Breast Surgery review reports lower contemporary figures. Recurrences generally appear within the first two to three years, and positive surgical margins are a key risk factor (Clinics, 2025).
- Is a phyllodes tumour breast cancer?
Most are not. The benign grade covers roughly 60% to 75% of cases (Annals of Breast Surgery, 2021). They are also not carcinomas: the neoplastic part is the stroma rather than the duct lining, so the malignant grade behaves more like a sarcoma than like a typical breast carcinoma.
- Can a core needle biopsy tell a phyllodes tumour from a fibroadenoma?
Often it cannot. The 2016 Histopathology consensus review states that separating a benign phyllodes tumour from a cellular fibroadenoma is especially problematic on core biopsies, citing a study with only 60% concordance against excision. Reported false-negative rates run 25% to 30% (Clinics, 2025), which is why many reports read as a fibroepithelial lesion that cannot be further characterised.
- Do phyllodes tumours spread to the lymph nodes?
Rarely. Reported nodal metastasis rates are under 1% to 3.8%, because spread is haematogenous rather than lymphatic, going mainly to the lungs, bones and brain (Clinics, 2025). Routine axillary dissection is therefore not recommended (Histopathology, 2016), and enlarged nodes, reported in about 17% of patients, are usually reactive (Annals of Surgical Treatment and Research, 2026).
A lump that grows fast and reads benign is not a contradiction to explain away; it is two documents read together, and the second only becomes final when the whole tumour is out.
Written by Healz Team · Filed under Health Insights