Neuroendocrine Tumour Grading (WHO)
On this page
Direct answer
Grading a gastroenteropancreatic neuroendocrine neoplasm under the WHO framework uses two numbers: mitotic count per 10 high-power fields and Ki-67 proliferation index. G1 means under 2 mitoses per 10 HPF and Ki-67 under 3%; G2 spans 2-20 mitoses and 3-20%; G3 exceeds 20% on either — and when the two disagree, the higher grade applies. The 2019 WHO edition then draws the line that changed practice: well-differentiated tumours remain neuroendocrine tumours (NET) even at G3, whereas poorly differentiated small- and large-cell carcinomas are neuroendocrine carcinomas (NEC) by definition — different biology, different drugs. Lung and gastrointestinal tract rules also differ: pulmonary carcinoids are graded by mitoses and necrosis alone into typical and atypical carcinoids, without Ki-67 in the formal definition.
What you must remember
- The G1-G3 numbers: Ki-67 cut-offs under 3, 3-20 and over 20%; mitoses under 2, 2-20 and over 20 per 10 HPF — grade discordance resolves upward.
- NET G3 versus NEC: NET G3 retains organoid architecture, chromogranin-strong expression and usually wild-type p53/RB; NEC shows smudged necrosis, single-file cytology, p53 aberrant or RB loss, and Ki-67 often over 50-70% — the distinction that routes patients to temozolomide or peptide-receptor therapy versus platinum-etoposide.
- Counting method: Ki-67 counted manually in hot spots over approximately 500-2000 cells; printed estimates overestimate — viva examiners specifically ask how you count.
- Lung grading: typical carcinoid under 2 mitoses per 2 mm² with no necrosis; atypical carcinoid 2-10 mitoses or necrosis; large-cell neuroendocrine carcinoma over 10 mitoses with necrosis and large-cell neuroendocrine morphology.
- Core immunostains: chromogranin A (dense-core granules), synaptophysin, and INSM1 as the robust nuclear marker; CD56 is supportive but nonspecific.
- Phaeochromocytoma-paraganglioma scoring: malignant-prediction in PPGL uses the PASS score (adaptive features, hyperplasia, nuclear atypia; PASS ≥4 flags malignant potential) and the GAPP grading — histology-based malignancy prediction where no marker is definitive.
- Merkel cell carcinoma: CK20 perinuclear dot-positivity with Merkel cell polyomavirus (CM2B4) staining and dot-like synaptophysin — a skin NEC with viral aetiology.
- Functional correlation: grade drives surveillance intervals and nucleotide-imaging choices — DOTATATE PET for well-differentiated NET, FDG PET for NEC, a pairing examiners increasingly test.
A pancreatic tumour with Ki-67 of 35%
A 55-year-old undergoes distal pancreatectomy for a 4 cm pancreatic head-body tumour. Histology shows well-formed nests and trabeculae with salt-and-pepper chromatin, minimal necrosis. Ki-67 runs at 35%, mitoses 4 per 10 HPF. The reflex label "NEC" would be wrong: this is a well-differentiated pancreatic NET, grade G3. Confirmatory stains show strong diffuse chromogranin, retained RB, and p53 wild-type pattern — no NEC genotype. The therapeutic fork opens exactly here: pancreatic NEC is treated with cisplatin-etoposide, while NET G3 is managed with capecitabine-temozolomide or peptide-receptor radionuclide therapy where somatostatin-receptor imaging is positive. Mislabelling the tumour sends the patient down the wrong chemotherapy track, which is why the 2019 separation of grade from differentiation was the decade's most consequential change in this field.
The same logic runs in reverse for a lung biopsy with nested morphology, 6 mitoses per 2 mm² and a punctate necrotic focus — atypical carcinoid, not LCNEC, because the cells are uniform, nucleoli small, and Ki-67 under 20%; reclassifying it upward would wrongly deny surgical curative intent.
Where students slip
The commonest error is treating Ki-67 over 20% as synonymous with NEC; the correct reading pairs the number with differentiation — morphology and p53/RB status arbitrate. The second slip is transplanting gastrointestinal cut-offs into the lung: bronchial carcinoids are graded by mitoses and necrosis only, so quoting "3% Ki-67 makes it G1" in a pulmonary question exposes the error. Third, marker over-reliance: TTF-1 is positive in many pulmonary small-cell carcinomas yet does not prove origin; extrapulmonary small-cell carcinomas can also stain, so primary-site assignment stays clinical. Indian laboratory angle: manual Ki-67 counting is the rule in most Indian laboratories without digital image analysis, making the hot-spot method — eyeball the slide, choose the darkest zone, count 500 cells — precisely the viva question asked.
Frequently asked questions
What are the Ki-67 and mitotic cut-offs for G1, G2 and G3?
Under 3% and under 2 mitoses per 10 HPF for G1; 3-20% and 2-20 for G2; over 20% or over 20 mitoses for G3, with discordant results assigned the higher grade.
How is a well-differentiated NET G3 distinguished from a NEC?
By morphology plus genotype — organoid well-differentiated architecture, strong chromogranin and wild-type p53/RB define NET G3, whereas NEC is poorly differentiated with p53 or RB abnormalities.
What separates atypical from typical carcinoid in the lung?
Atypical carcinoids have 2-10 mitoses per 2 mm² or necrosis; typical carcinoids stay under 2 mitoses and lack necrosis.
Which immunostains are first-line for neuroendocrine differentiation?
Chromogranin A and synaptophysin, with INSM1 as a robust third marker; CD56 is supportive only.
What is the PASS score used for?
Phaeochromocytoma of the adrenal gland scaled score — histological features totted up with a score of 4 or more indicating malignant potential in an organ with no reliable single marker.