Understanding Differentiation, Ki‑67, Mitotic Count, Hotspots, Pathology Workflow, and Primary–Metastasis Heterogeneity in Neuroendocrine Neoplasms (NENs)

Before you read this…

This article discusses pathology concepts such as Ki-67, grading, heterogeneity, and biopsy findings in neuroendocrine tumours (NETs). It is provided for educational purposes only and does not interpret any individual pathology report or scan result.

Ki-67 values, tumour grade, and sampling limitations can vary between different biopsies and over time. Their meaning depends on the full clinical context, including imaging, symptoms, and multidisciplinary review.

Only your own specialist team can explain what your specific Ki-67, grade, or pathology findings mean for you. No treatment decisions should be made based on this article alone.

Neuroendocrine Neoplasms (NENs) are biologically complex. They evolve over time, behave differently in different parts of the body, and often show internal variation even within a single tumour. To understand why Ki‑67 varies, why grades change, and why metastases often behave differently from primaries, we need to start with the foundation of NEN pathology: differentiation. It’s important to understand the difference between differentiation and grade. Differentiation = what the tumour looks like – Grade = how fast it is proliferating.

1. Differentiation: the foundation of NEN pathology

Differentiation describes how closely tumour cells resemble normal neuroendocrine cells.

Well‑differentiated NET

Cells still look “NET‑like”
Architecture is organised (nests, trabeculae, rosettes)
SSTR expression usually preserved
Growth tends to be slower
Graded by Ki‑67 (G1, G2, G3 NET)

Poorly differentiated NEC

Cells look nothing like normal neuroendocrine cells
Architecture is chaotic: disorganised sheets, necrosis, marked atypia
SSTR expression often lost
Extremely fast growth
Ki‑67 usually 50–90%
Diagnosis based on morphology, not Ki‑67 alone

⭐ Differentiation = what the tumour looks like

⭐ Grade = how fast it is proliferating

These are separate but linked concepts. However, it follows that poorly differentiated means it is proliferating fast.

2. Ki‑67 and mitotic count: how NETs are graded

For well‑differentiated NETs, grade is determined by Ki‑67. However, WHO 2021 Thoracic tumours indicates Mitotic Count determines grade of Lung/Thymic NENs:

For NEC, Ki‑67 is usually very high, but the diagnosis depends on poor differentiation, not the number.

More info in here too (but some of the stuff below will not be included at this link

2.1 Thoracic NETs: The Important Exception

Lung and thymic NETs (typical and atypical carcinoid) are not graded by Ki‑67. Ki‑67 may be additionally reported but does not determine grade.

According to WHO 2021 (Lung Blue Book):

Typical Lung NET: <2 mitoses per 2 mm², no necrosis
Atypical Lung NET: 2–10 mitoses per 2 mm², focal necrosis
LCNEC/SCLC: high mitotic rate + extensive necrosis *
there are a group of aggressive atypical Lung NETs (unintuitively known as “supra carcinoids” that are currently grouped with LCNEC for categorising. This may be reviewed in the next edition of Thoracic Tumours, e.g. a Grade 3 well differentiated Lung NET)

3. Intra‑tumour heterogeneity: why different parts of the same tumour may have different Ki‑67

NETs are mosaics of different biological behaviours.

For example, different regions may show:

2% Ki‑67
6% Ki‑67
12% Ki‑67
a small pocket at 18%

These example variations could arise from:

clonal diversity
microenvironmental differences (hypoxia, fibrosis, inflammation)
uneven blood supply
evolving subclones

This is why hotspot selection is essential. Hotspot selection in NET biopsies involves identifying the most proliferative area within the biopsy sample using Ki‑67 immunostaining. However, in biopsies (e.g. core and fine needle), there can be limitations as the hotspot may not appear in that sample.

4. How pathologists examine a tumour (the real workflow)

4.1 When a whole tumour is available:

4.1.1 Multiple blocks are taken

Centre, edge, invasive front, unusual areas — often 5–20 blocks.

4.1.2 Pathologist scans all slides at low power

Looking for areas that look more proliferative:

crowded nuclei
more mitoses
higher cellularity
less fibrosis
more atypia

This is trained pattern recognition, not guesswork.

4.1.3 They identify the single most proliferative hotspot

This is the area most likely to drive clinical behaviour.

4.1.4 Ki‑67 is is normally counted only in the hotspot

Not averaged across the tumour.

4.1.5 Mitotic count is done the same way

Highest mitotic region = the one that matters.

4.2. Core Needle Biopsy (CNB)

This is the most common biopsy type for NETs, especially liver metastases. A core biopsy provides:

a thin cylinder of tissue
usually 1–2 cm long
representing one tiny region of the tumour

4.2.1 What the pathologist can do

Assess differentiation (NET vs NEC)
Perform Ki‑67/mitotic count staining
Identify the best hotspot within that core
Perform immunohistochemistry (chromogranin, synaptophysin, SSTR2, etc.)

4.2.2 What the pathologist cannot do

Examine multiple regions of the tumour as per whole tumour above.
Compare centre vs edge vs invasive front
Identify the true highest‑grade hotspot if it wasn’t sampled
Evaluate full architectural patterns
Assess heterogeneity

4.2.2.1Why this matters

A core biopsy samples only one region, so it may miss the aggressive pocket. This is why:

primaries are often under‑graded
metastases often appear “higher grade”
Ki‑67 can rise on repeat biopsy. Mitotic count measurement is less efficient on CNB.
dual‑tracer PET helps guide biopsy to the most aggressive lesion

4.3 Fine‑Needle Aspiration (FNA)

FNA provides cells, not tissue architecture. This is the most limited sample type.

4.3.1 What FNA can do

Confirm neuroendocrine origin
Provide cytology (cell appearance)
Allow limited immunostaining
Sometimes allow Ki‑67 counting (if enough cells are present)

4.3.2 What FNA cannot do

Assess architecture (no tissue structure)
Reliably distinguish NET vs NEC
Provide accurate Ki‑67 in many cases
Identify hotspots
Evaluate heterogeneity

4.3.3Why this matters

FNA is not ideal for Ki‑67. It is used only when no other biopsy is feasible. Mitotic count measurement is less efficient on FNA, perhaps unsuitable as there is little tumour architecture.

5. Why Ki‑67 is sometimes reported as ranges or thresholds

5.1 Biological variation

Hotspot fields may show 6%, 8%, 9% → may be reported as 5–10%.

5.2 Small or fragmented biopsies

Exact counting may not be possible → “>3%”, “3–5%”, “5–10%”.

5.3 Category‑based lab protocols

Some labs use bands like <3%, 3–20%, >20%.

5.4 Grade‑signalling

“>3%” means G2, should not be assumed to be anything up to 100%”.

6. Differentiation + Ki‑67: how they interact

This is the key conceptual link with two examples:

⭐ A tumour with Ki‑67 40–60% is almost always poorly differentiated NEC

→ chaotic architecture → aggressive behaviour → platinum chemotherapy → FDG‑avid → SSTR‑negative

⭐ A tumour with Ki‑67 25–55% can be well‑differentiated NET G3

→ preserved architecture → SSTR expression → PRRT may still work → different prognosis

This is why differentiation must be assessed before interpreting Ki‑67.

7. What ENETS, WHO, and NANETS recommend

To counter the issues listed in paragraphs 5 and 6. all major guidelines state:

✔️ Report the exact Ki‑67 percentage

✔️ Count at least 500–2000 tumour cells

✔️ Identify the true hotspot

✔️ Avoid categorical reporting where possible

Because NET behaviour is non‑linear, and small changes matter.

8. Why liver metastases often have higher Ki‑67 than the primary

8.1 The liver selects for more aggressive clones

Normally, only the fitter, faster‑growing cells survive the journey. This doesn’t mean every liver met will be higher than the primary but it’s common.

8.2 The liver microenvironment promotes proliferation

Growth factors, cytokines, angiogenesis → higher Ki‑67.

8.3 Clonal evolution

Metastases represent a later evolutionary stage.

8.4 Sampling bias

The primary biopsy may miss the hotspot.

8.5 Ki‑67 is dynamic

It can rise over time due to progression or treatment pressure.

9. Primary vs metastasis: what the data show

Across multiple NET subtypes (e.g. small bowel, pancreas, lung):

⭐ Grade concordance:

~50–70% of patients have the same grade in primary and metastases.

⭐ Grade discordance:

~30–50% show different grades.

⭐ Direction of change:

Most discordance = upgrading in metastases (G1 → G2, G2 → G3)
Downgrading is less common.

⭐ Ki‑67 differences specifically:

40–60% of patients have higher Ki‑67 in liver metastases than in the primary.
A minority have similar Ki‑67.
A small fraction have lower Ki‑67 in metastases.

⭐ Within‑patient heterogeneity among metastases:

Different metastases can have different Ki‑67 values and even different grades. This matches dual‑tracer PET patterns:

SSTR+/FDG−
SSTR+/FDG+
SSTR−/FDG+

10. Clinical and advocacy implications

Never assume the primary’s grade applies forever.
Biopsy metastases when possible.
Ask for exact Ki‑67, not just (e.g.) “>3%”.
Enquire about dual‑tracer PET (or separate scans) to visualise heterogeneity where applicable (e.g. high Grade 2/G3).
Expect mixed responses to treatment.
Recognise that differentiation and grade are separate but both essential at grade 3.

BONUS SECTION: AI in Pathology — A Neutral, Evidence‑Based Overview

AI is entering pathology, but not as a replacement — as a support tool.

What AI can do today

Ki‑67 counting AI can count thousands of nuclei consistently and reduce inter‑observer variability.
Hotspot detection AI can scan whole slides and highlight areas with higher proliferation.
Mitotic figure detection AI can identify mitoses more consistently than humans.
Quality control AI can flag out‑of‑focus regions, staining artefacts, and tissue issues.
Pattern recognition AI can detect subtle features that correlate with differentiation or risk.

What AI cannot do

Diagnose NET vs NEC Differentiation requires human morphological interpretation.
Integrate clinical, radiological, and molecular data Pathologists synthesise information that AI cannot.
Take responsibility Pathologists remain legally and professionally accountable.
Replace judgement NETs are subtle, heterogeneous, and biologically complex.

Balanced conclusion

AI will make pathology more consistent and reproducible — especially for Ki‑67 counting — but it will not replace the pathologist. It is a tool, not a diagnostician.

In NETs, where:

small Ki‑67 differences matter
hotspots are subjective
differentiation is critical
heterogeneity is common

AI is likely to become a valuable assistant, not a decision‑maker.

Disclaimer

I am not a doctor or any form of medical professional, practitioner or counsellor. None of the information on my website, or linked to my website(s), or conveyed by me on any social media or presentation, should be interpreted as medical advice given or advised by me.

Neither should any post or comment made by a follower or member of my private group be assumed to be medical advice, even if that person is a healthcare professional. Some content may be generated by AI which can sometimes be misinterpreted. Please check any references attached.

Please also note that mention of a clinical service, trial/study or therapy does not constitute an endorsement of that service, trial/study or therapy by Ronny Allan, the information is provided for education and awareness purposes and/or related to Ronny Allan’s own patient experience. This element of the disclaimer includes any complementary medicine, non-prescription over the counter drugs and supplements such as vitamins and minerals.