19.4 — Staging, Grading and What the Numbers Mean

"Stage 4" is the phrase people fear most in oncology, and its meaning varies enormously by cancer.

Stage 4 testicular cancer is cured in the majority. Stage 4 pancreatic cancer usually is not.

The stage describes how far the cancer has spread. It does not, by itself, describe the outlook — that depends on which cancer, what drives it, and what treatments exist.

Staging: how far it has spread

The TNM system, used for most solid cancers.

T — the primary tumour. T1 to T4, by size or by depth of invasion depending on the organ. Tis means carcinoma in situ — confined above the basement membrane (Chapter 19.3).

N — regional lymph nodes. N0 to N3, by number and location of involved nodes.

M — distant metastasis. M0 or M1. There is no M2 — it has either spread distantly or it has not.

These combine into an overall stage from I to IV, and the combinations differ by cancer type.

Broadly:

Stage 0 — in situ. Not yet invasive, and effectively always curable.

Stage I — small, localised.

Stage II — larger, or minimal local spread.

Stage III — regional lymph node involvement, or substantial local invasion.

Stage IV — distant metastasis.

And two qualifiers appear on reports:

Clinical stage (prefix c) — based on examination and imaging before treatment.

Pathological stage (prefix p) — based on what was actually found in the removed specimen. More accurate, and it frequently differs from the clinical stage.

Which is why a stage can change after surgery, and why that is not an error.

Different systems exist for some cancers. Lymphoma uses the Ann Arbor system. Gynaecological cancers use FIGO staging. Leukaemias are not staged in the same way at all, because they are disseminated from the outset.

Grading: how abnormal the cells look

Staging is about extent. Grading is about behaviour.

A pathologist examines the cells and assesses how closely they resemble the normal tissue — how well differentiated they are — and how rapidly they are dividing.

Grade 1 — well differentiated. Resembles normal tissue, slow-growing. Grade 2 — moderately differentiated.Grade 3 — poorly differentiated. Little resemblance, rapid division. Grade 4 — undifferentiated.

And specific systems exist for specific cancers.

The Gleason score in prostate cancer — grading the two commonest patterns and adding them, giving scores from 6 to 10. It is now reported alongside a Grade Group from 1 to 5, because the old scale's minimum of 6 misleadingly suggested a middling grade.

The Nottingham system in breast cancer.

And grade and stage together predict behaviour far better than either alone. A small high-grade tumour may be more dangerous than a larger low-grade one.

The other things that matter now

And this is where cancer description has changed most.

Stage and grade are anatomical and morphological. Molecular characterisation is increasingly what determines treatment.

Receptor status in breast cancer:

Oestrogen and progesterone receptors — if positive, hormonal treatment works. HER2 — if positive, trastuzumab and related drugs work.

Triple-negative — none of the three, which removes those options and is associated with a worse prognosis, though immunotherapy and PARP inhibitors have improved it substantially.

Driver mutations in lung cancer — EGFR, ALK, ROS1, KRAS and others, each with a matching drug (Chapter 19.2).

Mismatch repair status — predicting immunotherapy response (Chapter 2.2).

Tumour mutational burden.

And gene expression panels — such as the 21-gene recurrence score in breast cancer, which identifies women who can safely avoid chemotherapy.

That last one is worth emphasising as a genuine advance. The TAILORx trial showed that a large proportion of women with early hormone-receptor-positive breast cancer derive no benefit from chemotherapy, and the test identifies them. Thousands of people a year avoid chemotherapy they would previously have received.

De-escalation — giving less treatment safely — is as significant as escalation, and it receives far less attention.

Reading survival statistics

And this section matters, because these numbers are frequently misunderstood in ways that cause unnecessary distress.

Five-year survival — the proportion alive five years after diagnosis.

And it is not a life expectancy. Someone with 90 percent five-year survival is not expected to die in year six. For many cancers, surviving five years without recurrence means being cured.

Relative survival — comparing with people of the same age and sex without cancer. Preferable, because it removes deaths from other causes.

Median survival — the point at which half the group has died. Half are still alive, and some by a long way.

And the distribution matters more than the median. Stephen Jay Gould, diagnosed with mesothelioma and given a median survival of eight months, wrote "The Median Isn't the Message" after realising the distribution was right-skewed — a long tail of survivors. He lived twenty years.

The essay is worth reading and its point is statistical rather than sentimental: a median tells you nothing about where in the distribution any individual sits.

Three specific traps:

Survival statistics are historical. A five-year figure published today describes people diagnosed at least five years ago and treated with what was available then. For cancers where treatment has changed rapidly — melanoma, lung cancer — the published figures substantially understate current outcomes.

Lead time and length time bias make screen-detected cancers look more survivable than they are (Chapter 16.1).

And averages conceal enormous variation — by stage, molecular subtype, age, fitness and treatment.

Which is why the honest answer to "how long have I got" is that population statistics cannot answer it for an individual, and why good oncologists give ranges and scenarios rather than numbers.

What determines prognosis

Stage — the largest single factor for most cancers.

Grade and molecular features.

Site — pancreatic and brain cancers remain difficult; thyroid and testicular are usually curable.

Performance status — a simple 0 to 4 scale of how much the person is up and about. It predicts outcome and tolerance of treatment better than almost any laboratory measure, and it is why oncologists ask about daily activity in detail.

Age and other conditions.

And treatment response, which frequently outweighs the baseline predictors.

What being told a stage should mean

A practical note, because this is a conversation most people will have.

Ask what the stage means for this cancer specifically, rather than assuming.

Ask whether the intent of treatment is curative or to control the disease. This is the single most important question, and it is frequently not stated plainly.

Ask what the molecular testing showed, because it may determine the treatment more than the stage does.

Ask what the alternatives are, including doing less.

And a second opinion is reasonable and normal, particularly for rare cancers or where the options are finely balanced.

Multidisciplinary team meetings — where surgeons, oncologists, radiologists and pathologists review each case together — are now standard, and they measurably improve decisions. It is reasonable to ask what the team recommended.

Response to treatment

Terms used in reports and trials:

Complete response — no detectable disease. Partial response — substantial shrinkage. Stable disease — neither growing nor shrinking meaningfully. And this can be a good outcome in advanced disease.Progressive disease — growing.

Remission — no evidence of disease. Complete remission is not the same as cure, because undetectable is not the same as absent.

Cure — used when the risk of recurrence has fallen to background. The interval differs by cancer: five years for many, and far longer for breast cancer, which can recur after twenty (Chapter 19.3).

And "no evidence of disease" is the phrase oncologists prefer, because it is accurate about what is known.

What the next page fixes

Chapter 19.5 goes through the common cancers one by one — how each presents, what is done about it, and what the realistic outlook is.