Cancer staging is the process of mapping how far a cancer has spread—its size or depth at the primary site (T), lymph node involvement (N), and distant metastasis (M)—then grouping those findings into a stage that guides treatment and estimates prognosis. For most solid tumors, stages run from I to IV; higher numbers mean more extensive disease, and stage IV generally means the cancer has spread to distant organs. Staging is a snapshot taken before treatment begins, and it is revised when new evidence changes the picture.
Staging is not a single test. It is a synthesis of physical examination, imaging, biopsy, surgical findings, and pathology, assembled into a common language so that a patient in one hospital can be discussed meaningfully at another. That language matters because stage determines whether surgery comes first, whether chemotherapy or radiation is added, and what outcomes a patient can reasonably expect. This explainer goes one level deeper than a glossary: why the system is built the way it is, where the numbers come from, and where the system breaks down.
The TNM scaffold and why three axes
TNM separates three anatomically distinct questions because each carries different prognostic weight and different treatment implications. The T category describes the primary tumor: its size, its depth of invasion into surrounding tissue, and whether it has grown into adjacent structures. The N category describes regional lymph nodes—whether cancer cells have reached the lymph nodes that drain the tumor's territory, and how many. The M category is binary in most systems: no distant spread (M0) or distant spread (M1).
The logic is that these are not interchangeable. A large tumor that has not breached its capsule behaves differently from a small tumor that has colonized a lymph node, and both differ from a tumor that has seeded the liver. Each axis is scored, then the combination is collapsed into an overall stage group. The American Joint Committee on Cancer (AJCC) publishes the staging manuals that most U.S. registries and hospitals use; the current edition is the eighth, effective for cases diagnosed since 2018.
T is usually numbered 1 through 4, with subcategories (T1a, T1b) that reflect finer distinctions. N runs from N0 to N3 depending on the cancer. Prefixes add clinical meaning: cTNM is clinical staging based on examination and imaging before treatment, pTNM is pathologic staging based on tissue removed at surgery, and ypTNM describes staging after neoadjuvant therapy. A lowercase "y" tells the reader the tumor has already been treated, which changes how the numbers should be interpreted.
From TNM to stage groups: the lookup table logic
Stage groups are not a formula. They are lookup tables specific to each cancer type, built from outcome data on large patient cohorts. The AJCC convenes site-specific expert panels that analyze survival data and assign stage groupings so that patients within a group have reasonably similar prognosis. That is why stage IIA in breast cancer and stage IIA in colon cancer are not the same disease and do not mean the same thing.
The grouping problem is combinatorial. In principle, dozens of T, N, and M combinations exist. The panels collapse them into four roman-numeral stages plus stage 0 (carcinoma in situ, where abnormal cells have not invaded beyond their origin layer). Some cancers add stage IVA and IVB to distinguish, for example, spread to a single distant organ from spread to multiple sites.
The eighth edition introduced a significant change: many cancer types now include biologic and molecular factors alongside anatomy. Prostate cancer staging incorporates the Gleason grade group and prostate-specific antigen (PSA) level. Breast cancer staging incorporates tumor grade, estrogen and progesterone receptor status, HER2 status, and multigene panel scores. The practical consequence is that two patients with identical T and N findings can receive different stage groups because their tumor biology differs.
Quantified context: what the numbers actually track
Stage is a proxy for tumor burden and biologic aggressiveness, and the survival gradients attached to it are steep. For most solid tumors, five-year relative survival for localized disease sits in the range of roughly 70 to 99 percent depending on the cancer type, falls to roughly 30 to 70 percent for regional disease, and drops to roughly 5 to 35 percent for distant disease. These ranges are wide because "cancer" is not one disease; the gradient within a single cancer type is what matters clinically.
The gradient is also why staging systems get revised. When a new edition is adopted, stage definitions shift, and survival statistics computed under the old edition are not directly comparable to those under the new one. Researchers refer to stage shifts when comparing trends across editions, because part of an apparent improvement in stage-specific survival can come from reclassification rather than from better treatment.
Node counts illustrate how granular staging has become. In colon cancer, the number of examined lymph nodes affects both N assignment and confidence in that assignment; pathology guidelines call for examination of at least 12 regional nodes to reliably distinguish N0 from node-positive disease. In breast cancer, sentinel lymph node biopsy samples the first draining node rather than all axillary nodes, and the size and number of involved nodes, plus whether involvement is microscopic or macroscopic, determine N subcategory.
How the stage is actually determined, step by step
Staging proceeds through a sequence that runs from least to most invasive. History and physical examination come first, then laboratory tests, then imaging selected for the cancer type—CT for many abdominal and thoracic tumors, MRI for brain and some pelvic tumors, PET-CT when distant spread is suspected, bone scan for cancers with a tropism for bone. Tissue diagnosis follows: a biopsy establishes the cancer type and grade, and for many cancers the biopsy specimen is also tested for molecular markers that now feed into staging.
Surgical staging comes last and is often the most definitive. When a patient goes to the operating room, the surgeon directly inspects and samples lymph nodes and nearby structures, and the pathologist examines everything removed. This is why cTNM and pTNM can diverge; imaging underestimates spread in a meaningful fraction of cases, particularly for small nodal deposits that fall below the resolution of CT or MRI.
The sequence has a counterintuitive feature. More imaging is not always better. Staging tests carry false-positive rates, and an ambiguous finding can trigger an invasive confirmatory procedure that turns out to be unnecessary. Guidelines therefore recommend staging tests selectively, matched to the cancer type and the probability of finding spread, rather than as a blanket panel.
Edge cases where the system strains
Not every cancer uses TNM. Leukemias are staged by different systems entirely, or not staged in the conventional sense, because they are diseases of blood and marrow rather than solid masses with a primary site. Lymphomas use the Ann Arbor system, which classifies by the number of lymphoid regions involved and whether disease is on one side or both sides of the diaphragm. Brain tumors are graded rather than staged, because gliomas rarely spread beyond the central nervous system and the grading reflects the tumor's biology and expected behavior.
Some cancers have staging systems that ignore size. In certain settings, the mere presence of distant spread defines stage IV regardless of how small the primary tumor is. In others, a cancer can be stage IV by virtue of spread without any identifiable primary tumor at all.
There is also the problem of restaging. A cancer that recurs after treatment is staged again, often under a separate schema, because the biology of a recurrence is not identical to the biology of the original presentation. And there is the problem of staging in patients who received treatment before surgery. The ypTNM system exists precisely because neoadjuvant therapy can shrink a tumor and sterilize lymph nodes, so the post-treatment pathology understates the original extent of disease.
The most important caveat is population-level. Stage-specific survival figures describe groups, not individuals. A stage III diagnosis with a favorable molecular profile and good performance status can behave better than a stage II diagnosis with adverse features. Clinicians use stage as one input among several, including age, comorbidity, functional status, and tumor biology.
Common misconceptions, corrected
The most persistent misconception is that stage equals outcome. Stage is a summary of anatomic extent at a moment in time; outcome depends on how the tumor responds to treatment and on its molecular drivers. A second misconception is that stage never changes. Staging is revised after surgery, after neoadjuvant therapy, and at recurrence. A third is that higher stage always means a different treatment category. Some stage III cancers are treated with surgery first; some stage II cancers receive chemotherapy before surgery.
Another misconception is that a lower stage is always better news. For some cancers, stage for stage, a smaller primary tumor with nodal spread does worse than a larger tumor without it—which is exactly why N and T are scored separately rather than combined into a single size measure.
Stage IV does not always mean incurable, and stage I does not guarantee cure. Staging tests are selected, not exhaustive; more scanning can produce false positives. A stage assigned at diagnosis may be revised after surgery or after neoadjuvant treatment. Survival statistics attached to a stage describe groups, never an individual. Different cancer types use different staging systems; roman numerals are not interchangeable across diseases. Recurrence is restaged under separate criteria and should not be compared directly to the original stage.
Comparison table: staging systems by cancer type
| Cancer type | Staging system | Key inputs beyond anatomy | Overall stages |
|---|---|---|---|
| Breast | AJCC TNM with biologic factors | Grade, ER/PR, HER2, multigene score | 0, I–IV |
| Colon and rectum | AJCC TNM | Depth of invasion, node count, metastases | 0, I–IV |
| Prostate | AJCC TNM plus grade group and PSA | Gleason grade group, PSA level | I–IV |
| Lung (non-small cell) | AJCC TNM | Tumor size, nodal station, distant spread | 0, I–IV |
| Lymphoma | Ann Arbor | Number of lymphoid regions, diaphragm side | I–IV |
| Brain glioma | WHO grading | Histology and molecular markers | Grade 1–4 (not stage) |
| Leukemia | Disease-specific (e.g., phase, risk) | Marrow and blood findings, genetics | Not conventional stage |
The table makes an important point: the word "stage" hides at least three different constructs. TNM groups anatomy and biology; Ann Arbor groups anatomic regions; WHO grading groups biologic behavior. Reading a stage without knowing which system produced it invites error.
Timeline table: how staging unfolds
| Timepoint | What happens | Staging label |
|---|---|---|
| Before any treatment | Examination, labs, imaging, biopsy | Clinical (cTNM) |
| Before neoadjuvant therapy | Multidisciplinary review confirms baseline stage | Clinical, documented |
| After neoadjuvant therapy, before surgery | Response assessed on imaging | Restaging (imaging-based) |
| At surgery | Nodes sampled, tumor removed | Pathologic (pTNM) |
| After neoadjuvant therapy and surgery | Post-treatment pathology reviewed | ypTNM |
| At recurrence | New imaging and biopsy as indicated | Restaging for recurrence |
| Long-term follow-up | Stage at diagnosis remains the reference | Original stage retained |
Practical actions and thresholds
For patients, the actionable steps are about documentation and questions rather than self-assessment. Ask for the stage in TNM notation, not only the roman numeral, because the TNM detail carries the information. Ask whether the stage is clinical or pathologic, and whether it was assigned before or after any treatment. Ask which edition of the staging manual was used, because comparisons across editions are unreliable.
Request the staging summary in writing, including T, N, M, and the stage group. Confirm whether the stage is clinical, pathologic, or post-neoadjuvant (y). Ask which imaging and biopsies supported the assignment. Ask whether molecular or grade factors modified the stage group. Ask what the stage implies for the recommended treatment sequence. Ask how the stage will be reassessed after treatment.
For clinicians and readers of the literature, the thresholds that matter are the ones embedded in the systems: 12 examined lymph nodes in colon cancer, sentinel node sampling in breast cancer, the distinction between microscopic and macroscopic nodal disease, and the grade group and PSA cut points in prostate cancer. These thresholds are not arbitrary; they were chosen because below them, staging error rates rise.
Limitations and what staging cannot tell you
Staging cannot predict response to a specific therapy. It was never designed to. It summarizes extent and, in modern editions, some biology, but it does not encode the full molecular profile that determines whether a targeted agent or immunotherapy will work. Two patients with the same stage can have completely different treatment trajectories.
Staging also cannot resolve the problem of lead-time bias. Earlier detection moves diagnosis to an earlier point in the disease course, which inflates survival time without necessarily extending life. Stage-specific survival improvements across eras therefore require careful interpretation.
Finally, staging systems lag behind science. Molecular classifiers are being developed for many cancers, and the AJCC has been incorporating them incrementally, but the roman-numeral framework remains fundamentally anatomic. The eighth edition's inclusion of biologic factors is a step toward a more biology-driven system, not its completion.
What staging does well is provide a shared, reproducible shorthand that lets a diagnosis be communicated, compared, and used to select among established treatment pathways. Read it as a map of extent at a point in time, not a verdict. The map is useful precisely because it is limited, and knowing its limits is what separates a working understanding from a thin one.