Better scans can make cancer survival statistics improve without anyone living longer
Doctors grade how far a cancer has spread from stage I, a tumour confined to one spot, to stage IV, where it has reached distant organs. That number steers treatment. But when scanners become sharper, patients get shuffled between stages, and the averages for every group can shift even though nobody's outcome has changed.
Staging weighs a tumour's size, whether it has invaded neighbouring organs, how many nearby lymph nodes are involved, and whether it has seeded elsewhere. The dominant scheme for solid tumours is TNM, for tumour, node and metastasis, whose universally accepted version comes from the UICC and shares its definitions with the American AJCC. A small c or p before the code marks whether it is clinical, based on exams, imaging and biopsies before surgery, or pathologic, based on a pathologist examining the removed tissue under a microscope.
The two often disagree. Pathologic staging is generally considered more accurate because the whole tumour can be inspected, yet it has its own blind spots: a single cancerous cell among healthy ones can be missed, and only sampled slices of lymph nodes are examined. Chemotherapy or radiation given before an operation can shrink a tumour so that the later pathologic stage understates the original. And many tumours are never removed, so a pathologic stage is not always available.
Not every cancer fits one mould. Most brain tumours are graded rather than staged. Colon cancer was long described with the Dukes letters A to D, melanoma adds measures of how deeply it has penetrated the skin, testicular cancer folds in blood markers, and Hodgkin lymphoma appends A or B according to symptoms such as fever. Stage 0 means abnormal cells still sitting in their original place. For early breast and prostate cancers with low risk of spread, specialist bodies advise against routine PET, CT or bone scans, judging that the harms of extra testing outweigh the benefits.
Changes to the rules or the technology cause stage migration, which underlies the statistical oddity called the Will Rogers phenomenon. Automation is creeping in: a Western Australian registry project built rule-based language software that staged 87 to 90 per cent of melanoma, breast and colorectal cases the same way as trained staff.
Source: Cancer staging
