The pathologist's report says what every surgeon hopes to see. No tumor on ink. The sentinel nodes are both clear. By every measurement the surgeon can make, the cancer is gone. And yet a small fraction of women with exactly this pathology will recur — locally, or at a distant site — years later. Nothing was missed by neglect or carelessness. So the question becomes: what does a clear margin actually prove?

Every principle of cancer surgery exists to reduce the probability of residual disease through a different mechanism. Adequate margins address the resection edge. Nodal assessment establishes stage and may add therapeutic value. Tumor containment prevents seeding the surgical field. En bloc resection removes invaded structures as one piece rather than cutting through cancer. And anatomic planes — following embryologic boundaries — match the routes lymphatics actually use. All five principles aim at a single inference: did we leave any cancer behind?

The pathologist examines the inked surface of the removed specimen. She samples sections through that surface and reports whether tumor cells are present at the ink. What she cannot examine is the tissue still in the patient, immediately adjacent to where the specimen was removed. R0 is a statement about what was examined. The tissue a millimeter beyond the cut is inferred from outcomes data, not measured directly. This is why margin requirements are empirically calibrated: the no-tumor-on-ink standard for invasive breast cancer was established by randomized trials, not by cell-level geometry.

The failure mode of sentinel node biopsy is the failure mode of any sampling strategy: the sample may not represent the whole. If the tracer is blocked by prior surgery or inflammation, the node the surgeon removes may not be the true first node in the drainage pathway. A negative result in a node that is not the real sentinel misses disease that is actually present. This failure mode is uncommon when the technique is performed correctly — but it is not zero. This is exactly why long-term recurrence data, not just node pathology, had to validate the approach before it replaced full dissection.

A lymph node that contains cancer cells tells you something true about the patient's disease. That is the staging value, and it is real. The question of whether removing that positive node, and the rest of the basin, extends the patient's survival is entirely different. It was historically assumed that it did — the reasoning was intuitive: cancer in the nodes might spread further; remove the nodes and you remove that source of spread. For decades the positive sentinel node triggered completion dissection as reflex standard practice. But intuition is not evidence.

The MSLT-II trial tested the therapeutic assumption directly in melanoma. Patients with sentinel-node-positive melanoma were randomized to immediate completion lymph node dissection versus observation with ultrasound surveillance. The result was unambiguous: completion dissection did not improve melanoma-specific survival. It significantly increased lymphedema. The same lesson emerged in breast cancer. The ACOSOG Z0011 trial showed survival was equivalent whether or not the axilla was dissected. And AMAROS showed that axillary radiation achieved equivalent regional control with less morbidity than surgery.

These trials forced a conceptual separation that intuition had resisted. The nodes contained disease — that was true. Removing the nodes did not extend survival — that was also true. The two facts are compatible if the cancer cells in those nodes were already controlled by systemic therapy, or had already seeded distant micrometastatic disease before surgery. Bernard Fisher articulated the principle in the context of mastectomy versus breast conservation: surgery achieves local control, and local control is not the same as systemic cure. The ceiling on survival is set by micrometastatic biology — cells that have already left the primary site before any operation begins.

Bernard Fisher's NSABP B-06 trial randomized women with early breast cancer to radical mastectomy, total mastectomy, or lumpectomy plus radiation. Survival was equivalent across all three groups. The Halsted radical mastectomy, standard for eighty years, offered no survival advantage over removing only the tumor. This pattern has recurred across cancer types: pancreatic, bladder, head and neck cancer. The principle is not that surgery should be as limited as possible. It is that surgery should be as limited as the evidence supports. Minimizing morbidity is not a compromise with oncologic goals — it is, in the evidence, usually the way to achieve the same goals with less harm.