For Ki67 in a small bowel neuroendocrine tumour, stain the largest mesenteric deposit

Design

retrospective cohort with digital image analysis of hotspot Ki67, manually validated in 20 cases, multivariate Cox regression

Population

75 jejunoileal well-differentiated neuroendocrine tumours with nodal metastasis or mesenteric deposit; mean age 61

Primary outcome

progression, and disease-specific death, by Ki67 index in primary, node and deposit

Effect

deposit Ki67 HR 1.22 per 1% (95% CI 1.02-1.46); primary and nodal Ki67 not significant; node or deposit exceeded primary in 68%

Grading well-differentiated neuroendocrine tumours turns on the Ki67 index, and in jejunoileal tumours it has never been clear which tissue to stain: these are often multifocal with nodal metastases and mesenteric deposits, and convention has been to use the primary or the largest focus. Seventy-five jejunoileal neuroendocrine tumours with at least one nodal metastasis or mesenteric deposit were stained in the largest primary, the largest node and the largest deposit, with hotspot areas quantified by digital image analysis and 20 cases manually counted for validation.

In 51 cases (68%) a node or a deposit had a higher Ki67 than the primary — so the primary systematically understates the index. On multivariate analysis only the tumour deposit Ki67 predicted progression: hazard ratio 1.22 for each 1% increase (95% CI 1.02-1.46, P = 0.032). Primary Ki67 did not predict it, nodal Ki67 did not, and neither did the highest Ki67 in the case nor the Ki67 of the largest focus — the two conventions most likely to be in use. For overall survival no Ki67 index was significant, but the size of the largest tumour deposit was (HR 1.51 per 1 cm, 1.08-2.11).

The change is specific and cheap: when a resection of a jejunoileal neuroendocrine tumour includes a mesenteric tumour deposit, take the Ki67 block from the largest deposit. It costs nothing beyond block selection, and it replaces a convention this study found to be uninformative with one that was prognostic.

Two cautions belong in the report. The grade derived this way may be higher than the grade the primary would have given, which has treatment implications the multidisciplinary team should see rather than discover. And the study is a single retrospective cohort with six disease-specific deaths — enough to change which block you take, not enough to change a grading threshold.

• Take the Ki67 block from the largest mesenteric tumour deposit when a jejunoileal neuroendocrine tumour resection contains one
• State in the report which tissue the Ki67 index was measured in — primary, node or deposit
• Expect a deposit or nodal index higher than the primary; this happened in about two-thirds of cases
• Record the size of the largest mesenteric deposit, which predicted overall survival here
• Do not substitute the highest Ki67 in the case or the largest focus — neither predicted progression

Why it matters

The block a pathologist happens to choose for Ki67 has been determining the grade, and in two-thirds of cases the primary reads lower than the metastasis.

Don't overread it

A single retrospective cohort with 21 recurrences and six disease-specific deaths supports changing block selection, not changing grading thresholds.

The statistics, in plain English

A hazard ratio of 1.22 per 1% rise in Ki67, with an interval from 1.02 to 1.46, only just clears 1.0 — the direction is secure, the magnitude is not, and the finding rests on 21 recurrences. That the primary and nodal indices were not significant does not prove they carry no information; with this number of events, only the strongest predictor is likely to survive multivariate analysis. Because several Ki67 measures were tested in the same cohort, the one that emerged could partly reflect that multiplicity, which is why the authors' framing is about tissue selection rather than about a new threshold.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.