OTP, CD44 and Ki-67 on the biopsy identified 17 of 19 lung carcinoid relapses; WHO identified 2

Design

Retrospective registry cohort with matched biopsy and resection specimens, three-pathologist revision

Population

98 patients with stage I–III lung neuroendocrine tumours undergoing curative resection, 2003–2021

Primary outcome

Identification of relapse risk on preoperative biopsy, and inter-observer agreement

Effect

Panel flagged 17/19 relapses vs 2/19 for WHO; NPV 0.96 vs 0.82; κ 0.673 vs 0.276

Lung neuroendocrine tumours are classified on resection into typical and atypical carcinoid, but the decision that increasingly matters — whether a sublobar resection is adequate — is taken before the resection exists. That forces the classification onto a preoperative biopsy, which is where it performs worst.

Ninety-eight patients with stage I–III tumours who went on to curative resection were drawn from the Dutch pathology registry, with matched biopsy and resection specimens, and three pathologists revised every case to WHO 2021. Nineteen relapsed over a median 83 months. A biomarker panel scored on the biopsy — OTP H-score at least 50, CD44 H-score at least 30, and Ki-67 proliferation index under 5% defining low risk — flagged 17 of those 19 as high risk. WHO classification of the same biopsies flagged 2. Negative predictive value was 0.96 against 0.82.

The reproducibility figures are as important as the prognostic ones. Inter-observer agreement was κ 0.673 for the panel and 0.276 for WHO, and biopsy-to-resection concordance was 0.584 against 0.169. The panel is not merely more predictive; it is more stable between observers and between specimens, which is what makes it usable in a real department rather than a study.

• The low-risk definition is all three together: OTP ≥50, CD44 ≥30, Ki-67 <5%
• Ki-67 was scored by eyeball hot-spot assessment, not image analysis — the result does not require digital pathology
• The intended use is preoperative, to inform whether sublobar resection is reasonable
• A high-risk panel result is not a diagnosis of atypical carcinoid; keep the WHO category in the report
• Three antibodies is a modest addition where neuroendocrine work-up is already being done

Why it matters

It says the classification we use to counsel these patients preoperatively misses almost every relapse it is being asked to predict.

Don't overread it

Retrospective, 98 patients and 19 events, from a single national registry. A negative predictive value of 0.96 rests on those 19 relapses, and the panel has not been tested prospectively against a resection-guided decision.

The statistics, in plain English

Nineteen events is few, so the 89% versus 11% contrast is built on small numbers and its confidence interval would be wide — but the gap is so large that imprecision is unlikely to reverse it. The kappa figures are the more robust comparison, because every case contributes rather than only the relapses: 0.673 is substantial agreement, 0.276 is fair, and the difference reached P < 0.001.

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.