Design: Systematic review without meta-analysis
Population: 359 studies of pulmonary carcinoid and large cell neuroendocrine carcinoma
Primary outcome: Use, method and concordance of central pathology review
Effect: Central review in 29.5%; median disagreement 17.3% (range 6.3–36.7%)

Classifying lung neuroendocrine tumours — typical and atypical carcinoid, large cell neuroendocrine carcinoma — is hard, especially on small biopsies. This systematic review examined 359 studies of pulmonary carcinoid and large cell neuroendocrine carcinoma for how central pathology review was used.

Only 106 (30%) described central review at all. The number of reviewing pathologists ranged from one to 19, and consensus and disagreement procedures varied widely. Just 21 studies reported concordance: unanimous diagnosis had a median of 59%, disagreement a median of 17%, and kappa values ranged from 0.21 to 0.98.

The authors propose a structured reporting form. For practice, the message is that a substantial fraction of these diagnoses are contestable even among experts, and that matters when the distinction between atypical carcinoid and large cell neuroendocrine carcinoma drives treatment.

Count mitoses and assess necrosis on the whole section, and record both in the report.
Use Ki-67 as support, not as the defining criterion, for carcinoid grading.
Be cautious grading lung neuroendocrine tumours on small biopsies; say so in the report.
Seek a second opinion when the atypical carcinoid versus large cell neuroendocrine carcinoma distinction changes treatment.

Why it matters

Research on lung neuroendocrine tumours, and the treatment decisions built on it, may rest on diagnoses experts would not agree on.

The statistics, in plain English

A median disagreement of 17% means about one in six cases was classified differently by different experts. Kappa ranges from 0 (chance agreement) to 1 (perfect); 0.21 is poor and 0.98 near-perfect, showing how inconsistent the reported agreement was.