The Three-Dimensional Hormonal Profile in Patients with PitNETs
Description
This dataset contains clinical and pathological information from 1096 surgical patients with PitNETs (Pituitary Neuroendocrine Tumors). It includes the following fields: patient number, admission date, discharge date, pathology report version (categorized as pre-2017 or 2017/2022 versions), gender, age at admission, pathological classification according to WHO 2022, hormone pattern reflected in the discharge diagnosis, hormone pattern reflected in the pathological report, and hormone pattern reflected in the ICD-10 coding. This dataset is suitable for analyzing the epidemiological characteristics of PitNETs, the consistency between pathological classification and hormone expression, the accuracy of diagnostic coding, and the classification differences between the different pathology versions.
Files
Steps to reproduce
Pituitary neuroendocrine tumours were characterized using three diagnostic methods: (1) clinical diagnosis, established by physicians based on a comprehensive evaluation of tumor characteristics and endocrine function at the time of discharge, prior to the availability of pathological confirmation; (2) standardized classification on the basis of ICD coding systems; and (3) definitive pathological subtyping according to the 2017 or 2022 WHO criteria, incorporating transcription factor expression and hormone immunohistochemical profiles. For comparative analysis, the patients were reclassified into seven hormonal profile groups (nonfunctioning/no hormone, PRL, GH, ACTH, TSH, multihormonal (MULTI), and LH/FSH) across all three diagnostic approaches. The pathological diagnoses were realigned to the seven groups to ensure consistency: null cell tumours were categorized as nonfunctioning, while mammosomatotroph tumours, mature/immature plurihormonal PIT1-lineage tumours, mixed somatotroph-lactotroph tumours, and plurihormonal tumours without distinct lineages were consolidated into the multihormonal group. Similarly, mixed-lineage cases were assigned to the multihormonal group. Notably, three acidophil stem cell tumour cases did not conform to any of the seven predefined groups, while all other pathological diagnoses could be mapped to one of these seven hormonal profile groups. All statistical analyses were performed using R software (version 4.3.0). Using the final clinical diagnosis from the discharge summary as the gold standard, we calculated diagnostic accuracy metrics—including sensitivity, specificity, positive and negative predictive values (PPVs and NPVs) with 95% confidence intervals (CIs)—for the ICD-10 code combinations assigned to identify nonfunctioning and functioning PitNETs. The Youden index and F1 score were also computed to provide a more comprehensive assessment of each coding strategy. Inter-rater agreement between clinical diagnoses and ICD-10 codes was evaluated using Cohen’s kappa (κ) statistic. To examine the association between evolving pathological classification standards and ICD-10 coding accuracy, diagnostic accuracy metrics were analyzed separately according to two pathology reporting styles influenced by the updated WHO classification for pituitary tumours. Group comparisons were performed using chi-square tests (with and without continuity correction) and Fisher’s exact test to assess differences in code performance metrics.