2023 World Conference on Lung Cancer (Posters)-P1.20. Improved Diagnosis of Lung Neuroendocrine Tumors: Role of Proliferative Activity and Deep Learning

Pdf Summary

The study conducted by Mathian et al. aimed to assess the current and emerging criteria for the histopathological classification of lung neuroendocrine tumors (LNETs), specifically focusing on the limitations and potential improvements in classification. The researchers utilized a large cohort of over 300 patients diagnosed with LNETs and employed deep learning algorithms to analyze whole slide images (WSI) and measure the expression of Ki-67 and Phospho-histone H3 (PHH3) proteins.<br /><br />The results of the study indicated that the manual and automatic assessments of Ki-67 could identify two distinct groups of LNETs with different prognostic values. Ki-67 was found to detect rare aggressive cases among patients diagnosed as typical carcinoids (TC). On the other hand, PHH3 did not help identify a subgroup of TC with a higher risk of relapse and did not significantly improve the concordance of diagnoses between readers.<br /><br />The researchers also highlighted the limitations of the current morphological criteria for the classification of LNETs, particularly the poor reproducibility of mitotic figure counts. However, they found that applying a wide range of thresholds to manual or automatic measurements of Ki-67 or PHH3 could successfully divide LNETs into two groups with prognoses similar to the current classification.<br /><br />While the current morphological classification of LNETs is moderately reproducible, it has limitations in predicting patients' response to treatment. The study suggests that emerging molecular markers, such as CD44 or OTP, could complement the current morphological criteria to better predict aggressiveness and identify potential successful treatments. Further studies using multi-omic data analysis are needed to better understand the biology of LNETs and explore new avenues for treatment development.<br /><br />This research was supported by various funding sources, including the Partner University Fund, the French Research Agency, and the Neuroendocrine Tumor Research Foundation. The study demonstrated the potential of deep learning algorithms and molecular markers to improve the classification and prognosis of LNETs, paving the way for further advancements in the field.

Asset Subtitle

Sylvie Lantuejoul

Meta Tag

Speaker Sylvie Lantuejoul

Topic Pathology & Biomarkers: Artificial Intelligence in Pathology

Keywords

histopathological classification

lung neuroendocrine tumors

LNETs

limitations

potential improvements

deep learning algorithms

whole slide images

Ki-67

Phospho-histone H3

proteins