Diagnosis of invasive encapsulated follicular variant papillary thyroid carcinoma by protein-based machine learning

Aug 3, 2026

Upload by Levi

Diagnosis of invasive encapsulated follicular variant papillary thyroid carcinoma remains challenging despite established criteria for follicular-pattern thyroid tumors and a recent World Health Organization reclassification. A supervised machine-learning approach was developed using proteomics from 46 thyroid tissue samples, followed by performance testing. A random forest classifier using five protein biomarkers distinguished the invasive encapsulated follicular variant from non-malignant samples with perfect training performance (AUC 1.00; accuracy 1.00). Protein-level receiver operating characteristic analyses supported differentiation across related groups, integrating high-throughput proteomics with machine learning.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Background

Methods

Results

Conclusions

Keywords

pISSN 2383-7837 ∙ eISSN 2383-7845 ORIGINAL ARTICLE Journal of Pathology and Translational Medicine 2025; 59: 39-49 [https://doi.org/10.4132/jptm.2024.09.14](https://doi.org/10.4132/jptm.2024.09.14) Diagnosis of invasive encapsulated follicular variant papillary thyroid carcinoma by protein-based machine learning Truong Phan-Xuan Nguyen1, Minh-Khang Le2, Sittiruk Roytrakul3, Shanop Shuangshoti1,4, Nakarin Kitkumthorn5, Somboon Keelawat1,6 Graphical abstract 􀀑ĂĐŬŐƌŽƵŶĚ DĞƚŚŽĚƐ • dŚǇƌŽŝĚ ĐĂŶĐĞƌ ŝƐ ƚŚĞ ŵŽƐƚ ĐŽŵŵŽŶĞŶĚŽĐƌŝŶĞ ŵĂůŝŐŶĂŶĐǇ͘ • ZĞĐĞŶƚ ĐůĂƐƐŝĨŝĐĂƚŝŽŶƐ ĞŵƉŚĂƐŝǌĞ ƚŚĞ ŶĞĞĚĨŽƌ ĂĐĐƵƌĂƚĞ ĚŝĨĨĞƌĞŶƚŝĂƚŝŽŶ ŽĨ ŝŶǀĂƐŝǀĞĞŶĐĂƉƐƵůĂƚĞĚ ĨŽůůŝĐƵůĂƌ ǀĂƌŝĂŶƚ ƉĂƉŝůůĂƌǇƚŚǇƌŽŝĚ ĐĂƌĐŝŶŽŵĂ ;/􀀜&sWd􀀒Ϳ ĨƌŽŵ ŽƚŚĞƌƚŚǇƌŽŝĚ ƚƵŵŽƌƐ͘ dŽ ĚĞǀĞůŽƉ Ă ŵĂĐŚŝŶĞͲůĞĂƌŶŝŶŐ ŵŽĚĞů ƵƐŝŶŐ ƉƌŽƚĞŽŵŝĐƐ ĚĂƚĂ ƚŽŝĚĞŶƚŝĨǇ /􀀜&sWd􀀒 ĂĐĐƵƌĂƚĞůǇ • hƚŝůŝǌĞĚ ůŝƋƵŝĚ ĐŚƌŽŵĂƚŽŐƌĂƉŚǇ ͲƚĂŶĚĞŵ ŵĂƐƐ ƐƉĞĐƚƌŽŵĞƚƌǇ ;>􀀒 Ͳ D^ͬD^Ϳ ĨŽƌ ƉƌŽƚĞŽŵŝĐ ĂŶĂůǇƐŝƐ͘ • ZĞƐƵůƚƐ ZK􀀒 ĂŶĂůǇƐĞƐ 􀀄 dƌĂŝŶ ƐĂŵƉůĞƐ ;ŶсϯϲͿ ϭ͘ϬϬ Ϭ͘ϳϱ Ϭ͘ϱϬ Ϭ͘Ϯϱ Ϭ͘ϬϬϬ͘ϬϬ Ϭ͘Ϯϱ Ϭ͘ϱϬ Ϭ͘ϳϱ ϭ͘ϬϬ ϭ Ͳ ^ƉĞĐŝĨŝĐŝƚǇ ^ĞŶƐŝƚŝǀŝƚǇ 􀀑 dĞƐƚ ƐĂŵƉůĞƐ ;ŶсϭϬͿ Ϭ͘ϬϬ Ϭ͘Ϯϱ Ϭ͘ϱϬ Ϭ͘ϳϱ ϭ͘ϬϬ ϭ Ͳ ^ƉĞĐŝĨŝĐŝƚǇ Calibration ƉůŽƚƐ 􀀒 dƌĂŝŶ ƐĂŵƉůĞƐ ;ŶсϯϲͿ 􀀘 dĞƐƚ ƐĂŵƉůĞƐ ;ŶсϭϬͿ 􀀄ĐƚƵĂů WƌŽďĂďŝůŝƚǇ ϭ͘ϬϬ Ϭ͘ϳϱ Ϭ͘ϱϬ Ϭ͘Ϯϱ Ϭ͘ϬϬ • KƵƌ ŵŽĚĞů ƐŚŽǁĞĚŚŝŐŚ ZK􀀒 􀀄h􀀒Ɛ сϭϬϬĂŶĚ ŐŽŽĚ ĐĂůŝďƌĂƚŝŽŶƐŝŶ ƚŚĞ ƚǁŽ ŐƌŽƵƉƐ;ƚŚĞ ƚƌĂŝŶŝŶŐ ĂŶĚŝŶƚĞƌŶĂů ƚĞƐƚͿ͘ 􀀄 ƚŽƚĂů ŽĨ ϰϲ ƚŚǇƌŽŝĚ ƚŝƐƐƵĞƐĂŵƉůĞƐ ǁĞƌĞ ƐƚƵĚŝĞĚ͕ ŝŶĐůƵĚŝŶŐďŽƚŚ /􀀜&sWd􀀒 ĂŶĚ ŶŽŶͲ/􀀜&sWd􀀒ƐƉĞĐŝŵĞŶƐ͘ • 􀀄 ZĂŶĚŽŵ &ŽƌĞƐƚ ĐůĂƐƐŝĨŝĞƌ ƵƐŝŶŐĨŝǀĞ ŬĞǇ ƉƌŽƚĞŝŶ ďŝŽŵĂƌŬĞƌƐ ;􀁿􀀜􀀑ϭ͕ EhWϵϴ͕ 􀀒Ϯ􀀒Ϯ>͕ EW􀀄Wϭ͕ <􀀒E:ϯͿ ǁĂƐĚĞǀĞůŽƉĞĚ ĨŽƌ ĚŝĂŐŶŽƐŝƐ͘ Confusion ŵĂƚƌŝĐĞƐ 􀀜 ZĞĨĞƌĞŶĐĞ ŶŽŶ Ͳ / 􀀜 &sWd􀀒 /􀀜&sWd􀀒 dƌĂŝŶ ƐĂŵƉůĞƐ ;ŶсϯϲͿ & Ϭ Ϯϲ ϭϬ Ϭ ZĞĨĞƌĞŶĐĞ ŶŽŶ Ͳ / 􀀜 &sWd􀀒 /􀀜&sWd􀀒 ŶŽŶͲ/􀀜&sWd􀀒 /􀀜&sWd􀀒 WƌĞĚŝĐƚŝŽŶ dĞƐƚ ƐĂŵƉůĞƐ ;ŶсϭϬͿ ϭ Ϯ ϳ Ϭ ŶŽŶͲ/􀀜&sWd􀀒 /􀀜&sWd􀀒 WƌĞĚŝĐƚŝŽŶ • dŚĞ ŵŽĚĞů ĂĐĐƵƌĂƚĞůǇƉƌĞĚŝĐƚĞĚ Ăůů ƚŚĞƐĂŵƉůĞƐ ǁŝƚŚŽƵƚ ĂŶǇĞƌƌŽƌƐ ŝŶ ƚƌĂŝŶŝŶŐƐĂŵƉůĞƐ ĂŶĚ ĂůŵŽƐƚ ĂůůƚŚĞ ĐĂƐĞƐ ŝŶ ƚĞƐƚŝŶŐƐĂŵƉůĞƐ͘ 􀀒KE􀀒>h^/KE^ dŚŝƐ ƐƚƵĚǇ ĚĞŵŽŶƐƚƌĂƚĞĚ ƚŚĂƚ ƚŚĞ ŝŶƚĞŐƌĂƚŝŽŶ ŽĨ ŚŝŐŚͲƚŚƌŽƵŐŚƉƵƚ ƉƌŽƚĞŽŵŝĐƐ ǁŝƚŚ ŵĂĐŚŝŶĞ ůĞĂƌŶŝŶŐ ĐĂŶ ĞĨĨĞĐƚŝǀĞůǇ ĚŝĨĨĞƌĞŶƚŝĂƚĞŝŶǀĂƐŝǀĞ ĞŶĐĂƉƐƵůĂƚĞĚ ĨŽůůŝĐƵůĂƌ ǀĂƌŝĂŶƚ ŽĨ ƉĂƉŝůůĂƌǇ ƚŚǇƌŽŝĚ ĐĂƌĐŝŶŽŵĂ ĨƌŽŵ ŽƚŚĞƌ ĨŽůůŝĐƵůĂƌ ƉĂƚƚĞƌŶ ƚŚǇƌŽŝĚ ƚƵŵŽƌƐ͘ EŐƵǇĞŶ d Ğƚ Ăů͘ :ŽƵƌŶĂů ŽĨ WĂƚŚŽůŽŐǇ ĂŶĚ dƌĂŶƐůĂƚŝŽŶĂů DĞĚŝĐŝŶĞ pISSN 2383-7837 ∙ eISSN 2383-7845 ORIGINAL ARTICLE Journal of Pathology and Translational Medicine 2025; 59: 39-49 [https://doi.org/10.4132/jptm.2024.09.14](https://doi.org/10.4132/jptm.2024.09.14) Diagnosis of invasive encapsulated follicular variant papillary thyroid carcinoma by protein-based machine learning Truong Phan-Xuan Nguyen1, Minh-Khang Le2, Sittiruk Roytrakul3, Shanop Shuangshoti1,4, Nakarin Kitkumthorn5, Somboon Keelawat1,6 1Department of Pathology, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand 2Department of Pathology, University of Yamanashi, Chuo City, Japan 3Functional Proteomics Technology Laboratory, National Center for Genetic Engineering and Biotechnology, National Science and Technology Development Agency, Pathumthani, Thailand 4Chulalongkorn GenePRO Center, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand 5Department of Oral Biology, Faculty of Dentistry, Mahidol University, Bangkok, Thailand 6Precision Pathology of Neoplasia Research Group, Department of Pathology, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand Background: Although the criteria for follicular-pattern thyroid tumors are well-established, diagnosing these lesions remains challenging in some cases. In the recent World Health Organization Classification of Endocrine and Neuroendocrine Tumors (5th edition), the invasive encapsulated follicular variant of papillary thyroid carcinoma was reclassified as its own entity. It is crucial to differentiate this variant of papillary thyroid carcinoma from low-risk follicular pattern tumors due to their shared morphological characteristics. Proteomics holds significant promis

FAQs

Why is diagnosing invasive encapsulated follicular variant papillary thyroid carcinoma challenging?

The follicular-pattern criteria are established, but distinguishing this invasive encapsulated variant from low-risk follicular pattern tumors is difficult due to shared morphological characteristics.

What machine-learning approach and biomarker set were used in the study?

The study used a supervised model and a random forest classifier based on five protein biomarkers: ZEB1, NUP98, C2C2L, NPAP1, and KCNJ3.

How well did the model perform in distinguishing the target carcinoma?

In training samples, the classifier achieved AUCs of 1.00 and accuracy rates of 1.00 for separating the invasive encapsulated follicular variant from non-malignant samples.

Aug 3, 2026

Upload by Levi