Pattern recognition of hematological profiles of tumors of the digestive tract: an exploratory study

Frontiers in Medicine

Medicine

Marina Majar¹  Miguel Machado¹  Miguel A. Santos-Silva²  Joao C. Sousa²  Joana Reis³  Nuno Sousa⁴ 

[1] Clinical Pathology Department, Hospital de Braga, Braga, Portugal;Life and Health Sciences Research Institute (ICVS), School of Medicine, University of Minho, Campus de Gualtar, Braga, Portugal;ICVS/3B's – PT Government Associate Laboratory, Braga/Guimarães, Portugal;Life and Health Sciences Research Institute (ICVS), School of Medicine, University of Minho, Campus de Gualtar, Braga, Portugal;ICVS/3B's – PT Government Associate Laboratory, Braga/Guimarães, Portugal;Clinical Academic Center-Braga (2CA), Braga, Portugal;Life and Health Sciences Research Institute (ICVS), School of Medicine, University of Minho, Campus de Gualtar, Braga, Portugal;ICVS/3B's – PT Government Associate Laboratory, Braga/Guimarães, Portugal;Clinical Academic Center-Braga (2CA), Braga, Portugal;Association P5 Digital Medical Center (ACMP5), Braga, Portugal;

关键词: cancer;    pattern recognition;    hematological profile;    anatomical location;    digestive tract tumor;   

DOI  :  10.3389/fmed.2023.1208022

 received in 2023-04-18, accepted in 2023-07-31,  发布年份 2023

来源: Frontiers

【 摘 要 】

AimsIn this study, we aimed to apply laboratory blood analysis to identify the hematological (based on hemoglobin concentration, erythrocytes, hematocrit, and RDW count) profiles associated with the most prevalent forms of digestive tract malignancies. Furthermore, we aimed to evaluate how these profiles contributed to distinguishing these tumors at diagnosis.MethodsWe collected data from the date of ICD-10 diagnostic coding for C15 esophagus, C16 stomach, C18 colon, and C19 rectum tumors of 184 individuals. The statistical analysis and data visualization approaches, notably the heat map and principal component analysis (PCA), allowed for creating a summary hematological profile and identifying the most associated parameters for each pathologic state. Univariate and multivariate data modeling and ROC analysis were performed in both SPSS and Python.ResultsOur data reveal unique patterns based on tumor development anatomical location, clustering the C18 colon and C19 rectum from the C15 esophagus and C16 stomach. We found a significant difference between C16 stomach carcinoma and the other tumors, which substantially correlated with raised RDW in conjunction with low hemoglobin concentration, erythrocytes, and hematocrit counts. In contrast, C18 colon carcinoma had the higher red blood cell count, allowing for the best classification metrics in the test set of the binary logistic regression (LR) model, accounting for an AUC of 0.77 with 94% sensitivity and 52% specificity.ConclusionThis study emphasizes the significance of adding hematological patterns in diagnosing these malignancies, which could path further investigations regarding profiling and monitoring at the point of care.

【 授权许可】

Unknown   
Copyright © 2023 Santos-Silva, Sousa, Majar, Machado, Reis and Sousa.