A group from The First Affiliated Human of Xi’an Jiaotong University, Xi’an, China etc. has reported new biomarkers from bronchial alveoli lavage to identify adenocarcinoma, squamous carcinomas, and small cell lung cancer.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7840895/
Lectin microarrays were used for the marker screening, and bronchoalveolar lavage fluid was used as a sample. The goal was to identify lung adenocarcinomas (ADC), lung squamous carcinomas (SCC), and small cell lung cancer (SCLC) from benign pulmonary diseases (BPD). The goal was also to distinguish between early stage lung cancer (LC-ES) and advanced stage long cancer (LC-AS).
From the results of the lectin microarray, significant differences were found in 15 lectins, and a logistic regression analysis was performed by combining several lectins to identify the differences between various lung cancers.
As a result
To discriminate lung cancer from good lung disease, ECA, GSL-I, and RCA120 were used (cutoff value: 0.754, ALC: 0.961, sensitivity: 0.918, specificity: 0.939),
to discriminate ADC, DBA, STL, UEA-I, BPL were used (cutoff value: 0.569, ALC: 0.619, sensitivity: 0.706, specificity: 0.586),
to discriminate SCC, PNA was used (cutoff value: 0.578, ALC: 0.693, sensitivity: 0.800, specificity: 0.667),
to discriminate SCLC, STL, BS-I, PTL-II, SBA, PSA were used (cutoff value: 0.728, ALC: 0.718, sensitivity: 0.721, specificity: 0.684),
and also
to detect early cancer, MAL-II, LTL, GSL-I, RCA120, PTL-II, PWM were used, resulting in (cutoff value: 0.668, AOC: 0.856, sensitivity: 0.829, specificity: 0.810).
In order to distinguish various diseases, it is advantageous not to rely on one variable (one lectin), but to distinguish the characteristic profile for each disease using multiple variables (multiple lectins). In my view, I think that it will become a trend to adopt AI instead of using conventional statistical analysis to improve discrimination ability of the differences in disease profiles.