PhD Scientific Days 2024

Theoretical and Translational Medicine II.

Predicting Immunotherapy Efficacy with Machine Learning in Gastrointestinal Cancers: a Systematic Review and Meta-Analysis

Előadó neve

Vágó-Szincsák, Sára, MSc

Neptun code

H1ZP61

Előadó munkahelye

Institute of Translational Medicine

Előadó telefonszáma

+36203335216

Előadó e-mail címe

vago.sara@phd.semmelweis.hu

Az előadás címe

Predicting Immunotherapy Efficacy with Machine Learning in Gastrointestinal Cancers: a Systematic Review and Meta-Analysis

Szerző(k) neve és munkahelye

Sara Szincsak¹, Peter Kiraly², Gabor Szegvari¹, David Dora³, Zoltan Lohinai⁴

1: Institute of Translational Medicine, Budapest, Hungary
2: Pulmonology Hospital of Torokbalint, Torokbalint, Hungary
3: Department of Anatomy, Histology, and Embryology, Semmelweis University,Budapest, Hungary
4: Institute of Translational Medicine, Semmelweis University, Budapest, Hungary

Bemutatás módja

Szóbeli

Szekció

Theoretical and Translational Medicine II.

Language of the presentation

English

Preferred session

Theoretical and Translational Medicine

Összefoglaló szövege

Introduction
Machine learning (ML) algorithms hold the potential to overperform the selection of patients for immunotherapy (ICI) compared to previous biomarker studies. We analyzed the predictive performance of ML models and compared them to traditional clinical biomarkers (TCB) in the field of gastrointestinal (GI) cancers.
Aims
Our aims are to collate studies on RNA signatures predictive of immunotherapy response using ML methods in GI cancers and to compare the predictive scores to standard clinical biomarkers.
Methods
A systematic search of PubMed was conducted to identify studies applying different ML algorithms to GI cancer patients treated with ICI using tumor RNA gene expression profiles and with outcomes response to immunotherapy or survival. We compared the ML methodology details and predictive power inherent in the published gene sets using 5-fold cross-validation and logistic regression (LR) on an available well-defined ICI-treated metastatic gastric cancer (GC) cohort (n=45). A set of standard clinical ICI biomarkers (MLH, MSH,CD8 genes, PMS2 and PD-L1)) and de-novo calculated principal components (PC) of the original datasets were also included as additional points of comparison.
Results
Nine articles were identified as eligible to meet the inclusion criteria. There were five GC, one colorectal and three pan-cancer studies. Classification and regression models were used to predict ICI efficacy. We validated the predictive power of applied ML algorithms on RNA signatures, using their reported receiver operating characteristics (ROC) analysis area under the curve (AUC) values on the validation GC dataset. In two cases our method has outperformed the published results (reported/LR comparison: 0.74/0.831, 0.67/0.735). Besides the published studies, we have included two benchmarks: a set of TCB and using principal components based on the whole dataset (PCA, 99% explained variance, 40 components). Interestingly, a study using a selected gene set (immuno-oncology panel) with AUC=0.83 was the only one that overperformed the TCB (AUC=0.8) and the PCA (AUC =0.81) results.
Conclusion
We found an immuno-oncology panel with an AUC=0.83 that overperformed the clinical benchmark or the PC results.
Funding
Semmelweis 250+ Excellence PhD Scholarship (EFOP‐3.6.3‐VEKOP‐16‐2017‐00009)
Hungarian National Research, Development and Innovation Office ((#146775)

University

Semmelweis University

Supervisor

Lohinai Zoltan MD, PhD

Publication of my abstract

I do not give consent to the publication of my abstract on the website of the congress.

Kind

Szabad

Status

elfogadva

Accepted presentation method

szóbeli

Előadás fájl jóváhagyás

nem rendelkezett róla

Előadó

7486

Start

12:15

End

12:25