Integrating MALDI-MSI-Based Spatial Proteomics and Machine Learning to Predict Chemoradiotherapy Outcomes in Head and
Marta Grzeski1, Patrick Moeller Jensen2, Benjamin-Florian Hempel1,3
1Imaging Mass Spectrometry Unit, Berlin Institute of Health, Charité - Universitätsmedizin Berlin, 13353 Berlin, Germany.
Researchers identified peptide signatures in head and neck squamous cell carcinoma (HNSCC) tissue linked to treatment outcomes. This MALDI-MSI proteomic approach aids risk stratification for HPV-negative HNSCC patients receiving platinum-based chemoradiotherapy.
Area of Science:
- Oncology
- Proteomics
- Medical Diagnostics
Background:
- Head and neck squamous cell carcinoma (HNSCC) often presents at advanced stages, posing challenges for risk stratification and treatment response prediction due to intratumoral heterogeneity (ITH).
- Accurate prognostication is crucial for tailoring treatment strategies in HPV-negative HNSCC patients undergoing chemoradiotherapy.
Purpose of the Study:
- To identify peptide signatures in HNSCC tissue associated with treatment outcomes in HPV-negative, advanced-stage patients receiving 5-fluorouracil/platinum-based chemoradiotherapy (CDDP-CRT).
- To evaluate the potential of matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) coupled with machine learning for prognostic profiling.
Main Methods:
- Integration of MALDI-MSI of tryptic peptides from formalin-fixed, paraffin-embedded HNSCC tumor sections with univariate statistics and machine learning.
- Development and testing of classification models using peptide profiles from 31 HPV-negative HNSCC patients treated with CDDP-CRT, with validation on an independent cohort treated with mitomycin C-based CRT (MMC-CRT).
Main Results:
- Classification models achieved balanced accuracies of 71% (unrestricted) and 72% (feature-restricted) in predicting recurrence or progression (RecPro) versus no evidence of disease (NED) in the CDDP-CRT cohort.
- The models demonstrated specificity for platinum-based therapy, showing no prognostic performance in the MMC-CRT cohort.
- The feature-restricted model yielded higher specificity (92%) but lower sensitivity (52%) compared to the unrestricted model.
Conclusions:
- MALDI-MSI-based proteomic profiling can identify patients at higher risk of recurrence after CDDP-CRT.
- This proteomic approach shows potential for personalized risk assessment and treatment planning in HPV-negative HNSCC.
- Findings suggest that peptide signatures are specific to platinum-based chemoradiotherapy regimens.
More Related Videos
07:15Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
07:13Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025