CAN THE EXTRACTED CHEMICAL INFORMATION FROM FFPE SAMPLES USING LA-REIMS IMAGING SUPPORT PATHOLOGICAL DIAGNOSIS?

Significance of the Topic

The use of formalin-fixed paraffin-embedded (FFPE) tissue blocks represents a cornerstone in clinical pathology and research archives worldwide. Recovering molecular information from these stored samples enables retrospective studies and biomarker discovery, but chemical alterations during FFPE processing often hamper analytical insights. Laser-assisted rapid evaporative ionization mass spectrometry (LA-REIMS) imaging offers a minimal-preparation approach to map phospholipid distributions directly on tissue sections, potentially bridging histopathology and metabolic profiling.

Objectives and Study Overview

• Assess the feasibility of LA-REIMS imaging to extract phospholipid signals (600–900 m/z) from human kidney FFPE sections with clear cell renal carcinoma.
• Compare chemical contrasts between healthy and tumorous regions against standard pathological annotations.
• Evaluate classification performance via multivariate statistics and supervised image classification.

Methodology and Instrumentation

• Tissue Samples: Ten 10 µm FFPE kidney sections annotated for tumor and healthy regions.
• LA-REIMS Imaging: Boxed setup with 2940 nm optical parametric oscillator, motorized XYZ stage, aerosol suction to REIMS source on a Xevo G2 QTof-MS (negative mode).
• Data Processing: PCA for spectral overview, unsupervised k-nearest neighbors (kNN) clustering, and supervised linear support vector classifier (LSVC) via custom Abstract Model Builder (AMX) software. Cross-validation performed by full group-out.

Main Results and Discussion

FFPE preparation induced a marked drop in signal intensity and metabolic complexity, particularly affecting PE(38:4) and PE(36:2) peaks. Despite reduced signal-to-noise, LA-REIMS imaging produced clear chemical contrast in regions with sufficient lipid retention. Unsupervised kNN clustering closely matched histological boundaries, while LSVC models based on four key phospholipid peaks (699.5, 701.5, 723.5, 885.5 m/z) achieved classification rates above 90% in six of eight usable samples. Variability in FFPE quality led to outlier cases with negligible signal.

Benefits and Practical Applications

LA-REIMS imaging on FFPE tissue enables rapid, label-free chemical mapping without additional sample preparation. Combined with multivariate classification, it supports in situ differentiation of tumorous versus healthy tissue, offering a complementary tool for pathology and biomarker research.

Future Trends and Potential Applications

• Integration of signal amplification or alternative fixation protocols to enhance lipid detection from archived samples.
• Expansion to other tissue types and molecular classes beyond phospholipids.
• Development of automated workflows for clinical decision support and high-throughput retrospective studies.

Conclusion

LA-REIMS imaging of FFPE kidney sections demonstrates that meaningful phospholipid patterns can be retrieved for pathological classification despite processing-induced signal loss. With optimized sample handling and advanced statistical models, this approach holds promise for augmenting histopathology with spatially resolved metabolic information.

References

• Waters Corporation. Fully Automated Chemical Imaging with LA-REIMS. Poster, 2023.
• Waters Corporation. DESI And/or LA-REIMS? Adjacent Automated Ambient Techniques for the Precise Identification of Cancer Tissue. Poster, 2023.