In modern oncology, the selection of an effective therapy for a patient still presents significant challenges due to the complexity and variability of the disease. In spite of the efforts, innovation and continuous improvements, the success rates of therapeutic choices remain suboptimal in many cases; the ability to monitor treatment response early in time and predict accurately patient-specific outcomes represents therefore an urgent and unmet need in clinical practice. Conventional approaches for therapy monitoring, such as imaging techniques, are often limited by late feedback, while immunohistochemistry for evaluation of tumor microenvironment lacks of quantitative references and threshold for resolutive assessments. In this framework, Raman spectroscopy emerges as a promising technique, offering label- free, non-destructive, and chemically specific insights into biological samples. In particular, this thesis investigates the use of spontaneous Raman spectroscopy applied to patient-derived head and neck squamous cell carcinoma (HNSCC) slice cultures. A home-built Raman microspectroscopy system was employed to acquire hyperspectral maps of HNSCC samples subjected to different treatment conditions on different timepoints. A dedicated preprocessing workflow was designed, together with custom Python tools for background subtraction, artifact identification and pixel-wise annotation of maps. Statistical analyses were subsequently performed using both univariate and multivariate methods. The results demonstrate that Raman spectroscopy is able to capture biochemical differences stemming from different tissue composition or associated with therapeutic interven- tions. By means of principal component analysis, providing programmed death ligand 1 (PDL1) protein was found to induce detectable modifications in DNA and protein-related Raman bands, while clustering approaches highlighted the possibility of discriminating tumor, tumor stroma, skeletal muscle and other healthy tissue structures across patients. These findings confirm the validity of Raman spectroscopy as a tool for the characterization ex-vivo patient-derived slice cultures, with the potential of revealing insights also on changes produced by medical treatment occurring in time.
Nell’oncologia moderna, la selezione di una terapia efficace per i pazienti presenta sfide rilevanti a causa della complessità e della variabilità della malattia. Nonostante le innovazioni e i continui progressi, i tassi di successo rimangono in molti casi non ottimali; la capacità di monitorare tempestivamente la risposta al trattamento e prevedere con precisione gli esiti specifici per un paziente rappresenta quindi una necessità urgente e ancora insoddisfatta in ambito clinico. Gli approcci convenzionali per il monitoraggio di terapia, come le tecniche di imaging, sono spesso limitati da feedback tardivi, mentre l’immunoistochimica per la valutazione del microambiente tumorale manca di riferimenti quantitativi e soglie per valutazioni risolutive. In questo contesto, la spettroscopia Raman emerge come una tecnica promettente, caratterizzata da specificità chimica, non distruttività e assenza di label per marcare i campioni biologici. In particolare, questa tesi indaga l’uso della spettroscopia Raman spontanea applicata a colture di sezioni di carcinoma squamoso della testa e del collo (HNSCC) derivate da paziente. Un sistema di microscopia Raman sviluppato ad-hoc è stato utilizzato per acquisire mappe iperspettrali di campioni HNSCC sottoposti a diverse tipologie di trattamento e a diversi istanti temporali. È stata progettata una pipeline di preprocessing, insieme a codici Python personalizzati per la sottrazione del background, l’identificazione di artefatti e l’annotazione pixel-per-pixel delle mappe. Successivamente sono state eseguite analisi statistiche con approcci univariati e multivariati. I risultati dimostrano che la spettroscopia Raman è in grado di cogliere differenze biochimiche derivanti dalla diversa composizione tissutale o associate a interventi terapeutici. Tramite l’analisi delle componenti principali , è stato osservato che il trattamento con proteina PDL1 induce cambiamenti nelle bande Raman relative a DNA e proteine, mentre approcci di clustering hanno evidenziato la possibilità di discriminare tumore, stroma tumorale, muscolo scheletrico e altre strutture tissutali sane tra i diversi pazienti. Questi risultati confermano la validità della spettroscopia Raman come strumento per la caratterizzazione di colture di sezioni derivate da pazienti, con il potenziale di rivelare anche cambiamenti nel tempo indotti da un eventuale trattamento.
Raman imaging of patient-derived tumour slice cultures: towards personalized therapy monitoring
Agrimi, Chiara
2024/2025
Abstract
In modern oncology, the selection of an effective therapy for a patient still presents significant challenges due to the complexity and variability of the disease. In spite of the efforts, innovation and continuous improvements, the success rates of therapeutic choices remain suboptimal in many cases; the ability to monitor treatment response early in time and predict accurately patient-specific outcomes represents therefore an urgent and unmet need in clinical practice. Conventional approaches for therapy monitoring, such as imaging techniques, are often limited by late feedback, while immunohistochemistry for evaluation of tumor microenvironment lacks of quantitative references and threshold for resolutive assessments. In this framework, Raman spectroscopy emerges as a promising technique, offering label- free, non-destructive, and chemically specific insights into biological samples. In particular, this thesis investigates the use of spontaneous Raman spectroscopy applied to patient-derived head and neck squamous cell carcinoma (HNSCC) slice cultures. A home-built Raman microspectroscopy system was employed to acquire hyperspectral maps of HNSCC samples subjected to different treatment conditions on different timepoints. A dedicated preprocessing workflow was designed, together with custom Python tools for background subtraction, artifact identification and pixel-wise annotation of maps. Statistical analyses were subsequently performed using both univariate and multivariate methods. The results demonstrate that Raman spectroscopy is able to capture biochemical differences stemming from different tissue composition or associated with therapeutic interven- tions. By means of principal component analysis, providing programmed death ligand 1 (PDL1) protein was found to induce detectable modifications in DNA and protein-related Raman bands, while clustering approaches highlighted the possibility of discriminating tumor, tumor stroma, skeletal muscle and other healthy tissue structures across patients. These findings confirm the validity of Raman spectroscopy as a tool for the characterization ex-vivo patient-derived slice cultures, with the potential of revealing insights also on changes produced by medical treatment occurring in time.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_10_Agrimi_Tesi_01.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
11.37 MB
Formato
Adobe PDF
|
11.37 MB | Adobe PDF | Visualizza/Apri |
|
2025_10_Agrimi_Executive Summary_02.pdf
accessibile in internet per tutti
Descrizione: Executive summary
Dimensione
968.45 kB
Formato
Adobe PDF
|
968.45 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/243968