Introduction: Perineural invasion (PNI) is a recognized negative prognostic factor in oral squamous cell carcinoma (OSCC), associated with an increased risk of locoregional recurrence, lymph node metastasis, and reduced overall survival. However, PNI is a pathological feature that can only be confirmed after surgical resection. Objective: The aim of this study was to develop a predictive tool for PNI based on the integration of clinical, endoscopic (both white light and Narrow Band Imaging, NBI), and laboratory data, in order to estimate its likelihood preoperatively and guide more tailored therapeutic decisions. Methods: A multicenter retrospective analysis was conducted on 154 surgically treated OSCC patients. All patients underwent preoperative endoscopic assessment using white light and NBI. Collected variables included clinical data (pain, functional impairment, bleeding), endoscopic features (brownish spots, tortuous vessels, depapillation, growth pattern), laboratory parameters (Neutrophil-to-Lymphocyte Ratio, NLR), and histopathological findings. A multivariate logistic regression analysis was performed to identify independent predictors of PNI, followed by the development of a clinical nomogram and decision tree. Model validation was carried out using the ROC curve, calibration plot, and Nagelkerke R² index. Results: PNI was detected in 55.2% of cases. Independent predictors significantly associated with its presence included advanced T stage (OR = 8.05, p < 0.001), presence of brownish spots on NBI (OR = 13.71, p < 0.001), depapillation (OR = 3.33, p = 0.020), pain (OR = 3.00, p = 0.041), and G3 histological grading (OR = 9.08, p = 0.004). The vascular pattern classifications by Takano and Ni were not found to be significantly predictive. NLR showed predictive value in the subgroup of patients with advanced cT stage. The model demonstrated good discriminative ability, with an area under the ROC curve (AUC) of 0.894, sensitivity of 77.9%, specificity of 82.4%, positive predictive value of 77.9%, and negative predictive value of 82.4%. The calibration curve confirmed good agreement between predicted and observed probabilities. The Nagelkerke R² value of 0.595 indicated a solid explanatory power of the model. Conclusions: The proposed predictive model, based on the analysis of clinical and bioendoscopic variables, proved effective in identifying OSCC patients at high risk of PNI in the preoperative setting. Its application may support clinicians in refining surgical strategies, suggesting more aggressive treatment approaches even in clinically early-stage patients. The clinical implementation of this tool could enhance therapeutic planning, reduce recurrence risk, and optimize the balance between oncologic radicality and functional preservation.
Introduzione: l’invasione perineurale (Perineural Invasion, PNI) è un fattore prognostico negativo nel carcinoma squamoso del cavo orale (Oral Squamous Cell Carcinoma, OSCC), correlato a una maggiore incidenza di recidive locoregionali, metastasi linfonodali e ridotta sopravvivenza. Tuttavia, la PNI è un dato patologico che è possibile ottenere solamente in seguito a trattamento chirurgico. Scopo dello studio: questo studio si propone di sviluppare uno strumento predittivo di PNI basato sull’integrazione di dati clinici, endoscopici (sia a luce bianca che con tecnica Narrow Band Imaging, NBI) e laboratoristici, per stimarne la probabilità in fase preoperatoria e orientare più precisamente le decisioni terapeutiche. Metodi: è stata condotta un’analisi retrospettiva multicentrica di pazienti affetti da OSCC trattati chirurgicamente. Sono stati inclusi 154 pazienti. I pazienti sono stati sottoposti a valutazione endoscopica preoperatoria in luce bianca e narrow band imaging (NBI). I dati raccolti includevano informazioni cliniche (dolore, limitazione funzionale, sanguinamento), endoscopiche (brownish spots, tortuous vessels, depapillazione, pattern di crescita), laboratoristiche (Neutrophil-to-Lymphocyte Ratio, NLR) e istopatologiche. È stata costruita una regressione logistica multivariata per identificare i predittori indipendenti di PNI e successivamente sono stati sviluppati un nomogramma e un albero decisionale clinico. La validazione del modello è stata effettuata mediante curva ROC, curva di calibrazione e indice di Nagelkerke R². Risultati: La PNI è stata riscontrata nel 55,2% dei casi. I fattori più fortemente associati alla sua presenza sono risultati: stadio T avanzato (OR=8.05, p<0.001), presenza di “brownish spots” alla NBI (OR=13.71, p<0.001), depapillazione (OR=3.33, p=0.020), dolore (OR=3.76, p=0.019) e grading G3 (OR=9.08, p=0.004). Le classificazioni del pattern NBI secondo Takano e Ni non sono risultate significativamente predittive. Il modello ha mostrato buone performance discriminative: l’area sotto la curva ROC (AUC) è risultata pari a 0.894, con sensibilità del 77.9%, specificità dell’82.4%, valore predittivo positivo del 77.9% e valore predittivo negativo dell’82.4%. La curva di calibrazione ha confermato una buona coerenza tra probabilità predetta e osservata. Il valore di Nagelkerke R² pari a 0.595 indica una buona capacità del modello di spiegare la variabilità del fenomeno studiato. Conclusioni: Il modello predittivo proposto, basato sull’analisi di variabili cliniche e bioendoscopiche, si è dimostrato efficace nell’identificare i pazienti con elevato rischio di PNI. Questo approccio potrebbe supportare i professionisti nel personalizzare più precisamente la strategia chirurgica, suggerendo, nei casi ad alto rischio, l’adozione di trattamenti più aggressivi, anche nei pazienti in stadio clinico precoce. L’implementazione clinica di questi strumenti potrebbe migliorare la pianificazione terapeutica, ridurre il rischio di recidiva e ottimizzare il bilancio tra radicalità oncologica e conservazione funzionale.
Ruolo della bioendoscopia integrata a dati clinici e radiologici nella predizione dell’invasione perineurale nel carcinoma del cavo orale: uno studio multicentrico su 154 pazienti
MORETTI, FRANCESCO
2024/2025
Abstract
Introduction: Perineural invasion (PNI) is a recognized negative prognostic factor in oral squamous cell carcinoma (OSCC), associated with an increased risk of locoregional recurrence, lymph node metastasis, and reduced overall survival. However, PNI is a pathological feature that can only be confirmed after surgical resection. Objective: The aim of this study was to develop a predictive tool for PNI based on the integration of clinical, endoscopic (both white light and Narrow Band Imaging, NBI), and laboratory data, in order to estimate its likelihood preoperatively and guide more tailored therapeutic decisions. Methods: A multicenter retrospective analysis was conducted on 154 surgically treated OSCC patients. All patients underwent preoperative endoscopic assessment using white light and NBI. Collected variables included clinical data (pain, functional impairment, bleeding), endoscopic features (brownish spots, tortuous vessels, depapillation, growth pattern), laboratory parameters (Neutrophil-to-Lymphocyte Ratio, NLR), and histopathological findings. A multivariate logistic regression analysis was performed to identify independent predictors of PNI, followed by the development of a clinical nomogram and decision tree. Model validation was carried out using the ROC curve, calibration plot, and Nagelkerke R² index. Results: PNI was detected in 55.2% of cases. Independent predictors significantly associated with its presence included advanced T stage (OR = 8.05, p < 0.001), presence of brownish spots on NBI (OR = 13.71, p < 0.001), depapillation (OR = 3.33, p = 0.020), pain (OR = 3.00, p = 0.041), and G3 histological grading (OR = 9.08, p = 0.004). The vascular pattern classifications by Takano and Ni were not found to be significantly predictive. NLR showed predictive value in the subgroup of patients with advanced cT stage. The model demonstrated good discriminative ability, with an area under the ROC curve (AUC) of 0.894, sensitivity of 77.9%, specificity of 82.4%, positive predictive value of 77.9%, and negative predictive value of 82.4%. The calibration curve confirmed good agreement between predicted and observed probabilities. The Nagelkerke R² value of 0.595 indicated a solid explanatory power of the model. Conclusions: The proposed predictive model, based on the analysis of clinical and bioendoscopic variables, proved effective in identifying OSCC patients at high risk of PNI in the preoperative setting. Its application may support clinicians in refining surgical strategies, suggesting more aggressive treatment approaches even in clinically early-stage patients. The clinical implementation of this tool could enhance therapeutic planning, reduce recurrence risk, and optimize the balance between oncologic radicality and functional preservation.| File | Dimensione | Formato | |
|---|---|---|---|
|
Moretti_Francesco_1217411.pdf
accesso aperto
Dimensione
1.15 MB
Formato
Adobe PDF
|
1.15 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/87282