Development of machine learning models for the prediction of long-term feeding tube dependence after oral and oropharyngeal cancer surgery.

PURPOSE: To predict the necessity of enteral nutrition at 28 days after surgery in patients undergoing major head and neck oncologic procedures for oral and oropharyngeal cancers. MATERIAL AND METHODS: Data from 193 patients with oral cavity and oropharyngeal squamous cell carcinoma were retrospectively collected at two tertiary referral centers to train (n = 135) and validate (n = 58) six supervised machine learning (ML) models for binary prediction employing 29 clinical variables available pre-operatively. RESULTS: The accuracy of the six ML models ranged between 0.74 and 0.88, while the measured area under the curve (AUC) between 0.75 and 0.87. The ML algorithms showed high specificity (range 0.87-0.96) and moderate sensitivity (range: 0.31-0.77) in detecting patients with ≥28 days feeding tube dependence. Negative predictive value was higher (range: 0.81-0.93) compared to positive predictive value (range: 0.40-0.71). Finally, the F1 score ranged between 0.35 and 0.74. CONCLUSIONS: Classification performance of the ML algorithms showed optimistic accuracy in the prediction of enteral nutrition at 28 days after surgery. Prospective studies are mandatory to define the clinical benefit of a ML-based pre-operative prediction of a personalized nutrition protocol.

Predicting the need for prophylactic gastrostomy in major head and neck cancer surgery: The PEG score.

BACKGROUND: To elaborate a preoperative score to predict the necessity of enteral nutrition at 28 days postoperatively in patients undergoing head and neck surgery. METHODS: A total of 424 patients with oral cavity, oropharyngeal, laryngeal, and hypopharyngeal carcinoma were retrospectively enrolled and analyzed to identify preoperative predictors of prolonged postsurgical enteral feeding which were used to create a prediction model with an easy-to-use nomogram. RESULTS: Five preoperative variables (body mass index, previous radiotherapy, preoperative dysphagia, type of surgery, flap reconstruction) were found to be independent predictive factors and were used to create a prediction model named PEG score together with the related nomogram. Accuracy, F1, and the area under the curve (AUC) were 0.74, 0.83, and 0.74. Different decision thresholds can be used to vary the sensitivity and specificity. CONCLUSIONS: The PEG score showed high prediction performances for modeling the need for enteral nutrition at 28 days postoperatively. Prospective studies are needed to define a personalized nutrition protocol.

Can High-Frequency Intraoral Ultrasound Predict Histological Risk Factors in Oral Squamous Cell Carcinoma? A Preliminary Experience.

Despite advancements in multidisciplinary care, oncologic outcomes of oral cavity squamous cell carcinoma (OSCC) have not substantially improved: still, one-third of patients affected by stage I and II can develop locoregional recurrences. Imaging plays a pivotal role in preoperative staging of OSCC, providing depth of invasion (DOI) measurements. However, locoregional recurrences have a strong association with adverse histopathological factors not included in the staging system, and any imaging features linked to them have been lacking. In this study, the possibility to predict histological risk factors in OSCC with high-frequency intraoral ultrasonography (IOUS) was evaluated. Thirty-four patients were enrolled. The agreement between ultrasonographic and pathological DOI was evaluated, and ultrasonographic margins' appearance was compared to the Brandwein-Gensler score and the worst pattern of invasion (WPOI). Excellent agreement between ultrasonographic and pathological DOI was found (mean difference: 0.2 mm). A significant relationship was found between ultrasonographic morphology of the front of infiltration and both Brandwein-Gensler score ≥ 3 (p < 0.0001) and WPOI ≥4 (p = 0.0001). Sensitivity, specificity, positive predictive value, and negative predictive value for the IOUS to predict a Brandwein-Gensler score ≥3 were 93.33%, 89.47%, 87.50%, and 94.44%, respectively. The present study demonstrated the promising role of IOUS in aiding risk stratification for OSCC patients.

Immunonutrition in major oncologic head and neck surgery: Analysis of complications, plasmatic equilibrium, and costs.

BACKGROUND: Malnutrition, in patients with solid tumors, is associated with a worse clinical outcome and about 40% of patients affected by head and neck cancers (HNC) are malnourished at the time of cancer diagnosis. We investigated the potential benefit of a standardized immunonutritional protocol (INP) to patients with HNC receiving major ablative surgery. METHODS: An observational study was conducted enrolling 199 patients: 50 treated with the INP and 149 with standard enteral nutrition. Complication rates, need for medications, and costs were considered as outcomes. RESULTS: INP played a protective role in development of major surgical complications (OR 0.23, p = 0.023), albumin administration (RR 0.38, p = 0.018), and antibiotic duration (p < 0.001) and is cost-effective in patients with moderate or severe malnutrition (-6083€ and -11 988€, p < 0.05). CONCLUSIONS: Our study supports the utility of INP, and accurate nutritional screening can help to identify malnourished patients who would receive the most benefits from this protocol.

Videomics of the Upper Aero-Digestive Tract Cancer: Deep Learning Applied to White Light and Narrow Band Imaging for Automatic Segmentation of Endoscopic Images.

Introduction: Narrow Band Imaging (NBI) is an endoscopic visualization technique useful for upper aero-digestive tract (UADT) cancer detection and margins evaluation. However, NBI analysis is strongly operator-dependent and requires high expertise, thus limiting its wider implementation. Recently, artificial intelligence (AI) has demonstrated potential for applications in UADT videoendoscopy. Among AI methods, deep learning algorithms, and especially convolutional neural networks (CNNs), are particularly suitable for delineating cancers on videoendoscopy. This study is aimed to develop a CNN for automatic semantic segmentation of UADT cancer on endoscopic images. Materials and Methods: A dataset of white light and NBI videoframes of laryngeal squamous cell carcinoma (LSCC) was collected and manually annotated. A novel DL segmentation model (SegMENT) was designed. SegMENT relies on DeepLabV3+ CNN architecture, modified using Xception as a backbone and incorporating ensemble features from other CNNs. The performance of SegMENT was compared to state-of-the-art CNNs (UNet, ResUNet, and DeepLabv3). SegMENT was then validated on two external datasets of NBI images of oropharyngeal (OPSCC) and oral cavity SCC (OSCC) obtained from a previously published study. The impact of in-domain transfer learning through an ensemble technique was evaluated on the external datasets. Results: 219 LSCC patients were retrospectively included in the study. A total of 683 videoframes composed the LSCC dataset, while the external validation cohorts of OPSCC and OCSCC contained 116 and 102 images. On the LSCC dataset, SegMENT outperformed the other DL models, obtaining the following median values: 0.68 intersection over union (IoU), 0.81 dice similarity coefficient (DSC), 0.95 recall, 0.78 precision, 0.97 accuracy. For the OCSCC and OPSCC datasets, results were superior compared to previously published data: the median performance metrics were, respectively, improved as follows: DSC=10.3% and 11.9%, recall=15.0% and 5.1%, precision=17.0% and 14.7%, accuracy=4.1% and 10.3%. Conclusion: SegMENT achieved promising performances, showing that automatic tumor segmentation in endoscopic images is feasible even within the highly heterogeneous and complex UADT environment. SegMENT outperformed the previously published results on the external validation cohorts. The model demonstrated potential for improved detection of early tumors, more precise biopsies, and better selection of resection margins.