Leveraging Artificial Intelligence (AI) Clinical Decision Support Software to Improve Treatment Plan Quality in Head and Neck Cancer Patients

Prospective Study of the Impact of Artificial Intelligence Organ-at-risk Dose Prediction on Head and Neck Radiotherapy Plan Quality

Introduction:Assessment of skeletal muscle function (SMF) is of clinical relevance in the prediction of treatment outcome and to decide on optimal management of head & neck cancer (HNC) patients. Handgrip strength (HGS) & handgrip endurance (HGE) are considered as surrogate marker for whole-body skeletal muscle function. Further, SMF depends substantially on the body composition (BC). Hence in this study, we compared BC, HGS and HGE between HNC patients and healthy controls and also analysed the association of HGS, HGE with body composition in HNC patients.Methods:A cross-sectional study, conducted in 44 subjects in the age between 18 to 60 years. Twenty-two were histologically proven HNC patients prior to cancer-specific treatment and twenty-two age and gender-matched healthy volunteers. The parameters recorded were Height, weight, waist circumference, hip circumference, HGS, HGE and BC. Hand-held dynamometer was used to measure HGS and HGE measured using a stopwatch. BC was estimated by whole-body bioelectrical Impedance analysis method using Bodystat Quad scan 4000 device.Result:Comparison of data between HNC patient & healthy control was done by Student's t test. HGS, HGE, lean body mass (LBM), fat-free mass index (FFMI), Phase angle (PA), body cell mass (BCM) and body cell mass index (BCMI) were found to be reduced significantly in HNC patients when compared to healthy subjects. Further, Pearson correlation analysis revealed a significant positive correlation of HGS & HGE with LBM, FFMI, PA, BCM & BCMI, whereas body fat mass index showed a negative correlation with HGS & HGE in HNC patients.Conclusion:Our findings revealed, a significant reduction in HGS, HGE in patient with HNC which denotes decreased skeletal muscle function and it is linearly associated with low muscle mass, body cell mass and phase angle.

Introduction: Cancer-related fatigue is physical, emotional, or cognitive exhaustion which can affect treatment adherence and quality of life, and predict survival. Head and neck cancer (HNC) patients experience high levels of fatigue due to the degree of radiotherapy (RT), but that alone does not fully explain it. Recent studies suggest that fatigue may be related to a patient's personalized metabolic and inflammatory response, suggesting a patient's gene expression (GE) may be used to predict and monitor fatigue - a precursor to managing it. In this analysis, we examined whether GE is predictive of pre-RT patient reported fatigue using cross-validated penalized Lasso regression - a machine learning approach. Study population: From Emory University Clinics, 44 HNC patients donated blood samples before undergoing RT. GE was assessed using an Affymetrix Clariom S Human microarray which measures gene transcripts for roughly 24,000 genes; each probe was log-transformed, normalized, and standardized. The validated 20-item self-report multidimensional fatigue inventory questionnaire measured each patient's continuous fatigue score. Methods: To predict fatigue, we used leave-one-out cross validation (CV). This means we built a Lasso regression model using GE probes from 43 of 44 subjects and used that model to predict fatigue for the remaining subject; this process was repeated 44 times until all subjects had a predicted fatigue score. We chose penalized Lasso regression because the ‘penalty' performs variable selection and mitigates collinearity between the GE probes; the penalty was chosen by an extra layer of CV not described. We compared the predicted fatigue scores to the corresponding patient reported fatigue scores using R2 (higher values mean stronger correlation and better prediction). Results: To test the approach, we allowed the Lasso regression to build a model based on all subjects and predict fatigue on all subjects. This prediction was expected to be high, and it was (R2=0.98). However, the approach was less successful predicting fatigue during the leave-one-out CV prediction (R2=0.15). The probes most influential predicting fatigue are linked to genes involved in tryptophan metabolism (precursor to the neurotransmitter serotonin), double strand break DNA repair, and transforming growth factor beta receptor signaling (inflammation cytokine). Conclusions: Gene expression may predict fatigue in head and neck cancer patients, but there is room for improvement. The model suggests that a patient's personalized DNA repair process, metabolic and inflammatory response may play key roles in patient fatigue. Integrating more data focused on these biological processes, for instance patient metabolomics, may improve the model performance. Future efforts include collecting a larger sample, trying alternative machine learning methods, and testing the robustness of the model over time. Citation Format: Ronald C. Eldridge, Andrew H. Miller, Deborah W. Bruner, Jonathan J. Beitler, Kristin A. Higgins, Evanthia C. Wommack, Linh Kha Huynh, Nabil F. Saba, Dong M. Shin, Canhua Xiao. Predicting fatigue levels of head and neck cancer patients with gene expression using machine learning [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4257.

e24020 Background: The Vulnerable Elders Survey (VES-13) is one of several tools that can identify older patients who are vulnerable (if the score is ≥ 3 out of 10) and more likely to benefit from Comprehensive Geriatric Assessment (CGA) prior to cancer treatment. The optimal cutpoint of the VES-13 to identify those whose final oncologic treatment plan would change after CGA is unclear. We hypothesized that patients with high positive scores (7-10) will have a higher likelihood of a change in the final oncologic treatment plan compared to low positive patients (score 3-6). Methods: Retrospective review of a customized database of all patients seen for pre-treatment assessment (solid tumor and lymphoma) in the geriatric oncology clinic at the Princess Margaret Cancer Centre from June 2015 to June 2019. Various VES-13 score cutpoints were compared with the final treatment plan to identify those individuals whose treatment was modified after CGA. Area under the curve was calculated and subgroups of patients treated locally or systemically were also examined to determine if performance varied by type of patient. Results: 386 patients with mean age 81, 58% males were included. Gastrointestinal cancer was the most common site 31% and 60% were planned to receive curative treatment. The final treatment plan was modified in 50% with VES-13 scores 7-10, 46.7% with scores 3-6 and 26.8 % for scores < 3 (P = 0.002; Table). The optimal VES-13 cutoff was between 3-6 (C-statistics 0.57-0.59).The VES-13 performed similarly in those considering local treatment (surgery with/without radiation) vs. chemotherapy. Modified final treatment for local therapy with VES-13 scores < 3 was 11.4 % compared to 42.9% with scores ≥ 3 was 42.9% (p-value < 0.001), whereas for systemic therapy it was 32.4% and 62.5%, respectively (p-value = 0.002). Conclusions: Although high positive VES-13 scores (7-10) had slightly higher likelihood of having the final oncologic treatment plan modified, there was no strong advantage compared to the conventional cutpoint of 3 or higher. The VES-13 performed similarly in predicting treatment change after CGA for local and systemic treatment plans. Further studies are required to identify the optimal frailty screening tool and cutpoint. [Table: see text]

The aims of this study were to identify the factors and reasons impacting discordance with the treatment plan in head and neck cancer (HNC) patients and compare the differences between the concordance group and the discordance group. This secondary analysis was conducted from population-based data from Taiwan collected from January 1, 2016, to June 30, 2018. Logistic regression analysis was used to identify the factors related to discordance with the treatment plan. We examined 1095 HNC patients, 12.1% of whom were discordant with treatment. Patients with advanced cancer stage, old age, and treatment plans of best supportive care (BSC) or surgery combined with radiation (RT), chemotherapy (CT), or chemoradiation (CCRT) were more likely to have discordance with their treatment plan. Of the 133 patients who were discordant with their treatment plan, the top reasons were as follows: "patients or their family considered patients' poor physical condition (chronic disease or unstable systemic disease), difficulty in enduring any condition likely to cause physical discomfort from disease treatment," "inconvenient transportation," and "disease progression." Patients' cancer stage, age, and types of treatment plans recommended significantly influenced discordance with treatment plan. Poor physical condition was the major reason for discordance with the treatment plan. Patients in the concordance group were significantly more likely than those in the discordance group to be younger than 65years, have less advanced cancer stage, and be recommended to receive surgery rather than any other regimen. Multidisciplinary team care can help patients make positive decisions about treatment.

BackgroundRadiotherapy (RT) is frequently used in the treatment of head and neck cancer, but different side-effects are frequently reported, including a higher frequency of radiation-related caries, what may be consequence of direct radiation to dental tissue. The intensity-modulated radiotherapy (IMRT) was developed to improve tumor control and decrease patient’s morbidity by delivering radiation beams only to tumor shapes and sparing normal tissue. However, teeth are usually not included in IMRT plannings and the real efficacy of IMRT in the dental context has not been addressed. Therefore, the aim of this study is to assess whether IMRT delivers lower radiation doses to dental structures than conformal 3D radiotherapy (3DRT).Material and methodsRadiation dose delivery to dental structures of 80 patients treated for head and neck cancers (oral cavity, tongue, nasopharynx and oropharynx) with IMRT (40 patients) and 3DRT (40 patients) were assessed by individually contouring tooth crowns on patients’ treatment plans. Clinicopathological data were retrieved from patients’ medical files.ResultsThe average dose of radiation to teeth delivered by IMRT was significantly lower than with 3DRT (p = 0.007); however, only patients affected by nasopharynx and oral cavity cancers demonstrated significantly lower doses with IMRT (p = 0.012 and p = 0.011, respectively). Molars received more radiation with both 3DRT and IMRT, but the latter delivered significantly lower radiation in this group of teeth (p < 0.001), whereas no significant difference was found for the other dental groups. Maxillary teeth received lower doses than mandibular teeth, but only IMRT delivered significantly lower doses (p = 0.011 and p = 0.003). Ipsilateral teeth received higher doses than contralateral teeth with both techniques and IMRT delivered significantly lower radiation than 3DRT for contralateral dental structures (p < 0.001).ConclusionIMRT delivered lower radiation doses to teeth than 3DRT, but only for some groups of patients and teeth, suggesting that this decrease was more likely due to the protection of other high risk organs, and was not enough to remove teeth from the zone of high risk for radiogenic disturbance (>30Gy).

Comparison of different treatment planning approaches using VMAT for head and neck cancer patients with metallic dental fillings

FDG-PET/MR is a hybrid imaging modality used for the staging and restaging of advanced head & neck cancer (HNC) patients. Their treatment typically involves radiation therapy, which requires previous dental focus assessment. The aim of this study was to analyze if staging FDG-PET/MR is a valuable tool for oral focus assessment. For this purpose, FDG-PET/MR findings, such as metabolic activity of periapical radiolucencies and marginal periodontitis, were retrospectively compared with conventional standardized dental focus assessment, including dental radiographs and clinical assessment of 124 teeth in seven patients. Increased FDG uptake of periapical lesions was found in one out of 23 lesions. Increased FDG uptake of the marginal periodontium was recorded in one out of 34 lesions. In summary, standardized dental focus assessment by panoramic radiography and periapical radiographs may be enriched by information from FDG-PET/MR, showing active inflammation in dental foci. However, many dental foci have no correlate in FDG-PET/MR. The treatment decision for oral foci may benefit from the visualized presence or absence of metabolic activity on FDG-PET/MR.

Authoritative research demonstrating efficacy of traditional dysphagia therapy for Head & Neck Cancer (HNC) patients is limited. A 2019 survey reported speech-language-pathologists (SLPs) have started using Manual Therapy (MT) to prevent or rehabilitate dysphagia in HNC patients. This application of MT is supported theoretically but no research has established efficacy. Further, specific contents of MT protocols employed in this setting remain unknown. In the absence of HNC dysphagia specific MT protocols, this study aimed to better understand MT protocols employed by SLPs to prevent and treat dysphagia in HNC patients during and after Radiation Therapy (RT). An internet-based questionnaire for SLPs who use MT with HNC patients was developed and tested for face/content validity. It was sent to SLPs practicing in the USA, twice, through three national listservs (ASHA-SIG13, ASHA-SIG3, University of Iowa Voiceserv). Of 64 respondents, 44 completed the survey. Of the 44, 15(34%) provided proactive MT during RT, 37(84%) provided proactive MT after RT (to prevent dysphagia), and 44(100%) provided reactive MT after RT (to treat dysphagia). 40(91%) were trained in MT through a CE course and 25(57%) had HNC-specific MT training. The most common MT techniques were laryngeal manipulation (LM) and myofascial release (MFR). During RT, MT protocols are gentler and highly tailored, with simple home programs of mild intensity. After RT, protocols are more regimented and aggressive, but still highly customized, with more diverse home programs of at least moderate intensity. MT for HNC patients lacks a standard protocol or approach, but MFR and LM, or components of those techniques, are used most frequently. Given the frequency with which MFR and LM are employed to treat dysphagia during and post-RT, and the lack of empirical evidence supporting or refuting their use, a collaboratively designed RCT is warranted to establish the safety and efficacy of MT for HNC patients.

IntroductionHigh quality diagnostic imaging can provide increased diagnostic accuracy and help guide medical decision-making and management, however challenges for radiology in resource-limited settings are numerous. Diagnostic imaging and teleradiology have financial and logistical implications, so evidence of impact is crucial. We sought to test the hypothesis that the implementation of computed radiography with teleradiology consultation support will significantly change diagnoses and treatment plans in a resource limited setting.MethodPaired before-after study to determine the therapeutic impact of an add-on diagnostic test. ‘Preliminary Plan’ and ‘Final Plan’ forms allowed direct comparison of diagnosis and treatment plans at initial consultation and following radiography and teleradiology. Consecutive consenting patients were included until the sample size (600) was reached. Changes in both diagnosis and treatment plan were analysed in the whole cohort, with sub-analyses of children aged <5 years, and cases of chest radiography.ResultsFinal analysis included 536 cases. Diagnosis changed following radiography and teleradiology in 62% of cases, and treatment plans changed in 61%. In chest radiography cases, 70% of diagnoses and 62% of treatment plans changed, while in children <5 years 66% of diagnoses and 58% of treatment plans changed.Reduced final treatment plans were most common for exploratory surgery (72% decrease), surgical orthopaedic intervention (62% decrease), and TB treatment (52% decrease), allowing more conservative medical or surgical management in 61 cases. Increased final treatment plans were highest in the orthopaedic and interventional surgery and referral categories. Of 42 cases requiring interventional surgery in the final plan, 26 (62%) were identified only after radiography and teleradiology. 16 additional cases were indicated for orthopaedic surgery, 10 cases required patient transfer, and TB treatment was indicated in 45 cases. A change in the original prescription plan occurred in 41% of 536 cases, with one or more prescriptions stopped in 28% of all cases.ConclusionWe found that computed radiography with teleradiology had significant clinical value in this resource-limited setting, with the potential to affect both patient outcomes and treatment costs through providing improved diagnostics and avoiding unnecessary treatments and medications.

BackgroundHead and neck cancer (HNC) patients undergoing radiotherapy (RT) may experience anatomical changes during treatment, which can compromise the validity of the initial treatment plan, necessitating replanning. However, ad hoc replanning disrupts clinical workflows and increases workload. Currently, no standardized method exists to quantify anatomical variation that necessitates replanning.PurposeThis project aimed to create geometrical metrics to describe anatomical changes in HNC patients during RT. The usefulness of these metrics was evaluated by a univariate analysis and through machine learning (ML) models to predict the need for replanning.MethodsA cohort of 150 HNC patients treated at McGill University Health Centre was analyzed. Based on the shapes of the RT structures (body, PTV, mandible, neck, and submandibular contours), we developed 43 metrics and automatically calculated them through a Python pipeline that we called HNGeoNatomyX. Univariate analysis using linear regression was conducted to obtain the rate of change of each metric. We also obtained the relative variation of each metric between the pre‐treatment and replanning‐requested scans. Fraction‐specific ML models (incorporating information available up to and including the specific fraction) for fractions 5, 10, and 15 were built using metrics, clinical data, and feature selection techniques. Model performance was estimated with repeated stratified 5‐fold cross‐validation resampling technique and the area under the curve (AUC) of the receiver operating characteristic (ROC) curve.ResultsUnivariate analysis showed that body‐ and neck‐related metrics were most predictive of replanning need. Our best specific multivariate models for fractions 5, 10, and 15 yielded testing scores of 0.82, 0.70, and 0.79, respectively. Our models early predicted replanning for 76% of the true positives.ConclusionsThe created metrics have the potential to characterize and distinguish which patients will necessitate RT replanning. They show promise in guiding clinicians to evaluate RT replanning for HNC patients and streamline workflows.

NTCP Modeling of Late Effects for Head and Neck Cancer: A Systematic Review.

Introduction Although post radiotherapy (RT) otological complications are complex and common, there is lack of data on its severity and spectrum. We intend to assess the incidence and severity of such ototoxicities in head neck cancer (HNCA) patients. Materials and Methods One year prospective study was conducted on thirty cases of HNCA receiving radiotherapy. Audiometry and tympanometry changes were noted and severity is assessed using Common Toxicity Criteria (CTC) -v2.0 of the National Cancer Institute (NCI). Fisher's Exact Test with two-sided p value was used for comparison of the tympanometric changes. Results The mean age was 55.4 years. The commonest site of HNCA was oral cavity (30%) with 46.6 % in stage II. 40% of cases were in grade II, followed by 27% cases in Grade III. The commonest complaint was pain (86.6%), hearing loss (80%) and 40% post-RT cases reporting Type B curve (relative risk 0.67). Conclusion Post-irradiation pain and OME are a major concern in HNCA patients having grade II (NCI-CTC). Early detection and prophylactic measures will improve the quality of life in such group of patients. This result validates the need for formulating otological diagnostic strategy &amp; therapeutic measures in patients receiving RT to mitigate post-irradiation ototoxicities.

Surviving and thriving: Assessing quality of life and psychosocial interventions in mental health of head and neck cancer patients

Role of the vulnerable elders survey-13 screening tool in predicting treatment plan modification for older adults with cancer