ICHNO-ECHNO 2022 - Abstract Book

S100

ICHNO-ECHNO 2022

Materials and Methods Patients with HNC, including mucosal, complex cutaneous, salivary gland malignancies and unknown primary were included prospectively in the OncoLifeS databiobank. At baseline, patients underwent a geriatric screening, including a multidomain assessment of physical, functional, cognitive, and socio-environmental status and frailty screening. Patients were asked to complete the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC- QLQ-C30) and the Head and Neck 35 (EORTC-QLQ-H&N35) questionnaires at baseline and at 3, 6, 12 and 24 months. Univariable and multivariable logistic regression analyses were performed to identify factors associated with non-response to QLQ for each time point. Results Of 369 included patients, 291 were analysed after exclusion of patients treated with palliative intention and missing baseline data. After removing deceased patients and patients with relapse per time point, response to QLQ rates were 100% at baseline, 86.8% at 3 months, 77.9% at 6 months, 81.1% at 12 months and 77.6% at 24 months. In univariable analysis, age above 65 years, impaired functional status (measured by the Activities of Daily Living, Instrumental Activities of Daily Living and the Timed Up & Go tests), impaired cognitive status (assessed by the Mini Mental State Examination), dependent living situation (requiring help at home or nursing home) and frailty (assessed by the Geriatric 8 and Groningen Frailty Indicator) were consistently significant factors for non-response to QLQ during the 24 months follow-up. Frailty status was a significant factor in multivariable models adjusted for age, sex and department of follow-up as well. Conclusion Impaired functional, cognitive and socio-environmental status, and frailty are associated with lower response rate to QLQ in HNC patients. This may introduce bias in clinical studies focusing on patient reported outcome measures. Therefore, researchers and clinicians should be aware of this when interpreting these studies. 1 Prince of Wales Hospital, Radiation Oncology, Sydney, Australia; 2 University of New South Wales, POW Clinical School, Faculty of Medicine, Sydney, Australia; 3 University of Queensland, School of Human Movement and Nutritional Sciences, St Lucia, Australia; 4 Royal Brisbane and Women's Hospital, Nutrition and Dietetics, Brisbane, Australia; 5 Prince of Wales Hospital, Department of Radiology, Sydney, Australia; 6 University of New South Wales, Graduate School of Biomedical Engineering, Sydney, Australia; 7 Tamworth Base Hospital, Radiation Oncology, Tamworth, Australia Purpose or Objective Computed tomography (CT)-defined sarcopenia has emerged as an independent prognostic indicator in patients with head and neck cancer (HNC), and is becoming an important component of nutritional assessment. The gold standard technique involves evaluation of skeletal muscle at the level of the third lumbar (L3) vertebra, however abdominal scans are not always available in HNC. Several studies have used muscle at the third cervical (C3) vertebra to estimate muscle at L3, however, poor agreement was found in sarcopenia diagnosis between actual and predicted muscle in our largely obese/overweight Australian population. We aimed to re-evaluate the use of C3 and propose a new prediction model more suited to our heterogeneous demographic of patients with HNC. Materials and Methods This retrospective study included all adult patients who presented to the head and neck clinic in our facility (January 2013 to August 2021), with newly diagnosed HNC of the larynx, hypopharynx, nasopharynx, oropharynx or oral cavity, and had a diagnostic positron emission tomography-computed tomography scan. The cross-sectional area (CSA) of skeletal muscle was measured at the level of L3 and C3 in the CT component of each scan and compared. The prediction model for CSA at L3 from C3 measures was determined using multivariate linear regression which identified clinical factors (age, sex, and weight) as predictors of CSA at L3, and trained using 5-fold cross-validation on a randomly selected 80% of the cohort. The model was tested on the remaining 20% of the dataset for correlation of actual and predicted CSA measures, with sensitivity and specificity of sarcopenia diagnosis using pre-defined sex and BMI-specific cut-off values. Bland-Altman plots were constructed to analyse agreement. Results The scans of 111 patients were used for model development. The majority of patients were male (86%), 73% had oropharyngeal cancer, and 60% had a BMI>25. Strong correlation between actual and predicted CSA ( r =0.896, p <0.001), and actual and predicted skeletal muscle index (SMI) ( r =0.849, p <0.001) was found on the test sample following cross-validation of the model. SMI values were then stratified for BMI, and the prediction model demonstrated a sensitivity of 80.0% and specificity of 85.7% for sarcopenia diagnosis. Bland-Altman plots demonstrated good agreement with mean percent difference (bias) in SMI=0.97% (SD 9.6, 95%CI -2.9 to 4.9%), limits of agreement (-17.9 to 19.9%). Conclusion Accurately diagnosing sarcopenia in patients with HNC is important for determining appropriate nutritional interventions. As many patients present as overweight or obese, this can be difficult without effective CT imaging methods when abdominal scans are not available. Our prediction model has a higher sensitivity and specificity for our population than other models using CSA of skeletal muscle at C3. Future testing in larger populations with similar demographic characteristics is likely required. PO-0161 Sarcopenia in patients with head and neck cancer – an alternate skeletal muscle prediction model B. Vangelov 1,2 , J. Bauer 3,4 , D. Moses 5,6 , R. Smee 1,2,7