ESTRO 38 Abstract book

S789 ESTRO 38

AUC 0.9 and p=0.02, AUC 0.91) and standard deviation (SD) (p=0.2, AUC 0.9 and p=0.01, AUC 0.94) at baseline and week 2, kurtosis (p=0.01, AUC 0.92) and SD (p=0.01, AUC 0.96) at week 4 and change in minimum ADC between baseline and week 2 (p=0.02, AUC 0.94) and baseline and week 4 (p=0.02, AUC 0.94) significantly predicted for local recurrence. For prediction of any recurrence, ADC min (p=0.02, AUC 0.87) and SD (p=0.01. AUC 0.85) at baseline, max (p=0.03, AUC 0.77) and SD (p=0.02, AUC 0.81) at week 4 were significant. On LASSO logistic regression, ADC minimum and SD at baseline were retained for any recurrence. No DCE parameters were predictive of A number of ADC parameters at various time points appear predictive of local and any recurrence. No DCE parameters correlated with recurrence. EP-1455 Post chemoradiotherapy FDG-PET parameters predict for recurrence in anal cancer: a prospective trial M. Jones 1 , G. Hruby 2 , U. Metser 3 , S. Sridharan 4 , A. Capp 4 , M. Kumar 4 , S. Gallagher 4 , N. Rutherford 5 , C. Holder 6 , C. Oldmeadow 6 , J. Martin 4 1 WP Holman Clinic- Royal Hobart Hospital, Radiation Oncology, Hobart, Australia ; 2 Royal North Shore Hospital, Radiation Oncology, St Leonards, Australia ; 3 Princess Margaret Cancer Centre, Cancer Clinical Research Unit, Toronto, Canada ; 4 Calvary Mater Newcastle, Radiation Oncology, Waratah, Australia ; 5 Calvary Mater Newcastle, Nuclear Medicine, Waratah, Australia ; 6 Hunter Medical Research Institute, Statistics, New Lambton Heights, Australia Purpose or Objective To investigate the prognostic significance of serial F-18 fluorodeoxyglucose-positron emission tomography (PET) parameters in squamous cell carcinoma of the anal canal (AC). Material and Methods A multi-centre prospective trial of 19 patients with non- metastatic AC treated per protocol underwent PET imaging before and 12 weeks following chemo- radiotherapy (CRT). PET parameters from the primary tumour were extracted and correlated with recurrence. Results With a median follow up of 15.8 months, 3/19 patients relapsed locally and 5/19 had any recurrence. On post-CRT PET, the median standard uptake value (SUV) within a volume of interest (VOI) bounded by an SUV of 3 correlated with local recurrence (p value = < 0.01, AUC 1.0). The mean SUV 3 did not reach significance (p value = 0.06, AUC 0.91). On LASSO logistic regression, post-CRT SUVmax was retained on the grouped post model for both local and any recurrence, but not in the final model. No pre-CRT PET parameters reached significance. Conclusion Patients with a metabolic PET response had a very low risk of recurrence or progression. 12-week post-CRT PET parameters extracted from the primary tumour correlated with local recurrence. Further studies in larger cohorts are warranted to confirm this finding. EP-1456 Clinical impact of re-irradiation with carbon ion radiotherapy for locally recurrent rectal cancer F. Valvo 1 , A. Barcellini 1 , V. Vitolo 1 , M.R. Fiore 1 , A. Iannalfi 1 , B. Vischioni 1 , A. Facoetti 2 , M. Bonora 1 , S. Ronchi 1 , R. Petrucci 1 , E. D'Ippolito 1 , A. Mirandola 3 , S. Molinelli 3 , S. Russo 3 , G. Viselner 3 , G. Magro 3 , D. Maestri 3 , M. Ciocca 3 , L. Preda 4 , R. Orecchia 5 1 National Centre of Oncological Hadrontherapy CNAO, Radiation Oncology, Pavia, Italy ; 2 National Centre of Oncological Hadrontherapy CNAO, Radiobiology, Pavia, Italy ; 3 National Centre of Oncological Hadrontherapy CNAO, Department of Medical Physics, Pavia, Italy ; recurrence. Conclusion

multivariable logistic regression was used to identify predictors of DP. Classification methods included: logistic regression (LR) gradient boosting (GBM), adaptive boosting (ADA), and random forest (RF). Patient characteristics with a high correlation (rho > 0.6) and factors in which missing data >30% were excluded from the models. 10-fold cross validation was used to assess the models. Classifier performance was assessed by receiver operating characteristic curves. Results 113 patients with 226 treated lesions were included; 79 patients and 203 lesions treated for OM and 34 patients and 23 lesions for OP. 31 (27%) of the treated patients had DP within 90 days. 2 patients died within 90 days. In univariable analysis, SBRT location (non-lung/liver), higher CEA, shorter DFI, greater number of prior systemic therapy lines and lower mean PTV BED were significantly associated with early DP (p < 0.05). In multivariable analysis, shorter DFI and higher CEA were significant predictors of DP (p < 0.05). Performance of the various classifiers is shown in Table 1. All ML classifiers were significantly better at identifying patients with DP compared to the logistic regression model (p < 0.05). There was no statistically significant difference in performance between the various ML classifiers. Table 1. Classifier performance. Conclusion The ability to predict patients at risk of DP would assist clinicians in identifying patients who may benefit minimally from SBRT for OM/OP disease. External validation would be of benefit. EP-1454 Multi-parametric MRI as a biomarker in anal cancer: a prospective trial M. Jones 1 , G. Hruby 2 , M. Kumar 3 , A. Capp 3 , S. Sridharan 3 , C. Coolens 4 , P. Stanwell 5 , J. Arm 6 , S. Gallagher 3 , C. Holder 7 , C. Oldmeadow 7 , J. Martin 3 1 Royal Hobart Hospital, Radiation Oncology, Hobart, Australia ; 2 Royal North Shore Hospital, Radiation Oncology, St Leonards, Australia ; 3 Calvary Mater Newcastle, Radiation Oncology, Waratah, Australia ; 4 Princess Margaret Cancer Centre, Radiation Oncology, Toronto, Canada ; 5 The University of Newcastle, School of Health Sciences, Callaghan, Australia ; 6 Calvary Mater Newcastle, Radiology, Waratah, Australia ; 7 Hunter Medical Research Institute, Statistics, New Lambton Heights, Australia Purpose or Objective To investigate the role of multi-parametric magnetic resonance imaging (MRI) as a potential biomarker in squamous cell carcinoma of the anal canal (AC). Material and Methods A multi-centre cohort of 20 patients with non-metastatic AC were enrolled on a prospective clinical trial. Multi- parametric MRIs, incorporating diffusion weighted (DW) and dynamic contrast enhanced (DCE) sequences, were performed at time of staging, during the second and fourth week of chemo-radiotherapy (CRT) and 8 weeks post treatment. MRI parameters were extracted from the primary tumour and correlated with recurrence. Results With a median follow up of 17 months, 3/20 patients had a local recurrence, and 5/20 had any recurrence. Apparent diffusion coefficient (ADC) skewness (p=0.04, LR RF Sensitivity 0.55 0.49 0.25 0.48 Specificity 0.73 0.87 0.96 0.92 PPV 0.22 0.31 0.51 0.51 NPV 0.92 0.90 0.93 0.93 AUC 0.61 0.70 0.74 0.76 GBM ADA