Abstract Book

S9

ESTRO 37

these outcomes would not be generalizable to patients with macroscopic cervical involvement. Our results suggest that cervical stromal invasion by itself may not be a significant risk factor of locoregional recurrence. In randomized trials, adjuvant radiotherapy was never shown to impact survival despite improving locoregional control. U.S. population-based studies using the SEER and National Cancer Database suggest that early stage patients at high-intermediate or high risk of recurrence may actually have improved overall survival from the use of adjuvant radiotherapy with either VBT or EBRT when controlling for age disease risk factors, and comorbid conditions. Currently dose is most commonly delivered as monotherapy with 7 Gy for 3 fractions to 0.5cm depth, though an ABS survey of U.S. radiation oncologists indicate marked heterogeneity with dose-fractionation regimens. Most commonly, dose is prescribed to the proximal 3-5cm or the proximal 1/3-1/2 of the vagina but there is no consensus. The ABS recommends treating the proximal 3-5cm of the vagina. The American Society of Radiation Oncology published their executive summary in 2014 addressing many controversial topics in the post- operative management of endometrial cancer patients and referral to this document may be useful to navigate such controversies. Finally, it is my approach to estimate the risk of recurrence, especially vaginal cuff recurrence, with observation and with adjuvant VBT for early stage patients. With a detailed discussion of side effects as well, the patient can make a decision based upon the risks and benefits of adjuvant VBT. This approach is supported by a survey performed by the Dutch Gynecologic Oncology Group. They found that the median minimal improvement in local control with adjuvant VBT was 0% for patients and 8% and physicians (p<0.001). Most patients would choose adjuvant VBT even with no benefit in local control. The vast majority of both patients and physicians prefer joint decision-making rather than the onus lying solely with either the patient or the physician. We believe that a discussion with the patient regarding the risks and benefits of adjuvant radiotherapy by the radiation oncologist encourages patient autonomy and informed decision making. SP-0025 Risk groups and decision making - The PORTEC experience R. Nout 1 1 Leiden University Medical Center LUMC, Department of Radiotherapy, Leiden, The Netherlands Abstract text The majority of endometrial cancer (EC) patients present at an early stage (80% FIGO stage I) and have a good prognosis (overall survival stage I: 85%). Total hysterectomy and salpingo-oophorectmy is the mainstay of treatment. Risk factors are used to help identify patients who may benefit from adjuvant (postoperative) therapy. Four major randomised trials investigated the role of adjuvant radiotherapy for intermediate risk EC (Aalders, PORTEC-1, GOG99, ASTEC/EN5). All trials found a significant (threefold) reduction of pelvic recurrence with postoperative external beam radiotherapy, however without overall survival benefit and at the cost of increased (mainly gastro-intestinal) morbidity. Both in PORTEC-1 and GOG99 a so called ‘high-intermediate risk’ (HIR) group was identified, showing the largest benefit from postoperative radiotherapy. In PORTEC-1 the risk of locoregional recurrence decreased from 23% to 5% at 5- years in patients with two out of three risk factors: age >60 years, grade 3, myometrial invasion >50%. GOG99 demonstrated the benefit after surgical lymph node staging and included lymph vascular invasion (LVSI) as risk factor in addition to age, grade and depth of myometrial invasion. Implementation of these HIR risk

factors led to a substantial decrease in the indications for radiotherapy at the time. Low-intermediate risk patients have 95% recurrence free survival and their excellent local control is not further improved with vaginal brachytherapy. For HIR patients, the majority (75%) of recurrences are located in the proximal vagina. The randomised PORTEC-2 trial demonstrated similar vaginal control (vaginal recurrence rate of 2% at 5-years) after vaginal brachytherapy with less morbidity and impact on patient reported QoL and symptoms compared to external beam radiotherapy. More recently the TCGA group comprehensively assessed molecular alterations in endometrial cancer and found four groups with distinct prognosis: POLE -mutant ultramutated, MSI hypermutated, p53-mutant copy-number high (serous like), and a copy number low group without a specific mutation. The POLE - mutant group had a very low risk of recurrence despite their ultramutated phenotype and the p53 mutant group had an increased risk of disease progression. Subsequent studies have confirmed the distinct prognostic impact of these four groups, including an analysis from the combined PORTEC-1 and PORTEC-2 cohorts. In addition to these molecular features analysis of the impact of presence and grading of LVSI pointed out that substantial LVSI in contrast to no or mild LVSI was a strong independent risk factor for regional and distant recurrence. Furthermore, risk groups based on combined molecular and clinico-pathologic features were found to have a higher prognostic accuracy than either alone. This concept has been carried forward in the on-going randomised PORTEC-4a trial that investigates molecular profile-based versus standard recommendations for adjuvant therapy, with the overall aim is to further decrease both over- and undertreatment in early stage endometrial cancer. SP-0026 Machine learning in medical imaging: challenges and opportunities M. De Bruijne 1 1 ErasmusMC Rotterdam, Departments of Radiology- Nuclear Medicine- and Medical Informatics, Rucphen, The Netherlands Abstract text This lecture introduces current state-of-the-art techniques in machine learning in medical imaging and highlights the potential of these techniques for computer-aided diagnosis, quantitative image analysis, and therapy planning. We will discuss challenges to introduce these techniques in clinical practice, related to a requirement of annotated data for model training and to the difficulty of modeling variations in scan protocols. Examples of successful applications of machine learning in radiology will be shown. SP-0027 Challenges of MRI that have prevented adoption for dose painting and will persist for radiomics C. Chung 1 1 MD Anderson Cancer Center, Radiation Oncology, Houston, USA Abstract text There is great potential for MRI to guide and improve radiotherapy, well beyond the immediately notable superior soft tissue contrast. Used to its full potential, MRI has the ability to provide biological information that can characterize tumor and surrounding tissues based on cellular density, tissue perfusion and metabolism and Symposium: Is radiomics going to deliver on the promises that dose painting made?