ESTRO38 Congress Report

E S T R O C O N F E R E N C E

Milan, Italy 26-30 April 2019

CONGRESS REPORT #38

Targeting optimal care, together

WWW.ESTRO.ORG

Table of Contents

INTRODUCTION

3

RTT

CLINICAL

34

4

5 46

35 36

INTRODUCTION 1. Implementation of plan of the day adaptive radiotherapy: Compliance to guidelines (E38-1524) 2. Dosimetric benefit of a clinically applied adaptive plan selection strategy for rectal cancer (E38-0922) 3. MERINO study: Defining a standardised delineation method for repeated GTV assessment using DW MRI (E38-2121) INTRODUCTION 1. Analysis of biomarkers for late radiotherapy toxicity in the REQUITE project (E38-1219) 2. Targeting TEMPRSS2:ERG fusion to achieve tumor- specific radiosensitization in prostate cancer (E38-1769) 3. Immunological contextural basis of a prognostic radio- mics signature in head and neck cancers (E38-0734) RADIOBIOLOGY

INTRODUCTION 1. HEAD & NECK Stem cell sparing IMRT for head and neck cancer patients: a double-blind randomized controlled trial (E38-2054) See under awards 2. RECTAL Organ preservation after chemoradiotherapy for rectal cancer: 5-year results of the GRECCAR2 trial (E38-1982) 3. PROSTATE Hyprofractionated vs conventional radiotherapy for prostate cancer: 7 yr results from the HYPROtrial (E38-1311) 4. BREAST Antihormones with or without irradiation in breast cancer: 10-year results of the ABCSG 8A trial (E38-1591) 5. CRANIAL Phase III trial of Prophylactic Cranial Irradiation with or without Hippocampus Avoidance in SCLC (E38-2563 ) INTRODUCTION 1. MRI-only proton therapy treatment planning with synthetic CT images generated using deep learning (E38-2116) 2. Radiotherapy-related lymphopenia affects overall survival in patients with lung cancer (E38-1852) 3. MLC tracking for lung cancer SABR is clinically feasible: results of first-in-human clinical trial E38-0898 4. When we have to apply volume corrections in dosimetry? (E38-1668) 5. Correcting CBCT images for dose calculation using a U-shaped deep convolutional neural network (E38-0312) PHYSICS INTRODUCTION 1. BrachyView: A Real-time In-body HDR Source Tracking System with Simultaneous TRUS Image Fusion (E38-0696) 2. Bi-objective optimization of dosimetric indices for HDR prostate brachytherapy within 30 seconds (E38-1275) See under awards 3. Clinical outcomes of focal salvage high-dose-rate brachytherapy for radiorecurrent prostate cancer (E38-0129) See under awards 4. Single dose high dose-rate (HDR) brachytherapy as monotherapy for localised prostate cancer (E38-0833) 5. 2 nd Conservative Treatment for 2nd Breast Tumor Event: GEC-ESTRO Breast Cancer WG updated results (E38-1734) 6. Contact X-Ray Brachytherapy (CXB) after local excision (LE) for early rectal adenocarcinoma (E38-0234) 7. Performance of ring vs ovoids and intracavitary vs intra- cavitary-interstitial in the EMBRACE study (E38-1266) BRACHYTHERAPY

38

6

39

8

9

40

41 42

11

43

44

12

13 14

AWARDS GRANTED AT ESTRO 38

45

46 47 48 50 52 53

LIST OF AWARDS 1. Donal Hollywood Award 2. Jack Fowler University of Wisconsin Award 3. ESTRO Varian Award 4. Elekta Brachytherapy Award 5. GEC-ESTRO Best Juniour Presentation Elekta Award

16

17

18

19

ACKNOWLEDGEMENTS

55

22

23 24

46

46

26

28

29

30

TABLE OF CONTENTS | Congress report

2

Introduction

We imagine that ESTRO 38 was a great event. What were, in your view, the reasons for this success? I really believe ESTRO 38 was a great success from the scientific point of view, due to the huge number of different topics belonging to all different fields of our discipline and the outstanding level of the

To get this vision statement we will be hardly working on these four strategy themes: • From Research To Practice • Strengthening the Society

• Strengthening the profession • Strenghtening partnership

Next annual ESTROmeetings will certainly be a keymoment in achieving the various results included in these strategic themes. In what way was the congress an asset to professionals in radiation oncology? What did it bring to their knowledge and practice? I believe that the ESTRO annual conference guarantees a mix of innovation and practice, of absorbing knowledge and sharing experiences, of networking, of science and fun; and it is indeed this mix that makes our conference a fixed “rendez-vous” for somany radiation oncology professionals from Europe and around the globe. ESTRO38was also themoment where the new visual identify of ESTRO was unveiled, allowing for a consistent brand visibility both on ESTRO’s digital and physical presence. The new website will allow the radiation oncology community to easily find the information they are looking for and the full integration with our databases – which will be finalized by end of 2019 – will further enhance the user experience. When is the next meeting? ESTRO 39 is scheduled next year, 3-7 April 2020 in Vienna, Austria. It will be another occasion for all RadiationOncology professionals to get together once again! ESTRO 39 will convey the theme ‘Translating research and partnership into optimal health’

UMBERTO RICARDI

contributions; multidisciplinary sessions, including several joint symposia with other scientific societies and for the first time we had a joint symposium with EACR.These always represented an added value, emphasizing the importance of the conference theme statement: targeting optimal care, TOGETHER. We had around 2250 abstracts this year, confirming the interest of people to be actively involved in our annual meeting: this aspect is certainly one of the key elements in making the annual conference so successful. The number of registered delegates and square meters of exhibition area were the two other elements making ESTRO 38 a great success. We had over 3.700 participants registered, excluding industry representatives. This is an increase of 6% compared to ESTRO 37. On the exhibition front, we hosted 94 industry partners, 14 national associations and 13 start-ups – an increase on all fronts! On what did the sessions focus? Give a few examples As mentioned before, the topics were broad and covering all aspects of radiation oncology disciplines. Naming a few sessions means ignoring somany others but among those sessions that emphasised the conference theme: “Targeting optimal care, Together”, I would mention those focusing on MR-guided radiotherapy and on combinations of immunotherapy with radiotherapy. Sessions on oligometastatic disease and particle therapy were also perfectly in line with the conference theme. A special mention goes to two other sessions: randomized clinical trials and late breaking abstract ones, where more than ten randomized clinical trials (RCT) have been presented in overcrowded large rooms: I hope to see full papers relating to these RCT published very soon in the Green Journal. Has the scientific programme been developed to reflect the ESTRO vision? What about the newly proposed vision? Yes, it was. The scientific programme of the conference was centred on our visionwhich states that all European patients must have proper access to state of the art radiotherapy. ESTRO’s new vision statement for 2030 ‘Radiation Oncology. Optimal Health for All, Together.’ emphasizes the ambition of the Society to further reinforce radiation oncology as a core partner in multidisciplinary cancer care and to guarantee accessible and high-value radiation therapy for all cancer patients who need it.

Looking forward to meeting you there! Umberto Ricardi ESTRO 38 Chair

Congress report | INTRODUCTION

3

CLINICAL

5

INTRODUCTION

2. RECTAL Organ preservation after chemoradiotherapy for rectal cancer: 5-year results of the GRECCAR2 trial (E38-1982)

6

3. PROSTATE Hyprofractionated vs conventional radiotherapy for prostate cancer: 7 yr results from the HYPROtrial (E38-1311)

8

4. BREAST Antihormones with or without irradiation in breast cancer: 10-year results of the ABCSG 8A trial (E38-1591)

9

5. CRANIAL Phase III trial of Prophylactic Cranial Irradiation with or without Hippocampus Avoidance in SCLC (E38-2563 )

11

CLINICAL | Congress report

4

Introduction Clinical Track

The seven-year results from the Dutch HYPRO trial of conventional (39 fraction of 2 Gy) versus hypofractionated (19 fractions of 3.4 Gy) radiotherapy for intermediate- to high-risk prostate cancer showed equivalence for relapse- free and overall survival, and suggested a local control benefit for hypofractionation in patients with Gleason > 7 tumours. The 10-year results of the Austrian ABCSG 8A trial on antihormonal treatment with or without whole breast irradiation (WBI) in low-risk breast cancer confirmed that WBI significantly improved local control and DFS. Finally, themulticenter Dutch/Belgiumphase 3 trial on prophylactic cranial irradiationwith or without hippocampus avoidance in patients with small cell lung cancer provided interesting first results on neurocognitive functions and brain recurrences. Claus Rödel Chair, SAG Clinical Radiotherapy

This year’s ESTRO annual congress featured more than a dozen prospective, randomised phase II or III clinical trials presented either at the randomised clinical trials proffered paper session or the late- breaking proffered paper session. The sheer number is impressive and highlights the ongoing success

CLAUS RÖDEL

of clinicians throughout ESTRO countries to provide high- level evidence for the use of radiation therapy alone or as a treatment component withinmultimodality approaches. Moreover, results of innovative early clinical trials (phase I and II) combining radiotherapy with targeted agents, including immune modulation, were presented for various tumor entities. Technical advances have been made for MR-guided adaptive radiation and high-precision radiotherapy, including SBRT, particularly in the treatment of oligometastatic disease. Five of the prospective randomised clinical trials – obviously a subjective selection – will be presented and discussed in more detail in the following pages. These include an innovative, double-blind randomised trial from the UMCG University Medical Center Groningen, The Netherlands. This group tested parotid gland stem cells-sparing IMRT for head and neck patients, based on previous findings in mice, rats, and humans, which showed that stem and progenitor cells mainly reside in the region of the parotid gland containing the major ducts (van Luijk et al., 2015). The Groningen group subsequently hypothesized that reducing dose to these high stem cell density regions may prevent or alleviate xerostomia. Although they could not show that parotid gland stimulated salivary flow at 12 months (primary endpoint) was significantly improved in the experimental stem cell sparing IMRT arm, this study is a prime example of excellent, hypothesis-driven, well- designed and executed clinical research. Large national and international groups reported their (updated) randomised data on rectal, breast, prostate, and lung cancer at the ESTRO 38 meeting. Three abstracts reported randomised data on rectal cancer therapy. These include the five-year results of the French GRECCAR2 organ preservation trial of chemoradiotherapy followed by local excision versus total mesorectal excision surgery in low T2/T3 rectal cancer patients, supporting the long-term oncologic safety of limited surgery for selected rectal cancer patients with good response after chemoradiotherapy. The GRECCAR2 study is further described in the next section. Bujko et al. provided an update of the Polish randomised phase III study of total neoadjuvant therapy (TNT), i.e. 5x5 Gy followed by consolidation chemotherapy prior to radical surgery, versus long-course chemoradiotherapy in patients with fixed T3 or T4 rectal cancer. After seven years of follow- up, no significant differences were noted for local and distant control, disease-free (DFS) and overall survival. The German Rectal Cancer Study Group (Fokas et al.) presented first results of the randomised CAO/ARO/AIO-12 phase II trial on TNT sequences: Upfront chemoradiotherapy followed by consolidation chemotherapy was less toxic and more effective in terms of pathological complete response rates than the sequence induction chemotherapy followed by chemoradiotherapy and surgery.

Congress report | CLINICAL

5

Clinical

2. RECTAL Organ preservation after chemoradiotherapy for rectal cancer: 5-year results of the GRECCAR2 trial (E38-1982)

Véronique Vendrely, Philippe Rouanet, Jean-Jacques Tuech, Alain Valverde, Bernard Lelong, Michel Rivoire, Jean-Luc Faucheron, Jafari Mehrdad, Guillaume Portier, Eric Frison, Julien Asselineau, Quentin Denost, Eric Rullier

Bordeaux University Hospital (France) and the GRECCAR

Context of the study Rectal preservation, using a local excision (LE) or an observational strategy has been increasingly debated for good responders after neoadjuvant CRT to decrease surgical morbidity and increase quality of life. The French GRECCAR 2 trial was the first phase III trial investigating this strategy. Although LE was not superior to TME due to a high rate of completion surgery, which increasedmorbidity and side-effects at 2 years, this trial highlighted the good pathological response with no positive mesorectal node. Further follow-up was needed to validate the oncological safety of the strategy: here are presented the 5-year results FromMarch 2007 to September 2012, 148 patients clinically good responders after CRT for T2T3 rectal cancer, size ≤ 4 cm, were randomized in 74 LE vs. 74 TME, 3 were excluded and 145 analyzed. In the LE group, 26 had a completion TME for ypT2-3, which was part of the protocol. Kaplan-Meier and Cox regressions were used to estimate and compare local and metastatic recurrence and survival at 5 year What were the three main findings of your research? 1. Overall (84% vs. 82%; HR=0.92, p=0.845) and disease- free (70% vs. 72%; HR=0.87, p=0.682) survivals were not different between the LE and the TME groups at 5 years 2. Moreover, there was no difference either regarding local recurrence (7% vs. 7%; hazard ratio (HR)=1.41, p=0.599) nor metastatic recurrence (18% vs. 19%; HR=0.86, p=0.734) between the LE and the TME group at 5 years. 3. Interestingly, in both groups themetastatic risk remained high (almost 20%) even for good responders with initial small tumors such as T2-3 rectal cancers less than 4 cm in diameter. What impact could your research have? The 5-year results of the GRECCAR 2 trial confirm the oncological safety of the rectal preservation strategy in a selected population of good responders after RCT for small T2-3 rectal cancers. Further questions have been raised regarding evaluation, accurate selection of patients and place of complementary TME that lead to build a new randomized trial: GRECCAR 12. This randomized phase 3 trial is designed to test the hypothesis that adding neoadjuvant chemotherapy to CRT will optimize the tumor response and increase the number of patients who will benefit from organ preservation. Moreover, given the need for improved response evaluation, this trial will focus on MRI response evaluation and its correlation with clinical and pathological response. of the GRECCAR 2 trial. Overview of abstract

Is this research indicative of a bigger trend in oncology? Most of ongoing trials focus currently on the optimization of tumor response in order to increase the chance of rectal preservation. There are several options to increase the response rate: radiotherapy dose escalation using contact-therapy or HDR brachytherapy, intensification of concomitant chemotherapy, neoadjuvant chemotherapy before irradiation or additional chemotherapy after irradiation. Finally, accurate evaluation of tumor response remains a major issue regarding rectal preservation.

CLINICAL | Congress report

6

Figure 1: Flow chart of the trial

Figure 2: 5-year recurrence-free Survival

Congress report | CLINICAL

7

Clinical

3. PROSTATE Hyprofractionated vs conventional radiotherapy for prostate cancer: 7 yr results from the HYPRO trial (E38-1311)

Luca Incrocci, Kim C. de Vries, Ruud C. Wortel, Esther Oomen-de Hoop, Wilma D. Heemsbergen, Floris J. Pos

Erasmus MC Cancer Institute, Rotterdam, The Netherlands Netherlands Cancer Institute/Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands

Context of the study In the early 90’s multiple studies hypothesised an α/β for prostate cancer as low as 1.5 Gy. This opened the way for several hypofractionation trials such as the CHiPP, PROFIT, RTOG 0415 and HYPRO. These trials tried to answer the question whether it is safe to use hypofractionation in low and intermediate-high risk localized prostate cancer. Long term results (5-10 years) show that hypofractionation is safe for both outcome and toxicity. National and international guidelines consider hypofractionation now as the new standard of care for localized prostate cancer. Overview of abstract Patients with intermediate-high risk prostate cancer were randomly assigned to either hypofractionated radiotherapy (64·6 Gy, 19 fractions/3·4 Gy, three fractions/week) or conventionally fractionated radiotherapy (78·0 Gy, 39 fractions/2·0 Gy, five fractions/week). The HYPRO trial did not show inferiority in toxicity, nor superiority in relapse free survival at 5 year. The use of long-termandrogen deprivation therapy in 66% of the patients might have obscured a potential benefit of hypofractionated radiotherapy at 5-year follow-up. We update here the outcome results at 7 years. What were the three main findings of your research? 1. At 7-year follow-up relapse free survival and overall survival demonstrated no difference between hypofractionation versus conventional fractionation. 2. None of the examined subgroups seemed to have an advantage of hypofractionation in term of relapse free survival. 3. A trend toward better local control was found after hypofractionation. Local control was statistically significant superior in patients with Gleason ≥8 tumors. What impact could your research have? The HYPRO trial is the biggest randomized trial which included intermediate and high risk prostate cancer, showing its safety in terms of relapse free survival in this group of patients. Our results show a better local control in patients with Gleason score ≥8 prostate cancer, although the study was not powered for this question. In this group of patients it may be beneficial to give a radiobiological higher dose to the prostate. Due to novel imaging techniques at present it is easier to diagnose patients who have only local disease. Therefore, excluding N+M+ patients, this effect might even be bigger. Is this research indicative of a bigger trend in oncology? The randomized hypofractionation trials have opened the trend to reduce the number of fractions in external beam

radiotherapy for prostate cancer, not only convenient for the patients but also enabling to make better use of hospital resources. Multiple new (ultra)hypofractionation trials are now being conducted, such as the HYPO-RT-PC and the PACE-B trial, that reduce the number of fractions to 4 or 5. The preliminary results of these trials, mainly in low and intermediate risk patients, are promising. In the past the question has been how low the α/β ratio for prostate cancer might be. Now, the question is how far we can go in reducing the number of fractions.

CLINICAL | Congress report

8

Clinical

4. BREAST Antihormones with or without irradiation in breast cancer: 10-year results of the ABCSG 8A trial (E38-1591)

Gerd Fastner MD 1 , Felix Sedlmayer MD 1 , JoachimWidder MD 2 , Martina Metz MD 3 , Hans Geinitz MD 4 , Karin Kapp MD 5 , Lidija Sölkner MSc 6 , Richard Greil MD 7 , Raimund Jakesz MD 6 , Werner Kwasny MD 8 , Dietmar Heck MD 9 , Vesna Bjelic-Radisic MD 10 , Marija Balic MD 11 , Herbert Stöger MD 11 , Ursula Wieder MD 12 , Ronald Zwrtek MD 13 , Dagmar Semmler MD 13 , Wilfried Horvath MD 14 , Elisabeth Melbinger-Zeinitzer MD 15 , Martin Wiesholzer MD 16 , Viktor Wette MD 17 , Michael Gnant MD6, 18 1 Department of Radiotherapy and Radio-Oncology, Paracelsus Medical University, University Hospital Salzburg, Landeskrankenhaus, Salzburg, Austria, 2 Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria, 3 Department of Radiooncology and Radiotherapy, Wiener Neustadt Hospital, Wiener Neustadt, Austria, 4 Department of Radiation Oncology, Ordensklinikum Barmherzige Schwestern Linz, Linz, Austria, 5 Department of Therapeutic Radiology and Oncology, Comprehensive Cancer Center (CCC), Medical University of Graz, Graz, Austria, 6 Austrian Breast and Colorectal Cancer Study Group,Vienna, Austria, 7 Department of Medical Oncology, Paracelsus Medical University, Salzburg, Austria, 8 Department of Surgery, Wiener Neustadt Hospital, Wiener Neustadt, Austria, 9 Department of Surgery, Ordensklinikum Linz Barmherzige Schwestern, Linz, Austria, 10 Department of Gynecology, Medical University of Graz, Graz, Austria, 11 Department of Internal Medicine, Division of Oncology, Medical University of Graz, Graz, Austria, 12 Department of Surgery, Hanusch Spital, Vienna, Austria, 13 Department of Surgery, Mistelbach Hospital, Mistelbach, Austria, 14 Department of Surgery, Guessing Hospital, Guessing, Austria, 15 Department of Surgery, Wolfsberg Hospital, Wolfsberg, Austria, 16 Department of Internal Medicine I, Division of Nephrology and Hematology, University Hospital St. Poelten, Karl Landsteiner University of Health Sciences, St. Pölten, Austria, 17 Department of Surgery, Krankenhaus der Barmherzigen Brüder St Veit an der Glan, St. Veit an der Glan, Austria, 18 Department of Surgery and Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria Context of the study

were noted. In our risk analysis the omission of WBI turned out to be the main negative predictor for IBR.Unclassified tumor grading Gx had significant influence, which may be attributable to an unknown portion of a more aggressive undifferentiated grading G3 carcinoma (Figure 3). What impact could your research have? Radiotherapy (RT) still maintains its high potential to significantly improve LC and DFS in combination with AH compared to AH alone also for prognostically favourable breast cancer patients after long term FUP. In the light of current knowledge this does not necessarily address only WBI, since partial breast treatments (PBI) have turned out to be competitive in these low risk settings. PBI are currently carried out either with postoperative hypofractionated external beam radiotherapy (EBRT), intraoperative techniques (electrons (IOERT), 50-kV orthovolt) or interstitial brachytherapy, respectively. Therefore, a carefully performed patient selection on the basis of prognostic risk factors like patients` age, histopathological tumor characteristics as well as modern molecular genetic tests should help to decide about de-escalation of RT in speciic cases. The total omission of RT should only be considered in frail elderly patients who are not resilient for any radio-oncological intervention. Is this research indicative of a bigger trend in oncology? The results of our trial have shown that in postmenopausal “low-risk” breast cancer patients with positive hormonal receptors, RT remains of clinical relevance independently of antihormonal medication. However, our findings corroborate to the necessity of individually tailored and accurate treatment strategies. This includes all involved disciplines from surgery, medical oncology to radio-oncology in order to avoid overtreatment and to minimize treatment related toxicity. Especially in low risk patients, modern RT techniques like hypofractionated WBI or PBI reduce both overall treatment time and irradiated tissue volumes, apparently without compromising oncologic outcome.

It is still amatter of debate weather elderly postmenopausal women with a good prognosis of hormone sensitive breast cancer disease who receive antihormones (AH) profit froman additional course of postoperative whole breast irradiation (WBI) after breast preserving surgery (BPS). Most of the clinical trials, which investigated this issue, published their data after a median observation period of five years. So far, long term experiences are scarce and have been reported only twice by US and UK study groups. Hence, the Austrian Breast cancer and colorectal study group (ABCSG) decided to re-evaluate their data of 2007 (ABCSG 8A trial) after a 10 year follow up (FU), respectively. Overview of abstract The purpose of our research (updated ABCSG 8A trial) was to investigate 10 year long term data of hormonal receptor positive breast cancer patients after BPS (n= 869) who received consecutively AHwith (n= 439, group 1) or without WBI (n=430, group 2). As respective clinical endpoints were defined: local control (LC) in the affected breast, overall survival (OS; i.e. death due to any cause), disease-free survival (DFS; i.e. any clinical event due to breast cancer) and metastases-free survival (MFS; i.e. probability for no distant metastases). Furthermore, we were looking for possible independent risk factors, whichmight be predictive for local breast cancer recurrence. What were the three main findings of your research? After a median FU of 9.89 years, 10 patients developed an in-breast tumor recurrence (IBR) in group 1 (withWBI) and 31 in group 2 (without WBI), which translated into an actuarial 10-year LC of 97.5% and 92.5 %, respectively. As depicted in Figure 1, differences between both groups turned out to be statistical significant (p=0.0004). Moreover, with an 10-year DFS of 94%, breast cancer disease recurred less frequent in group 1 compared to 88.4% in group 2 (p= 0.01) (Figure 2). For all other remaining endpoints no significant differences

Congress report | CLINICAL

9

Figure 1: Local Control

Figure 2: Disease free survival

Figure 3: Independent risk factors for IBR

CLINICAL | Congress report

10

Clinical

5. CRANIAL Phase III randomized trial of Prophylactic Cranial Irradiation with or without Hippocampus Avoidance in SCLC: primary endpoint on neurocognition (E38-2563) J. Belderbos 1 , D. De Ruysscher 2 , K. De Jaeger 3 , F. Koppe 4 , M. Lambrecht 5 , Y. Lievens 6 , E. Dieleman 7 , j. Jaspers 8 , J. Van meerbeeck 9 , J. Ubbels 10 , M. Kwint1, M. Kuenen 11 , S. Deprez 12 , M.B. De Ruiter 11 , K. Sikorska 13 , H. Van Tinteren 13 , S.B. Schagen 11 . 1 The Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands, 2 Maastricht University Medical Center- School For Oncology And Developmental Biology Grow, Radiation Oncology, Maastricht, The Netherlands, 3 Catharina Hospital, Radiation Oncology, Eindhoven, The Netherlands, 4 Institute Verbeeten, Radiation Oncology, Tilburg, The Netherlands, 5 UZ Gasthuisberg, Radiation Oncology, Leuven, Belgium, 6 Ghent University Hospital and Ghent University, Radiation Oncology, Gent, Belgium. 7 Amsterdam UMC- Location AMC, Radiation Oncology, Amsterdam, The Netherlands, 8 Erasmus MC Cancer Institute- Erasmus MC University Medical Center, Radiation Oncology, Rotterdam, The Netherlands, 9 Antwerp University Hospital, Department of Pulmonology & Thoracic Oncology, Antwerp, Belgium, 10 University of Groningen- University Medical Center Groningen, Radiation Oncology, Groningen, The Netherlands, 11 The Netherlands Cancer Institute, Division of Psychosocial Research And Epidemiology, Amsterdam, The Netherlands, 12 University Hospitals Leuven- KUL, Imaging and Pathology, Leuven, Belgium, 13 The Netherlands Cancer Institute, Department Of Biometrics, Amsterdam, The Netherlands.

Context of the study Prophylactic cranial irradiation (PCI) is standard treatment in patients with Small Cell Lung cancer (SCLC) and remission following treatment of the primary tumour, to prevent brain metastases. PCI results in a significant decrease in the incidence of brain metastases and improves survival. Adverse effects of PCI include fatigue, nausea, neurocognitive deterioration (such as learning and memory), and ataxia. These side effects may bemild and transient, but can also be progressive and persistent. Memory dysfunction is a major complaint following PCI, and previous studies demonstrated a significant decline in hippocampus related memory function following PCI. We hypothesized that sparing the hippocampus would improve neurocognitive decline after PCI. We therefore initiated amulticentre randomized phase III trial (NCT01780675) to investigate memory functioning after PCI with or without Hippocampus Avoidance (HA) in SCLC patients with a response to chemotherapy, or after chemo-radiation of the lung. Overview of abstract Memory functioning of patients included in the trial was assessed by a neuropsychological test battery at baseline, 4, 8, 12, 18 and 24 months after the brain irradiation with or without HA. The primary endpoint was a decline in the total recall score of the Hopkins Verbal Learning Test– Revised (HVLT–R), assessed at 4 months and a decline of ≥ 5 points was considered a failure. Secondary objectives were other cognitive outcomes/quality of life, radiological brain abnormalities on MRI (baseline, 4 and 12 months) and evaluation of the incidence and location of brain metastases following standard PCI (25 Gy in 10 fractions) compared to HA-PCI with a mean biological hippocampus dose ≤ 6.1 Gy (α/β=2Gy). What were the three main findings of your research 1. In both treatment arms the total recall score of the HVLT-R dropped at 4 months at least 5 points compared to baseline in 28% of the patients. 2. Based on the total recall score of HVLT-R (where a 5 point decline or greater was considered a failure) this prospective phase III trial did not show a significant

difference in memory decline between patients who received HA PCI and patients who were treated with conventional PCI at 4, and also not at 8 months 3. The incidence of brain recurrences was not increased

in the hippocampus avoidance region. What impact could your research have?

Having improved knowledge on the neurocognitive effects of PCI in SCLC helps to better inform patients about side effects, as it may also help us to identify those at particular risk for cognitive decline. The translational research part in this trial, including extensive fMRI and biomarkers, may give thus far unknown insights in the effects of radiation to the brain and on brain functioning in general. This knowledge will be used to design new trials aiming at decreasing side- effects of PCI. Is this research indicative of a bigger trend in oncology? There is increasing awareness about the impact of cancer treatments on neuro-cognitive functioning of patients. Our trial is the first worldwide to investigate the role of hippocampus sparing in PCI. It shows that we still lack fundamental knowledge in brain functioning to optimize the balance between the delivery of an effective cancer treatment such as PCI and the prevention of side effects. Other groups are investigating pharmacological interventions like memantine to reduce neuro-cognitive decline or try to avoid PCI by MRI surveillance of the brain.

Congress report | CLINICAL

11

PHYSICS

13

INTRODUCTION

14

1. MRI-only proton therapy treatment planning with synthetic CT images generated using deep learning (E38-2116)

16

2. Radiotherapy-related lymphopenia affects overall survival in patients with lung cancer (E38-1852

17

3. MLC tracking for lung cancer SABR is clinically feasible: results of first-in-human clinical trial E38-0898

18

4. When we have to apply volume corrections in dosimetry? (E38-1668)

19

5. Correcting CBCT images for dose calculation using a U-shaped deep convolutional neural network (E38-0312)

PHYSICS | Congress report

12

Introduction Physics track

to achieve a coherent k Q

data set in the revised TRS- 398

The development of the physics tracks for ESTRO 38 has highlighted the major current trends in radiotherapy physics with a big increase in numbers of abstracts submitted on predictive modelling, radiomics and adaptive radiotherapy. There were also a very high number of submissions

protocol.

Finally Landry et al evaluated three different deep learning based correction strategies using neural networks for image correction in CBCT in terms of their photon and proton dose calculation accuracy. VMAT and proton pencil beam scanning (PBS) plans were optimized and recalculated on the corrected CBCT images. They concluded that training directly on corrected CBCT image slices is optimal for PBS proton dose calculations, while for VMAT all three correction strategies provided sufficient accuracy. Additionally the correction times were found to be fast enough for online adaptive RT workflows.

CATHARINE CLARK

on protons across all the different topics, but especially in planning, measurement and toxicity, suggesting that the implementation of proton treatment in Europe well underway. A significant increase in the use of MRI in pre- treatment planning, intra-fraction motion management and in radiomics was also seen. Overall extremely varied work was submitted, from fundamental dosimetry to radiobiological modelling to advanced imaging techniques. This led to very good number of abstracts submitted to all the topics and meaning that overall 826 abstracts were considered. This large number has meant that the number of proffered paper sessions has been increased for 2019. Over the following pages you will find five abstracts which have been selected to give you a taste of the breadth of the topics to come in the two physics tracks: The first was by Barragan Montero et al and looked at MRI only proton treatment planning with synthetic CT images generated using deep learning. This work used neural networks to quickly generate and validate the conversion to CT from MR images and using these images for proton planning. The differences between the plans on the synthetic and original CT were minimal implying that the synthetic CT could be used alone. Abravan et al used datamining to investigate the relationship between dose and the incidence of lymphopenia by identifying anatomical regions where the dose received by the tissue correlated with the lymphopenia. This work used an extensive dataset of 562 lung cancer patients to develop a model and identify the regions with lymphopenia of ≥G3. Multivariate analysis to identify the dose parameters and themodel was then validated on 301 oesophageal patients. The regression calculation showed that lymphopenia of ≥G3 predicted overall survival. MLC tracking for lung cancer SABR was investigated by Booth et al in a trial to determine clinical feasibility and measure the target and normal tissue doses in comparison with conventional SABR treatment. They showed that real-time adaptation using MLC tracking was feasible in seventeen lung cancer patients. Furthermore reductions in treated volumes up to 47% were seen, which translated to reductions in lung dose. Zink et al led a group of European national standards laboratories who have investigated the volume averaging correction, k vol , for a range of ion chambers in both flattened and non-flattened beams. The results showed that kvol, even for flattened beams, may be not negligible and the contribution to k Q should be indicated. They recommend that k Q should therefore be corrected for volume averaging

Catharine Clark Chair, SAG Radiation Physics

Congress report | PHYSICS

13

Physics

1. MRI-only proton therapy treatment planning with synthetic CT images generated using deep learning (E38-2116) Ana María Barragán Montero 1,2 , Samaneh Kazemifar 2 , Kevin Souris 1 , Robert Timmerman 3 , Steve Jiang 2 , Xavier Geets 1 , Edmond Sterpin 1 , Amir Owrangi 2 . 1 UCLouvain- Institute of Experimental and Clinical Research, Molecular Imaging, Radiotherapy and Oncology MIRO, Brussels, Belgium, 2 UT Southwestern Medical Center, Medical Artificial Intelligence and Automation MAIA, Dallas, USA. 3 UT Southwestern Medical Center, Radiation Oncology, Dallas, USA.

Context of the study Magnetic resonance imaging (MRI) is often used in radiation therapy treatments to accurately contour the target volume and organs at risk (OARs), due to the superior soft tissue contrast in comparison with computed tomography (CT) images. The use of MRI images is especially crucial in treatment sites located in the abdomen and brain, where the tumour volume is mainly surrounded by soft tissue. However, CT images are still required in order to retrieve information needed for dose calculation (i.e., electron density values for conventional radiotherapy with photons and stopping powers for ion therapy). Therefore, the current treatment planning workflow for these sites relies on the contouring of the target and OARs on MRI and the posterior transfer of contours to CT via image registration. However, MRI-CT co-registration introduces systematic geometrical uncertainties up to 2-3 mm, which shift high dose regions away from the target and lead to a geometric miss that compromises tumour control. This problem has recently led to the concept of MRI-only based treatment planning, where pseudo or synthetic CT images for dose calculation are generated directly from the MRI scan. In addition, treatment planning based on MRI-only would reduce radiation dose, patient time, and hospital resources. MRI-only treatment planning is then a very attractive concept that is gaining popularity. However, the accurate generation of Hounsfield unit (HU) maps fromMRI images is not straight-forward. The rise of deep learning methods and the promising results obtained in medical imaging applications provides a good environment to apply this kind of methods to the synthetic CT generation problem. In this study, we evaluate the dosimetric accuracy of synthetic CT images generated with a deep learning method based on generative adversarial networks (GANs), for treatment planning of scanned proton therapy. Overview of abstract MRI-only treatment planning is becoming popular due to increased soft-tissue contrast and the reduction of radiation dose, among other advantages. Since CT images are still needed for dose calculation, different methods have been proposed to convert MRI to synthetic CT (sCT) images. Specifically for proton therapy, sCT must have extremely good quality, because small HU errors could greatly impact the dose distribution. This work explores the application of deep learning to this problem, since it could potentially improve the quality of the generated sCT images in comparison with the existing methods, given its ability to capture complex relationships between images.

What were the three main findings of your research? 1. High quality images: Our deep learning model based on generative adversarial networks (GANs) was able to learn how to accurately transform MRI to synthetic CT images, with a mean absolute error over all test patients as low as 47.2 ± 11.0 HU. 2. Easy implementation: The model generated the sCT images in less than 10 s, without requiring any human intervention and using a single MRI sequence (T1- weighted). 3. High dosimetric accuracy: The dosimetric evaluation showed that the dose differences were mostly below 1 to 2% of the dose prescription, which outperforms the current existing models for synthetic CT generation. What impact could your research have? So far, themodels presented in the literature requiredmanual pre- or post- processing of the synthetic CT images, such as manual insertion of air-cavities or bone segmentation, in order to minimize proton range differences and ensure a reasonable dosimetric accuracy. In contrast, our model removes all human intervention and achieves a high accuracy, even in cases with complex air-bone interfaces close to nasal cavities. This, together with the superior speed for the sCT generation (less than 10s), brings closer the clinical implementation of MRI-only treatment planning. Thus, potentially reducing the uncertainties due to MRI to CT contour propagation inherent to the current treatment planning workflow, which may improve clinical outcomes in those tumours located in soft-tissue regions. Is this research indicative of a bigger trend in oncology? Indeed, the application of deep learning methods in the domain of medical imaging is becoming more and more trendy, motivated by the excellent results obtained so far. This work is just another example of this trend, which justifies the investigation of deep learning for new applications, and eventually, the clinical implementation of some of them in the near future. However, before bringing these models to the clinical environment, the scientific communitymust carefully validate themwith databases that are representative of the targeted patient population, and build quality assurance tools to ensure its correct performance for each single case and/or patient.

PHYSICS | Congress report

14

Figure 1. DVH (a) for one of the test patients representing the dose computed in the CT (solid line) and SCT (dotted line), for the CTV and the OARs including right optic nerve (RON), left optic nerve (LON), brainstem, and optic chiasm. The corresponding dose distributions (at one slice on the centre of the target) are presented in (b) for the CT and (c) for SCT.

Congress report | PHYSICS

15

Physics

2. Radiotherapy-related lymphopenia affects overall survival in patients with lung cancer (E38-1852)

Azadeh Abravan 1,2 , Corinne Faivre-Finn 1,2 , Jason Kennedy 2 , Alan McWilliam 1,2 , Marcel van Herk 1,2

1 Division of Cancer Sciences, The University of Manchester, Manchester, United Kingdom. 2 Radiotherapy Related Research, The Christie NHS Foundation Trust, Manchester, United Kingdom.

Context of the study Lymphocytes, one of the subtypes of white blood cells, are one of the most radiosensitive cells in the human body. Lymphopenia, a drop in lymphocyte counts, can be observed during or early after a course of radiotherapy in cancer patients. Lymphopenia can lead to opportunistic infection, poor quality of life, and impact negatively on patients’ survival. However, a comprehensivemethodology to identify the cause of this adverse effect has not yet been established. In our study, we employed for the first-time image-based data mining approaches to identify regions of the body associated with severe lymphopenia (i.e. grade 3 or higher, according to Common Terminology Criteria for Adverse Events v4.0) in lung cancer patients treated with radiotherapy. Overview of abstract We first showed that lymphopenia during treatment is related to overall survival in lung cancer patients receiving curative-intent radiotherapy with/without chemotherapy. We then went on to identify regions in the body that are associated with lymphopenia after radiotherapy. To achieve this, wematched patients who developed lymphopenia and those who did not, based on their tumour volume, baseline lymphocyte count, prescribed radiotherapy dose (all of which are known to affect lymphocyte counts), and tumour histology. After matching, identified patients were used in image-based data mining. This methodology analysed the planned dose distribution across all patients to identify regions where dose is associated with lymphopenia. Our data mining approach found a region including heart, lung, and thoracic vertebrae where the difference in radiotherapy dose between lymphopenia yes/no groups was significant. Multivariablemodel for predicting lymphopenia, using radiotherapy dose parameters and other known factors, was proposed and further validated in a cohort of oesophageal cancer patients. What were the three main findings of your research? Firstly, lung cancer patients who develop lymphopenia during treatment have significantly worse overall survival compared to those without lymphopenia. Secondly, data mining from matched patients showed that irradiation of lung, heart, and thoracic vertebrae is associated with lymphopenia. Thirdly, after multivariable modelling, lymphopenia is found to be related to the radiotherapy dose delivered to thoracic vertebrae and heart, in combination with the baseline lymphocyte counts and radiotherapy duration. The same result is observed in the oesophageal cancer cohort. What impact could your research have? Radiotherapy-related side-effects can have a significant

negative impact on patients’ survival. In recent years, the thoracic oncology community has become increasingly aware of the impact of heart irradiation on morbidity and mortality in lung cancer patients. Emerging data has also highlighted that lymphopenia can adversely affect prognosis due to an increased risk of opportunistic infections. Identifying regions of the body associated with severe lymphopenia in lung cancer patients treated with radiotherapy will lead to personalized treatment and improved outcome. This can be overcome by reducing dose to identified regions, frequent monitoring of lymphocyte counts during therapy, and use of prophylactic antibiotics for patients at higher risk. Reducing the risk of lymphopenia may also be of importance in the era of immunotherapy and radiotherapy combinations. Is this research indicative of a bigger trend in oncology? In the field of Oncology, ‘learning fromevery patient treated’ is considered a complementary alternative to randomized clinical trials and is a fast-growing field. Retrospective data mining allows valuable information to be extracted from large volumes of routine data. It is important to develop and improve techniques to employ these massive amounts of information produced every day, intelligently. Particularly, methods of handling and maintain the rich 3-dimensional dosimetric information of every patient. Careful analysis of such data would lead us to better understand the mechanisms that affect individual patients following their care andwill allowdevelopment of new treatment strategies.

PHYSICS | Congress report

16

Physics

3. MLC tracking for lung cancer SABR is clinically feasible: results of first-in-human clinical trial (E38-0898)

Jeremy Booth, Vincent Caillet, AdamBriggs, Nick Hardcastle, Kathryn Szymura, Ricky O’Brien, Ben Harris, TomEade, Paul Keall

Northern Sydney Cancer Centre, Australia

Context of the study Real-time adaptive radiotherapy provides activemonitoring and correction for motion of lesions during treatment delivery. The technology has been available on specialist linacs but is not commercially available for standard c-arm linacs. The potential benefits of real-time adaptation are higher accuracy, lower target dose volumes, and lower organ at risk dose. In this study we investigated the feasibility of using real-time adaptive radiotherapy on a standard c-arm linac to treat lung cancer SABR patients. Overview of abstract This study demonstrated that standard radiotherapy linacs can be used tomonitor and adapt to respiration-induced lung cancer motion during high dose radiotherapy. Seventeen patients were treated with a real-time adaptation technique known as MLC tracking. Comparison planning against standard technique showed reduced target volumes, lower lung dose and higher accuracy determined as mean beam to tumour alignment. What were the three main findings of your research? Real-time adaptive radiotherapy can be delivered with high accuracy on standard linacs. The study shows that withMLC tracking the planning target volume (volume of lesion plus lungmargin receiving the prescription dose was reduced by an average of 25% (range 2% - 47%), the surrounding lung tissue dose was reduced by an average of 14% (range 9% - 42%) and the overall system accuracy is 1.2mm. The top row of the figure below shows planning dose is smaller for MLC tracking than standard planning which uses an Internal Target Volume (ITV). Across the bottom row, the figure shows that delivery is more accurate; with MLC tracking the dose cloud (indicating the prescribed dose) conforms to red line (targeted volume). With standard ITV-based treatment the colour cloud is missing part of the targeted volume indicating tumour under-dose.

Figure Dose clouds for MLC tracking planned (top left) and delivered (bottom left) show high fidelity in delivery under motion and smaller dose clouds representing less organ at risk exposure. Dose clouds for same patient planned with standard ITV-based method (top left) and simulated delivered dose (bottom right) shows dose cloud not extending to edge of the target red contour representing under-dose. What impact could your research have? This research concludes that standard linacs can effectively deliver real-time adaptive radiotherapy and that this technique offers high accuracy. Enabling real- time adaptive radiotherapy is complementary to online adaptive techniques and could be used reduce radiation to nearby healthy organs, to increase the viable tumour size for stereotactic radiotherapy, improve safety for patients and facilitate re-treatment or increased tumour doses. Is this research indicative of a bigger trend in oncology? With improvements incomputational power and radiotherapy technology, we are seeing increases in radiotherapy treatment plan adaptation to daily anatomical changes which may be due to patient shape, organ deformation/ filling, internal tumour size/shape/position and even patient condition. We push this further by showing it is feasible to adapt to the tumour position as it moves during treatment delivery.

Congress report | PHYSICS

17