Search

Aim Approximately 5% of all mucosal head and neck (H&N) squamous cell cancers (SCC) arise from the hypopharynx. Patients with hypopharyngeal SCC (HPSCC) tend to have a poor prognosis compared with other subsites with reported 5-year survival of 27% in the UK. Most patients (80%) have stage III/IV disease at presentation. There are very few HPSCC-specific studies and this subsite is not well represented in more general H&N SCC trials. Thus, deciding the best treatment plan is difficult and relies on the expertise of an experienced multi-disciplinary team (MDT). Patient fitness for treatment complicates matters further. The incidence of H&N cancer increases with age and is closely correlated with deprivation. The aim of this series was to review outcomes of patients with HPSCC in our cancer network. Method This retrospective study included all patients with a histological or radiological diagnosis of HPSCC made from August 2016 to August 2018. They were identified from the cancer network MDT database. Subsites included pyriform fossa, post cricoid and posterior pharyngeal wall. Data including patient demographics, treatment details, toxicity and disease control were extracted from case records. Results 118 patients were evaluable. 8 (6.7%) patients had a radiological diagnosis, the remainder were biopsy proven HPSCC. The probability of survival at 24 months was higher in patients of good performance status (PS 0-1: 41.7%, 95% CI 29.7-53.2% Vs. PS ≥2: 27.5%, 95% CI 13.9-43.0%). Patients aged >70 years had a lower probability of survival at 24 months compared to those

A new method for automatic liver tumour segmentation from computed tomography (CT) scans based on deep neural network is presented. Two cascaded deep convolu-tional neural networks are used to segment the CT image of the abdominal cavity. The first U-net is used for coarse segmentation to obtain the approximate position of the liver and tumour. Using this as a prediction the original image is cropped to reduce its size in order to increase the segmentation accuracy. The second modified U-net is employed for accurate segmentation of the actual liver tumours. Residual modules and dense connections are added to U-net to help the network train faster while pro-ducing more accurate results. In addition, multi-dimensional information fusion is introduced to make the network more comprehensive in restoring information. The Liver Tumour Segmentation (LiTs) dataset is used to evaluate the relative segmenta-tion performance obtaining an average dice score of 0.665 based our method.

Background FOLFIRINOX is a standard treatment for metastatic pancreatic cancer patients. The effectiveness of neoadjuvant FOLFIRINOX in patients with borderline resectable pancreatic cancer (BRPC) remains debated. Methods We performed a systematic review and patient-level meta-analysis on neoadjuvant FOLFIRINOX in patients with BRPC. Studies with BRPC patients who received FOLFIRINOX as first-line neoadjuvant treatment were included. The primary endpoint was overall survival (OS), and secondary endpoints were progression-free survival, resection rate, R0 resection rate, and grade III–IV adverse events. Patient-level survival outcomes were obtained from authors of the included studies and analyzed using the Kaplan-Meier method. Results We included 24 studies (8 prospective, 16 retrospective), comprising 313 (38.1%) BRPC patients treated with FOLFIRINOX. Most studies (n = 20) presented intention-to-treat results. The median number of administered neoadjuvant FOLFIRINOX cycles ranged from 4 to 9. The resection rate was 67.8% (95% confidence interval [CI] = 60.1% to 74.6%), and the R0-resection rate was 83.9% (95% CI = 76.8% to 89.1%). The median OS varied from 11.0 to 34.2 months across studies. Patient-level survival data were obtained for 20 studies representing 283 BRPC patients. The patient-level median OS was 22.2 months (95% CI = 18.8 to 25.6 months), and patient-level median progression-free survival was 18.0 months (95% CI = 14.5 to 21.5 months). Pooled event rates for grade III–IV adverse events were highest for neutropenia (17.5 per 100 patients, 95% CI = 10.3% to 28.3%), diarrhea (11.1 per 100 patients, 95% CI = 8.6 to 14.3), and fatigue (10.8 per 100 patients, 95% CI = 8.1 to 14.2). No deaths were attributed to FOLFIRINOX. Conclusions This patient-level meta-analysis of BRPC patients treated with neoadjuvant FOLFIRINOX showed a favorable median OS, resection rate, and R0-resection rate. These results need to be assessed in a randomized trial.

The aim of this study was to evaluate the long-term safety of the omission of immediate neck dissections (IND) in patients with human papillomavirus (HPV)-positive head and neck squamous cell carcinoma (HNSCC) achieving a less than complete nodal response on 12-week FDG PET-CT.Patients with HPV-positive, node-positive HNSCC that were treated with radical (chemo) radiotherapy (RT) between January 2013 and September 2019 were identified. PET-CT responses were classified as complete (CR), incomplete (ICR) or equivocal (EQR) nodal responses. Clinical outcomes were obtained.347 patients were identified. Median follow-up was 43.9 (IQR, 30.8-61.2) months. 62.8% (218/347) achieved a CR, 23.4% (81/347) EQR and 13.8% (48/347) ICR nodal response. 70 of 81 (86.4%) patients with an EQR and 25 of 48 (52.1%) with an ICR had no residual disease during follow up (a pathologically negative ND if surgery undertaken or no subsequent neck or distant relapse clinically/radiologically). Median survival of the EQR and CR groups were not reached, and despite the omission of IND in 95% of the EQR group there was no statistically significant differences in overall survival (OS) between the groups, p = 1.0. Median survival of ICR was not reached. However, OS for ICR group was significantly worse than that of CR, and EQR, both p 0.001.The omission of IND in those achieving an EQR nodal response does not compromise long-term survival. This supports the safety of extended surveillance in patients with HPV-positive disease and an EQR on 12-week FDG PET-CT.

A novel encoder-decoder deep learning network called TwoPath U-Net for multi-class automatic brain tumor segmentation task is presented. The network uses cascaded local and global feature extraction paths in the down-sampling path of the network which allows the network to learn different aspects of both the low-level feature and high-level features. The proposed network architecture using a full image and patches input technique was used on the BraTS2020 training dataset. We tested the network performance using the BraTS2019 validation dataset and obtained the mean dice score of 0.76, 0.64, and 0.58 and the Hausdorff distance 95% of 25.05, 32.83, and 37.57 for the whole tumor, tumor core and enhancing tumor regions.

Funder: DH | National Institute for Health Research (NIHR); doi: https://doi.org/10.13039/501100000272, Funder: Cancer Research UK (CRUK); doi: https://doi.org/10.13039/501100000289, The coronavirus disease 2019 (COVID-19) pandemic epicentre has moved to the USA and Europe, where it is placing unprecedented demands on healthcare resources and staff availability. These service constraints, coupled with concerns relating to an increased incidence and severity of COVID-19 among patients with cancer, should lead to re-consideration of the risk-benefit balance for standard treatment pathways. This is of particular importance to pancreatic cancer, given that standard diagnostic modalities such as endoscopy may be restricted, and that disease biology precludes significant delays in treatment. In light of this, we sought consensus from UK clinicians with an interest in pancreatic cancer for management approaches that would minimise patient risk and accommodate for healthcare service restrictions. The outcomes are described here and include recommendations for treatment prioritisation, strategies to bridge to later surgical resection in resectable disease and factors that modify the risk-benefit balance for treatment in the resectable through to the metastatic settings. Priority is given to strategies that limit hospital visits, including through the use of hypofractionated precision radiotherapy and chemoradiotherapy treatment approaches.

BackgroundPatients with cancer are considered at higher risk of COVID-19 infection and increased severity of infection. Anti-cancer treatment may further increase those risks.The aim of this work is to report early outcomes in patients with head and neck cancer (HNC) treated during the pandemic.Materials and MethodsA retrospective cohort study in a UK tertiary level oncology centre between 1st March and 23 June 2020, including patients with HNC who were either newly diagnosed, had developed new recurrent/metastatic disease, or were already scheduled to receive treatment during that period.Results200 patients were evaluated. Median age was 64 years, 65.5% had multiple co-morbidities, 77.5% were current or ex-smokers and 59.5% lived in areas of deprivation. 99 patients were treated with 6 weeks of radical (chemo) radiotherapy. Systemic anti-cancer treatment was delivered to 40 patients.2 (1.0%) patients with HNC had confirmed COVID-19 infection; 1 patient prior to primary radical RT - no delay to treatment was required and RT was completed as planned, 1 patient acquired COVID-19 after primary surgery but recovered well and started adjuvant RT 9.7 weeks after surgery.The proportion of patients receiving supportive care only (19.5%) was in keeping with that pre-COVID-19. The proportion of patients not completing (chemo) radiotherapy (3.4%) or with gaps in treatment (14.1%) was similar to pre-COVID-19. 30-day mortality after radical (chemo)radiotherapy was 2.3%, no higher than in previous years.ConclusionsIt is feasible and safe to deliver standard treatment for patients with HNC during the COVID-19 pandemic.

Background and purpose Following resection of pancreatic cancer, risk of positive margins and local recurrence remain high, especially for borderline-resectable pancreatic cancer (BRPC). We aimed to establish the maximum tolerated dose of a margin-intensified five-fraction stereotactic body radiotherapy (SBRT) regimen designed to treat the region at risk. Materials and methods We conducted a prospective multicentre phase-1 rolling-six dose-escalation study. BRPC patients received pre-operative SBRT, with one dose to the primary tumour and an integrated boost to the region where tumour was in contact with vasculature. Four dose-levels were proposed, with starting dose 30 Gy to primary PTV and 45 Gy to boost volume (PTV_R), in five daily fractions. Primary endpoint was maximum tolerated dose (MTD), defined as highest dose where zero of three or one of six patients experienced dose-limiting toxicity (DLT). Results Twelve patients were registered, eleven received SBRT. Radiotherapy was well tolerated with all treatment completed as scheduled. Dose was escalated one level up from starting dose without encountering any DLT (prescribed 32.5 Gy PTV, 47.5 Gy PTV_R). Nine serious adverse reactions or events occurred (seven CTCAE Grade 3, two Grade 4). Two patients went on to have surgical resection. Median overall survival for SBRT patients was 8.1 months. The study closed early when it was unable to recruit to schedule. Conclusion Toxicity of SBRT was low for the two dose-levels that were tested, but MTD was not established. Few patients subsequently underwent resection of pancreatic tumour after SBRT, and it is difficult to draw conclusions regarding the safety or toxicity of these therapies in combination.

Highlights • Dose calculations between synthetic CT and planning CT acquired at the same treatment fraction have small differences. • The safe omission of repeat CT during H&N RT is feasible when CBCT is available. • Synthetic CT verification appears to overestimate the need to re-plan, therefore not putting patients at risk. • Repeat CT should be used for replanning, being guided by CBCT and synthetic CT., Objectives The study aimed to assess the suitability of deformable image registration (DIR) software to generate synthetic CT (sCT) scans for dose verification during radiotherapy to the head and neck. Planning and synthetic CT dose volume histograms were compared to evaluate dosimetric changes during the treatment course. Methods Eligible patients had locally advanced (stage III, IVa and IVb) oropharyngeal cancer treated with primary radiotherapy. Weekly CBCT images were acquired post treatment at fractions 1, 6, 11, 16, 21 and 26 over a 30 fraction treatment course. Each CBCT was deformed with the planning CT to generate a sCT which was used to calculate the dose at that point in the treatment. A repeat planning CT2 was acquired at fraction 16 and deformed with the fraction 16 CBCT to compare differences between the calculations mid-treatment. Results 20 patients were evaluated generating 138 synthetic CT sets. The single fraction mean dose to PTV_HR between the synthetic and planning CT did not vary, although dose to 95% of PTV_HR was smaller at week 6 compared to planning (difference 2.0%, 95% CI (0.8 to 3.1), p = 0.0). There was no statistically significant difference in PRV_brainstem or PRV_spinal cord maximum dose, although greater variation using the sCT calculations was reported. The mean dose to structures based on the fraction 16 sCT and CT2 scans were similar. Conclusions Synthetic CT provides comparable dose calculations to those of a repeat planning CT; however the limitations of DIR must be understood before it is applied within the clinical setting.

Radiotherapy-induced xerostomia (RIX) is a common and untreatable side effect of radiotherapy to the head and neck. Visco-ease™ mouth spray (Lamellar Biomedical Ltd), a new product that is made from lamellar body mimetics, reduces the viscosity of saliva ex vivo. The purpose of this study was to evaluate its safety and effectiveness in the treatment of RIX in 43 patients with cancer of the head and neck. They were randomised into the Visco-ease™ or placebo groups, and asked to complete the Groningen radiotherapy-induced xerostomia (GRIX) questionnaire each week. The primary endpoint was a change in GRIX score from baseline to end of treatment. There was no difference in scores between the two groups, and none of the patients had device-related serious adverse events. Visco-ease™ oral spray was safe and tolerable but no better than placebo in reducing RIX in this group of patients.

Background: Despite being relatively rare pancreatic cancer is one of the highest causes of death. Even within the potentially resectable group outcomes are poor. We present our initial experiences utilising a neoadjuvant approach to localised pancreatic cancer, evaluating survival, response rates and tolerability.\ud \ud Methods: This was a retrospective analysis of a prospectively maintained database. Patients from 2012 to 2015 referred to a busy regional Hepato-Pancreatic Biliary (HPB) MDT were included. Patients were classified according to respectability criteria (utilising NCCN guidelines) and a treatment plan agreed. Systemic therapy with either FOLFIRINOX or Gem/Cap was delivered followed by chemoradiotherapy if disease remained localised. Toxicity, response, pathological outcomes and survival were all recorded.\ud \ud Results: A total of 85 patients were included in the study: 45 had initially resectable disease; 19 required a response for resection and 21 had locally advanced inoperable disease; 34 patients underwent resection. The median survival for the potentially resectable group was 22.2 months while for those undergoing resection it was 37 months.\ud \ud Conclusions: We have demonstrated that a neoadjuvant approach is deliverable and tolerable. In addition we have demonstrated impressive survival results in patients undergoing resection with no detriment in outcome for those not proceeding to surgery.

Introduction and background:\ud \ud A significant proportion of patients with intermediate and high risk squamous cell cancer of the oropharynx (OPSCC) continue to relapse locally despite radical chemoradiotherapy (CRT). The toxicity of the current combination of intensified dose per fraction radiotherapy and platinum based chemotherapy limits further uniform intensification. If a predictive biomarker for outcomes from CRT can be identified during treatment then individualised and adaptive treatment strategies may be employed.\ud \ud Methods/design:\ud \ud The MeRInO study is a prospective observational imaging study of patients with intermediate and high risk, locally advanced OPSCC receiving radical RT or concurrent CRT Patients undergo diffusion weighted MRI prior to treatment (MRI_1) and during the third week of RT (MRI_2). Apparent diffusion coefficient (ADC) measurements will be made on each scan for previously specified target lesions (primary and lymph nodes) and change in ADC calculated. Patients will be followed up and disease status for each target lesion noted. The primary aim of the MeRInO study is to determine the threshold change in ADC from baseline to week 3 of RT that may identify the sub-group of non-responders during treatment.\ud \ud Discussion:\ud \ud The use of DW-MRI as a predictive biomarker during RT for SCC H&N is in its infancy but studies to date have found that response to treatment may indeed be predicted by comparison of DW-MRI carried out before and during treatment. However, previous studies have included all sub-sites and biological sub-types. Establishing ADC thresholds that predict for local failure is an essential step towards using DW-MRI to improve the therapeutic ratio in treating SCC H&N. This would be done most robustly in a specific H&N sub-site and in sub-types with similar biological behaviour. The MeRInO study will help establish these thresholds in OPSCC.

A new neural network for automatic head and neck cancer (HNC) segmentation from magnetic resonance imaging (MRI) is presented. The proposed neural network is based on U-net, which combines features from different resolutions to achieve end-to-end locating and segmentation of medical images. In this work, the dilated convolution is introduced into U-net, to obtain larger receptive field so that extract multi-scale features. Also, this network uses Dice loss to reduce the imbalance between classes. The proposed algorithm is trained and tested on real MRI data. The cross-validation results show that the new network outperformed the original U-net by 5% (Dice score) on head and neck tumour segmentation.

Background: There is uncertainty over optimal management of locally advanced non-metastatic oesophageal and gastric (OG) adenocarcinomas which are deemed irresectable at time of diagnosis due to local tumour or nodal burden. Current practice in our regional centre is to administer chemotherapy in a “downstaging” strategy in the hope of achieving tumour shrinkage to allow radical treatment. Patients without sufficient response to downstaging are treated palliatively. The aim of this study was to review our single unit outcomes of this treatment strategy. Methods: Data was collected retrospectively from electronic patient records on all cases discussed at regional MDT over a 32-month period (January 2015–August 2017). Results: A total of 44 patients [70.5% male, median age 70 years, 13 (29.5%) oesophageal, 12 (27.3%) junctional and 19 (43.2%) gastric] were included in the study. Thirty-six (81.8%) of patients received the full number of planned cycles of chemotherapy; toxicity and disease progression (both 6.8% of cases) were the most common reasons for early cessation of treatment. Seventeen (38.6%) patients underwent resection and an R0 resection was achieved in 13 (76.5%) of these patients. After median follow up of 16.8 months, the median overall survival (OS) in the resection vs. palliative cohorts was 42.6 vs. 16.4 months (P Conclusions: Our data show that a downstaging approach can be successfully implemented (R0 resection achieved) in up to a third of patients with good survival results. Further prospective data identifying patient and pathological characteristics predicting response to treatment are needed to optimise selection into a downstaging programme.

Surveillance PET-CT scans at 12 weeks post-radiotherapy for head and neck cancer can be used to omit neck dissections with no detriment in overall survival. Human Papillomavirus (HPV) driven tumours behave differently on conventional imaging after radiotherapy but it is unknown if this effect is seen on PET-CT and if HPV status affects the accuracy of PET-CT. We aimed to determine the negative and positive predictive values (NPV and PPV) of 12 week surveillance PET-CT in HPV positive and negative tumours, and investigate predictors of relapse in equivocal responders.A retrospective cohort study in a UK tertiary level oncology hospital, between 2013 and 2016 included adults with oropharyngeal squamous cell carcinoma, or HPV positive head and neck squamous cell cancers of unknown primary, treated with radiotherapy.The PPVs of 12 week PET-CT in HPV positive and negative disease are 30% and 81.8% respectively (p 0.01). The NPVs of 12 week PET-CT in HPV positive and negative disease are 92.9% and 55.6% respectively (p 0.01). 67% of HPV positive patients with equivocal responses on 12 week PET-CT achieved complete response by 24 weeks. Equivocal responses in HPV positive disease had statistically similar survival to patients with complete responses. Comparing disease and imaging characteristics, there were no predictors of residual tumour.HPV positive tumours have a poor PPV of 30% on 12 week surveillance PET-CTs and take longer to achieve complete response. A period of further surveillance can be considered instead of an immediate neck dissection in this group of patients.

A novel algorithm for automatic head and neck 3D tumour segmentation from magnetic resonance imaging (MRI) is presented. The proposed algorithm pre-processes the MRI data slices to enhance quality and reduce artefacts. An intensity standardisation process is performed between slices, followed by cancer region segmentation of central slice, to get the correct intensity range and rough location of tumour regions. Fourier interpolation is applied to create isotropic 3D MRI volume. A new location-constrained 3D level set method segments the tumour from the interpolated MRI volume. The proposed algorithm is tested on real MRI data. The results show that the novel 3D tumour volume extraction algorithm has an improved dice score and F -measure when compared to the previous 2D and 3D segmentation method.

Pre-trial radiotherapy quality assurance (RTQA) for the SPARC trial of preoperative pancreas SBRT highlighted overgenerous target contouring. We aimed to assess for improvement following a training workshop and appraise the effectiveness of a novel web-conferencing format for on-trial RTQA.

BackgroundOesophageal cancer remains a common cause of cancer mortality worldwide. Increasingly, oncology centres are treating an older population and comorbidities may preclude multimodality treatment with chemoradiotherapy (CRT). We review outcomes of radical radiotherapy (RT) in an older population treating squamous cell carcinoma (SCC) oesophagus.MethodsPatients over 65 years receiving RT for SCC oesophagus between 2013 and 2016 in the West of Scotland were identified. Kaplan-Meier and Cox-regression analysis were used to compare overall survival (OS) between patients treated with radical RT and radical CRT.ResultsThere were 83 patients over 65 years treated with either RT (n=21) or CRT (n=62). There was no significant difference in median OS between CRT versus RT (26.8 months vs 28.5 months, p=0.92). All patients receiving RT completed their treatment whereas 11% of CRT patients did not complete treatment.ConclusionSurvival in this non-trial older patient group managed with CRT is comparable to that reported in previous trials. RT shows better than expected outcomes which may reflect developments in RT technique. This review supports RT as an alternative in older patients, unfit for concurrent treatment.

709 Background: Olaparib is a potent inhibitor of PARP-1, which has a critical role in signalling DNA single strand breaks (SSB) as part of the base excision repair pathway, and may have radio-sensitizing effects due to impaired resolution of radiation induced SSB. We hypothesize that O may potentiate the effects of X-CRT in pts with LAPC. Methods: Eligible pts with LAPC, ECOG < 1, tumor diameter < 6cm, with stable disease (SD) or response after 12 weeks’ induction chemotherapy, were treated with 1 of 4 escalating doses of O given bid po starting on day -3, and then in combination with X (830 mg/m2 bid) and radiation (50·4 Gy in 28 fractions) all administered Mon-Fri. Dose limiting toxicities (DLT) were determined on clinical and lab toxicity assessments (NCI-CTC AE v4.03) performed weekly from the start of O until completion of O plus X-CRT (i.e. 6 weeks). Dose escalation continued with a rolling-six design until the Maximum Tolerated Dose (MTD) was reached. Blood samples for PK analyses of O and PD measurement (inhibition of PARP activity) were collected on day -3 (O monotherapy) and during week 1 of O + X-CRT. Results: 18 pts, (9 m, 9 f, ECOG 0/1 [n=6/12]), age range 49-81 (median=70) years, with histologic (14) or cytologic (4) proven LAPC, had received induction chemotherapy with gemcitabine [GEM] (n=2), GEM + X (12), or FOLFIRINOX (3) with partial response (n=4) or stable disease (14). Pts received 50 (3), 100 (4), 150 (6), or 200 (5) mgs bid of O with X+CRT. DLTs were observed in 2 pts (both at 200mgs bid): 1 pt with grade 3 nausea (on optimal anti-emetics) and grade 3 fatigue, 1 pt with grade 3 anorexia. 6 pts were subsequently recruited at 150mgs bid with no DLTs. No pts had complete response, 2 pts had partial response (1 pt each at 100 and 150 mgs bid) and 1 pt (at 100 mgs bid) had progressive disease; the remaining 14 pts had SD. Conclusions: The recommended dose (RP2) of O is 150mgs bid when given in combination with X + CRT in LAPC. Recruitment of up to 12 pts with borderline operable LAPC at the RP2 is ongoing. PK analyses of O, PD studies (PARP inhibition – PBMCs; cytokeratin 18 – serum; γH2AX foci – hair follicles), and exploratory predictive marker studies (tumor – NGS; RNA exome sequencing) are ongoing. Clinical trial information: ISRCTN10361292.

A novel algorithm for automatic 3D segmentation of magnetic resonance imaging (MRI) data for detection of head and neck cancerous lymph nodes (LN)) is presented in this paper. The proposed algorithm pre-processes the MRI data slices to enhance quality and reduce artefacts. A modified Fuzzy c-mean process is performed through all slices, followed by a probability map which refines the clustering results, to detect the approximate position of cancerous lymph nodes. Fourier interpolation is applied to create an isotropic 3D MRI volume. A new 3D level set method segments the tumour from the interpolated MRI volume. The proposed algorithm is tested on synthetic and real MRI data. The results show that the novel cancerous lymph nodes 3D volume extraction algorithm has over 0.9 Dice similarity score on synthetic data and 0.7 on real MRI data. The F-measure is 0.92 on synthetic data and 0.75 on real data.

Despite radical chemo-radiotherapy (CRT), many patients with intermediate and high risk SCC oropharynx (OPSCC) relapse. Treatment related toxicity limits further uniform intensification across the patient group. If a predictive biomarker for outcomes from CRT can be identified during treatment, individualised and adaptive treatment strategies may be employed for the non-responders. This is the 1st study to use DW MRI for early response assessment in a specific H&N sub-site with sub-type of similar biological behaviour.

Background Survival from lung cancer remains poor in Scotland, UK. It is believed that comorbidity may play an important role in this. The goal of this study was to determine the value of a novel comorbidity scoring system (SCSS) and to compare it with the already established Charlson Comorbidity Index and the modified Glasgow Prognostic Score (mGPS). We also wished to explore the relationship between comorbidity, mGPS and Performance Status (PS). In addition we investigated a number of standard prognostic markers and demographics. This study aimed to determine which of these factors most accurately predicted survival. Methods Between 2005 and 2008 all newly diagnosed lung cancer patients coming through the Multi-Disciplinary Teams (MDTs) in four Scottish Centres were included in the study. Patient demographics, World Health Organization/Eastern Cooperative Oncology Group performance status, clinico-pathological features, mGPS, comorbidity and proposed primary treatment modality were recorded. Univariate survival analysis was carried out using Kaplan–Meier method and the log rank test. Results This large unselected population based cohort study of lung cancer patients has demonstrated that a number of important factors have significant impact in terms of survival. It has gone further by showing that the factors which influence survival are different, depending upon the stage of cancer at diagnosis and the potential treatment strategy. The novel comorbidity scoring system, the SCSS, has compared very favourably with the more established CCI. Conclusion This study has identified that a variety of factors are independent prognostic determinants of outcome in lung cancer. There appear to be clear differences between the early and late stage groups.

A combined interpolation - 3D Level Set Method (I+3DLSM) based segmentation process is presented. The performance in terms of accuracy of the 3-dimensional (3D) level set method (LSM) in the segmentation of throat regions from highly anisotropic magnetic resonance imaging (MRI) volumes, with and without an interpolation step is evaluated. Qualitative and quantitative results from real MRI data suggest that performing interpolation, to reconstruct isotropic MRI volumes, prior to 3D LSM improves the accuracy of the segmentation results, compared to interpolation post 3D LSM and no interpolation at all.

Aims To carry out statistical validation of a newly developed magnetic resonance imaging (MRI) auto-contouring software tool for gross tumour volume (GTV) delineation in head and neck tumours to assist in radiotherapy planning. Materials and methods Axial MRI baseline scans were obtained for 10 oropharyngeal and laryngeal cancer patients. GTV was present on 102 axial slices and auto-contoured using the modified fuzzy c-means clustering integrated with the level set method (FCLSM). Peer-reviewed (C-gold) manual contours were used as the reference standard to validate auto-contoured GTVs (C-auto) and mean manual contours (C-manual) from two expert clinicians (C1 and C2). Multiple geometric metrics, including the Dice similarity coefficient (DSC), were used for quantitative validation. A DSC≥0.7 was deemed acceptable. Inter- and intra-variabilities among the manual contours were also validated. The two-dimensional contours were then reconstructed in three dimensions for GTV volume calculation, comparison and three-dimensional visualisation. Results The mean DSC between C-gold and C-auto was 0.79. The mean DSC between C-gold and C-manual was 0.79 and that between C1 and C2 was 0.80. The average time for GTV auto-contouring per patient was 8 min (range 6–13 min; mean 45 s per axial slice) compared with 15 min (range 6–23 min; mean 88 s per axial slice) for C1. The average volume concordance between C-gold and C-auto volumes was 86.51% compared with 74.16% between C-gold and C-manual. The average volume concordance between C1 and C2 volumes was 86.82%. Conclusions This newly designed MRI-based auto-contouring software tool shows initial acceptable results in GTV delineation of oropharyngeal and laryngeal tumours using FCLSM. This auto-contouring software tool may help reduce inter- and intra-variability and can assist clinical oncologists with time-consuming, complex radiotherapy planning.

TPS4166 Background: There is increasing evidence suggesting benefit from a neoadjuvant approach to PC. However, the optimal regimen is unclear and will likely require a precision medicine approach, where patient and tumor attributes define therapy. Platinum-containing regimens have shown survival benefit for PC, with occasional exceptional responders, but biomarkers (BM) of response are not well defined and treatment decisions are often based on patient performance status (PS) and co-morbidity. Tumors with defects in BRCA1/2and other Fanconi Anemia genes show defective DNA damage response (DDR), conferring potential selective sensitivity to DNA-damaging agents (e.g. platinum) and newer targeted agents. We have shown that DDR deficiency (DDRd) is present in up to 20% of PC. This study aims to exploit DDRd as a therapeutic vulnerability, with integrated analysis to define candidate BM for FA and AG response. Methods: PRIMUS-002 will enroll patients registered on the Precision-Panc Master Protocol who are molecularly profiled using the Precision-Panc Clinical Cancer Genome including a novel DDRd assay, and the transcriptome with longitudinal sampling (pre-, during, and post-treatment). Patients receive either FA ( nab-paclitaxel 150mg/m2IV,oxaliplatin 85mg/m2, folinic acid 350mg flat dose, fluorouracil infusion 2400mg/m2continuous IV infusion), orAG ( nab-paclitaxel 125mg/m2, gemcitabine 1000 mg/m2) for 3 months,based on patient age and PS.Following initial safety analysis, chemoradiation may be introduced. The primary endpoint is disease progression (DP) during neoadjuvant therapy. The study is designed to detect a 20% difference in DP between the BM+ve (10%) and BM –ve (30%) in patients treated with FA (90% power, 5% 1-sided level of statistical significance)., Exploratory translational endpoints include surrogate therapeutic response assessment using CA19.9, PET-CT SUV, DWI-MRI and ctDNA. Current Enrolment: 2 patients enrolled to date: 1 to receive FA and 1 to AG treatment. Clinical trial information: ISRCTN34129115.

Definitive chemo-radiotherapy (dCRT) has been advocated as an alternative to surgical resection for the treatment of locally advanced oesophageal cancer (OC). We have retrospectively reviewed 4 years' experience of patients (pts) who underwent contemporary staging and were treated with concurrent chemo-radiotherapy (dCRT) or single modality radical radiotherapy (RT) with curative intent.Retrospective analysis permitted identification of consecutive patients who underwent contemporary staging prior to non-surgical treatment for locally advanced oesophageal carcinoma. The primary outcomes were overall survival (OS) and disease-free survival (DFS), adjusted for baseline differences in age, tumour staging and histological cell type. All patients were treated with either dCRT or single modality RT within a single centre between 2009 and 2012.We identified 235 patients in total [median age 69.8 years, male =130 pts, female =105 pts, adenocarcinoma (ACA) =85 pts, squamous =150 pts]. A total of 190 pts received dCRT and 45 patients were treated with RT. All patients were staged with CT of chest, abdomen and pelvis, 226 patients underwent endoscopic ultrasound (EUS), and 183 patients had PET-CT. Patients treated with dCRT demonstrated longer OS (27 vs. 25 months respectively, P=0.02) and DFS (31 vs. 16 months respectively, P=0.01) compared to those treated with RT. More advanced tumour stage (stage 3 vs. stage 1/2) at presentation conferred poorer OS (32 vs. 38.2 months, P=0.02) and DFS (11 vs. 28 months, P=0.013). We demonstrated an acceptable toxicity profile with only 77 patients (32.8%) suffering grade 3 toxicity and 9 patients (4.2%) experiencing grade 4 toxicity by CTC criteria. The NG/PEG feeding rates were 4% across all treated patients.This retrospective analysis is in keeping with current treatment paradigms emphasising the importance and safety of concurrent CRT in maximising curative potential for patients undergoing non-surgical treatment of OC. Although retrospective, in comparison to similar retrospective series from both our centre and historical literature, this data suggest improvements in OS and DFS, possibly due to improved patient selection through the use of more effective tumour staging.

TPS4158Background: Platinum-containing regimens (e.g. FOLFIRINOX) have shown survival benefit for pancreatic cancer (PC), with some “exceptional responders”; biomarkers (BM) of response are not well defined. Tumors with mutations in BRCA1/2 and other Fanconi Anemia genes, are preferentially sensitive to DNA-damaging agents (e.g. platinum) through synthetic lethality mechanisms, as they are defective in DNA damage response (DDR). We have shown that DDR deficiency may be present in 20% of PC. PRIMUS-001 aims to exploit defective DDR as a therapeutic vulnerability, with integrated study of BM of DNA-damaging agent response. Methods: PRIMUS-001 will enroll patients who are profiled molecularly using Precision-Panc NGS Diagnostic (BM via bespoke clinical grade assay). It is a multicentre, randomized (1:1), open-label, phase II trial comparing FOLFOX-A (nab-paclitaxel 150mg/m2 IV, oxaliplatin 85mg/m2, folinic acid 350mg flat dose, fluorouracil infusion 2400mg/m2 continuous IV infusion) vs AG (nab-paclitaxel 125...