A Retrospective Audit of the use of Genomic Profiling in a Breast Cancer Screening Centre

ABSTRACTS 2

Cambridge University Hospitals NHS Trust, UK Hutchison-MRC Research Centre, Cambridge, UK 4 Nottingham University Hospitals NHS Trust, UK 3

Background: Germline CDH1 (E-cadherin) mutation is present in 40% of individuals fulfilling the criteria for Hereditary Diffuse Gastric Cancer (HDGC) and confers 80% lifetime risk of gastric cancer. Current guidelines advise prophylactic gastrectomy for CDH1-positive individuals but endoscopic surveillance is recommended for patients delaying gastrectomy, due to comorbidity or psychosocial reasons, and for individuals from CDH1-negative families fulfilling HDGC criteria (van der Post et al J Med Genet 2015). This study aimed to determine the endoscopic yield of signet ring cell foci (SRCF) according to CDH1 mutation status. Method: A prospective cohort study of 54 CDH1+ve and 31 CDH1
ve patients fulfilling HDGC criteria undergoing 175 endoscopies (CDH1+ve: 103; CDH1
ve: 72) according to the Cambridge HDGC protocol (van der Post et al 2015). The median (IQR) follow up, in months, was 16.0 (3.5e39.0) (CDH1+ve: 11.0 (3.0e28.3); CDH1-ve: 27.0 (7.0e48.0)). Data on patient age, gender, ethnicity, H. pylori status and PPI use was collected. The outcome was time to detection of SRCF from first endoscopy. KaplaneMeier analysis with log rank test and Cox regression on CDH1 status, age and ethnicity were performed. Results: Between the CDH1+ve and
ve group, there was no significant difference in gender (p ¼ 0.637), H. pylori status (p ¼ 0.094) or PPI use (p ¼ 0.810), but there was a significant difference in median age (34.5 vs 45, p ¼ 0.026) and ethnicity (18.5% vs 0% Asian, p ¼ 0.012). CDH1+ve patients had significantly higher progression to endoscopic detection of SRCF (p ¼ 0.000) than CDH1
ve patients, with over 12fold increased risk of SRCF on endoscopy (HR 12.2, 95% CI 2.8e52.8, p ¼ 0.001). On Cox regression, age (p ¼ 0.811) and ethnicity (p ¼ 0.733) were not associated with risk of SRCF. In the CDH1
ve group, only 2 related individuals ever had SRCF detected. Conclusion: There is very low endoscopic yield of SRCF in CDH1
ve patients. The value and frequency of surveillance in this group should be reconsidered. http://dx.doi.org/10.1016/j.ejso.2016.07.113

399. A comparison of the prognostic value of the adaptive, Tlymphocyte and generalised inflammatory infiltrate in patients with primary operable colorectal cancer Fiona Ross, Campbell Roxburgh, Donald McMillan, Paul Horgan, James Park University of Glasgow, UK Background: The local inflammatory response (LIR) is an important determinant of disease progression and outcome in patients with colorectal cancer (CRC). Various measures of the LIR have been described; whereas some, such as the KlintrupeM€akinen (KM) grade describe the density of the generalised LIR, others such as Immunoscore reflect T-lymphocyte density. The aim of this study was to examine whether differing approaches offer complimentary prognostic value when used in combination in patients undergoing CRC resection. Method: Patients who underwent resection of stage IeIII CRC from 1997e2008 were identified from a prospectively maintained database. The generalised LIR at the invasive margin (IM) was assessed on H&E sections using KM grade (low grade vs. high grade). The density of mature (CD3+) and cytotoxic (CD8+) T-lymphocytes at the IM was assessed using immunohistochemistry and classified as low or high. Relationship with five-year cancer-specific survival was examined. Results: 246 patients were included in the final analysis. KM grade, CD3+and CD8+densities were strongly related (P < 0.001). The KM grade stratified survival from 91% to 71% at five years (P < 0.001), whereas CD3+ and CD8+ density both stratified survival from 87% to 71% (P ¼ 0.012 and P < 0.001 respectively).

S247 Combined assessment of CD3+ and CD8+ density and KM grade offered differing prognostic value: CD3+ density stratified survival of patients with a high KM (97% to 73%, P ¼ 0.002) but not a low KM (P ¼ 0.224). Conversely, CD8+ density stratified survival of patients with a low KM (81% to 66%, P ¼ 0.015) but not those with a high KM (P ¼ 0.159). Conclusion: In the present study, the prognostic value of markers of the adaptive immune response, as measured by CD3+ and CD8+ T-lymphocyte density, differed depending on the density of the generalised LIR. This would suggest that not only type of immune cell and density, but also the immune contextures are important determinants of outcome in patients with CRC. http://dx.doi.org/10.1016/j.ejso.2016.07.114 401. Characteristics of older patients receiving cancer chemotherapy Mary Denholm, Francesca Holt, Pippa Corrie Cambridge University Hospital NHS Foundation Trust, UK Background: Due to an ageing population, increasing numbers of older patients are being referred for cancer chemotherapy. The proportion of patients over the age of 75 entering clinical trials is tiny so little is known about the true risks and benefits of treating older people. We undertook a review of our local chemotherapy dataset, in order to benchmark our current practice and consider future requirements of an oncogeriatric service. Method: The electronic medical records of patients over 75 years old receiving chemotherapy at Addenbrooke’s hospital between November 2014 and November 2015 were reviewed. The following patient characteristics were determined: age, sex, ECOG performance status (PS), cancer diagnosis, treatment intent, number of medications, updated Charlson Co-morbidity Index (CCI), independence with personal activities of daily living (PADL). Results: 311 patients receiving chemotherapy had an average age of 80 years, range: 75 95 years, 53% were male. 161 (52%) patients had a solid tumour diagnosis, 150 (47%) had a haematological malignancy. 274 (88%), were treated with non-curative/palliative intent, 37 (12%) with curative intent. PS was recorded in 139 (45%) cases and the split was 0 (21%), 1 (62%), 2 (14%), 3 (3%). 208 (67%) patients were independent with PADL. The average number of medications taken was 5. The average CCI was 0.9. Conclusion: The majority of older patients receiving chemotherapy were being treated with palliative intent. ECOG PS was not available for over half of patients, which may reflect poor documentation, or difficulty categorising older people using this scale. One third of patients were not independent with PADL, while polypharmacy and the CCI of 0.9 are all triggers for justifying comprehensive geriatric assessment in at least some of these patients. This review supports the need for better tools to assess older patients receiving chemotherapy and closer integration of oncology and geriatric services in the future. http://dx.doi.org/10.1016/j.ejso.2016.07.115 402. A Retrospective Audit of the use of Genomic Profiling in a Breast Cancer Screening Centre Peter Lion, Susan Cleator, S. Shousha Imperial College Healthcare NHS Trust, UK Background: The survival benefit from adjuvant chemotherapy for ER positive and HER2 negative N0/N1 breast cancer is often small and risks can outweigh benefits. A number of genomic profiling tools have been developed to help guide clinical decision making for these patients. Funding is currently available in the UK for Oncotype DX profiling for ‘intermediate risk’ N0 cases; in our unit it is offered to patients with an absolute benefit from third generation chemotherapy of 3%e8%. The

S248 Endopredict assay has also been shown to identify cases with N0 and additionally N1 disease at low risk of recurrence with endocrine therapy alone within. Our unit currently additionally has funding to undertake Endopredict testing on all postmenopausal cases with N1 disease considered a suitable candidate for chemotherapy and with an estimated absolute benefit from 3rd generation chemotherapy of 3. Methods: Data was collected retrospectively for all patients diagnosed and treated for screen-detected or symptomatic breast cancers at our centre over a 12 month period. Patients with ER negative, Her2 positive, N2 and metastatic disease were excluded. Additionally, cases with significant comorbidities or who were over the age of 70 were excluded. Of the remaining patients, the number eligible for Oncotype DX testing was assessed. Additionally the number of post-menopausal women with N1 disease considered eligible for Endopredict was assessed. Results: Of the 586 new diagnoses made at our centre, 366 patients completed their treatment locally. From 366 patients, 102 had ER positive/ HER2 negative, N0 disease. Of these, 15(4%) were eligible and suitable for Oncotype DX testing under current NICE guidance. We additionally identified 13(3.6%) post-menopausal patients with N1 nodal disease who were suitable for Endopredict. A further 6(1.6%) patients were excluded from Endopredict only on the grounds of being pre-menopausal. In addition, if the upper limit survival benefit criteria had been removed, a further 14 (3.8%) patients would have become eligible. Conclusion: Under current guidance, the proportion of patients with N0 disease selected for genomic profiling within NICE guidance is small. However the numbers would be significantly increased by changes to the selection criteria. http://dx.doi.org/10.1016/j.ejso.2016.07.116

413. Clinical characteristics and treatment outcome of small cell cancer of the bladder; results from a 10-year experience from 3 UK cancer centres Caroline Chau1, Rosalind Jarvis2, Chern Lee3, Hannah Markham3, Wessam Al-Utayem2, Neerja Agarwal1, Tom Geldart2, Simon Crabb3 1 Porstmouth Hospital NHS Trust, UK 2 Poole Hospital NHS Foundation Trust, UK 3 University Hospital Southampton NHS Foundation Trust, UK Background: Small cell carcinoma (SCC) of the bladder is rare, accounting for less than 1% of all urinary bladder cancers. It is aggressive and outcomes are often poor. A National Institute of Health and Care Excellence (NICE) guideline on bladder cancer was published in 2015 but specifically excluded the management of SCC of bladder. There is therefore no standard therapeutic approach for this patient cohort defined for UK practice. Method: We conducted a retrospective case record analysis of all patients presenting with SCC of the bladder in 3 cancer centres in South Central England between 1/1/2006 and 1/1/2016. We report on the patient characteristics, treatment received and clinical outcome. Data cut-off date is 6/3/2016. Results: 52 eligible patients were identified. 42/52 patients (81%) were male, with a sex ratio of 3.8:1. Median age was 76.9 (range 41e93). With a median follow up of 16.9 months (range 4.0e101.0), 40/52 patients (77%) have died. Median overall survival was 12.9 months (95% CI 12.0e13.7 months). 34/52 (65%) had pure SCC and 18/52 had mixed transitional cell carcinoma (TCC) with SCC. 28/52 (54%) presented with localised disease (N0M0); within this group 12/28 (43%) had cystectomy, 6/28 (21%) received radiotherapy, 3/28 (11%) received chemotherapy only, 5/28 (18%) were deemed unfit for treatment, and 2/28 (7%) declined treatment. 30/52 (58%) received primary chemotherapy; the regimens used were cisplatin/ gemcitabine, cisplatin/ etoposide and carboplatin/ etoposide. At data cut-off date, 7/52 (13%) were relapse free. Median time from cystectomy to relapse was 8.33 months (range 1.71e27.5), with 87% relapsed

ABSTRACTS within 12 months. There was no standardised treatment strategy for these patients between the 3 cancer centres. Conclusion: The prognosis for SCC of the bladder is poor. Practice appears to be less consistent than for TCC. National guidance is needed to help streamline appropriate and timely care for these patients. http://dx.doi.org/10.1016/j.ejso.2016.07.118

423. The implementation of focused performance benchmarking as a method of regional quality improvement in breast cancer care Elizabeth Appleton1, Naik Jay2 1 Leeds Teaching Hospitals, UK 2 Mid Yorks NHS Trust, UK Background: In the drive for high quality, cost-effective healthcare, performance benchmarking has become an integral part of quality improvement across the NHS. The NICE quality standard for breast cancer contains 13 auditable standards, each with a defined set of quality measures allowing benchmarking. Audit across all standards is time-consuming, and can be difficult to coordinate across care providers. We were therefore interested in implementing and assessing the feasibility of focused benchmarking against these standards across the Yorkshire Cancer Network (YCN), as a method of facilitating more continuous analysis of services. Method: All standards were audited in the Mid Yorkshire NHS trust (MYHT) over the preceding two years, this enabled definition of key outcomes for a future audit. An audit tool was adapted from NICE, which aimed to enhance wider participation through accessible data entry, and standardise outputs. Engagement was then encouraged through Network Meetings. Results: Data was collected locally and analysed centrally for a relatively large number of patients (312) in a comparatively short time frame, across four NHS trusts. Clinical outcomes were largely comparable across sites, and results were discussed at a Network meeting. Participation encouraged communication and collaboration between care providers, highlighting areas for development, and allowing comparison of working practices to drive quality improvement. Although each audit is limited, the outcomes were adapted to reflect the learning process and through collaboratively defined priorities. Limitations: Engagement was a key challenge, with not all eligible trusts participating. An element of inter-observer variability was also identified in the cross-site data collection. Conclusion: We have shown that it is feasible to establish a focused, cancer network-wide audit of breast cancer care, enabling a more continuous, evolving audit process in the future and encouraging collaboration. As automatic data collection increases, the key value may be in ensuring the most clinically relevant data is collated and analysed. http://dx.doi.org/10.1016/j.ejso.2016.07.119

424. Differentiating tumour progression from pseudoprogression in patients with glioblastoma using multiparametric MRI imaging: Data from Barts Health NHS trust London Rachel Lewis, Ashwin Bhandari, Edward McKintosh, Piers Plowman, Joe Lansley, Jane Evanson, Anant Krishnan Barts Health NHS Trust, London, UK Background: Conventional imaging using contrast-enhanced MRI is not sufficiently accurate to differentiate between progression and pseudoprogression in patients with glioblastoma (GBM), who have had radiotherapy with concurrent temozolomide. Advanced MRI techniques currently help identify transforming low grade gliomas. Dynamic susceptibility contrast perfusion MRI provides information on tumour blood volume and flow, which increases with tumour progression. Proton MR Spectroscopy adds further information about