Disclosure Information Rebecca Fitzgerald holds patents for Cytosponge and related assays which have been licensed by the Medical Research Council to Covidien (now Medtronic)
Precision early diagnosis of oesophageal cancer using pill on a string Rebecca Fitzgerald MD. FMedSci Hon. Consultant Gastroenterologist, Cambridge University NHS Hospitals Trust Professor of Cancer Prevention and MRC Programme Leader MRC Cancer Unit, Hutchison-MRC Research Centre University of Cambridge
Life history of cancer Cell S Nik-Zainal et al 149;994-1007 2012
Main Types of Oesophageal Cancer Squamous cell carcinoma Men= women Tobacco smoke Alcohol Thermal injury Micronutrient deficiency Adenocarcinoma Men> women Reflux Obesity Dysphagia, weight loss, GI bleeding. F. Lordick. Oesophageal Cancer: ESMO Clinical Practice Guidelines. Ann Oncol (2016) 27 (suppl 5): v50-v57
Global annual incidence of oesophageal cancer Males Adenocarcincoma Squamous cell carcinoma Females Smyth, E. C.Fitzgerald RC et al. (2017) Oesophageal cancer Nat. Rev. Dis. Primers doi:10.1038/nrdp.2017.48
Trends for oesophageal adenocarcinoma Male Female UK Data on 114,494 EACs diagnosed in England between 1971 and 2013 courtesy Judith Offman
Survival from oesophageal cancer 480,000 new diagnoses per annum globally 450,000 cancer deaths per annum globally Overall survival 13% (worse in developing countries) Improving to 40% if detected early enough for curative therapy chemo, radiotherapy, surgery
Oesophageal cancer subtypes and precursor lesions Barrett s to adenocarcinoma Squamous dysplasia to Squamous cell carcinoma
Pathogenesis of oesophageal cancer Smyth, E. C. et al. (2017) Oesophageal cancer Nat. Rev. Dis. Primers doi:10.1038/nrdp.2017.48
Barrett s progression rates and survival Non-dysplastic Barrett s LGD HGD Intramucosal cancer Advanced Cancer Progression rate %/year 5-year survival 0.3 0.5-10 13-30 >90% 65% <15% (Desai Gut 2011, Hvid-Jensen NEJM, 2011, Duits, Gut 2014, Bhat, JNCI 2011)
NICE approved treatments for early cancer and dysplasia in 2015 NICE ruling for LGD 2 randomised controlled trials: Dysplasia trial Shaheen et al NEJM 2009; LGD SURF trial JAMA 2014
Reasons for LATE diagnosis of oesophageal adenocarcinoma - Patients and doctors wait until alarm" symptoms - Most patients at risk are not investigated - Endoscopy is invasive and expensive - White light endoscopy has limited resolution to detect focal dysplasia Oesophageal cancer is a public health concern and we need to develop an alternative safe, minimally invasive, affordable test Chief Medical Officer report 2008
Paradigm shift for early diagnosis in the esophagus: Cytosponge Technology Objective biomarker assays for diagnosis and risk stratification Non-endoscopic cell collection Collect along entire oesophagus and minimise sampling bias
William Croone 1680 "for the support of a lecture and illustrative experiment for the advancement of natural knowledge
Device considerations Low cost Ease of administration Stability of device (transportation and storage) Acceptability Kadri S.Fitzgerald RC BMJ 2010; 341: c4372 (BEST1) Ross-Innes Fitzgerald PLOS Medicine 2015; doi: 10.1371 (BEST2)
Biomarker critical for a clinical Cytosponge test Gene expression level(dct) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0-0.1 NE BE GC NE BE GC NE BE GC DDC HOXB5 Stomach Barrett s Sq. Oes TFF3 Antibody to TFF3 Lao-Sirieix et al. GUT, 2009
Biomarker considerations Storage and transport of sample Assay design and applicability (Immunohistochemistry vs sequencing) Assay to GCLP Scoring and assessment Cost of assay
Studies BEST1 (500 patients) Study design Primary care Patients with reflux Endoscopy as gold standard BEST2 (1,110 patients) Study design Hospital setting Barrett s cases and controls (reflux) Endoscopy as gold standard Main Outcomes - Feasibility in primary care - Safety - Acceptability Main Outcomes - Safety - Acceptability - Accuracy Kadri S.Fitzgerald RC BMJ 2010; 341: c4372 (BEST1) Ross-Innes Fitzgerald PLOS Medicine 2015; doi: 10.1371 (BEST2)
Studies BEST1 (500 patients) Study design Primary care Patients with reflux Endoscopy as gold standard BEST2 (1,110 patients) Study design Hospital setting Barrett s cases and controls (reflux) Endoscopy as gold standard Main Outcomes - Feasibility in primary care - Safety - Acceptability Main Outcomes - Safety - Acceptability - Accuracy Kadri S.Fitzgerald RC BMJ 2010; 341: c4372 (BEST1) Ross-Innes Fitzgerald PLOS Medicine 2015; doi: 10.1371 (BEST2)
Summary of data > 2,000 patients Safe Acceptable 80% preferred Cytosponge to endoscopy Often tolerated better than endoscopy (p=0.0003) median 6 (IQR 5-8)
Cytosponge-TFF3 performance Sensitivity 79.9% Sensitivity increases with length (84% 2 cm and 87% 3 cm) Sensitivity 94% if remove inadequate samples (as per cervical screening) Specificity 92-94% Kadri S.Fitzgerald RC BMJ 2010; 341: c4372 (BEST1) Ross-Innes Fitzgerald PLOS Medicine 2015; doi: 10.1371 (BEST2)
Health Economics analyses Benaglia T Gastroenterology. 2013 Jan; 144:62-73 2 microsimulation models using BEST2 trial data Screening patients with GERD by Cytosponge with follow-up endoscopy confirmation compared with endoscopy has favourable QALYs and ICERs CISNET Consortium Heberle CR.Hur C et al Clin Gastroenterol Hepatol 2017
Barrett s ESophagus Trial 3 https://www.best3trial.org/ Randomised Trial in Primary Care n = 10,000 Normal care for reflux or Offer of Cytosponge + usual care Primary objective Number of cases Barrett s diagnosed across two arms Secondary objectives Cost-effectiveness of the Cytosponge TM test vs standard of care Acceptability to patients and primary care clinicians 23
Where could Cytosponge fit into the clinical pathway? - Possible cancer symptoms endoscopy - Higher risk groups (Barrett s, male sex, white race, persistent heartburn, abdominal obesity, smokers, family history) - Investigation for all reflux/dyspepsia consulters Cytosponge - Population screening (as performed for colon, cervix)? Cytosponge Encourage early consultation First cancer symptom First GP visit GP diagnostic tests Hospital referral endoscopy Diagnosis & treatment plan Start treatment Primary care Secondary care
What next if TFF3 positive? Heartburn Symptoms Screening with Cytosponge TM test TFF3 negative test Patients discharged from screening programme TFF3 positive test Endoscopy
Using mutations to track Cancer Evolution Jos Jonkers Cancer Discovery 2012;2:486-488
Is Cytosponge sampling sufficient to identify the invading clone(s)? Cancer SJ Leedham.NA Wright Gut 2008; 57:1041-1048
Cytosponge captures entire clonal architecture 1 Clone 1 Clone 2 Clone 3 Clone 4 Clone 5 Clone 6 Clone 1 Clone 2 Clone 3 Clone 4 Clone 5 Clone 6 X axis for each clone chr 1-23 Y axis Variant Allele Fraction for each mutation Ross-Innes et al Nature Genetics 47; 2015
Molecular genetic heterogeneity Secrier M.Fitzgerald RC Nature Genetics 2016
Recurrent TP53 driver mutations 6/8=75% TP53 70-80% cases have TP53 mutation Dulak et al Nature Genetic 2014 (TCGA) Weaver et al Nature Genetics 2014 (ICGC) Murugaesu N et al. Cancer Discovery 2015
Driver gene mutations in 551 oesophageal adenocarcinomas ICGC data Frankell A Fitzgerald RC unpublished
Mutation prevalence in Barrett s Never-dysplastic BE (f/up to 10 yrs) BE with high grade dysplasia Oesophageal adenocarcinoma Weaver et al Nature Genetics 2014
Dramatic increase copy number Alterations in invasive disease EAC: Sum of copy number gain/loss Barrett s: Sum of copy number gain/loss N=30 normal/barrett/s/adenocarcinoma trios Ross-Innes et al Nature Genetics 47; 2015
Reconstructing the life-history of Oesophageal Adenocarcinoma Life Oesophageal ICGC data in The evolutionary history of 2,658 Cancers biorxiv under review Nature Genetics 2017
Ordering of mutations and implications for biomarkers Screening window for curative intervention Never-dysplastic BE BE with high grade dysplasia Early invasive EAC Metastatic EAC Boundary Boundary Boundary Recurrently mutated genes TP53 SMAD4 (12%) Copy number (>80%) (70-80%) Multiple other mutated genes e.g. ARID1A, SMARCA4?Diversity measures
Risk stratification panel Given the sample is called HIGH RISK: the probability of being a true negative: 13% (5-27%) the probability of being a true positive: 87% (73-95%) 9 negative + 59 HGD were classified as high risk" Given the sample is called LOW RISK: the probability of being a true negative: 162/162 (96-99.99%) the probability of being a true positive: 0/162 (0.01-4%) In our data set: 162 negatives + 0 HGD were classified as low risk (Age, BMI, Barrett s length, atypia, copy number, p53 status) (BEST2 n=468)
Ongoing research with diagnostic test Improve risk stratification Automated slide reading for TFF3 One platform for diagnosis and stratification e.g. DNA methylation + mutation Near bedside test Incorporate other diagnostic information
Methylation panel on Cytopsonge 143 Controls + 169 NDBO TFF3 BEST1 TFF3 BEST2 Methylation AUC 83.6% 86.2% 90.4% Chettouh H.Fitzgerald RC et al Gut 2017
Precision diagnosis Heartburn Symptoms Screening with Cytosponge TM test TFF3 negative test Patients discharged from screening programme TFF3 positive test Risk stratification using panel of biomarkers Low risk High risk Repeat Cytosponge TM every 3-5 yrs Endoscopy + biopsy Standard management based on clinical result
Global perspective Reasons for late diagnosis: - Lack of education - Affordability - Access Cancer patients need better care, not just more technology Richard Sullivan, CS Pramesh, CM Booth Nature; Comment Sept 2017
High incidence areas of squamous cell carcinoma in China and Iran Feicheng Yanting
Screening: 40 years since Wilson and Junger and now in a genomic age Evidence based - Should respond to an unmet need - Objectives of screening clearly defined at the outset - Scientific evidence to define the target population - Scientific evidence for programme effectiveness Implementation - Quality assurance - Implementation strategy Patient centred - Patient considerations for equity and informed choice - Evidence that benefits outweigh harms Adapted from Anderman Dery et al Bulletin of WHO 2017
The future of cancer prevention? Self testing Bill Gates and Jess Bezos pour millions into new cancer test
Multi-analyte cancer blood test When weighted for actual incidence in the U.S., we estimate the sensitivity of CancerSEEK to be 55%. Joshua D. Cohen et al. Science 2018;science.aar3247
Early Detection Economics The jury is still out on the question of whether Precision Cancer Medicine will save or cost money to the NHS (the literature suggests more) Cancer Research UK 2014 Report suggesting that if cancers presently diagnosed at Stage 3 and 4 were halved by 2024, the saving to the NHS would be 257M/year. 6
Acknowledgments