Novel methods for the detection, risk stratification and management of early colorectal cancer and pre-malignant lesions

Johnstone, Mark S. (2025) Novel methods for the detection, risk stratification and management of early colorectal cancer and pre-malignant lesions. PhD thesis, University of Glasgow.

Abstract

Colorectal cancer (CRC) is the 4th most common cancer in the United Kingdom (UK) with approximately 44,000 cases each year, and with 16,800 deaths, is the 2nd most common cause of cancer-related mortality. Outcome is directly linked to stage at diagnosis with 5-year survival estimated to be 90.9% for those with stage I disease, 84.3% for stage II, 65.0% for stage III and 10.5% for stage IV(1). Pathway to diagnosis in Scotland can be via symptomatic referral to an outpatient clinic, emergency symptomatic presentation or via the Scottish Bowel Screening Programme or surveillance colonoscopy. Screening in Scotland utilises a quantitative faecal immunochemical test (FIT) in individuals aged 50 to 74 years, followed by colonoscopy for those patients testing positive, at a faecal haemoglobin (f-Hb) threshold of 80 µg Hb/ g of faeces(2). Screening in this fashion increases the number of early-stage cancers diagnosed, reduces CRC-mortality and may reduce the incidence of CRC through the removal of precursor polyps(3-8). Unfortunately, screening only accounts for 19% of CRCs diagnosed in the West of Scotland(9). While the symptoms of CRC, including rectal bleeding, persistent change in bowel habit, abdominal pain and weight loss, are commonly present at the time of diagnosis, considerable symptomatic overlap exists with other significant bowel disease (advanced polyps and inflammatory bowel disease (IBD)) and functional bowel disorders. Therefore, the positive predictive value of such symptoms for CRC is low. Indeed, in a study of 384,510 colonoscopies performed in the UK, rectal bleeding and anaemia were associated with the highest adjusted positive predictive values (aPPV) for CRC (2.5% and 2.1% respectively), while all other symptoms were associated with a CRC aPPV of <1%(10). Conversely, FIT is a powerful objective biomarker of CRC-risk in symptomatic patients, with a linear relationship between f-Hb and CRC-risk observed(11). Consequently, in addition to its use in screening, FIT has now been widely embedded into symptomatic referral pathways as an effective means of triaging patients(11-22).

Chapter 1 provides a broad overview of the epidemiology, aetiology, signs and symptoms, investigation, staging and management of CRC.

In Chapter 2 the impact of integrating FIT into referral pathways from primary care to colorectal and gastroenterology was assessed in a cohort of 4968 symptomatic patients. Additionally, the association between CRC risk, symptoms, faecal haemoglobin (f-Hb) and anaemia were examined. GP referral and secondary care investigation patterns were indeed influenced by FIT, with a raised f-Hb correlating both with the decision to refer to secondary care and to perform colonoscopy. While rectal bleeding showed a positive correlation with CRC risk, no individual symptom independently predicted CRC on multivariate analysis. Conversely, both f-Hb and anaemia independently predicted CRC risk and represent valuable objective markers of risk in symptomatic patients. The combined absence of a raised f-Hb or anaemia effectively excluded CRC in 99.96% of cases.

In chapter 3 the results of a multicentred study of 5,761 patients investigating the prevalence of repeat FIT testing in primary care are presented. The study aimed to examine the relationship between serial f-Hb concentrations and CRC risk in symptomatic patients. Consecutive FIT tests submitted within 12 months of each other accounted for 9.1% of all FIT tests submitted to the laboratories of the Scottish health boards under investigation. CRC prevalence amongst patients with such serial FIT measurements was 0.7%, lower than the CRC rate observed in single FIT symptomatic cohorts. Patients with two f-Hb measurements <10µg/g had a significantly lower CRC risk (0.1%) than those with at least one f-Hb ≥10µg/g. As the number of FIT tests performed within a year rose, the likelihood of having a positive test rose, while the CRC rate fell. Performing two FITs within a year for patients with persistent symptoms therefore seems to be an effective safety netting practice, while performing more than two within this timeframe is unlikely to be beneficial.

In chapter 4, a cohort of 1,272 symptomatic patients who underwent FIT testing followed by colonoscopy was used to explore demographics and alternative lower gastrointestinal pathologies associated with a raised f-Hb. In addition to CRC, advanced adenomas, non advanced polyps and IBD independently predicted a raised f-Hb, as did older age, deprivation, use of oral anticoagulants and self-reported rectal bleeding. Deprivation independently predicted a raised f-Hb in patients with no pathology found at colonoscopy.

In chapter 5 attention was turned to the bowel screening programme. A cohort was established of 770 patients who underwent potentially curative resection for CRC. Patients were grouped based on diagnosis via screening or symptomatic pathways and the impact of important covariables, comorbidity and the systemic inflammatory response (SIR), on outcome was assessed. Patients with screen-detected disease had tumours of an earlier stage, were significantly less comorbid as measured by the American Society of Anaesthesiologists (ASA) score and had a significantly lower SIR as compared to non-screen-detected patients. Despite this, after adjusting for numerous covariables, non-screen-detection and a raised SIR independently predicted poorer overall and cancer specific survival.

In chapter 6 a prospective observational study of the management and outcomes of 236 patients with T1 polyp CRCs is presented. Male sex, older age, distally located lesions and pedunculated morphology were more likely to be managed with polypectomy only, while proximally located lesions and larger polyp size were more likely to proceed directly to formal colorectal resection. Younger age, requirement for piecemeal polypectomy and an involved polypectomy margin were associated with a higher chance of progressing to formal colorectal resection after polypectomy. Poor differentiation independently predicted lymph node involvement, submucosal venous invasion (SMVI) and mucinous-subtype predicted recurrence and SMVI predicted CRC-specific survival. Although 64.4% of polypectomy only patients had margin involvement or other high-risk factors, zero developed recurrence. Of 94 with polypectomy margin involvement, only 5 had confirmed residual tumour. Overall, lymph node metastases (7.1%), recurrence (4.2%) and cancer-specific mortality (3.0%) were rare. Surveillance following local excision of T1 CRC polyps may be safe for many patients.

Chapter 7 presents preliminary data from the Integrated Technologies for Improved Polyp Surveillance (INCISE) project; a large, retrospective, multi-partner collaborative aiming to use patient characteristics, digital pathology, immunohistochemistry (IHC), genomic and transcriptomic features of index polyp tissue to predict metachronous polyp risk and refine post-polypectomy surveillance. The INCISE cohort is formed of 2,643 patients who underwent polypectomy during screening colonoscopy followed by subsequent surveillance colonoscopy. In this particular sub-study, the most recent British Society of Gastroenterology post-polypectomy surveillance risk criteria (BSG-2020) were retrospectively applied to the cohort, such that 51.5% of patients would no longer qualify for surveillance. After a median 36 months, the metachronous advanced polyp/ CRC rate in BSG-2020 high risk patients was found to be only marginally greater than low risk individuals (16.3% vs 13.0%). Furthermore, while BSG 2020 risk stratification group was associated with a significant difference in overall metachronous lesion rate, it did not differentiate advanced and non-advanced metachronous lesions and was not significantly associated with late metachronous lesions detected after 2 years from index polypectomy, suggesting that current surveillance protocols would benefit from refinement.

In chapter 8 a review of established risk factors for metachronous polyp development is given, including patient demographics and comorbidities, chemopreventive medications and conventional polyp pathology. This precedes the results of a formal systematic review of all studies exploring genomics, transcriptomics, immunohistochemistry or features of the microbiome as novel biomarkers of metachronous polyp risk, following colorectal polypectomy. 4,165 paper titles, 303 abstracts and 215 full manuscripts were reviewed, with 25 papers included in the final study. 49 mutations/ SNPs/ haplotypes in 23 genes/ chromosomal regions (KRAS, APC, EGFR, COX1/2, IL23R, DRD2, CYP2C9/24A1/7A1, UGT1A6, ODC, ALOX12/15, PGDH, SRC, IGSF5, KCNS3, EPHB1/ KY, FAM188b, 3p24.1, 9q33.2, 13q33.2) were found to predict metachronous adenoma / advanced adenoma risk, while the expression levels of 6 proteins correlated with metachronous adenoma (p53, β-catenin, COX2, Adnab-9, ALDH1A1) or sessile serrated polyp (ANXA10) risk.

Chapter 9 explored the utility of COX2 and p53 expression as potential biomarkers for predicting metachronous polyp or CRC risk in INCISE cohort patients. 1,236 of 2,643 INCISE patients had tissue retrieved for immunohistochemical assessment, with 859 of those randomised to the training cohort and 377 to the test cohort. Formalin-fixed, paraffin-embedded tissue blocks of index polyp tissue were used to construct tissue microarrays (TMA) with four cores available per patient. The cores were stained with COX2 and p53 antibody and cytoplasmic COX2 expression and nuclear p53 expression were quantified digitally using QuPath software. On univariate analysis high cytoplasmic COX2 expression (p=0.019) predicted shorter time to detection of any metachronous lesion, as did key demographics such as increasing age (p=0.034) and pathological parameters such as increased polyp number at index colonoscopy (p<0.001) and BSG 2020 high risk (p<0.001). While high cytoplasmic COX2 expression retained significance as independent predictor of shorter time to development of any metachronous lesion on multivariate analysis (p=0.016), the positive association between COX2 expression and metachronous polyp or CRC risk could not be replicated within test cohort patients and therefore does not appear to represent a useful biomarker for this purpose.

Chapter 10 presents the results of a smaller pilot study, examining β-catenin as a potential biomarker for metachronous polyp or CRC risk. The same TMA constructed for chapter 9 was stained with β-catenin antibody. Nuclear β-catenin expression was assessed using QuPath software amongst 339 INCISE patients. Low β-catenin expression was found to predict a shorter time to detection of any metachronous polyp or CRC on both univariate and multivariate cox regression. Work is ongoing to score β-catenin expression in the whole cohort and validate these preliminary findings.

Finally, chapter 11 summarises the main findings of the thesis and lays out potential future work.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QR Microbiology; Q Science > QR Microbiology > QR180 Immunology; R Medicine > RC Internal medicine > RC0254 Neoplasms. Tumors. Oncology (including Cancer)
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Cancer Sciences
• Supervisor's Name: Edwards, Professor Joanne and McSorley, Mr. Steven
• Date of Award: 2025
• Depositing User: Theses Team
• Unique ID: glathesis:2025-85445
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 11 Sep 2025 10:59
• Last Modified: 11 Sep 2025 11:02
• Thesis DOI: 10.5525/gla.thesis.85445
• URI: https://theses.gla.ac.uk/id/eprint/85445

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI