Aims Guaiac faecal occult blood tests are being replaced by faecal immunochemical tests (FIT). We investigated whether faecal haemoglobin concentration (f-Hb) was related to stage in progression of colorectal neoplasia, studying cancer and adenoma characteristics in an evaluation of quantitative FIT as a first-line screening test.
Methods We invited 66 225 individuals aged 50–74 years to provide one sample of faeces. f-Hb was measured on samples from 38 720 responders. Colonoscopy findings and pathology data were collected on the 943 with f-Hb≥400 ng Hb/ml (80 µg Hb/g faeces).
Results Of the 814 participants with outcome data (median age: 63 years, range 50–75, 56.4% male), 39 had cancer, 190 high-risk adenoma (HRA, defined as ≥3 or any ≥10 mm) and 119 low-risk adenoma (LRA). 74.4% of those with cancer had f-Hb>1000 ng Hb/ml compared with 58.4% with HRA, and 44.1% with no pathology. Median f-Hb concentration was higher in those with cancer than those with no (p<0.002) or non-neoplastic (p<0.002) pathology, and those with LRA (p=0.0001). Polyp cancers had lower concentrations than more advanced stage cancers (p<0.04). Higher f-Hb was also found in those with HRA than with LRA (p<0.006), large (>10 mm) compared with small adenoma (p<0.0001), and also an adenoma displaying high-grade dysplasia compared with low-grade dysplasia (p<0.009).
Conclusions f-Hb is related to severity of colorectal neoplastic disease. This has ramifications for the selection of the appropriate cut-off concentration adopted for bowel screening programmes.
- Colorectal Cancer
- Gi Neoplasms
- diagnostic screening
Statistics from Altmetric.com
Guaiac faecal occult blood tests are used in screening programmes for colorectal (bowel) neoplasia, reducing mortality.1 Many data have been published2 and it has been widely documented3–6 that gender, age and deprivation affect positivity rates. However, guaiac faecal occult blood tests are not specific for human faecal haemoglobin concentration (f-Hb) and are rapidly being replaced by faecal immunochemical tests (FIT) for haemoglobin, which have many advantages.7 Quantitative FIT allow the f-Hb concentration used as a cut-off to select those who warrant further investigation, usually colonoscopy, to be selected to give the characteristics deemed desirable by programme organisers.
Sensitivity and specificity can be altered using different cut-off f-Hb, sensitivity increasing as the cut-off is lowered.8 Moreover, at any cut-off f-Hb, the sensitivity and positive predictive value for significant neoplasia (cancer plus high-risk adenoma) is substantially higher and specificity and negative predictive value lower among men than women.9 More recently, it has been demonstrated that mean f-Hb in FIT-positive individuals is significantly lower in women compared with men and for younger compared with older participants.10 Few studies11–13 have demonstrated that f-Hb increases as disease becomes more serious, from the normal through low- and high-risk adenomatous polyps (LRA, defined as <3 or any <10 mm, HRA, defined as ≥3 or any ≥10 mm) to invasive cancer. Chen et al14 showed that, in those with f-Hb below the cut-off, f-Hb at first screening predicted subsequent risk of incident colorectal neoplasia. An accompanying commentary stated that there may be a continuum of increasing risk as f-Hb increases from zero.15
To examine this hypothesis further, we carried out an observational study examining colonoscopy and pathology findings in those who were above the f-Hb cut-off selected for use in an evaluation of FIT as a first-line test.
From 1 July 2010 to 12 January 2011, all eligible participants in the Scottish Bowel Screening Programme, aged 50–74 years, resident in NHS Tayside and NHS Ayrshire & Arran, were sent a FIT kit pack containing a single sample collection device (Eiken Chemical Company, Tokyo, Japan).
The samples were analysed for f-Hb using OC-Sensor Diana automated immunoturbidimetric analysers (Eiken Chemical Company). Analyses were carried out by trained staff: the laboratory has a comprehensive total quality management system and is accredited to ISO15189-based standards by Clinical Pathology Accreditation (UK).
All participants with f-Hb≥400 ng Hb/ml buffer were reported as positive and referred for colonoscopy. This cut-off f-Hb was chosen to achieve approximately 2% positivity as per the current programme, selected for the available colonoscopy resource. Data for colonoscopy outcomes and any subsequent pathology were downloaded from clinical IT systems. Data on colonoscopy were collected on the quality of the investigation (quality of preparation, completeness of colonoscopy) and on the number, size and localisation of colorectal cancers and adenomas, and whether biopsy was performed. Full pathological data were collected on all excised/biopsy specimens including polyp type, presence or absence of malignancy, stage of any cancer and, in all adenoma, the severity of dysplasia. Lesion size was recorded from pathology reports except when removed piecemeal, when colonoscopy measurement was used. f-Hb was collated into clinical outcome groups according to most serious diagnosis. Assignment as HRA was ≥3 adenomas, or any adenoma with a maximum diameter ≥10 mm, taken from recommendation from the British Society of Gastroenterology16 as used in Scotland. f-Hb from those with adenoma were further grouped according to characteristics relating to their most serious lesion: size (small, a maximum dimension of <10 mm or large, ≥10 mm), degree of dysplasia (high or low-grade, HGD or low-grade dysplasia (LGD)), villous nature (presence or absence) and location (proximal defined as the region of the colon up to and including the splenic flexure, or distal, the region thereafter).
MedCalc (MedCalc Software, Mariakerke, Belgium) statistical software was used for all calculations. The Mann–Whitney U test was used for comparison between the groups and median lesion size. Probability of p<0.05 was considered significant.
Table 1 summarises the number of participants in the study and their outcomes. Those undergoing colonoscopy had a mean age of 62.8 years and 56.4% were male subjects. Table 2 shows the number in each clinical outcome group at increasing f-Hb. Table 3 further classifies those with a neoplastic lesion according to features of the most serious adenoma or cancer detected along with the median (and 95% CI) f-Hb and p values.
f-Hb in those with cancer was significantly higher than in all other groups (p<0.02), apart from the HRA group (p=0.07). In addition, f-Hb was greater in HRA compared with LRA and diverticular disease and those with no pathology detected (p<0.005). There was no statistically significant difference between the LRA group and those without neoplasia.
Of the 39 participants with cancer, 29 had f-Hb>1000 ng Hb/ml and 10 had f-Hb that could be measured within the analytical working range of 0–1000 ng Hb/ml. These included four participants with a confirmed polyp cancer (Dukes’ stage A), two non-polyp Dukes’ A cancers, three Dukes’ B and one Dukes’ stage C1 rectal cancer with f-Hb of 442 ng/ml. In all, 36 of the cancers had staging available and 12 (six polyp; 27.3%), 11 (33.3%), 12 (36.4%) and 1 (3.0%) cancers were Dukes’ A, B, C1 and C2, respectively. The differences in median f-Hb between different stages were not statistically significant.
f-Hb was significantly higher in those with a large compared with a small adenoma (p<0.0001). Those with adenoma with HGD had higher f-Hb than with LGD (p<0.009). Although f-Hb in those who had an adenoma with a villous component was higher than in non-villous adenoma, this was not statistically significant (p=0.07) nor was the difference between those with their most serious adenoma found proximally and distally (p=0.08). No significant difference in f-Hb was found between those with multiple adenomas (defined as ≥3) and those with only one or two adenomas found (p=0.64).
The significant difference in f-Hb seen between HGD and LGD adenomas was not evident when holding size constant, with no difference between small LGD and HGD adenomas (p=0.88), or between large LGD and HGD adenomas (p=0.09). A multiple linear regression model using size, degree of dysplasia, presence or absence of villous component, site and number of adenomas as categorical explanatory variables showed adenoma size was the only characteristic to be significantly related to f-Hb (p<0.0001; all other variables p>0.1).
Table 4 shows the range and median size of lesions in those with hyperplastic polyp (HPP), adenoma or cancer detected and p values for comparisons with significant differences in bold. Significantly larger lesions were found between any adenoma and HPP; HRA and LRA; HGD and LGD adenomas; villous and non-villous adenomas; and more advanced cancers and polyp cancers. In addition, all cancers were significantly larger than HRA (p<0.0001) and indeed any adenoma (p<0.0001). Adenomas in the distal colon were larger than in the proximal region, in contrast to cancers, which were larger proximally.
Wide distribution of the f-Hb of FIT-positive participants existed, with considerable overlap between different clinical outcomes. Thus, a single f-Hb measurement on a sample from an individual might not be a reliable indicator of stage of neoplasia. However, there was statistical evidence of a relationship between increasing f-Hb concentration and stage of colorectal neoplasia. In addition to a significantly higher median f-Hb concentration in participants with any neoplasia detected compared with those with no or non-neoplastic pathology, increasing f-Hb concentration is related to the severity of the lesion among those with neoplasia detected.
Median f-Hb concentration was statistically significantly higher in HRA compared with LRA. Furthermore, median f-Hb in LRA was perhaps surprisingly low in comparison with those with no neoplasia detected, lower than in those with HPP (although not statistically significant). Indeed, relatively low concentrations of f-Hb in non-advanced adenomas have been documented with most having f-Hb under 75 ng Hb/ml.11 The higher median f-Hb seen with a large adenoma compared with a small adenoma, but not seen when comparing those who had ≥3 adenomas with only one or two adenomas detected, demonstrates that the difference between the HRA and LRA group can be attributed to lesion size. Indeed, the median maximum dimension of HRA was over double that of LRA and significantly greater (p<0.0001).
Further, the group with an adenoma displaying HGD had a significantly higher median f-Hb than with LGD. This may indicate that the severity of the dysplastic change can reflect propensity of the lesion to bleed. However, multivariate analysis showed no significant difference between HGD and LGD adenomas when size was taken into account. Thus, the grade of dysplasia and f-Hb was primarily related to adenoma size. Adenomas displaying HGD were significantly larger than those with LGD, and with 27.2% of participants with larger adenomas also displaying HGD compared with 7.5% in those with smaller lesions, increasing f-Hb is associated with larger lesions which, in turn, are more likely to display more severe dysplasia.
FIT are possibly less effective at detecting lesions located in the proximal colon than distally.17 Here, 77.8% of adenomas and 69.2% of cancers were located in the distal colon. Levi et al11 have also shown that f-Hb was similar between participants with advanced adenomas in the proximal and distal colon. Despite distal adenomas being significantly larger than proximal and median f-Hb appearing to be higher in those with distal adenomas, our findings confirm that this difference is not statistically significant (p=0.08). Moreover, our results confirm the findings of Ciatto et al13 that a relationship exists among increased f-Hb and increasing size, severe dysplasia, villousness and location in the left colon.
Although the median f-Hb in those with cancer was not statistically significantly different to those with HRA, they had a higher proportion with f-Hb above the cut-off f-Hb, with 74.4% with cancer having an f-Hb concentration greater than 1000 ng Hb/ml, compared with 58.4% with HRA, providing further evidence of a potential continuum of risk with increasing f-Hb. In addition, 90% of cancers with f-Hb within the analytical range (<1000 ng Hb/ml) were early stage and 40% of these were polyp cancers as compared with 53.8% of early stage cancers above the upper analytical limit.
Non-neoplastic pathology, particularly diverticular disease, was not associated with f-Hb significantly different from those where no pathology was detected. This is an important observation since it indicates that false positive results arising from benign disease are likely to be no more common than false positive results when no abnormality is found on colonoscopy.
This study has limitations. First, the distributions of f-Hb could not be fully assessed since the upper analytical limit was 1000 ng Hb/ml. Of the 813 examined here, 393 had an f-Hb>1000 ng/ml: quantitative f-Hb was not recorded for 48.3% of participants. Perhaps those who had large cancers had f-Hb far higher than 1000 ng Hb/ml. This might explain the lack of a statistically significant difference in f-Hb between cancer and HRA, despite the malignant lesions being significantly larger. Second, only participants with a positive result were referred for colonoscopy. Analysis of the relationship between f-Hb and disease could only be carried out on those with an f-Hb concentration above the cut-off. False negative f-Hb has not been taken into account. However, we plan to investigate this by identifying interval cancers.
A major strength of this study is that we have detailed the relationship between f-Hb and severity of disease in an average-risk, population. These findings, from ostensibly healthy, asymptomatic individuals, have potential implications for future selection of optimum cut-off f-Hb in bowel cancer screening, particularly where colonoscopy capacity is limited. With the aim of screening being to detect cancer and its precursors, our results reinforce the argument for the use of quantitative FIT with an adjustable cut-off concentration through confirming that cancer and HRA arise in lesions that are larger and more prone to bleed, the very lesions screening aims to detect.
Faecal haemoglobin concentration is related to severity of colorectal neoplasia with concentrations in cancer related to stage.
Higher faecal haemoglobin concentrations are found in high-risk compared with low-risk adenoma, with large compared with small adenoma and also with an adenoma displaying high-grade dysplasia compared with low-grade dysplasia.
Faecal haemoglobin concentration is related to lesion size.
Individual faecal haemoglobin concentrations could be important in determining priority for colonoscopy and for inclusion in risk-scoring algorithms for colorectal neoplasia.
Interactive multiple choice questions
This JCP review article has an accompanying set of multiple choice questions (MCQs). To access the questions, click on BMJ Learning: Take this module on BMJ Learning from the content box at the top right and bottom left of the online article. For more information please go to: http://jcp.bmj.com/education. Please note: The MCQs are hosted on BMJ Learning—the best available learning website for medical professionals from the BMJ Group. If prompted, subscribers must sign into JCP with their journal's username and password. All users must also complete a one-time registration on BMJ Learning and subsequently log in (with a BMJ Learning username and password) on every visit.
The authors thank the staff of the Scottish Bowel Screening Centre for their work in sending out the FIT kits and in the analysis of returned samples. The authors also thank Iain McElarney of Mast Diagnostics Division, Bootle, Merseyside, UK, for his input into the preparation of material for potential participants and in the setting up of the automated analytical systems.
Contributors JD obtained data, analysed data and wrote manuscript. RJCS and CGF initiated the project, led the gaining of funding and wrote the manuscript. PJMcD and JAS supervised the sample handling and faecal test analyses, participated in data collection and analysis and contributed to the manuscript. FAC and MB provided pathology data and contributed to the manuscript. RHD led the project in NHS Ayrshire & Arran and contributed to the manuscript.
Funding The additional resources required to undertake this evaluation were provided in part by the Scottish Government Health Directorates. Data analysis was in part supported by a grant from the Chief Scientist Office (grant no. CZH/6/4) to establish a Bowel Screening Research Unit. These sponsors approved the study design but had no role in the collection, analysis, or interpretation of data or in the writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. All authors are independent of the funders in terms of freedom to publish.
Competing interests CGF undertakes consultancy with Immunostics Inc, Ocean, NJ, and Mode Diagnostics, Glasgow, UK. No other competing interests.
Ethics approval Ethics approval was not sought. The work was approved by the Scottish Bowel Screening Programme Board and had Caldicott Guardian approval from both NHS Tayside and NHS Ayrshire & Arran.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.