Use of faecal markers in screening for colorectal neoplasia: a European group on tumor markers position paper
Several randomized controlled trials have shown that population-based screening using faecal occult blood testing (FOBT) can reduce mortality from colorectal neoplasia. Based on this evidence, a number of countries have introduced screening for colorectal cancer (CRC) and high-risk adenoma and many others are considering its introduction. The aim of this article is to critically review the current status of faecal markers as population-based screening tests for these neoplasia. Most of the available faecal tests involve the measurement of either occult blood or a panel of DNA markers. Occult blood may be measured using either the guaiac faecal occult blood test (gFOBT) or a faecal immunochemical test (iFOBT). Although iFOBT may require a greater initial investment, they have several advantages over gFOBT, including greater analytical sensitivity and specificity. Their use results in improved clinical performance and higher uptake rates. Importantly for population screening, some of the iFOBTs can be automated and provide an adjustable cutoff for faecal haemoglobin concentration. However, samples for iFOBT, may be less stable after collection than for gFOBT. For new centres undertaking FOBT for colorectal neoplasia, the European Group on Tumour Markers recommends use of a quantitative iFOBT with an adjustable cutoff point and high throughput analysis. All participants with positive FOBT results should be offered colonoscopy. The panel recommends further research into increasing the stability of iFOBT and the development of improved and affordable DNA and proteomic-based tests, which reduce current false negative rates, simplify sample transport and enable automated analysis.
Colorectal cancer (CRC) is the third most prevalent cancer worldwide, with an estimated one million new cases and a half million deaths each year.1, 2 After lung cancer, CRC is the second most common cause of death from cancer for men and women combined.
Although many screening tests are available for CRC and advanced adenoma, the most widely used is faecal occult blood testing (FOBT). The use of FOBT has been shown to reduce cancer mortality in four large randomized trials.3–7 Several expert groups therefore now recommend that all average-risk men and women should undergo screening for CRC and advanced adenoma (Table 1). In light of these recommendations, population screening for these lesions has recently been initiated in several countries.
Table 1. Groups and organizations recommending screening for colorectal neoplasia
This article critically reviews the current status of faecal markers in screening for CRC and high-risk adenoma (i.e., > 10 mm, ≥ 3 more adenomatous polyps of any size, significant villous component, or high grade dysplasia), and makes recommendations for their use in population-based screening. In preparing the article, the literature relevant to faecal screening tests for colorectal neoplasia was reviewed. Particular attention was given to systematic reviews, prospective randomised trials and guidelines published by expert panels.
Available Screening Tests for Colorectal Neoplasia
Two main types of screening tests are available for CRC and high-risk adenoma, bowel visualisation and measurement of markers, mainly in faeces.19, 20 Bowel visualisation techniques include colonoscopy, computed tomographic colonography, and flexible sigmoidoscopy (FS). These tests will not be discussed further in this article other than to state that, at this time, high quality colonoscopy is the most effective means of detecting CRC and high-risk adenoma. As a screening test for CRC, colonoscopy however, has not yet undergone evaluation in a large randomized controlled trial.
Compared with bowel visualisation, measurement of faecal markers has several advantages including being safe, noninvasive and less expensive to perform. An important attraction of FOBT in population-based screening is that they are simple to use and can be done at home, without a general practitioner or hospital appointment. The most widely used faecal based assays are the FOBT which depend for their efficacy on the premise that CRC and adenomas cause bleeding into the colon and that the stool sample used in the test has been exposed to this blood. Other faecal-based assays are based on a panel of DNA markers or miscellaneous faecal proteins.14, 21, 22
Two main types of FOBT exist: the guaiac-based faecal occult blood test (gFOBT) which uses the pseudoperoxidase activity of the haem moiety of intact or free haemoglobin, and the immunochemical faecal occult blood test (iFOBT) which detects the globin moiety in intact human haemoglobin or its early degradation products.21–24 The latter assay is also known as the faecal immunochemical test or FIT. Of the FOBT, gFOBT has been available for considerably longer than iFOBT and its clinical efficacy has been more widely investigated.20–24 Neither gFOBT nor iFOBT are specific for colorectal neoplasia, since any bleeding into the colon could result in a positive test. Indeed, the gFOBT is not even specific for blood. All positive FOBT screening results must therefore be followed-up with a complete colonoscopy.
Guaiac-Based Faecal Occult Blood Tests
Since gFOBT have been available for several decades, are relatively easy to perform and inexpensive, they have been widely used to screen for CRC and high-risk adenoma. The test is available in various formats but typically involves the application of two samples from each of three separately passed stools onto the paper in the test cards. The use of multiple samples improves the chance of detecting colorectal neoplasia (i.e., cumulative sensitivity) because bleeding may be intermittent, of low volume or not spread homogeneously throughout the stool.
Different gFOBT products exist which are highly variable with respect to sensitivity and specificity for colorectal neoplasia.24, 25 This variability primarily reflects the test characteristics, but is also influenced by the population studied and the conditions under which the samples are collected, transported and tested. Rabeneck et al.25 carried out a systematic review of the literature on repeated annual or biennial gFOBT and found that the reported sensitivity for CRC varied from 51% to 100%, specificity varied from 90.4% to 97% and the positive predictive value (PPV) from 2.4% to 17.0%.
Although gFOBT have multiple limitations (Table 2), their use in screening has clearly been shown to reduce mortality from CRC. After a systematic review of the literature, Hewitson et al.26 identified four randomized controlled trials. Meta-analysis of the four trials showed an average 16% reduction in the relative risk of CRC mortality in the screened population (relative risk [RR]: 0.84, 95% confidence interval [CI]: 0.78–0.92). Uptake was between 54% and 67% in the first round and between 60% and 78% for at least one of the rounds. The highest participation was in the Minnesota trial,4 which was not population-based but relied on volunteers recruited through a clinic. When the relative risk was adjusted for attendance at screening, the overall predicted relative mortality reduction was 25% amongst those screened (RR: 0.75; 95% CI: 0.66–0.84).26
Table 2. Advantages and disadvantages of different faecal assays in screening for colorectal neoplasia. gFOBT, guaiac faecal occult blood tests; iFOBT, immunochemical faecal occult blood tests
Although all of the randomised trials used the Hemoccult II test (Beckman, Fullerton, CA), several nonrandomized studies have used different gFOBT,27–36 commonly including HEMA-SCREEN (Immunostics, Ocean, NJ). Results from these studies are similar to those reported in the large randomized research trials, confirming the feasibility and efficiency of using gFOBT to screen for CRC and high-risk adenoma. Although gFOBT results do not provide a quantitative result which would enable the analytical sensitivity to be adjusted, the results of individual spots on the traditional six-window gFOBT can be used to generate screening algorithms such as that used in pilot studies in England and Scotland.32, 35 Using such an algorithm, albeit making the screening programme more complex, does allow the overall positivity rate to be adjusted so as to match availability of colonoscopy.
Immunochemical Faecal Occult Blood Tests (iFOBT)
1. The general principle of iFOBT is antibody detection of the globin component of human haemoglobin or its early degradation products. Although in use in Japan for over 20 years, iFOBT is now finding increasing application, especially in European countries, as it has several advantages over gFOBT (Table 2), (for review, see Halloran et al, in preparation and ref.37). Unlike the gFOBT, iFOBT uses only one or two faecal samples. The use of two samples results in increased sensitivity, but a marginal decrease in specificity. A disadvantage of iFOBTs is that samples are relatively unstable at ambient temperature, with increasing storage time contributing to sample deterioration.40–42 In contrast, samples for gFOBT are relatively stable when smeared immediately onto test cards.43
iFOBT can be divided into qualitative tests that provide a positive or negative test result and those that provide a quantitative result. Whilst the term “quantitative” is widely used in the literature, the crude sample collection technique, heterogeneous distribution of blood in faeces and sample instability makes true quantitation impossible. An advantage of quantitative assays is that they avoid subjective interpretation of test results.
A further advantage of quantification is that it allows adjustment of a cutoff haemoglobin concentration to meet local requirements (e.g., prevalence of CRC and high-risk adenoma and availability of colonoscopy). Decreasing the cutoff concentration generally increases sensitivity and decreases specificity, whereas increasing the cutoff concentration decreases sensitivity but increases specificity. Increasing sensitivity results in an increased detection rate but the decreased specificity means a greater number of subjects referred unnecessarily for colonoscopy44 Clearly, therefore, the positivity cutoff concentration used must be carefully selected. Several studies have addressed the impact of different cutoff concentration on positivity rates and performance of iFOBT. Most of these used the OC-Sensor iFOBT (Eiken Chemical Company, Japan) and concluded that a cutoff concentration of 75 ng Hb/mL provided both an adequate positivity rate and an acceptable trade-off between detection rate and number of unnecessary colonoscopies performed.45–48
In addition to the cutoff concentration, test performance of iFOBT would be expected to depend on the number of samples collected and analysed. In a multicenter study on 20,596 average-risk subjects, Grazzini et al.47 investigated both effect of using different cutoff concentrations (80 ng Hb/mL, 100 ng Hb/mL and 120 ng Hb/mL) and the use of one versus two samples. These different strategies were compared with the standard approach, that is, a single sample collected and using a cutoff concentration of 100 ng Hb/ml. None of the strategies investigated however, displayed a clear-cut superiority.46
Combining gFOBT with iFOBT Screening
One of the strategies recommended in the guidelines of the World Health Organization and World Organization for Digestive Endoscopy was the combination of an initial gFOBT and a subsequent iFOBT.22 This approach has been studied using a two-tier reflex approach in which individuals with a weak positive or equivocal result on gFOBT, (i.e., one to four of the six possible spots positive), were asked to complete a iFOBT with a tube collection device49 and a card collection device.50 This approach reduces the number of false positive results since those who are iFOBT positive have significantly more pathology than those with iFOBT negative results. A potential disadvantage of any multi-tier strategy, however, is that it will reduce uptake rates and programme sensitivity. Furthermore, it is limited by the sensitivity of the first test.
Key Differences Between iFOBT and gFOBT
Comparative analytical specificity
Although both gFOBT and iFOBT are purported to detect blood in faeces, iFOBT are specific for human blood. The gFOBT, being dependent on the pseudoperoxidase activity of the haem moiety of haemoglobin, is potentially susceptible to interference in ingested foods. Peroxidase-like activities may produce false positives and can be found in a variety of vegetables and certain meats, especially red meats. In practice, however, this interference appears to be minimal.51, 52 Furthermore, according to Sinatra et al.,53 potential interference can be eliminated by delaying colour development for at least 48 hr between sample collection and analysis. Whether these findings apply to the more sensitive guaiac tests such as Haemocult SENSA (Beckman Coulter, Fullerton, CA) is unclear.
Although normal dietary intake of vitamin C may not interfere with gFOBT, high intakes may produce false negative results because of the reducing properties of this molecule.54 Joint guidelines published by the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer and the American College of Radiology (ACS/USMSTFCC/ACR) therefore recommended that intake of vitamin C > 250 mg/day should be avoided during the 3 days before collection of samples for gFOBT.14
Conflicting reports exist as to whether aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs) interfere with gFOBT.56–58 The ACS/USMSTFCC/ACR however, recommends avoidance of NSAIDs for 7 days before sensitive gFOBT, unless subjects are receiving cardioprotective agents.14 Recently, it was demonstrated that anticoagulants and antiplatelet medications lower the positive-predictive-value of gFOBT for CRC and high-risk adenoma and these should be stopped if clinically feasible before stool collection.40
As well as having superior analytical specificity, iFOBT are also more specific than gFOBT for lower gastrointestinal (GI) tract bleeding and thus for colorectal neoplasia. This is because globin is degraded by proteases in the upper GI tract, leaving the haem moiety intact. Thus, gFOBT can theoretically detect bleeding from all parts of the GI tract, although the current evidence has been deemed insufficient to recommend for or against routine upper GI tract visualisation as a means of detecting gastric or oesophageal cancers for patients who are gFOBT positive and colonoscopy negative.58
Comparative clinical performance
Several population-based studies have investigated the sensitivity and specificity of iFOBT for CRC. After a systematic review of these studies published in 2008, the US Preventive Services Task Force concluded the following59:
Overall, iFOBT displayed higher sensitivity for CRC (61–91%) than that previously found for the Haemoccult II gFOBT.
Specificity of the iFOBT varied from 91% to 98%.
In most studies, the specificity of iFOBT appeared to be somewhat lower than that previously reported for Haemoccult II but better than that of Haemoccult SENSA.
Since the publication of this systematic review, several large studies have compared the clinical performance of iFOBT with gFOBT.60–63 In one of the largest randomized studies, van Rossum et al.60 directly compared the gFOBT Hemoccult II, with the iFOBT, OC-sensor assay, in a Dutch cohort of 20,623 previously unscreened subjects. The iFOBT used a single sampling tube and the cutoff concentration used for defining positivity was 100 ng Hb/mL, although all subjects with a concentration > 50 ng Hb/ml were referred for colonoscopy. The threshold for defining gFOBT positivity was blue discoloration on any of six stool samples within 30–60 seconds of applying the developing solution. No dietary restriction was undertaken. Positivity rates were significantly higher for iFOBT vis-à-vis gFOBT (5.5 for iFOBT and 2.40 for gFOBT, p < 0.01). The PPV for the tests were not significantly different however and the number of subjects having to undergo colonoscopy to find one cancer was similar with both tests. However, the intention-to-screen detection rates for cancers and advanced adenomas were significantly higher with the iFOBT than with the gFOBT.
In another Dutch randomized trial, Hol et al.61 compared the detection rates of advanced neoplasia using gFOBT, iFOBT and FS. Although detection rates for advanced adenomas and cancer combined were lower with iFOBT (2.4%) than with FS (8.0%), iFOBT gave a higher diagnostic yield compared with gFOBT (1.1%). Based on the intention to screen analysis, more patients with CRC were detected with iFOBT (2.8%) than with FS (1.6%). Other recent studies have also confirmed the superiority of iFOBT over gFOBT, especially for detecting high-risk adenomas.62–69
One of the most important factors determining the success of any screening programme is participation rate or test uptake. Participation rates have been reported to be significantly higher with iFOBT than with gFOBT.60, 61, 70 This increased uptake probably relates to simpler sample collection and handling techniques and the opportunity to use fewer samples, that is, one or two with iFOBT compared with six samples (two samples from three stools) with gFOBT.
A number of modelling studies have shown that use of iFOBT provide a more cost-effective strategy for CRC screening strategy than gFOBT,71–74 probably due to their increased sensitivity and higher uptake. Using data from a Dutch randomized controlled study, van Rossum et al.71 concluded that the hypothetical average subject would, on average, gain, after one round of iFOBT screening, 0.003 life years and save 27 Euro as compared with gFOBT, and would gain 0.003 life years and 72 Euro as compared with no screening. The health economics of colorectal neoplasia screening should be determined in each country, however, so as to take account of the costs of different tests, investment costs, participation rates, mailing systems, cost and capacity of laboratory staff, preferences of those undergoing screening and colonoscopy investigations.
Compared with gFOBT and iFOBT, the use of DNA markers to screen for colorectal neoplasia has been less widely investigated, especially in large population based studies. The rationale in using faecal DNA markers to screen for colorectal neoplasia is based on the finding that mutant DNA present in these neoplastic lesions can be shed into the GI lumen and excreted in stools. One of the most widely investigated DNA panels involves the measurement of 21 separate mutations in the K-RAS, APC, and P53 genes as well as the detection of BAT-26, and L-DNA (EXACT Science, Marlborough, MA).75 The performance of this panel was compared with that of a gFOBT in a large population-based study involving > 4,000 asymptomatic average risk subjects. The main conclusion from this study was that the DNA panel displayed a higher sensitivity than the gFOBT, without a reduction in specificity. However, the sensitivity of the gFOBT used in this investigation was lower than that reported in other studies, possibly because the gFOBT result was based on only a single time point test, whereas most earlier reports on gFOBT involved repeated testing.
More recently, Ahlquist et al.76 compared the same DNA panel with two different gFOBT, Haemoccult II and Haemoccult SENSA. In this multicenter study involving 2,497 asymptomatic subjects, the sensitivity for screen-relevant neoplasia was 20% for the DNA test, 11% for Haemoccult II and 21% for Haemoccult SENSA. Specificity was 96% for the DNA panel, 98% for Haemoccult and 97% for Haemoccult SENSA. A modified version of this DNA method (STD-2, EXACT Science) which involves measurement of K-RAS mutations, scanning of the APC mutator cluster region and vimentin gene methylation detected significantly more neoplasms than both gFOBT used. For the colonoscopically normal patients, the positivity rate was 16% with SDT-2, compared with 4% with Hemoccult (p = 0.010) and 5% with HemoccultSensa (p = 0.030).
European Group on Tumour Markers Recommendations (EGTM) on the Use of Faecal Tests in Screening for Colorectal Neoplasia
For colorectal neoplasia screening using faecal markers, the EGTM panel suggest the following. Some of these recommendations have been published by other groups (Halloran et al., in preparation).13, 14, 25
As iFOBT has several advantages over gFOBT (Table 2), we recommend a quantitative iFOBT with an adjustable cutoff faecal haemoglobin concentration and high throughput analysis for colorectal neoplasia screening in average-risk populations.
A system of centralized population-based screening should be used. Opportunistic screening cannot be controlled and is susceptible to several forms of selection bias.
We recommend that parallel pilot feasibility studies are undertaken where gFOBT is already practiced, before implementing iFOBT screening in a country or area. In existing programmes, using gFOBT screening, considerations about uptake rates, positivity rates and resources should be made before considering a change over to iFOBT.
Sensitivity for CRC should be > 50% and the overall sensitivity of the screening programme should be > 90%.
Because clinical specificity is important in any screening programme, the test and screening algorithm used should have an overall specificity > 85% after several repeated tests.
Screening using FOBT should probably be performed at least biennially, although studies are needed on the optimal screening interval.
Good quality colonoscopy must be offered to all FOBT-positive subjects and participants should be specifically informed about follow-up as well as about the complication rate, any costs or other disadvantages associated with the procedure.
Sample rehydration of gFOBT should not be performed. It increases the sensitivity but decreases the specificity.
The EGTM panel does not recommend dietary restriction before screening. Although certain dietary components can theoretically interfere in gFOBT, the evidence supporting this is weak. The panel recommends a delay of > 48 hr before analyzing gFOBT to allow degradation of endogenous peroxidase activity. However, as a high intake of vitamin C has been reported to interfere in gFOBT, we recommend considering restricting its intake to < 250 mg per day in the 3 days before, and during testing.14
Laboratories performing FOBT should use appropriate internal quality control and quality assurance strategies, participate in external quality assessment schemes and be accredited to the same standards as medical laboratories (e.g., to ISO 15,189).
In order for national screening programmes to monitor and maintain consistency of the quality of analysis, testing is best restricted to a small number of designated laboratories that have training expertise, experience in faecal testing and facilities for high volume analysis.
It is recommended to record the date that the sample is collected and to designate a maximum period between collection and analysis, so that unsuitable samples (e.g., due to heat-induced deterioration), can be excluded. This applies especially to the quantitative iFOBT.
Single office-based FOBT should not be used for colorectal neoplasia screening.
DNA-based tests are currently unsuitable for population screening. Further research into these tests should be undertaken and should focus on developing the optimum combination of markers, determining the value of repeated testing, simplifying and automating the test procedure, and reducing cost. Furthermore, we recommend that comparative studies with iFOBT should be performed.
The availability of iFOBT represents a major advance in screening for CRC and high-risk adenoma over the gFOBT. For countries introducing FOBT, iFOBT should be considered the standard for CRC screening. Although rarely used at present, faecal DNA testing is likely to evolve and improve in the future. As availability of a blood test might be expected to increase uptake, research is currently focusing in this area, especially on the measurement of mutant DNA, specific mRNAs and specific protein panels that are relevant to CRC.77–80 However, the benefit of being able to complete a FOBT at home without the need for a general practitioner, clinic appointment or the need for venesection should be regarded as a major advantage.