Objective: To investigate whether NSAIDs, clopidogrel and warfarin reduce the positive predictive value for significant neoplasia cancer or polyp 10mm in an asymptomatic population undergoing FOBT screening for colorectal cancer.
Methods: This retrospective study investigated patients who had a positive faecal occult blood sample and than subsequently had a colonoscopy at The West London Bowel Cancer Screening Centre at Charing Cross Hospital, from November 2007 to March 2011.
Drug information was taken from screening questionnaires and colonoscopy findings from pathology reports. The finding of a cancer or polyp 10mm was considered as a significant neoplasia.
Results: 1,139 patients were included in the study. 32%(357) of the patients had significant neoplasia. 33.2% of the patients were on at least one of the drugs being studied.
The 'Aspirin' 'Aspirin/Other NSAIDs' and 'Other NSAIDs/Clopidogrel' groups had a reduced PPV for the finding of Significant neoplasia(P< 0.05).
The 'Other NSAIDs' and 'Other NSAIDs/Clopidogrel' group had a reduced PPV for the finding of neoplasia (Cancer/polyp and polyp).
Conclusion: This is the first study to investigate the individual effect of these drugs on patients involved in the BCSP. These findings (along with other literature) suggest that changes to the current BCSP should be considered. Viable alternative strategies include the use of the Faecal Immunochemical Test to replace the current Guaiac FOBT and the use of CT Colonography instead of Optical Colonoscopy.
Introduction
Colorectal Cancer
Colorectal Cancer (CRC) is a major cause of morbidity and mortality worldwide, with the highest incidence in developed countries. In 2008 it was the third most common cancer in men and women in the UK1. It carries a 50% long term survival rate, however the United Kingdom continues to lag behind mainland Europe in the survival for CRC2.
Screening
CRC is a disease that fits into many of the Wilson and Junger criteria that determine its suitability to be screened3. It has a well established sequence of progression in the form of the adenoma to carcinoma sequence. Intestinal adenomatous polyps are recognised as pre-malignant lesions, and removal has been shown to reduce the incidence of colorectal cancer4. Furthermore, the current investigation and treatment modalities are both effective and acceptable to most patients. This allows treatment of the disease at a stage where it can be effectively cured.
The NHS Bowel Cancer Screening Programme (BCSP) has been fully rolled out from the 20th August 2010. It involves the detection of occult blood in the faeces using the Faecal Occult Blood Test (FOBT). Currently the guaiac-based test is used. This involves smearing faeces onto absorbent paper onto which hydrogen peroxide is applied. If blood is present, it has a peroxidase-like effect, rapidly breaking down hydrogen peroxide and causing a change in colour. Currently all individuals registered with a GP between the ages of 60-74 are invited to join the screening programme every 2 years. Patients are asked to collect 6 samples from 3 separate bowel motions. Once the FOBT kit is used it is sent to one of 5 regional hubs, each co-ordinating approximately 20 screening centres. Those with a positive FOBT are invited for colonoscopy in one of these Bowel Cancer Screening Centres.
Recent studies have established the effectiveness of the BCSP in reducing CRC mortality and have reinforced the original randomised controlled trials of Mandel5, Hardcastle6 and Kronborg7. Results from the first 3 rounds of the Department of Health pilot study, as well as 2 year data from the first English centre to start bowel cancer screening, have shown that approximately 40% of the cancers detected were Dukes A compared to 10% of controls. This demonstrates the dramatic down staging of cancer that can be achieved8,9. Meta-analyses have estimated a reduction in CRC mortality of between 14-16 %10,11.
Studies that have investigated the effect of anticoagulants on the FOBT have suggested that they could interfere with the accuracy of the test. This includes the use of the drugs; Aspirin and other NSAIDS, Clopidogrel or Warfarin. A well designed retrospective study in the USA found that these drugs reduced the positive predictive value (PPV) of the FOBT by 10%12. These findings were also replicated in a study based on the Scottish arm of the BCSP13. Interestingly, the effect of warfarin is very controversial, with some studies suggesting it reduces PPV14 and others finding no effect15. Furthermore, some have suggested that the use of warfarin and other anticoagulants could lead to cancers being detected at an earlier stage as they are more likely to bleed. However, there is limited evidence for this as yet.
Non-Steroidal Anti-Inflammatory Drugs (NSAIDs)
NSAIDS are used throughout the world and include the drugs aspirin, ibuprofen and naproxen. Their mechanism of action involves the inhibition of the enzymes COX 1, COX 2 or both. Inhibiting these enzymes reduces the synthesis of prostaglandins and thromboxane and leads to their analgesic, antipyretic and anti-platelet properties16.
Gastrointestinal irritation is the main adverse drug reaction associated with the use of NSAIDS. Prolonged use is associated with gastric ulceration and bleeding which can manifest as occult blood in the faeces17.
Aspirin's cardioprotective effects has meant that it is often continuously prescribed to patients suffering from ischaemic heart disease. In 2009, 33 million prescriptions for aspirin were made in the U.K, typically in the form of a 75 mg tablet once a day18. Gastric side effects have been shown to be reduced through the simultaneous prescription of proton pump inhibitors such as omeprazole19. Furthermore, the use of enteric-coated formulations are claimed to reduce the incidence of gastrointestinal bleeding.
Clopidogrel
This is an antiplatelet drug that irreversibly inhibits the P2Y12 ADP chemoreceptor on platelets. This prevents the aggregation of platelets by blocking activation of the glycoprotein IIB/IIIa pathway20. It is commonly prescribed as a prophylactic agent in coronary artery, peripheral vascular and cerebrovascular disease and in 2009 there were approximately 3.5 million prescriptions for clopidogrel18. Similar to the use of aspirin, gastrointestinal haemorrhage is a common adverse drug reaction with an incidence of 2% annually. The rate of haemorrhage is increased when co-administered with aspirin21.
Warfarin
Warfarin inhibits the vitamin K-dependant synthesis of the clotting factors II, VII, IX and X. In order to convert these precursors into active form, the enzyme gamma-glutamyl carboxylase must carboxylate a glutamic acid residue on their surface. Carboxylation requires vitamin K as a co-factor. However, warfarin diminishes available vitamin K in the tissue reducing the carboxylation and hence synthesis of active factors22. Approximately 8 million prescriptions for warfarin were made in the U.K in 2009 for indications such as atrial fibrillation and prosthetic heart valves18.
Aims
The importance of establishing the effect of certain variables on the positive predictive value of the FOBT is essential. Even a modest decrease in PPV will significantly increase the number of colonoscopies performed when applied to a population based screening test. This would then translate into increased morbidity and mortality from procedure related complications as well as increased cost23. The impact of these drugs on the FOBT is also important as unnecessarily stopping treatment can put patients under avoidable risk. The current study aims to accurately quantify the effect of these drugs, and therefore potentially offer guidance when determining what screening approach to use in patients on these drug.
Hypothesis
NSAIDs, clopidogrel and warfarin reduce the positive predictive value for Significant neoplasia(cancer or polyp 10mm) in an asymptomatic population undergoing FOBT screening for colorectal cancer.
Methods
2.1 Data Collection
This retrospective study investigated patients who had a positive faecal occult blood sample and then subsequently had a colonoscopy at The West London Bowel Cancer Screening Centre at Charing Cross Hospital, London, from the period of March 2009 to April 2011.
A drug history was taken prior to colonoscopy by a specialist Bowel Cancer Screening Nurse. If the patient stated that they were on a particular drug it was assumed they had not started this medication after the FOBT. The short amount of time between the FOBT and colonoscopy meant that this had a low chance of occurring. Patients were excluded if they were unsure of their drug history or had not been taking the anticoagulant drugs regularly.
Information from colonoscopy reports was recorded including the diagnosis, number of adenomas, size of the largest adenoma, polyp histology and the Dukes stage. Diagnosis from any incomplete colonoscopies was attained from any additional procedures including CT Colonography.
A 'significant neoplasia' was determined as a cancer or polyp 10mm, a 'normal result' was described as one without any pathology, and 'other' was defined as any pathology that was not a cancer or a polyp e.g. angiodysplasia, diverticulosis and IBD.
Prior to analysis, the results of the current study were combined with data collected from an earlier study that used the same methodology between November 2007 to February 2009. This was done in order to increase the number of samples.
2.2 Statistics
The PPV of the FOBT was compared between the drugs group and the controls group using a two-tailed, Fisher's Exact test, online on Graph-pad software. A p-value of less than 0.05 was taken as statistically significant.
Results
Over the total period of the study data was collected from 1,139 patients who had a positive FOBT and subsequent colonoscopy. 416 patients were from November 2007 to February 2009, 723 patients were from March 2009 to April 2011 (the current study). All tables and figures are displayed at the end of the project.
Figure 1 (page 25) summarises the pathology findings. 76 (7%) of the patients were found to have CRC. Over half of the patients had some from of polyp with half of those patients having polyps 10mm or larger (Significant adenoma). In total 357 (32%) of the patients had significant neoplasia(Cancer or polyp 10mm). Approximately one third (365) of the patients were diagnosed as having a normal colon, and 109 (9%) with another pathology including conditions such as diverticulosis, IBD and angiodysplasia.
Figure 2 (page 26) describes the drug treatments that the patients were on. 33.2% of the patients were on at least one of the drugs with NSAIDs being the largest group (27%). The majority of the NSAIDs group was made up of aspirin users.
The effect of the drugs on the PPV of the FOBT has been categorised into different groups. Findings related to significant neoplasia are particularly important as the BCSP is designed to detect this in the community.
Significant neoplasia:
Cancer or polyp 10mm vs. a normal result
Cancer or polyp 10mm vs. a normal result or other pathology
Cancer or polyp 10mm vs. a normal result or other pathology or polyp <10mm
Neoplasia:
Cancer or polyp vs. a normal result
Cancer or polyp vs. a normal result or other pathology
Polyp vs. a normal result
Polyp vs. a normal result or other pathology
The patients on the drugs being studied were separated into Aspirin, Other NSAIDs (non-aspirin NSAIDs), Clopidogrel and Warfarin groups. The patients that were not on any of these drugs were used as controls.
Tables 1 and 2 (page 29 and 30) describe the findings for the data collected for the period of November 2007 to April 2011.
Aspirin
The aspirin users had a significantly reduced rate of significant neoplasia compared to controls (27% Aspirin, 33% Controls P=0.042). This is displayed by Figure 3 (page 27).
Other NSAIDs
Patients in the Other NSAIDs group had a significantly reduced rate of neoplasia compared to controls (42% Other NSAIDs, 60% Controls, P=0.036).
All NSAIDs (Aspirin or Other NSAIDs)
Those individuals on aspirin or other NSAIDs had a significantly reduced rate of significant neoplasia (all 3 categories) compared to controls (P=0.035, 0.022 and 0.044).
Other NSAIDs and Clopidogrel
The Other NSAIDs and Clopidogrel group had a significantly reduced rate of both significant neoplasia and neoplasia in all but one of the categories studied (Significant neoplasiaP=0.046, 0.044 and Neoplasia P=0.01, 0.012, 0.017).
Figure 4 (page 28) compares the PPV of the different drugs compared to the control group. This figure allows for the comparison of the effect of the different drugs. Aspirin and NSAIDs lead to a reduction in the PPV of approximately 7% and clopidogrel 11%. Warfarin does not reduce the PPV.
Discussion
The findings of this study have shown that the accuracy of the FOBT is reduced in patients using certain drugs. A summary of the effect of each drug and a comparison with other literature is discussed below.
4.1 Aspirin and Other NSAIDs
The current study has shown that the use of NSAIDs leads to a significant decrease in the PPV of the FOBT for both significant neoplasia and any neoplasia.
There was a 7% reduction in finding significant neoplasia in those patients treated with aspirin (0.042). Although the Other NSAIDs group did not lead to statistically significant results, there was a 9% reduction of significant neoplasia. The low number of patients in this category (36) may have allowed for a type 2 error.
The effect of other NSAIDs can also be observed when this data is combined with those on aspirin. This led to two of the non-significant aspirin categories to become statistically significant (Table 1). The role of other NSAIDs is further reinforced as significant findings were found in relation to the outcome of any neoplasia (Table 2).
The results of this study clearly show that NSAIDs have a significant impact on the FOBT. This is in accordance with previous literature. The work by Clarke et al helps to confirm the findings of the current study24. This research was carried out in the Tayside population included in the national colorectal cancer pilot. Here, Clarke et al showed a significantly reduced diagnosis of carcinoma and neoplasia in those on NSAIDs, aspirin or warfarin. It should be noted that this study did not investigate the effect of each of these drugs in isolation.
The work by Sawhney et al12 in 2010 currently best matches the current research. 1,126 patients that underwent colonoscopy after FOBT were separated into five groups; low-dose aspirin, NSAIDs, warfarin, clopidogrel and controls. Table 3 (page 31) compares the results of this study with the current study. Sawhney et al also found a reduced rate of significant neoplasia in those patients on aspirin. However, this was at a greater effect. This study found a 10% reduction in PPV with a noticeably more significant P value, 0.003. This is in contrast to a 7% reduction and a P value of 0.042 in the current study. Furthermore, this study also found a significant reduction in PPV in those patients on other NSAIDs (0.003).
There are some potential explanations behind the more significant findings, such as with the samples used. The average age of participants in the Sawhney et al study was 68.6 years and almost all participants were males. This can be contrasted with recent epidemiological data which found 50% of participants in the U.K screening programme to be from the ages of 60-64 and with more women responding to the invitations than men25. It has been proven that patients of older age face a higher risk of GI bleeding26. This could have led to the considerably higher false positives in the Sawhney et al study.
The handling of the data is also different. The aspirin group includes all those patients on aspirin up to 325 mg/day. Whilst the concentration of drugs was not attained in the current study it is reasonable to suggest that many of the patients might have been on 75 mg aspirin OD. This concentration difference could have led to more severe GI toxicity and hence explain the more conclusive result achieved by Sawhney et al. It should also be noted that this study had many more patients on clopidogrel and other NSAIDs and so may explain why they were able to find more significant results in these categories. This argument is further strengthened as the proportion of patients using NSAIDs that have significant neoplasia was similar between the two studies(10.8% and 8.9% respectively).
The reduction in positive predictive value observed in those patients on regular NSAIDs can be explained through three main actions; gastro-intestinal toxicity, chemoprevention and anti-platelet effect.
GI toxicity
This occurs through both a direct and indirect pathway. NSAIDs cause damage directly to the gastric mucosa due to their acidic nature. Moreover, the highly acidic environment of the stomach retains the drugs in their non-ionized lipophilic form. This promotes migration into the epithelial cells of the gastric tissue. Once there, the NSAIDs dissociate into their ionized form, causing the trapping of hydrogen ions. This leads to gastric erosion and bleeding27. NSAIDs have also been linked with mucosal damage through the uncoupling of mitochondria and subsequent release of apoptotic factors inducing cytochrome C28.
The indirect effects of NSAIDs appears to be the predominant means of injury to the gastric mucosa. This is related to the inhibition of gastric prostaglandin formation in the GI tract. Arachidonic acid is converted to prostaglandins by the cyclooxygenase enzymes (COX enzymes). NSAIDs inhibit the COX enzymes which in turn reduce prostaglandin formation in the gastric mucosa. This leads to decreased protective bicarbonate and mucus secretion but increased gastric acid production. It has also been shown that there are two isoforms of the COX enzymes (COX-1 and COX-2) with different roles in the body. COX-1 is regarded as having housekeeping functions in most tissues including the gastric mucosa. However, COX-2 has been found to be specifically related to the inflammatory response. As a result of these differing functions, it has been suggested that COX-1 inhibition induces adverse effects such as gastric ulcer formation. However, COX-2 inhibition is related to the anti-inflammatory effects of NSAIDs27,29,30. It has subsequently been shown that selective COX-2 inhibitors (celecoxib), have a reduced relative risk of GI ulceration compared to other NSAIDs31. However, COX-2 selective inhibitors have been controversially linked to increased cardiovascular disease, which has prevented their wide scale use32.
Interestingly, a systematic review has shown that faecal blood loss increases in a dose dependant manner in aspirin users33. In fact, anti-inflammatory doses (>1,800 mg/day) caused five times the blood loss compared to placebos. This could explain why aspirin alone did not lead to strongly significant effects as many of the patients may have only been on 75 mg/day.
Chemoprevention
The reduced incidence of significant neoplasia in NSAID users may also be related to their chemopreventive properties. These drugs have been shown to exert effects in all the clinical stages of colorectal neoplasia. Although initial studies focused on hereditary syndromes34, more recent data has found benefit in the general population. A 2010 meta-analysis of five randomised controlled trials demonstrated that the use of aspirin (mean of 6 years) reduced both the incidence and mortality of colon cancer over a 20 year period35.
One well characterised mechanism behind the chemopreventive properties of NSAIDs revolves around the inhibition of COX-2 which has been found to be increased in cancerous tissue. COX-2 has been linked with the formation of carcinogenic compounds such as prostaglandin E2(PGE2). This is hypothesised to lead to increased tumour angiogenesis, anti-apoptotic activity, tumour metastasis as well as decreased immune surveillance. Reduced prostaglandin synthesis is not the only way NSAIDs may exert their protective effects. NSAIDs have been shown to block the activity of the anti-apoptotic NF-kB pathway which has been shown to be overactive in many different tumours36.
4.2 Antiplatelet and Anticoagulant Drugs
The use of clopidogrel or warfarin did not lead to a statistically significant reduction in the positive predictive value of finding significant neoplasia or any neoplasia. This could be due to their underlying action in the body and is discussed below.
Clopidogrel
There is some evidence from my study that suggests that clopidogrel does increase false positives. When the results of those patients on clopidogrel are combined with other NSAIDs, a stronger statistically significant result is achieved than with other NSAIDs alone. Furthermore, Figure 4 shows a greater reduction in PPV in clopidogrel users than those on NSAIDs. In fact, clopidogrel appears to lead to a 10% reduction in the PPV compared to 7% in NSAID users. However, this value was not found to be statistically significant due to the low sample. Sawhney et al12- also found the largest reduction in PPV in clopidogrel users. Importantly this was found to be statistically significant due to the larger sample (41).
Clopidogrel is an antiplatelet drug that can induce GI haemorrhage. However, its use has been shown to moderately lower the rate of gastrointestinal bleeding compared to aspirin37,38. A recent editorial suggests that clopidogrel does not induce the formation of new ulcers (like NSAIDs) but may convert asymptomatic ulcers into clinically apparent, bleeding ulcers. The mechanism behind this resides in the fact that platelet aggregation plays a critical role in healing due to the release of pro-angiogenic factors, such as vascular endothelial growth factor. Therefore, clopidogrel may only introduce false positives in select patients with a background of asymptomatic ulcers39. This is estimated to form at a rate of up to 2 percent per month in patients taking no medications40. Unfortunately this was very difficult to investigate due to the very few number of clopidogrel patients.
Warfarin
The issue of whether warfarin increases the false positive rate of the FOBT is controversial with much conflicting research. Although a significant result was obtained by Sawhney et al, the p value was borderline significant (0.05). Another notable study found a significantly increased diagnosis of significant adenoma but not cancer in warfarin users41. However, the current study and other studies42,43 failed to find any relationship. This lack of evidence may highlight warfarin as a potential help rather than hindrance. This has been suggested by Bin et al who found a higher proportion of Dukes A/B tumours in warfarin users (p=0.046)42.
This phenomenon may be explained by the fact that warfarin does not induce spontaneous bleeding in the gastric mucosa but increases bleeding of established neoplasia. The lack of samples meant that the current study was unable to investigate this. Moreover, as the BCSP has generally been shown to detect cancers at an early stage, this would probably need a large patient sample to prove.
Aspirin
Aspirin has both antiplatelet function and can induce GI bleeding. These actions would appear to counteract their impact on the FOBT, potentially increasing the likelihood of bleeding from neoplastic lesions but also leading to false positive FOBT results. From analysis of the current study and previous literature the latter effect appears to dominate.
4.3 Clinical Implications
There are several possible CRC screening tests available. The type of investigation can be classified into two categories. Stool based tests, which can detect blood (guaiac and immunochemical faecal occult blood tests) or faecal DNA. Structural examinations, which can be described as endoscopic (flexible sigmoidoscopy and colonoscopy) or radiological exams (computed tomography colonography and magnetic resonance colonography). So far only the FOBT44,45 and flexible sigmoidoscopy46 have demonstrated a reduction in mortality over a ten year period.
Possible changes to the BCSP are discussed below:
Individuals on certain drugs could be asked to stop their use before undertaking the gFOBT
An alternative initial screening option is considered.
An alternative secondary screening option is considered.
There is continued use of the gFOBT, but the approach to screening individuals on certain drugs is changed.
I) Temporarily stop treatment
If possible patients should be advised to temporarily stop the use of certain drugs before the FOBT. However, in patients in whom this is not practical other options need to be considered as this may place them under unnecessary harm.
II) Initial screening options, Guaiac FOBT (gFOBT) vs. Faecal Immunochemical Test (FIT)
The current BCSP uses a guaiac FOBT, however the FIT is considered to be superior. In relation to the BCSP this translates into a higher detection rate of significant adenoma and cancer compared to the gFOBT47,48. Furthermore, the impact of certain drugs on the positive predictive value of the FIT has recently been investigated. In a study of 2,376 patients who had FIT testing and subsequent colonoscopy, the PPV for significant neoplasia was not affected by ongoing anticoagulant or low dose aspirin therapy49. In fact, a previous study found that the sensitivity of the FIT is slightly increased with the use of low-dose aspirin, NSAIDs, and anticoagulants. These findings could be due to the mechanics of the FIT. It uses antibodies to detect the globin portion of human haemoglobin. This means that it is specific for detecting blood from the colon as any globin released from the gastric mucosa is degraded47. There are also other advantages of using the FIT over gFOBT. Whilst gFOBT only determines the presence or absence of blood, the FIT allows its quantification. This can then be used to determine a threshold value for referral to colonoscopy, allowing specific screening programmes to tailor the test according to their particular resources. Furthermore, the test is not affected by diet, unlike the gFOBT. The FIT is also easier to perform by patients, which has been reflected with higher uptake rates in two Dutch studies48,50.
III) Secondary screening options, Computed Tomographic Colonography (CTC) vs. Optical Colonoscopy
In the current BCSP, any individual with a positive gFOBT is referred for optical colonoscopy. Patients who are referred unnecessarily can be considered to be exposed to more harm than benefit as there are between 3-5 serious adverse events per 1000 procedures51. When these risks are applied to a screening program, this equates to a significant burden of mortality and morbidity. Furthermore, the benefits of colonoscopy decrease with advancing age and the existence of co-morbidities because of the increased complication rate and reduced life expectancy of the individual51.
CTC allows endo-luminal examination of the entire colon using computed tomography. Although it still requires complete bowel preparation and inflation of the bowel with CO2, it is considered much less invasive than optical colonoscopy with an extremely low complication rate52. Furthermore, the sensitivity for diagnosing polyps 10mm or larger has been found to be around 90%53, which is similar to optical colonoscopy (100%) 54. These characteristics would make it an ideal screening modality in patients in whom there is a suspicion of a false positive FOBT.
Introduction of CTC may also bring economic reward. A model used by Sweet A concluded that using CTC to follow up FOBT positive patients was significantly less costly than directing all FOBT positive patients to optical colonoscopy55.
However there are drawbacks of using CTC. The sensitivity for the detection of polyps more than 6mm is significantly reduced (78%). Furthermore, any abnormal lesions found under CTC require follow up with colonoscopy due to the need for biopsy and resection. The radiation exposure (10-12mSv) is also important to consider, as this itself can induce cancer54. This point is especially significant in individuals under surveillance as their risks are increased with repeated exposure. Although CTC allows the identification of extracolonic abnormalities, the extra diagnostic tests can place an unnecessary burden on patients. One study found that up to 12% of patients underwent additional testing due to an extracolonic finding, but only a few actually gained any real clinical benefit56.
Another argument against using CTC in the drugs cohort is that it would avoid approximately 75% of these patients from having an unnecessary colonoscopy (Figure 4). This number is not too different from those patients not on any drugs that have unnecessary colonoscopies anyway, approximately 66%.
IV) Increase the threshold for a positive FOBT in NSAID users
A criterion could be created to reduce the number of false positive FOBTs. The risk factors for the development of NSAID-associated gastroduodenal ulcers have been well documented (Table 4, page 32). This information can be used to create a criterion for the referral of patients on these drugs for optical colonoscopy. For example, a higher threshold can be used to register a positive FOBT reading in individuals that use NSAIDs who also have a number of risk factors. In the current BCSP 5/6 abnormal readings would warrant an individual to have a colonoscopy. However, the threshold could be moved to 6 abnormal readings. Research has also found that age, gender and family history of CRC all affect the positive predictive value of the FOBT, suggesting that these could also be factored into a criterion57-59.
4.4 Limitations
There are many potential improvements to the methods of the current study. Obtaining a larger sample would have been beneficial. Although there were statistically significant results, a larger sample size would have also allowed me to investigate the effect of specific NSAIDs and whether warfarin leads to cancers being diagnosed at an earlier stage. This would have been especially interesting with the use of COX-2 inhibitors which have been shown to cause reduced GI complications31-. Furthermore, the concomitant use of NSAIDs with anticoagulants is an established risk factor for the development of GI ulceration60. Although there was a 14% reduction in finding significant neoplasia in those patients on more than one drug, this was not found to be statistically significant with the current sample size.
It would also have been better to document the dosage of the drugs that were investigated. This is because larger doses of NSAIDs have been linked with increased gastric ulceration and may lead to more false positive results60. Data on the age and sex of the patients could also have been collected and used to better match the control group with those on drugs using a T-test. This is important as those on certain drugs may have been older and so were more predisposed to GI bleeding.
Information about drugs was taken from screening questionnaires carried out after the FOBT. Ideally, it would have been better to know what drugs the patients were on whilst doing their FOBT as there is a small chance their drug regimes may have changed from when they took the FOBT and their screening interview.
4.5 Further Research
More extensive research over the use of the FIT in the BCSP should transpire as this has the potential to replace the gFOBT. Although limited in number, current literature suggests that NSAIDs and anticoagulants do not affect the results of the FIT as they do with the gFOBT. Furthermore, the FIT has other benefits as mentioned in the discussion.
A trial investigating the use of CTC before an optical colonoscopy in patients at risk of false positive FOBT should also be undertaken. This should focus on cancer detection rates as well as financial consequences of such a programme.
There has been growing interest in DNA based faecal tests as an alternative screening modality. They detect specific DNA mutations that are associated with CRC development and have been found to be significantly more sensitive than gFOBT. The reason behind this is that mutated cells are shed continuously from neoplasia into the faeces whereas CRC and adenomas bleed intermittently61. However, more studies are needed to compare DNA testing to the FIT and its performance on average-risk populations.
Summary
The investigations used to screen for bowel cancer involve a balance between cancer detection (FOBT) and cancer prevention (colonoscopy). Prevention strategies normally involve the removal of pre-malignant lesions to prevent the development of cancer. However, these investigations are normally associated with a much higher complication rate and a reduced patient uptake. Research must focus on refining the accuracy of detection tests to ensure patients are only exposed to more invasive tests if absolutely necessary. This work, in corroboration with similar studies, suggests that an alternative screening strategy should be seriously considered in some patients.
Figures and Tables
Figure 1: This figure summarises the findings from the colonoscopy pathology reports. 32% of the patients had significant neoplasia (Cancer or polyp 10mm). Other pathologies included diverticular disease, IBD, and angiodysplasia.
Figure 2: Venn diagram that displays the proportion of patients in each drug category. 27% (314) of the patients were on NSAIDs, 3.16% (36) on Warfarin and less than 1% (9) on clopidogrel.
C:\Users\Alg\Documents\BSC stuff\BSC project\venn diagramn 2.bmp
Figure 3: This bar chart illustrates the significant drop in PPV of the FOBT in patients on aspirin compared to controls (0.042). There were approximately 7% fewer cases of significant neoplasia in the aspirin group.
Figure 4: Bar chart comparing the drop in PPV of the FOBT between the different drugs groups. NSAIDs have been shown to significantly reduce the PPV. Clopidogrel use is associated with the most significant reduction in PPV, but due to the low numbers this was not statistically significant. Warfarin did not lead to any significant fall in PPV.
Table 1: P values for the Fisher's Exact tests when observing the significant neoplasia categories. All the patients on drugs were compared to controls (patients not on NSAIDs, Clopidogrel or Warfarin).
P <0.05
Column1
Aspirin
Aspirin or NSAIDs
Other NSAIDs
Other NSAIDs or Clopidogrel
Clopidogrel
Warfarin
Multiple Drugs
Cancer or polyp 10mm vs. a normal result
0.090
0.035
0.150
0.046
0.161
1
0.246
Cancer or polyp 10mm vs. a normal result or other pathology
0.066
0.022
0.093
0.044
0.300
1
0.278
Cancer or polyp 10mm vs. a normal result or other pathology or polyp<10mm
0.042
0.044
0.366
0.254
0.725
1
0.240
Column1
Aspirin
Aspirin or NSAIDs
Other NSAIDs
Other NSAIDs or Clopidogrel
Clopidogrel
Warfarin
Multiple Drugs
Cancer or polyp vs. a normal result
0.312
0.105
0.052
0.01
0.069
1
0.446
Cancer or polyp vs. a normal result or other pathology
0.253
0.078
0.036
0.012
0.168
1
0.503
Polyp vs. a normal result
0.382
0.154
0.081
0.017
0.084
0.853
0.621
Polyp vs. a normal result or other pathology
0.306
0.111
0.055
0.019
0.186
0.861
0.648
Table 2. P values for Fisher's Exact tests when observing the neoplasia categories. All the patients on drugs were compared to controls (patients not on NSAIDs, Clopidogrel or Warfarin). P <0.05
Table 3: Comparison of Sawhney et al(12) with current study
Control
Aspirin
Other NSAIDs
Warfarin
Clopidogrel
Sawhney et al(12) (% with significant neoplasia)
30.50%
20.5% (0.003)
19.7% (0.003)
20.0% (0.05)
7.30% (0.02)
Current study Nov 2007-April 2011 (% with significant neoplasia)
33.90%
27.17% (0.042)
25.0% (0.271)
33.3%(0.946)
22.0% (0.463)
Table 4: This table describes the risk factors for the development of GI ulceration. This form of information can be used to predict those patients who are more likely to have false positives with the FOBT.
Established risk factors for the development of NSAID associated GI ulcers60
Advanced age
History of ulcer
Concomitant use of corticosteroids
Higher doses of NSAIDs and the use of multiple NSAIDs
Concomitant administration of anticoagulants
Serious Systemic disorders
7. Acknowledgements
I would like to thank my supervisors Dr Geoff Smith and Dr John Martin. I am also very appreciative of all the help and support from the Gastroenterology department at Charing Cross Hospital, especially the Bowel Cancer Screening Nurse Amy.
