• 3. Leeds Flexiscope Trial Results
  
• 4. Cases and Controls
  
• 5. N-acetyltransferase Genotype
  
• 6. Smoking
  
• 7. Measures of Meat Exposure
  

3.a. General Practice Recruitment

31 general practices were recruited out of 35 invited.  A total of 27,513 potential volunteers were available from the 161 general practitioners involved.  Of these, 775 potential volunteers were excluded according to the criteria in Table 15 (see Figure 14).

3.b Questionnaire response

The remainder 26,738 were sent an initial questionnaire (Appendix 3: Initial Questionnaire).

The results of this questionnaire are listed in Figure 14. Recruitment and response to initial questionnaire.  Of the 14,230 volunteers who were interested i.e. responded 'Yes, definitely' or 'Yes, probably', 13058 were randomised approximately 2:1 to the control (8689) or screening arm (4369) according to the trial protocol (Figure 7).

3.c. Attendance

4304 volunteers were invited to attend for screening flexible sigmoidoscopy from the 4369 that were randomised.  Of these, 2941 (68.3%) attended which was considerably better than the 60% attendance rate that had been predicted (Figure 7).  8 patients despite attending for screening did not undergo flexible sigmoidoscopy either because they did not tolerate the enema, found even digital examination too uncomfortable and declined further examination or had an alternate adverse event (see 3.g. Adverse events).  A further 27 volunteers underwent colonoscopy in a separate centre (Wharfedale Hospital) as part of a family history study leaving 2906 that underwent screening flexible sigmoidoscopy in Leeds.

Sex Distribution

• Males        1434  
• Females       1472

Age Distribution   

• Age range  55 - 66 (mean 59.9)

3.d. Screening findings and outcome

Examination Completion

A complete examination was defined as being beyond the rectosigmoid junction as judged by the endoscopist (Figure 15) such that the whole sigmoid colon was viewed.  The examination was judged complete in 86%, the remaining 14% being incomplete because of patient discomfort or poor bowel preparation.

Bowel Preparation

A high proportion (>90%) of volunteers took a self-administered phosphate enema at home prior to travelling to the endoscopy unit.  The remainder were given an enema in the department by one of the endoscopy nurses.  The reasons for this were principally anxiety about giving themselves the enema though a few were disabled or the enema had been damaged in the post (one person reported that their dog ate it!).

Bowel preparation was judged by the endoscopist as being excellent (no faecal matter or faecally stained fluid), good (some soiling with faecal fluid), adequate (some residue but not enough to inhibit inspection of the mucosa) or poor (significant soiling that limited the examination).  93% of examinations had adequate bowel preparation or better (Figure 16).

Sigmoidoscopy Findings

The distribution of pathology seen at screening flexible sigmoidoscopy is shown in Figure 17.  The overall adenoma rate in Leeds was 14% (including coincident adenomatous and metaplastic polyps).  A subsequent study performed between centres found that the only independent factor that correlated with adenoma rate was the time taken to withdraw the scope²⁶⁹.  This suggest that there is some inter-individual variation between endoscopists.  The range of adenoma rate varied between 9 and 15%.

The adenomas found at screening were categorised histologically as being tubular, tubulo-villous or villous (Figure 18) and by the degree of dysplasia (Figure 19).  Histological examination was undertaken by a single pathologist (Professor P. Quirke), who also was head of the trial pathologists group, to ensure that there was conformity in the histological reporting.

Figure 18. Histology of adenomas found at screening

Screening Outcome

The outcome after screening was dependent on the findings (Figure 20) combined with the informed consent of the volunteer:

• Immediate discharge - those with a normal examination, metaplastic polyps in the distal rectum or inadequate bowel preparation and refused repeat examination.
• Discharge after histology - those volunteers who have had a polyp biopsied and it is either a low risk adenoma or a metaplastic polyp.
• Repeat flexible sigmoidoscopy - if a polyp was low risk but there were doubts as to completeness of excision, if a polyp was seen but not excised or if the patient had poor prep and wished to return for repeat examination on another day.
• Refer for colonoscopy - either a high risk adenoma was found (>1cm diameter, villous histology, severe dysplasia), or a suspected carcinoma or there were more than 20 metaplastic polyps.
• Refer for surgery - a definite carcinoma or a large polyp not respectable endoscopically.

90% of volunteers were discharged following a single examination, this was 5% less than projected and this is due to repeat flexible sigmoidoscopy for poor bowel preparation and repeat examination to assess completeness of excision.  The 5.4% colonoscopy rate was as expected (Figure 20).

3.e. Colonoscopy findings

Attendance rates for colonoscopy were 91% (148 attended out of 163 referred for colonoscopy).  Examination completion to the caecum was achieved in 70% of cases (104 out of 148).  10% of the colonoscopies revealed significant pathology (adenoma or cancer) and 85% were normal (Figure 21).

3.f. Cancer outcomes

There were 11 cancers found in Leeds through the trial - 9 found at screening flexible sigmoidoscopy and a further 2 at colonoscopy.  5 out of 11 (55%) were Dukes' A (Figure 22) including 1 malignant polyp that at subsequent colonic resection showed no residual tumour.  There was 1 case that was unable to be classified because at operation the tumour was found to be inoperable and so no complete histological specimen was obtained.

Figure 22.  Dukes' classification of cancers

3.g. Adverse events

(This data was compiled by Rekha Patel of St. Mark's Hospital)

Before each adverse event summary the bold number in parenthesis grades the event by its perceived level of relatedness to the screening as follows:  (1) definite; (2) probable; (3) possible

Death following surgery

(2) This lady had undergone an anterior resection in November 1997 for removal of a large, carpet-like, severely dysplastic villous adenoma of the rectosigmoid junction. The anastamosis was covered with a loop ileostomy and the patient made an excellent post-operative recovery. Three months later she was admitted for closure of her ileostomy and this was done without any problems. Although there was a temporary small bowel obstruction following the procedure, by the seventh day she seemed to be making a good recovery. However, at 3am the following morning she had a cardiac arrest without any warning. At first it was felt that it was probably a pulmonary embolus, but this was discounted and the conclusion was that she probably had acute cardiac arrhythmia (March 1998).

Chest pain

(1) This patient, who suffered from angina and had previously undergone angioplasty, came for a flexible sigmoidoscopy in June 1997, but this was incomplete due to poor bowel prep. She then returned in November and the test had to be abandoned again because she was experiencing a tightness in the chest. She was treated in the Accident and Emergency department (November 1997).

(1) This patient came for a flexible sigmoidoscopy and had three small polyps removed by cold biopsy. On the questionnaire she filled out the next day () she reported that she had had a very bad night's sleep because of severe pains, which she said were due to wind but felt in her chest mostly. At three months she reported no further problems. (February 1998)

Bleeding requiring hospitalisation

Following flexible sigmoidoscopy

(2) Bleeding after the snaring of an 11mm polyp in the sigmoid colon at flexible sigmoidoscopy. This episode lasted seven days and required a stay of ten days in hospital.  The patient had colonoscopy one week after discharge at which the site of the previous polypectomy was observed to be healing.  (November 1998).

Bleeding Following Colonoscopy

(1) This woman was admitted for rectal bleeding following an incomplete colonoscopy at which a 16mm tubulovillous adenoma was completely removed. She lost five pints of blood, was treated with  a two unit blood transfusion and glucose, and spent three days in hospital. She had a barium enema five weeks after colonoscopy and no abnormality was seen except for diverticular disease. She was examined by the consultant surgeon on several occasions during the course of the following year; he noted that she had had episodes of rectal bleeding before entering the trial and he discharged her to routine polyp follow up in May 1998 after injecting her haemorrhoids with phenol had stopped the bleeding. (May 1997)

Bleeding not requiring hospital admission

Cases with no previous bowel complaint of any kind

(1) This woman had two tiny (2mm) polyps cold biopsied at her flexible sigmoidoscopy. At three months she reported that she'd had rectal bleeding immediately after the test. She described it as 'very little bleeding' , but said that it had lasted a few weeks and that she was still getting it occasionally (March 1997).

(3) This man had rectal bleeding within a month of a flexible sigmoidoscopy  at which no polyps were seen. He had a colonoscopy  privately two months later. There was nothing abnormal detected at this examination. He was not admitted. On review three months later he was seen to be fine and was discharged. (October 1997)

Those who had previously had some kind of bowel complaint

(3) This man, who suffered from piles, bled for one week after a completely normal flexible sigmoidoscopy (April 1997)

(2) This man had said that he already sometimes experienced bleeding per rectum. Three months after his flexible sigmoidoscopy,  at which four polyps were hot biopsied -one 8mm in diameter, the others all less than 5mm-  he reported slight discomfort and slight bleeding starting within 24 hours of the test. The soreness and the bleeding lasted for about 10 days (June 1997).

Hospitalisation for reasons other than bleeding

(3) This lady attended for flexible sigmoidoscopy but was not screened due to a severe headache after taking the enema, although the hospital were unable to confirm if this was related. She had the symptoms of a subarachnoid haemorrhage and was admitted overnight. She underwent a CT Scan and lumbar puncture, then discharged with no abnormality detected. (April 1997).

Fainting or vaso-vagal symptoms

Following the enema for flexible sigmoidoscopy

(1) This woman underwent a flexible sigmoidoscopy at which nothing abnormal was detected. On her post-examination questionnaire she reported that after taking the enema she felt unwell. She went to lay on her bed and fainted, and felt faint and sickly for 30 minutes after that. (April 1998)

(1) On the report for this man's flexible sigmoidoscopy the endoscopist noted that after his bowel preparation this man had experienced a vaso-vagal attack lasting three minutes. On his post-examination questionnaire the patient himself said that he'd fainted just after taking the enema and had some 'cramp-like discomfort' for some time, but that he recovered before the test itself commenced (May 1998).

(1) This woman informed us in her post examination questionnaire that she had fainted at home before the test. (November 1998).

(1) This woman had a vaso-vagal attack in the trial unit after taking her enema. It lasted only one minute and then resolved after the subject was put in the recovery position, but her sigmoidoscopy was not carried out because of this adverse reaction (April 1997).

(1) This man had a vaso-vagal attack in the trial unit after taking his enema. It lasted three minutes and he also felt unwell and experienced abdominal pain. His sigmoidoscopy was not done due to this adverse reaction. (September 1997).

(1) This man had a vaso-vagal attack at home after taking his enema. It lasted 15 minutes. The patient contacted his GP and was treated with iv fluids, but he did not then have a sigmoidoscopy. (November 1997).

(1) This woman had a vaso-vagal attack at home after taking her enema. It lasted five minutes. (October 1998).

Fainting under other circumstances

(3) This man reported in his post examination questionnaire that he had fainted the day after a Flexible Sigmoidoscopy at which a 3mm rectal polyp had been removed (July 1998).

4.a. Leeds Results

There were a total of 406 volunteers who were found to have at least 1 adenoma.  The study group of 376 was selected from this group according to the following criteria:

• Histologically confirmed adenoma of the left colon
• Caucasian
• Following informed consent agreed to blood taking and DNA analysis

The distribution of the number of polyps found is shown in Figure 23.  It was unusual to find more than 2 adenomas (4%), but rarely an individual was found with as many as 6 confirmed adenomas in the left colon alone. Interestingly, there were no polyps seen in the proximal colon on colonoscopy in that individual.

Figure 23. Distribution of number of adenomas found at screening

Small distal metaplastic polyps seen in the last 5cm of the rectum were very commonly found and there was debate as to their significance.  In light of this a consensus was reached that different endpoints may be used (Table 18).  Subsequent data analysis within this thesis was performed on Case group 2 and Control group 2 i.e. the presence of an adenoma (but no carcinoma) identified a case regardless of the presence of metaplastic polyps.  Controls had a 'clean' colon except for a few with metaplastic polyps in the distal rectum (MDRs)

348 controls were matched according to protocol by age, sex and G.P. practice (see 2.b.ii. Cases and controls).  The distribution by sex was 110 female and 238 male case / controls pairs.

There was no significant difference between the ages of the matched pairs.  Mean age was 60.6 years (s.d. = 2.8) for cases and 60.4 years (s.d. = 2.9) for controls.

4.b. Collaborative Results

Collaboration between the FlexiScope Trial Centres of Leeds, Cambridge and Portsmouth resulted in 887 matched cases and controls being recruited (Leeds = 348 , Cambridge = 302 and Portsmouth = 237 pairs).

Comparability Between Centres

The data from each of the 3 centres were analysed to ensure comparability.

There was no significant difference in those who had ever / never smoked (Table 19).  But analysis of smoking duration (Table 20) showed a significant difference with Leeds and Portsmouth having a greater proportion of smokers who had smoked for a longer duration (c² = 31.6, p < 0.001).

Intake of red meat (Table 21) however was lower in Cambridge compared to Leeds and Portsmouth and reached statistical significance (c² = 14.4, p = 0.025).  There was no significant difference for preference for well-cooked red meat (Table 22) between the centres.

There were differences in NAT1 genotype between the centres (Table 23) with Leeds having a high proportion of fast acetylators for NAT1 (27%) and Portsmouth having a low proportion (18%) (c² = 13.5, p = 0.001).  There was however no significant difference seen for NAT2 (Table 24).

5.a. NAT1 Genotype

5.a.i. Leeds NAT1 Results

The analysis of NAT1 genotype focused on the 'high risk' 'fast acetylator' allele NAT1*10 in comparison to the other alleles.  The putative 'fast acetylator' phenotype is associated with either heterozygous or homozygous NAT1*10 alleles and in the analysis these genotypes have been grouped together.

A total of 711 volunteers were genotyped for NAT1.  Table 25 shows the analysis of putative NAT1 phenotype and adenoma.  Analysis of the fast NAT1*10 allele using conditional logistic regression (n=327 matched pairs) was associated with a 1.24 times greater risk of adenoma but this did not reach statistical significance (95% C.I. = 0.87 - 1.78).

Analysis of NAT1 alleles showed no significant increased risk for any individual allele (Table 26).  In 3 of the DNA samples analysed, only 1 of the alleles could be determined with confidence due to technical problems with the assay and so these contribute only 1 allele to this analysis and therefore there is an odd number of total alleles.

5.a.ii. Collaborative NAT1 Results

The same analysis of NAT1 genotype was performed on the collaborative dataset.

A total of 1905 volunteers were genotyped for NAT1 (Table 27).  The fast NAT1*10 allele showed no significant increased risk of adenoma formation overall or in any centre individually.

Assessment of individual alleles showed no significant association with the risk of an adenoma.

5.b. NAT2 Genotype

5.b.i. Leeds NAT2 Results

NAT2 fast phenotype was considered by the presence of one or more F1 alleles.  Cross-tabulation of putative fast NAT2 phenotype showed a significant association with the risk of adenoma in Leeds (Table 29).  Analysis using conditional logistic regression (n=336 matched pairs) was associated with a 1.35 times greater risk of adenoma and this was close but did not reach statistical significance (odds ratio 1.35, 95% C.I. 0.99 - 1.85).

A total of 721 volunteers were genotyped for NAT2 (Table 29)

In light of the contention as to whether a homozygous 'fast' NAT2 (F1) allele confers faster acetylation and therefore greater adenoma risk than heterozygous 'fast' NAT2^218,219, an analysis was performed comparing slow (mutant / mutant), intermediate (mutant / F1) and fast (F1 / F1) acetylators.  There was no associated increased risk of developing an adenoma with the homozygous F1 allele (Table 30).

Analysis of individual alleles showed there was no significant single allele associated with adenoma formation (Table 31).

5.b.ii. Collaborative NAT2 results

Univariate analysis for NAT2 fast phenotype shows a significant correlation with adenomas in Leeds (p=0.042) (Table 29).  This was not the case in the other centres and the overall effect is null (Table 32).

A total of 1626 volunteers were genotyped for NAT2 (Table 32).

Analysis on the collaborative dataset was performed comparing slow (mutant / mutant), intermediate (mutant / F1) and fast (F1 / F1) acetylators.  There was no associated increased risk of developing an adenoma with the homozygous F1 allele (Table 33).

Similarly, the individual allele frequency showed no significant single allele was associated with the risk of an adenoma.

5.c. NAT1 / NAT2 Genotypes

5.c.1. Leeds NAT1 / NAT2 Results

Analysis of the combination of NAT genotypes was undertaken looking at different ways of combining NAT1 and NAT2 genotypes.

The effect of a subject having either fast NAT1 OR fast NAT2 putative phenotypes was analysed using a chi-squared test comparing these 'fast  phenotypes' to the remaining 'slow phenotypes'.  There was a significant increased risk of having an adenoma in the 'fast phenotype' group in Leeds (Table 35) and this was associated with 1.5 times the risk of adenoma using conditional logistic regression (n = 331 matched pairs, odds ratio = 1.50, 95% C.I. = 1.1 - 2.0).

Fast NAT1 AND fast NAT2 had no significant effect on adenoma risk and therefore in subsequent analyses investigating the combination of NAT genotypes the focus has been fast NAT1 OR fast NAT2.

5.c.ii. Collaborative NAT1 / NAT2 Results

Analysis was performed to assess the effect of fast NAT1 or fast NAT2 acetylator genotypes versus slow NAT1 and slow NAT2 for the collaborative data.  This demonstrated no significant difference despite the positive result for the Leeds data.

Stratifying by centre, the adenoma risk of fast NAT1 or fast NAT2 genotype compared to slow NAT1 and slow NAT2 was below unity in Cambridge (0.8) and Portsmouth (0.65).  Overall, there was no significant difference in the collaborative data (c² = 0.52, p = 0.47).

6.a. Smoking

6.a.i. Leeds Smoking Results

Self-reported data from the pre-screening medical questionnaire (Appendix 6: Pre-Screening Medical Form) demonstrated 61% of the case / control study population had smoked or were continuing to smoke.  Mean duration was 28.8 years (Range 1 - 53 years, Std. Dev. 12.6 years).  49% of women had smoked compared to 68% of men.

An analysis was performed using c² test to assess the effect of smoking on the risk of finding an adenomatous polyp in the left colon.  The cases and controls were therefore divided in to non-, ex- and current smokers and analysis was performed for the group as a whole and individually by sex. (Table 37).  The odds ratios for developing an adenoma according to sex and smoking exposure are shown in Table 38.

These data show a significant increased risk of developing an adenoma in people who were current smokers and this association was stronger in women.  Conditional logistic regression likelihood ratio (LR) test for interaction showed a significant increased in risk females for ex-smoking and current smoking when compared to males (c² = 10.1, p = 0.006).

Duration of smoking was coded as never, less than 20 years, between 20 and 35 years and greater than 35 years and analysed for the group as a whole  and individually by sex (Table 39).  The time intervals were chosen to give similar numbers in each of period and also to fit in with Giovannucci's theory that smoking for greater than 20 years increases adenoma risk and smoking for greater than 35 years increases cancer risk^162,163.

There was a correlation with smoking duration and adenoma formation overall and in women, but this was not significant in men.  The group of women at highest risk were those who had smoked the longest (odds ratio - 3.4, 95% C.I. 1.7- 6.8).  In men this correlation with duration of smoking was not apparent (Table 40).

6.a.ii. Collaborative Smoking Results

Of the combined collaborative case / control study population 61% had smoked or were continuing to smoke.  Mean duration was 29.1 years (Range 1 - 56 years, Std. Dev. 13.0 years).  46% of women had ever smoked compared to 68% of men.  This was almost identical to the Leeds data (6.a.i. Leeds Smoking Results). 

Identical analyses were performed on the collaborative data as the Leeds data.  The c² test was used to assess the effect of smoking on the risk of finding an adenomatous polyp in the left colon and are shown in Table 41.  The odds ratios for developing an adenoma according to sex and smoking exposure are shown in Table 42.

These data show a significantly increased risk of developing an adenoma in people who were current smokers.  This association was very strong for both sexes and greater than the risk of being an ex-smoker.  Ex-smokers had a significantly increased risk in woman (odds ratio = 1.93, 95% C.I. 1.36 - 2.78) but this was not the case in men (odds ratio = 1.17, 95% C.I. 0.91 - 1.50).

There was no significant difference due to sex for the effect of being a non-, ex- or current smoker on adenomas (LR Test for interaction between smoking duration and sex: c² = 3.72, p = 0.16).

There was a correlation with smoking duration and adenoma risk overall and the group at highest risk were those who had smoked the longest with a dramatic increase in risk in the group who had smoked for longer than 35 years (Table 44).  This effect showed no significant differences due to sex (LR Test: c² = 4.69, p = 0.20).