Symptoms do not predict colorectal cancer in an FOB screened population

Background and aims

Scotland has the highest incidence of bowel cancer in the United Kingdom with 3500 cases diagnosed annually. ¹ Early diagnosis at a favourable stage confers significant survival benefit. Urgent symptomatic referral is currently the main diagnostic pathway promoted by Scottish Government Referral Guidelines and Standards.^2,3 Unfortunately, despite this colorectal cancer tends to be diagnosed at a late and often advanced stage in Scotland and streamlined diagnostic pathways have yet to reverse this trend. ³

The National Bowel Screening Programme aims to detect cancer at an early or even precancerous stage in asymptomatic individuals using faecal occult blood (FOB) testing. The meta-analysis of FOB screening shows that it reduces mortality and improves survival by 16%. ⁴ Individuals in Scotland aged 50–74 are sent a FOB test kit every 2 years. Those testing positive are invited to attend for a colonoscopy. Implementation of the Scottish Bowel Screening programme was phased in across Scottish Health Boards and complete coverage was achieved by the end of 2009. Ayrshire and Arran Health Board began screening in September 2007.

It is a basic premise of cancer screening that asymptomatic individuals are tested. Symptomatic individuals are expected to present via symptomatic referral pathways. However, a previous study has shown that many people who have had a positive FOB test and attend for surveillance colonoscopy are in fact symptomatic. ⁵ This, therefore, is a population with a high prevalence of neoplasia in which we can study the possible predictive value of symptoms.

We aimed to:

measure the prevalence of lower gastrointestinal (GI) symptoms in FOB screen-positive patients in Ayrshire and Arran;

Methods

This is a prospective study collecting data on a consecutive series of two groups of patients, screened patients from the first 6 months of the Ayrshire and Arran Bowel Screening Programme and symptomatic patients referred urgently from primary care with a suspected diagnosis of cancer. All FOB screen-positive patients were asked about symptoms using a standard pro forma at pre-colonoscopy assessment. They were asked about overt rectal bleeding, altered bowel habit, abdominal pain, weight loss and anaemia.

We calculated the positive predictive values (PPVs) for neoplasia by dividing the number of true positive tests (cancers plus adenomas) by the number of true positives added to the number of false positives. We compared the neoplasia diagnostic rate in FOB screen-positive patients with symptomatic patients referred urgently with a suspected diagnosis of colorectal cancer attending a one-stop diagnostic colorectal clinic. This clinic had the highest diagnostic yield for neoplasia prior to the introduction of the Bowel Screening Programme and data were available for a similar number of patients.

Results

In the prevalent screening round, 378 colonoscopies were performed. Cancer was detected in 32 and polyps in 106. In all, 13 patients included in the list had both polyps and cancer. Overall, 125 patients had abnormal colonoscopies resulting in a 33% detection rate for neoplasia.

Of the 378 individuals, 198 (52%) were symptomatic and 180 (48%) were asymptomatic. The symptoms recorded were as follows: 126 had overt rectal bleeding, 79 had altered bowel habit, 24 had pain, 12 had weight loss, 4 had anaemia and 43 had other GI symptoms. Of these 198 symptomatic individuals, 129 had one symptom and 69 had two or more symptoms.

Table 1 compares the diagnostic rates in symptomatic and asymptomatic screened patients.

OPEN IN VIEWER

Table 1.

A comparison of detection rates for neoplasia between asymptomatic and symptomatic screened patients.

Asymptomatic	Symptomatic
Number	180	198
Adenoma	43 (24%)	50 (25%)
Cancer	18 (10%)	14 (7%)
PPV neoplasia	34%	32%

Looking only at the cancer cases, 18 were asymptomatic (56%) and 14 were symptomatic (44%). Of the 93 polyp cases , 43 (46%) were asymptomatic and 50 (54%) were symptomatic. Of the 253 cases that had no neoplasia, 119 (47%) were asymptomatic and 134 (53%) were symptomatic.

No statistical relationship was found between individual symptoms and diagnosis, combinations of symptoms and diagnosis or overall numbers of symptoms and diagnosis. Figure 1 shows the relatively even distribution of symptoms across the groups. OPEN IN VIEWER

By comparing with the screened patients, data of 604 patients attending a one-stop colorectal clinic at Crosshouse Hospital over 2 years were available. These patients made up the symptomatic referral group. In all, 181 (30%) patients failed to attend; therefore, diagnostic data were available only for 423 patients. In all, 169 had colonoscopy and 255 had flexible sigmoidoscopy. The findings were normal in 105 (25%), benign polyps in 67 (16%) and cancer in 22 (5%) giving a total neoplasia rate of 89 (21%). Table 2 compares the diagnostic rates in the FOB screened individuals to the symptomatic patients. There was a significantly greater diagnostic yield for colorectal neoplasia in the FOB screened individuals (p < 0.001, chi-square test).

OPEN IN VIEWER

Table 2.

A comparison of diagnostic rates in asymptomatic vs. symptomatic patients.

Screen detected	One-stop clinic
Number	378	423
Adenoma	93 (25%)	67 (16%)
Cancer	32 (8%)	22 (5%)
PPV neoplasia	33%	21%

Discussion

Most individuals attending for colonoscopy as part of the Scottish Bowel Screening Programme had symptoms. We are not screening a truly asymptomatic population and this makes bowel screening different from other population-based screening programmes in the United Kingdom. This is probably a reflection of the high prevalence of GI symptoms in the general population. A total of 52% of the individuals in this study had symptoms compared with 78% of the individuals in the Scottish pilot study in Tayside. ⁵ There may be regional variation as Ayrshire and Arran is a predominantly rural population, compared to the mixed rural and urban population of Tayside.

Despite 33% of our study population having neoplasia no symptom or group of symptoms was predictive of neoplasia. This mirrors the findings of the Scottish Pilot Study ⁵ and showing that no individual symptom or symptom complex can predict neoplasia in an FOB screened population. Certain symptoms or groups of symptoms may occur more frequently in individuals with cancer or large polyps^7,8 but there is no evidence that they can accurately predict neoplasia. Only one study on bowel cancer screened population had suggested that symptoms may be significant. ⁷ This study utilised screening endoscopy rather than FOB testing in patients aged 55–64 and rectal bleeding was reported more than twice as frequently in patients diagnosed with cancer. The risk of colorectal cancer was increased further if patients had bleeding plus altered bowel habit. However, despite these findings, most patients with neoplastic lesions in this study ⁷ were asymptomatic, therefore, it cannot be used to guide or prioritise individuals for screening.

If symptoms are not useful in diagnosing cancer in a screened population with a high prevalence of neoplasia, it is perhaps not surprising that efforts to streamline symptom-based referral to detect early bowel cancer has been disappointing. In this study, we have compared urgent symptomatic referred patients with screened individuals and shown that screening has the highest PPV for neoplasia.

Conclusion

Half of the FOB screen-positive individuals in the Bowel Screening Programme have bowel symptoms. Bowel symptoms in these FOB screen-positive individuals did not correlate with an increased rate of neoplasia. The PPV for neoplasia is higher both in symptomatic and asymptomatic screen-positive patients than in conventional symptomatic referrals. FOB bowel screening is an effective method for diagnosing neoplasia (cancer and adenomas), which may deliver an increased rate of early cancer detection and ultimately a reduced incidence of colorectal cancer by the identification and removal of premalignant adenomas.