Background: Screening for colorectal cancer by use of guaiac-based faecal occult blood tests (FOBT) reduces disease-specific mortality. However, due to imperfect specificity, about half of individuals positive for guaiac FOBT are negative for neoplasia on colonoscopy. We aimed to assess whether the testing of individuals positive for guaiac FOBT in a screening programme for colorectal cancer by use of a sensitive immunochemical FOBT could select more appropriately those who should receive colonoscopy.
Methods: We invited individuals who were guaiac FOBT positive in the second screening round of a pilot study in Scotland, UK, to give two samples, each from separate stools, for immunochemical FOBT while awaiting colonoscopy. Results were classed as: both samples negative (N/N), one sample negative and one positive (N/P), and both samples positive (P/P); data were assessed for sampling bias. We compared immunochemical findings with those from colonoscopy using odds ratios of positive samples (P/P) versus negative (N/N and N/P). Sensitivity, specificity, and positive and negative likelihood ratios for cancer, and for cancer and high-risk adenomatous polyps were also calculated.
Findings: 1486 participants were invited and 801 (54%) sets of duplicate samples were returned. We found no evidence of sampling bias with regard to sex, age, or degree of positivity on guaiac FOBT. Of 800 sets returned with consent and analysed, 173 (22%) were N/N, 129 (16%) were N/P, and 498 (62%) were P/P. Chi2 test showed a highly significant positive correlation between degree of positivity on guaiac FOBT and on immunochemical FOBT (p<0.003). 795 individuals had data for colonoscopy: one (<1%) of 171 N/N participants and one (<1%) of 127 N/P participants had colorectal cancer, compared with 38 (8%) of 497 P/P participants; adenomatous polyps were found in 28 (16%) N/N individuals, 24 (19%) N/P individuals, and 193 (39%) P/P individuals. Normal colonoscopy was less common in the P/P group (85 [17%]) than in the N/N (67 [39%]) and N/P (49 [39%]) groups. The odds ratio for P/P being associated with cancer was 7.57 (95% CI 1.84-31.4) and with high-risk adenomatous polyps was 3.11 (1.86-5.18). Sensitivity of a P/P result for cancer was 95.0% (81.8-99.1), and for cancer and high-risk adenomatous polyps was 90.1% (84.4-94.0); specificity was 39.5% (36.0-43.1) and 47.8% (43.9-51.8), respectively.
Interpretation: Immunochemical FOBT for individuals with positive guaiac FOBT could decrease substantially the number of false positives in a screening programme for colorectal cancer.