Getting the best out of faecal immunochemical tests and faecal calprotectin

Faecal calprotectin (FC) and the faecal immunochemical test for haemoglobin (FIT) are recommended for use in primary care where colorectal cancer (CRC) is not suspected.^{1 2} We are unclear how best to use these two biomarkers in younger patients where lower gastrointestinal symptoms are extremely common. Distinguishing irritable bowel syndrome from inflammatory bowel disease (IBD) from CRC on clinical grounds is often an uncertain exercise. Currently, there is no direct comparative evidence base on which to identify the optimal use of each biomarker. We undertook an illustrative comparison of the diagnostic accuracy of FC in a patient dataset at ‘low risk’ of CRC stratified by age and symptoms, and of FIT, from three existing published studies that included both CRC and IBD in their outcomes.^3–11 From an existing dataset of patients using the York Faecal Calprotectin Care Pathway (YFCCP), we identified 1919 patients fulfilling National Institute for Health and Care Excellence (NICE) DG30 criteria. Patients were stratified based on age, symptomatology and diagnosis, either (1) CRC or (2) a composite of CRC, significant adenomatous polyps (≥10 mm or high-grade dysplasia or ≥5 subcentimetre polyps) and IBD (organic enteric disease).⁸ Fishers’ exact test was used to compare the data for FC of ≥100 µg/g faeces against the previously published data using FIT≥10 µg Hb/g faeces.^9–11

Meaningful patient numbers for age ranges could be obtained only when evaluating those with change of bowel habit, although similar outcomes were obtained using all patients fulfilling DG30 criteria. The low prevalence of 0.6% for CRC meant that no useful data could be obtained when looking at CRC in isolation. When organic enteric disease is used (prevalence of 5.9%), the diagnostic performance of FC is improved compared with FIT. The sensitivity and specificity of FC are 87.5 (77.1–93.8) and 88.6 (86.6–90.4), respectively. When compared with FIT from McDonald et al (sensitivity 57.0 (45.8–67.6) and specificity 99.0 (96.3–99.9)), FC is more sensitive but less specific (p=0.002 and p<0.001, respectively). FC is significantly better than FIT in Godber et al (sensitivity 71.1 (55.5–83.2), p=0.05, and specificity 79.5 (75.4–83.1), p<0.001) and in Mowat et al (sensitivity 68.6 (58.7–77.5), p=0.004, and specificity 83.6 (80.6–86.4), p<0.001).

These differences relate to the improved performance of FC in patients in <50 years, and when applied to the York Health Economics Consortium analytical decision tree developed for the YFCCP evaluation, an estimate of cost benefit can be generated (tables 1 and 2).

Table 1.

Sensitivity and specificity of FC≥100 µg/g faeces compared with FIT≥10 µg Hb/g faeces for organic enteric disease in those with change of bowel habit using McDonald et al, Godber et al and Mowat et al as the comparators

OED	Disease prevalence (%)	Sensitivity % CI)	P value	Specificity % (CI)	P value	NPV % (CI)	PPV % (CI)
FIT≥10µ/g comparator: McDonald et al*	30.7	57.0 (45.8 to 67.6)		99.0 (96.3 to 99.9)		83.8 (78.5 to 88.0)	96.1 (85.4 to 99.4)
FIT≥10 µg/g comparator: Godber et al*	9.3	71.1 (55.5 to 83.2)		79.5 (75.4 to 83.1)		96.4 (94.0 to 97.9)	26.2 (18.9 to 35.1)
FIT≥10 µg/g comparator: Mowat et al*	13.6	68.6 (58.7 to 77.5)		83.6 (80.6 to 86.4)		94.4 (92.2 to 96.0)	39.8 (32.6 to 47.4)
FC≥100 µg/g
50–59 years	5.8	65.0 (41.0 to 84.0)	1.0	85.0 (80.6 to 88.6)	0.11	97.5 (94.8 to 98.9)	21.0 (12.1 to 33.5)
40–49 years	5.5	88.9 (63.9 to 98.1)	0.81	89.4 (85.3 to 92.5)	0.001	99.3 (97.2 to 99.9)	32.7 (20.4 to 47.7)
30–39 years	3.6	100 (62.9 to 100)	0.051	91.7 (87.2 to 94.7)	<0.001	100 (97.9 to 100)	31.0 (16.0 to 51.0)
18–29 years	8.7	100 (83.4 to 100)	<0.001	89.3 (84.7 to 92.6)	0.002	100 (98.0 to 100)	47.2 (33.5 to 61.2)

FC, faecal calprotectin; FIT, faecal immunochemical test for haemoglobin; NPV, negative predictive value; OED, organic enteric disease, a composite of colorectal cancer, significant adenomatous polyps and inflammatory bowel disease; PPV, positive predictive value.

Table 2.

YHEC modelled outcomes of FIT compared with FC based on sensitivity and specificity per thousand patients using McDonald et al, Godber et al and Mowat et al as the comparators

Faecal biomarker	Total costs	Correctly diagnosed non-OED	Correctly diagnosed OED	Unnecessary colonoscopies	Clinician attendance	Incremental costs
FIT≥10 µg/g comparator: McDonald et al*	£258 179	931	32	9	139	−£70 908
FIT≥10 µg/g comparator: Godber et al*	£360 117	752	43	188	304	£30 938
FIT≥10 µg/g comparator: Mowat et al*	£338 517	790	41	150	270	£9338
FC ≥100 µg/g	£329 179	812	49	125	270	Referent

FC, faecal calprotectin; FIT, faecal immunochemical test for haemoglobin; OED, organic enteric disease, a composite of colorectal cancer, significant adenomatous polyps and inflammatory bowel disease.

FIT promises to be transformative in the way primary care risk assesses patients presenting with lower gastrointestinal symptoms. However, in England, there is little guidance on how to apply NICE DG30. When used in a population where the prevalence of the target disease is low and the true specificity is uncertain, there is a risk of creating large numbers of false-positive patients that will be upstaged to ‘2-week wait’ referrals, threatening to overwhelm already stretched secondary care diagnostic services. Since FIT is unsupported by a lower age limit and so is applicable within the same clinical context as FC, it is particularly important to define how patients and the health economy can get the best out of these diagnostic biomarker technologies.

While an evidence base is awaited, a pragmatic approach might be considered based on the fact that 94% of CRC presents in patients aged ≥50 years. When looking at organic enteric disease in this ‘low-risk’ cohort aged <50 years, FC performs similarly or better than FIT when compared with three previously published studies.

As FIT is introduced in support of DG30, its clinical effectiveness and cost-effectiveness should be evaluated across all age groups, recognising the benefits of early diagnosis for all patients with organic enteric disease, and its performance should be benchmarked against existing FC pathways.