How well can stool tests detect microscopic inflammation in people with ulcerative colitis?

Details

Histological remission—remission assessed in tissue samples—has emerged as the optimal treatment outcome in ulcerative colitis (UC). This study explored how well the noninvasive stool biomarkers calprotectin (FC) and lactoferrin (FL) performed compared with two histological scoring systems, the Nancy Index (NI) and Riley Index (RI).

This was a retrospective diagnostic accuracy study using secondary analysis of patient data from 2002–2017, extracted from medical registries of clinics in Essen-Mitte, Germany. Patients with UC underwent colonoscopy, with biopsies from the rectum and sigmoid scored by 2 experienced pathologists using NI and RI. Patients provided a stool sample within 7 days before or after colonoscopy. Recommended cutoffs of FC >50 μg/g and FL ≥7.25 μg/g were evaluated against the reference standard of NI ≥2, using specificity, sensitivity, positive predictive value, negative predictive value, and accuracy (effectiveness).

The study included 226 patients with UC, with a mean age of 45.2 years (SD 13.3). The two histological indices were highly correlated (r = 0.980, P < 0.001). Stool biomarkers were moderately correlated with NI (FC: r = 0.383; FL: r = 0.420; both P < 0.001) and RI (FC: r = 0.395; FL: r = 0.424; both P < 0.001).

Biomarker concentrations were higher in patients with active histological disease (NI ≥2) than in those with inactive disease (NI ≤1). Median [interquartile range] concentrations in μg/g were FC 69.72 [20.07–254.38] and FL 18.59 [6.06–44.42] for active disease, compared with FC 12.35 [3.89–32.16] and FL 3.14 [0.75–11.05] for inactive disease (reported z = −6.60, P < 0.001).

Concentrations differed significantly across NI grades 0–4 (FC: H4 = 45.2; FL: H4 = 47.5; both P < 0.001). Grade 0 had lower median concentrations than grade 3 (FC: 10.94 vs 72.22; FL: 2.30 vs 29.10; both P < 0.001) and grade 4 (FC: 10.94 vs 67.00; FL: 2.30 vs 27.64; both P < 0.001). Grade 0 also had lower FC than grade 2 (10.94 vs 56.22, P = 0.001); this comparison was significant for FC only. Grade 1 had lower concentrations than grade 3 (FC: 17.49 vs 72.22; FL: 4.24 vs 29.10; both P ≤ 0.001) and grade 4 (FC: 17.49 vs 67.00; FL: 4.24 vs 27.64; both P < 0.001). These concentration comparisons are in μg/g.

Receiver operating characteristic (ROC) areas under the curve indicated moderate diagnostic accuracy: FC 0.76 (95% confidence interval 0.70–0.83) and FL 0.73 (95% confidence interval 0.66–0.80). Optimized cutoffs of FC ≥34.29 μg/g and FL ≥5.85 μg/g slightly improved accuracy compared with the manufacturers’ cutoffs: 69.9% vs 65.9% for FC and 71.7% vs 69.0% for FL.

The authors concluded that stool calprotectin and lactoferrin correlate with histological disease activity and distinguish patients in histological remission from those with evidence of moderate to severe activity. Their noninvasive nature and low cost support their use in clinical monitoring of patients with UC.