A longitudinal study of fecal calprotectin and the development of inflammatory bowel disease in ankylosing spondylitis

All patients fulfilling the modified New York criteria for AS who were registered at the rheumatology clinics at Sahlgrenska University Hospital and the hospitals of Borås and Alingsås were invited to participate in the study in 2009, hereafter called baseline [24, 25]. Exclusion criteria were diagnosed IBD, psoriasis, dementia, pregnancy, and difficulties in understanding the Swedish language. Altogether, 204 patients were included at baseline and 164 (80%) patients were reexamined at the 5-year follow-up in 2014. All patients gave their written informed consent. The study was approved by the regional ethics committee in Gothenburg and was carried out in accordance with the Helsinki declaration. A flowchart of the enrollment of the patients and the study procedures is shown in Fig. 1. The patients underwent the same assessments at both visits, including physical examinations and responding to questionnaires. Blood and stool samples were collected. The medical records of all the included patients at baseline were checked for ileocolonoscopies, episodes of gut inflammation, or diagnoses of IBD.

The Ankylosing Spondylitis Disease Activity Score based on C-reactive protein (ASDAS_CRP), Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), Bath Ankylosing Spondylitis Patient Global Score, and Bath Ankylosing Spondylitis Functional Index (BASFI) were used to assess disease activity and physical function. Back mobility was measured for the calculation of Bath Ankylosing Spondylitis Metrology Index (BASMI) [26].

Stool samples were collected by the patients and sent to the laboratory, where the samples were immediately frozen. After thawing of the samples, fecal calprotectin was analyzed using an enzyme-linked immunosorbent assay (ELISA) kit (Bühlmann Laboratories AG, Schönenbuch, Switzerland). The same ELISA kit and procedures were used at baseline and at 5-year follow-up. In the general population, fecal calprotectin is <50 mg/kg, according to the manufacturer of the kit.

Blood samples were analyzed for hemoglobin, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) using standard laboratory techniques. Serum calprotectin was analyzed at baseline by ELISA (PhiCal; Immundiagnostik AG, Bensheim, Germany) following the manufacturer’s instructions. According to the manufacturer, the normal range for serum calprotectin is 500–3000 ng/ml.

At baseline, all patients with fecal calprotectin ≥500 mg/kg were asked to send in a new stool sample. NSAID use was not stopped. If their fecal calprotectin remained ≥500 mg/kg, the patients were advised to undergo a clinical routine ileocolonoscopy (Fig. 1). At 5-year follow-up, all patients with fecal calprotectin ≥200 mg/kg were asked to send in another stool sample after pausing NSAID medication for 3 weeks. If their fecal calprotectin remained >200 mg/kg when retested, the patients were referred for ileocolonoscopy. Before the ileocolonoscopy, NSAIDs were again paused for 3 weeks. At 5-year follow-up, all ileocolonoscopies were performed by one endoscopist (HS) and followed the same procedure.

Data from all patients who participated at baseline (n = 204) were analyzed with the aim of determining the 5-year incidence of IBD and the predictors thereof. For all other data analyses, only the patients who participated both at baseline and at 5-year follow-up (n = 164) were included.

Statistical analyses were performed using IBM SPSS Statistics 23 software (IBM, Armonk, NY, USA). Descriptive statistics are presented as median and IQR. In comparisons between groups, the Mann-Whitney U test was used for continuous variables, and the chi-square test or Fisher’s exact test was used for categorical variables. The Wilcoxon signed-rank test was used for repeated measurements. Correlations were calculated using Spearman’s correlation (r _s). Owing to its skewed distribution, fecal calprotectin was log-transformed and used as an outcome in linear regression with a stepwise model. All tests were two-tailed, and p ≤ 0.05 was considered statistically significant.

In total, 164 patients, comprising 45% women and 55% men, completed the study and were examined at the two study visits (baseline and 5-year follow-up). The characteristics of the study population are provided in Table 1.

Fecal calprotectin was elevated (>50 mg/kg) in 70.7% (116 of 164) of the patients at baseline and in 63.4% (104 of 164) of the patients at 5-year follow-up. The levels of fecal calprotectin did not differ between baseline and follow-up (p = 0.660) (Table 1).

The change in fecal calprotectin between baseline and the 5-year follow-up was ≤50 mg/ml in 48% of the patients, 51–100 mg/kg in 16%, and 101–200 mg/kg in 16%. The median change between baseline and 5-year follow-up for calprotectin was 1.5 (IQR −52.7 to 52.7) mg/kg. The intraindividual correlation coefficient for fecal calprotectin at baseline and follow-up was r _S = 0.512 (p < 0.001) (Fig. 2a).

At both visits, fecal calprotectin was positively and significantly correlated with parameters reflecting higher disease activity and poorer function at the same visit, such as CRP, ESR, ASDAS_CRP, BASDAI, BASFI, and BASMI. In addition, fecal calprotectin at baseline was positively associated with higher levels of CRP, ESR, ASDAS_CRP, and BASFI at 5-year follow-up (Table 2).

The patients who were smoking at baseline (9%) had, in comparison with nonsmokers, significantly higher fecal calprotectin at 5-year follow-up (median [IQR] 107 [43–402] vs. 85 [45–221] mg/kg, p = 0.026). Baseline serum calprotectin was also positively correlated with fecal calprotectin, both at baseline (r_S = 0.252, p = 0.001) and at 5-year follow-up (r _S = 0.170, p = 0.030).

The medications used at baseline and 5-year follow-up are listed in Table 1. Fecal calprotectin was significantly higher in NSAID users than in nonusers at both study visits (baseline median [IQR] 105 [45–262] vs. 68 [44–93] mg/kg, p = 0.032; 5-year follow-up 93 [44–215] vs. 40 [23–120] mg/kg, p = 0.011) (Fig. 3a). The users of anti-tumor necrosis factor (anti-TNF) antibodies (infliximab, adalimumab, golimumab) had significantly lower fecal calprotectin than patients without TNF blocker therapy at both visits (baseline 43 [17–84] vs. 99 [54–256] mg/kg, p = 0.004; 5-year follow-up 48 [26–135] vs. 89 [40–210] mg/kg, p = 0.022), whereas no such difference was found between users of TNF receptor fusion proteins (etanercept) and nonusers of TNF blockers. At 5-year follow-up, the patients treated with TNF receptor fusion proteins had significantly higher fecal calprotectin than the patients treated with anti-TNF antibodies (200 [93–555] vs. 48 [26–135] mg/kg; p = 0.016) (Fig. 3b). Comedication with methotrexate and a TNF blocker was common. Methotrexate was mostly taken in low doses (5–15 mg weekly). At 5-year follow-up, the use of methotrexate among patients on anti-TNF antibodies was not significantly different from the use in patients taking etanercept (18 of 33 [54.5%] vs. 2 of 6 [33%]; p = 0.407 by Fisher’s exact test).

Although the patients treated with TNF blockers had a significantly lower frequency of NSAID intake than the patients without TNF blockers, the significant differences in fecal calprotectin between users and nonusers of disease-modifying antirheumatic drugs and/or TNF blockers remained after adjustment for NSAID use. Linear regression analysis was performed with log-transformed fecal calprotectin at 5-year follow-up as an outcome measure, and the following baseline values were used as covariates: sex, age, serum calprotectin, fecal calprotectin, CRP, ASDAS_CRP, BASFI, smoking, and use of NSAIDs and anti-TNF antibodies/TNF receptor fusion proteins. Only fecal calprotectin at baseline (β = 0.001, SE = 0.000, 95% CI 0.0008–0.0015, p < 0.001) and smoking at baseline (B = 0.241, SE = 0.121, 95% CI 0.002–0.479, p = 0.048) remained independently associated with higher fecal calprotectin at 5-year follow-up (R ² = 0.247).

A high frequency of gastrointestinal symptoms (diarrhea, presence of blood or mucus in stools, obstipation, abdominal pain, tenesmus, reflux symptoms, and epigastric pain) was reported by the participants both at baseline and at 5-year follow-up (Table 1). There were no significant associations between the levels of fecal calprotectin and gastrointestinal symptoms at any of the time points, however.

At 5-year follow-up, the patients who had fecal calprotectin ≥200 mg/kg (n = 40 [24%]) were asked to send in a new stool sample after pausing NSAID use for 3 weeks, which a total of 33 patients did. After the 3-week NSAID pause, fecal calprotectin dropped by 116 (−20 to 289) mg/kg from a range of 410 (295–510) mg/kg to 290 (120–480) mg/kg (p = 0.005) (Fig. 2b).

Ileocolonoscopies were performed in 8 patients at baseline and in 16 patients at 5-year follow-up (Fig. 1). The findings from the ileocolonoscopies are summarized in Table 3. Altogether, 10 (41.7%) of the 24 patients who underwent ileocolonoscopy had inflammatory changes in the gut biopsies. In addition, five patients (20.8%) had only macroscopic changes.

At 5-year follow-up, 3 patients (1.5%) were diagnosed with CD (corresponding to an annual incidence of 294 new cases per 100,000 person-years), and 1 patient (0.5%) was diagnosed with lymphocytic colitis. No cases of UC were found. Additionally, five patients (2.5%) demonstrated characteristics of subclinical CD with aphthous ulcerations and/or chronic inflammation in the distal ileum, and another six patients (2.9%) had ulcerations and/or chronic inflammation in the colon. In total, 15 patients (7.4%) demonstrated characteristics of inflammation in the large or small intestine during the 5-year follow-up period.

At baseline, the patients who developed CD had, in comparison with the other patients, higher fecal calprotectin (570 [271–570] mg/kg vs. 85 [43–230]; p = 0.014) and more often reported having diarrhea containing mucus (3 of 3 vs. 30 of 201; p = 0.004). No significant associations were found between the development of CD and the other baseline parameters. A receiver operating characteristic (ROC) curve was created with baseline fecal calprotectin as the test variable and development of CD as the categorical state variable. The area under the curve was 0.913 (95% CI 0.805–1.000; p = 0.014). Using a threshold for baseline fecal calprotectin of 266 mg/kg, based on this ROC curve, the sensitivity for development of CD was 100% and the specificity was 78.7%.