The natural clinical course of inflammatory bowel disease (IBD) is characterized by episodes of relapse and remission. The main treatment goal in IBD is to induce and maintain remission by effective suppression of gut inflammation. Despite successful medical treatment, subclinical inflammation may still exist in the bowel, leading to significant risk of relapse.1–3 Thus, detection of such inflammation and identifying patients at high risk of relapse are of great clinical significance, as they may have therapeutic implications. Currently, serological biomarkers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are widely used as noninvasive parameters for IBD. However, they have insufficient sensitivity and specificity for intestinal inflammation.4–6
Fecal calprotectin (FC) is an excellent marker of intestinal inflammation, as it reflects the migration of neutrophils through the inflamed bowel wall to the mucosa.7 FC appears to have good diagnostic precision in distinguishing IBD from non-IBD patients.8 Moreover, FC concentrations correlate well with histological and endoscopic inflammation.9, 10 Recently, extending the use of FC to predict relapse of ulcerative colitis (UC) and Crohn's disease (CD) patients has attracted significant interest and has been the focus of a growing number of publications.11–24 However, the results reported are variable and controversial. The objective of this study was to assess the overall capacity of FC in predicting relapse of IBD with a meta-analysis.
- Top of page
- MATERIALS AND METHODS
The initial search yielded 270 articles, of which 11 studies were eligible for analysis.12–17, 20–24 Of these publications, two studies20, 21 were excluded, as they were related to clinical course and treatment outcome instead of predicting IBD relapse; two studies22, 23 were excluded because FC concentrations used for prediction were not measured at remission; and one study conducted in a pediatric population was excluded.24 Therefore, six studies12–17 were included in the final analysis. The clinical characteristics and QUADAS scores of these studies are listed in Table 1.
Table 1. Study Characteristics of Included Studies
|Study/Year||Age (year)||Patients,* No. (IBD/ UC/CD)||FC Assay||Cutoff (IBD/UC/CD)||Standard of Relapse (UC/CD)||Test Results (Total/UC/CD)||Quality Score|
|Kallel et al, 2010||15-66||−/−/53||Calprest||340 μg/g||-/CDAI>150 or ΔCDAI>100||−/−/8||−/−/4||−/−/2||−/−/39||10|
|García-Sánchez et al, 2010||27-54||135/69/66||Calprest||150/120/200 (μg/g)||TW ≥11/ CDAI≥150||29/17/14||31/18/17||10/4/4||65/30/31||11|
|Gisbert et al, 2009||30-56||163/74/89||Calprest||167/164/169 (μg/g)||TW >11/ CDAI>150||18/9/9||34/16/18||8/4/4||103/45/58||10|
|D'Inca et al, 2008||15-80||162/97/65||Calprest||130/130/130 (mg/kg)||ET>4/ CDAI>150 or|| || || || || |
|ΔCDAI>50||39/26/13||35/18/17||18/11/7||70/42/28||12|| || || || || |
|Costa et al, 2005||24-54||79/41/38||Calprest||150/150/150 (μg/g)||UCAI>4/ CDAI>150||30/17/13||17/4/13||4/2/2||28/18/10||10|
|Tibble et al, 2000||16-77||80/37/43||Roseth||50 mg/L||HBI>4+ ΔHBI>2/ CDAI>150 + ΔCDAI>100||40/nk/nk||6/nk/nk||4/nk/nk||30/nk/nk||12|
Quality of Reporting and Study Characteristics
All the included six studies were prospective studies with enrollment of consecutive IBD patients. One study15 solely involved patients with CD; in the remaining five studies, both UC and CD patients were included. The criteria for remission and relapse were all based on clinical activity indices. FC levels were measured in clinical remission to predict relapse during subsequent 1-year follow-up in all six studies. In all, 672 IBD patients were included (UC = 318; CD = 354) in the six studies. Regarding FC assay, one study17 used an old method assay (Roseth) and the other five studies used the new quantitative enzyme linked immunoassay (Calprest). According to the literature, the results obtained by the old method were multiplied by a factor of 5 to be compared with the new method.8
Forest plots of the pooled sensitivity and specificity are shown in Figure 1. The sensitivity ranged from 0.68–0.91 (pooled sensitivity 0.78, 95% CI: 0.72–0.83), while specificity ranged from 0.62–0.91 (pooled specificity 0.73, 95% CI: 0.68–0.77). The PLR was 2.81 (95% CI: 2.09–3.78), NLR was 0.31 (95% CI: 0.2–0.47), and DOR was 10.31 (95% CI: 5.05–21.06). Chi-square values for sensitivity, specificity, PLR, NLR, and DOR were 11.75 (P = 0.038), 16.68, (P = 0.005), 12.95 (P = 0.024), 10.93, (P = 0.0529), and 13.82 (P = 0.016), respectively (Table 2). These indicated a significant heterogeneity across studies regarding sensitivity, PLR, NLR, and DOR, except for NLR.
Table 2. Pooled Results for Predictive Capacity of Fecal Calprotectin in IBD Relapse
| ||IBD||UC||CD||CD Involving Colona|
|Sensitivity (95%CI)||0.78 (0.72–0.83)||0.77 (0.67–0.85)||0.75 (0.64–0.84)||0.76 (0.59–0.88)|
|Heterogeneityb (P)||11.75 (0.04)||3.47 (0.32)||2.67 (0.61)||0.51 (0.92)|
|Specificity (95%CI)||0.73 (0.68–0.77)||0.71 (0.64–0.77)||0.71 (0.64–0.76)||0.77 (0.69–0.83)|
|Heterogeneity (P)||16.68 (<0.001)||3.22 (0.36)||21.14 (<0.001)||11.57 (0.01)|
|PLR (95%CI)||2.81 (2.09–3.78)||2.47 (1.92–3.19)||2.37 (1.56–3.61)||3.26 (1.78–5.95)|
|Heterogeneity (P)||12.95 (0.024)||2.94 (0.40)||12.51 (0.01)||8.28 (0.04)|
|NLR (95%CI)||0.31 (0.2–0.47)||0.36 (0.24–0.53)||0.41 (0.27–0.61)||0.34 (0.19–0.60)|
|Heterogeneity (P)||10.93 (0.05)||3.08 (0.38)||2.34 (0.67)||1.54 (0.67)|
|DOR (95%CI)||10.31 (5.05–21.06)||7.70 (3.93–15.09)||6.5 (3.15–13.43)||10.39 (3.75–28.81)|
|Heterogeneity (P)||13.82 (0.02)||3.60 (0.31)||5.47 (0.24)||4.21 (0.24)|
|AUC (SEM)||0.83 (0.05)||0.78 (0.04)||0.79 (0.05)||0.81 (0.05)|
As shown in Figure 2, the sROC curve showing sensitivity versus 1-specificity from individual studies is not positioned near the desirable upper left corner. The maximum joint sensitivity and specificity was 0.77 (SEM 0.04, standard error of the mean), with AUC of 0.83 (SEM 0.05).
Figure 2. Summary ROC curves for FC in predicting relapse of IBD. The size of the solid circle represents sample size of each study included in the meta-analysis. The regression sROC curves summarize the overall diagnostic accuracy.
Download figure to PowerPoint
The capacity of FC to predict relapse of UC and CD are shown in Table 2. It could not be concluded that the overall accuracy of FC for predicting relapse was inferior or superior in UC than CD. In the subgroup of 196 colonic involved CD patients (including colonic and ileocolonic CD), the pooled sensitivity, specificity, PLR, NLR, and DOR were 0.76, 0.77, 3.26, 0.34, and 10.39, respectively. The AUC was 0.81 (SEM 0.05). These data indicated that the overall predictive capacity appears stronger in patients with colonic and ileocolonic CD. Because of insufficiency of available data about CD cases involving ileum, the predictive capacity of FC was not evaluated in isolated small bowel CD.
As shown in Table 1, three publications had higher QUADAS scores (≥11). Due to the limited number of included studies, QUADAS scores were not used to perform meta-regression analysis to assess the effect of study quality on relative DOR of FC in predicting relapse.
The Egger test showed that the potential publication bias was significant (P < 0.05). This result indicated a potential for publication bias. Due to the limited number of included studies in the meta-analysis, funnel plots were not assessed.
- Top of page
- MATERIALS AND METHODS
Calprotectin is remarkably stable in stools for up to 7 days at room temperature, enabling sample collection at home even in cases of distant hospitals or reference centers.27 A reliable measurement can be performed on 1–2 g of stools by commercially available enzyme-linked immunoassay.28, 29 These facts allow FC to be a simple and noninvasive test for IBD patients. Two meta-analyses have been published in recent years on the use of testing FC levels in IBD.8, 27 Conclusions from these meta-analyses suggest that FC is a useful screening tool for distinguishing IBD from non-IBD diagnoses, and may thereby reduce the number of unnecessary endoscopic procedures. To the authors' knowledge, this is the first study employing meta-analysis to evaluate the diagnostic value of FC in predicting relapse of IBD, although some comments and reviews have been published on the subject.11, 18, 19
The results of the present meta-analysis demonstrate that the pooled sensitivity and specificity of FC level were 0.78 and 0.73, respectively, for patients with quiescent IBD in predicting relapse. The sROC curve showed an optimum joint sensitivity and specificity of 0.77, with AUC of 0.83. These indicate that overall predictive capacity of FC was good, although not as high as expected. DOR is the ratio of the odds of positive test results between the diseased and nondiseased groups. Higher DOR value represents better discriminatory performance.26 A DOR of 10.31 in our meta-analysis indicates that FC is helpful in predicting relapse of IBD. Compared with sROC and DOR, likelihood ratios including PLR and NLR are considered more clinically useful. PLR above 5 and NLR below 0.2 have been noted as providing strong diagnostic evidence.30 In our meta-analysis, the pooled PLR was 2.81, suggesting that relapse patients have about a 3-fold higher chance of higher FC level compared with nonrelapse patients. As for NLR, the value was 0.31, suggesting that a probability of relapse for the patient is 31% if the FC level was not above the cutoff value.
As UC and CD are diseases with different inflammatory patterns, the value of FC in predicting relapse may be different for these two diseases. Based on the studies to date about FC in predicting IBD relapse, FC seems to be a relatively sensitive and specific marker of relapse in UC. Nevertheless, the data on CD are conflicting at this point. In effect, Costa et al13 and D'Inca et al14 did not find a significant difference in median baseline FC level between relapse and nonrelapse CD groups, while the other four studies found that the median FC level was significantly higher in the relapse group.12, 15–17 This may partly result in conflicting data on specificity in different studies. In the Costa et al study,13 the specificity for predicting relapse in CD was only 43%, compared with 90.7% in the Kallel et al study.15 Taken together, the evidence suggests that FC is a stronger predictor of relapse in UC than in CD.13 To evaluate possible explanations for this controversy, the overall predictive values of FC for UC and CD relapse were compared in the present meta-analysis. As shown in Table 2, the sensitivity, PLR, and DOR for UC were slightly higher than those for CD. In contrast, the NLR and AUC were lower in UC. A similar result was observed for specificity. Therefore, it could not be concluded that the overall accuracy of FC for predicting relapse was inferior or superior in UC than CD.
It has been suggested that the phenotype of CD could influence the predictive value of FC. Some studies showed that FC predicts relapse, particularly in patients with colonic and ileocolonic CD, but not in those with ileum disease.14, 16 In our meta-analysis, a total of 196 CD cases involving the colon (including colonic and ileocolonic CD) were evaluated, whereas the cases involving ileum were not assessed because of insufficiency of available data in the included studies. As shown in Table 2, the overall predictive capacity appears stronger in patients with colonic and ileocolonic CD.
This meta-analysis has certain limitations. First, the pooling of diagnostic accuracy data inevitably contributed to sources of bias,31 which were revealed in the significant amount of statistical heterogeneity across studies. As studies that report positive results are more likely to be published, publication bias may also be introduced by inflation of diagnostic accuracy estimates. Second, the criteria of remission and relapse for all included studies were based on subjective clinical activity indices. Regarding CD, clinical relapse was defined using the Crohn's Disease Activity Index (CDAI) in each study. However, for UC variable activity indices were used, including the modified Truelove Witts score, Edwards and Truelove score, ulcerative colitis activity index score, and Harvey–Bradshaw index. The difference in these clinical activity indices may result in the heterogeneity across studies. Further prospective studies investigating FC in prediction of endoscopic relapse are required. Finally, due to lack of required data reported in the original publications, it was not possible to evaluate the effect of other CD phenotypes such as ileal disease on the predictive capacity of FC.
In conclusion, the current evidence suggests a potential role for FC in predicting relapse of quiescent IBD. Although the overall test performance including sensitivity and specificity is not as high as expected, measuring FC levels at clinical remission can be used to predict IBD relapse due to its simplicity and noninvasiveness.
- Top of page
- MATERIALS AND METHODS
We thank Dr. Javier P. Gisbert (Hospital Universitario de la Princesa, Madrid, Spain), Dr. Valle García-Sánchez (Hospital Universitario Reina Sofía, Córdoba, Spain), Dr. F. Costa (University of Pisa, Pisa, Italy), and Dr. D'incà (University of Padua, Padua, Italy) for providing additional information about their studies.