OUP user menu

Journal of Crohn's and Colitis: 10 (9)

Editor-in-Chief

Laurence J. Egan, Ireland

Associate Editors

Shomron Ben-Horin, IsraelSilvio Danese, ItalyPeter Lakatos, HungaryMiles Parkes, UKJesús Rivera-Nieves, USABritta Siegmund, GermanyGijs van den Brink, NLSéverine Vermeire, Belgium

6.585
5.586

Published on behalf of

High C-reactive protein in Crohn's disease patients predicts nonresponse to infliximab treatment

Fernando Magro, Eduardo Rodrigues-Pinto, João Santos-Antunes, Filipe Vilas-Boas, Susana Lopes, Amadeu Nunes, Claudia Camila-Dias, Guilherme Macedo
DOI: http://dx.doi.org/10.1016/j.crohns.2013.07.005 129-136 First published online: 1 February 2014

Abstract

Background Infliximab (IFX) is effective in treating Crohn's disease (CD) and C-reactive protein (CRP) is a useful biomarker in assessing inflammatory activity.

Aim Correlate CRP levels before beginning of IFX, at week 14 and CRP delta within the first year of IFX treatment.

Methods Retrospective study of CD patients undergoing treatment with IFX. Primary nonresponse (PNR) was defined as no symptomatic improvement and CRP persistently elevated; sustained response (SR) as symptomatic improvement for at least 1 year without therapeutic adjustment; response after therapeutic adjustment (RTA) as analytic and clinical response but requiring IFX dose/frequency adjustment or association with another drug.

Results Baseline CRP levels were higher in PNR compared with SR (26.2 mg/L vs 9.6 mg/L, p = 0.015) and RTA (26.2 mg/L vs 7.6 mg/L, p = 0.007). CRP levels greater than 15 mg/L at baseline predict PNR with 67% sensitivity and 65% specificity. Lower CRP levels at week 14 were more likely to predict SR relative to RTA (3.1 mg/L vs 7.6 mg/L p = 0.019) and PNR (3.1 mg/L vs 9.1 mg/L; p = 0.013). CRP levels greater than 4.6 mg/L at week 14 predict PNR with 67% sensitivity and 62% specificity. A higher CRP delta between beginning of treatment and week 14 is more likely to predict SR relative to RTA (5.2 mg/L vs 0.6 mg/L p = 0.027).

Conclusion CRP levels at week 14 were associated with SR in patients treated with IFX, independently of baseline CRP serum levels. High inflammatory burden at beginning of IFX treatment was correlated with a worse response.

Keywords
  • Crohn disease
  • Infliximab
  • C-reactive protein
  • Prognosis
  • Treatment outcome

1 Introduction

Inflammatory bowel disease (IBD) is an idiopathic, chronic process of the bowel associated with decreased quality of life and increased morbidity.1 It is believed to be the result of an imbalance between proinflammatory and anti-inflammatory mediators2 and tumor necrosis factor (TNF)-α is now known to play a pivotal role in the onset and perpetuation of the inflammation. Infliximab (IFX) is a chimeric IgG 1 monoclonal antibody with high specificity and affinity to TNF-α.38 Clinical benefit from IFX is reported in 60–70% of the patients in short and medium term,913 however, secondary loss of response is intrinsic to the use of therapeutic antibodies and up to 30–40% of patients who initially respond to IFX, lose response within the first year.1113

C-reactive protein (CRP), an acute phase reactant, is produced and released by hepatocytes in response to cytokine (interleukin-6, TNF-α and interleukin-1β) stimulation at the site of inflammation.14 CRP plays an important role as a noninvasive inflammatory marker in patients with IBD, however, there is remarkable heterogeneity in the CRP response between Crohn's disease (CD) and ulcerative colitis (UC). Whereas CD is associated with a strong CRP response, UC has only a modest to absent CRP response.1419 Besides that, in CD, a significant number of patients present with low CRP levels despite clinically active disease. So far, no clear cut-off values have been determined as predictive of remission/nonresponse/relapse. Different studies suggest that lower baseline CRP levels can predict nonresponse, as it may correspond to absence of inflammation. Increased CRP levels in CD have been associated with relapse. Once a wide range of CRP values are observed, with significant overlap between different degrees of severity, the dynamics of CRP evolution and the comparison of the CRP value with previous values in a given patient, might be more important than a particular cut off value for CRP.17,2026

Our aim is to correlate CRP levels before beginning of IFX, at week 14 and CRP delta between baseline and week 14 in patients that have primary nonresponse (PNR), sustained remission (SR) or response after therapeutic adjustment (RTA) within the first year of IFX treatment.

2 Material and methods

Retrospective study based on medical records from patients with CD undergoing treatment with IFX (followed at our institution) between January 2006 and February 2012. Diagnosis of CD was made by endoscopic, radiologic and histological criteria,27 and indication for IFX treatment was judged by senior IBD gastroenterologists taking into consideration clinical, biological, and endoscopic findings.

Our study was performed with 148 CD patients and their serum CRP was evaluated. Serial blood samples were taken before each IFX infusion, and CRP was measured as part of standard follow-up by using an enzyme-linked immunosorbent assay (Olympus CRPLatex Calibrator Normal Set®). A CRP level below 3 mg/L was considered normal. In total, there were available 1776 serial CRP measurements, with a median of 12 per patient. Week 14 was the time point selected to evaluate if the induction therapy was successful once it corresponds to the end of the induction and the beginning of maintenance treatment. All clinical data and CRP values were collected from the patients' records and were related to clinical outcome, namely relapse of disease activity and the need for therapy adjustment. Harvey–Bradshaw index (HBI) was used to assess patients' symptoms, being assessed by the senior IBD gastroenterologist at each clinic visit; those who scored 4 or less were considered to be in clinical remission and those who scored more to have clinical activity. All patients were submitted to colonoscopy in the three months previous to beginning of IFX treatment, with all patients having endoscopic activity, with ulcers. Other causes that could mimic disease activity, including infections (clostridium difficile, cytomegalovirus, bacterial, mycobacterium tuberculosis) were ruled out. Standard imaging with abdominal ultrasound or computed tomography was performed, whenever there was a suspicion of abscesses, perforating disease with phlegmone or any other complication. None of our patients had any of those complications. A protocol is followed before IFX beginning and during therapy, with patients starting IFX only after discussion between senior IBD gastroenterologists responsible for the IBD outpatient clinic. In all our patients, IFX was started due to the severity of luminal disease and not due to perianal disease or for prevention of post-operative recurrence.

All patients received a 5 mg/kg IFX infusion at baseline and then received subsequent infusions of 5 mg/kg at weeks two, six and then every 8 weeks. In case of loss of clinical response, optimization of anti-TNF treatment was always considered as primary intention (dose escalation to 10 mg/Kg or shortening infusion interval to every 6 weeks were made if patients remained symptomatic) and was a decision of the patient's gastroenterologist. Neither IFX trough levels nor IFX antibodies were available. Therapy adjustments were considered clinically successful if symptoms disappeared (HBI score ≤ 4) and if IFX could be continued thereafter without further adjustments (increase dose, shorten interval or addition of immunomodulators). PNR was defined as absence of symptomatic improvement and CRP persistently high (despite therapeutic adjustments); SR as symptomatic improvement for at least 1 year without need for any dose or interval adjustment or any other changes in therapy; RTA as analytic and clinical response after IFX dose modification and/or association of another drug (steroids, immunomodulators); short-term response (STR) as clinical improvement in those who have not yet completed one year of treatment. CRP delta was defined as the CRP variation between baseline and week 14 (i.e. week 14 CRP minus baseline CRP).

2.1 Statistics

All data were arranged, processed and analyzed with SPSS® v.18.0 data (Statistical Package for Social Sciences). Categorical variables were described through absolute and relative frequencies and continuous variables were described as mean and standard deviation, median, percentiles, minimum and maximum. Hypotheses were tested about the distribution of continuous variables with non-normal distribution, by using the nonparametric Mann–Whitney and Kruskal–Wallis tests, depending on the nature of the hypothesis. Model performances were assessed using the area under the ROC curve (ROC AUC). Logistic regression was used to determine the relationship between CRP delta and treatment failure regarding sustained response (adjusted to baseline CRP). For an effect size of 0.30 the sample power is around 80% for a confidence level of 95%. The effect size was used to calculate the power of the sample and was calculated using posthoc values. For that, average values of the variable of interest and of the different study groups were used. Software used was Gpower and family test T test, as well, as the test available — means: Wilcoxon–Mann–Whitney test (two groups). All hypotheses were tested at 5% level of significance.

3 Results

3.1 Population

Baseline demographic characteristics of all 148 patients with CD in treatment with IFX are summarized in Table 1. Most CD patients who started treatment with IFX had ileocolonic disease (39%), with non-stricturing, non-penetrating behavior (44%). IFX was started in all the patients due to severity of luminal disease, even in those with concomitant perianal disease. The primary indication for treatment was disease activity despite azathioprine full-dose treatment (mean dose of 2 mg/kg; thiopurine methyltransferase test was not available) (46% patients), followed by corticosteroid dependency (18% patients) and azathioprine intolerance (15% patients). Sixty-four percent of the patients have previously been treated with azathioprine and corticosteroids, and treatment was started as combination in 82% of the patients (Table 2).

View this table:
Table 1

Baseline characteristic of Crohn's disease (CD) patients on infliximab regarding disease duration, time since infliximab first infusion and Montreal classification.

CD (n = 148)
Female/male74/74
Median age (years: min–max)36 (14–68)
Median age at diagnosis (years: min–max)25 (9–61)
Median disease duration until 1st IFX (years; min–max)4 (0–29)
Median follow-up since diagnosis (years: min–max)7 (1–24)
Median follow-up since IFX (months: min–max)24 (0–134)
Montreal classification
Age at diagnosis (%)
A114 (9%)
A2119 (81%)
A315 (10%)
Disease location (%)
L1 + L1–455 (41%)
L2 + L2–425 (18%)
L3 + L3–456 (40%)
L41 (1%)
Behavior (%)
B1 + B1p41 + 21 (44%)
B2 + B2p34 + 11 (32%)
B3 + B3p18 + 16 (24%)
Perianal disease (%)48 (34%)
View this table:
Table 2

Indications for infliximab, previous treatment and combination treatment in Crohn's disease (CD).

CD (n = 148)
Indications for infliximab (%)
Corticosteroid dependency29 (20%)
Corticosteroid resistance10 (7%)
Thiopurine intolerance24 (16%)
Activity despite thiopurine full-dose85 (57%)
Previous treatment (%)
Immunomodulators26 (18%)
Corticosteroids13 (9%)
Corticosteroids + immunomodulators (subsequent use)104 (70%)
Adalimumab3 (2%)
Immunomodulators + adalimumab2 (1%)
Combination treatment with azathioprine (%)118 (80%)

In our study, 12 patients out of 148 had persistent low CRP levels despite having clinical and endoscopic activity. However, no correlation was found with age at diagnosis (p = 0.110), disease location (p = 0.561), disease behavior (p = 0.787) or response to IFX (p = 0.758).

Baseline CRP and week 14 CRP levels were not influenced by perianal disease (p = 0.622 and p = 0.797, respectively), disease location (p = 0.162 and p = 0.969, respectively) or disease behavior (p = 0.368 and p = 0.608, respectively).

3.2 Treatment response

Patients' response to IFX treatment is described in Fig. 1. Sixteen CD patients were responding to IFX but have not yet reached 1 year of treatment (short-term response). For long-term analysis, only patients on maintenance therapy for at least one year who were considered SR or those considered PNR or RTA were included. In CD, there was no significant association between IFX response and gender (p = 0.064) or age at diagnosis (p = 0.745) (Table 3). No significant association was found between location and response to IFX (p = 0.528), but patients with longer disease duration responded worse to IFX (either PNR or RTA) than those with shorter disease duration (p = 0.006) (Table 3).

Figure 1

Outcome to infliximab treatment in Crohn's disease patients — flowchart. IFX, infliximab; CD, Crohn's disease.

View this table:
Table 3

Patients' characteristics according to disease duration, C-reactive protein (CRP) levels at baseline and week 14, if on combination treatment and response to infliximab.

CharacteristicsPrimary non response (n = 17)Sustained response (n = 79)Response with therapeutic adjustment (n = 28)p-Value
Gender (%)0.064
Male5 (8.6%)43 (74.1%)10 (17.3%)
Female12 (18.2%)36 (54.5%)18 (27.3%)
Median age at diagnosis (years: min–max)25 (14–55)25 (11–55)25 (9–61)0.745
Median disease duration until 1st infliximab (years: min–max)6 (0–29)3 (0–27)9 (0–29)0.006
Median baseline CRP (mg/L: min–max)26.2 (1–144.3)9.6 (0.5–200.2)7.65 (0.9–46.9)0.02
Baseline CRP (%)0.582
< 3 mg/L2 (8%)17 (68%)6 (24%)
> 3 mg/L15 (16%)57 (62%)20 (22%)
Median CRP at week 14 (mg/L: min–max)9.1 (0.6–110.1)3.1 (0.1–29)7.6 (0.4–58.4)0.007
CRP at week 14 (%)0.053
< 3 mg/L3 (6%)39 (78%)8 (16%)
> 3 mg/L10 (14.3%)40 (57.1%)20 (28.6%)
Combination treatment with azathioprine (%)10 (10.3%)65 (67%)22 (22.7%)0.132

p-Value: overall comparison between groups.

    Ileocolonic disease was not statistically significant associated with a further dose increase (p = 0.514) or shortening of frequency (p = 0.212) regarding other disease locations, as well as, combination treatment did not influence the need of dose adjustment (p = 0.799) nor the need of frequency adjustment (p = 0.498).

    3.3 C-reactive protein levels and response

    3.3.1 Baseline

    The median baseline CRP value was 9.0 mg/L. Baseline CRP values were higher in PNR patients when compared to those with SR (26.2 mg/L vs 9.6 mg/L; p = 0.015) or RTA (26.2 mg/L vs 7.6 mg/L; p = 0.007) (Fig. 2). There was no statistical difference between the patients with SR and the patients with RTA (9.6 vs 7.6; p = 0.432). CRP values greater than 15 mg/L at baseline predict PNR with 67% sensitivity and 65% specificity (Fig. 3). The negative predictive value (NPV) was 93%.

    Figure 2

    C-reactive protein levels (mg/L) at baseline median and interquartile range (Mann Whitney test). CRP, C-reactive protein; PNR, primary nonresponse; SR, sustained response; RTA, response after therapeutic adjustment.

    Figure 3

    Receiver operating characteristic curves for baseline serum C-reactive protein and primary non response at week 14. CRP, C-reactive protein.

    Baseline CRP levels did not influence hospitalization (9 mg/L [0.1–125.0] vs 9 mg/L [0.6–200.2], p = 0.805).

    3.3.2 Week 14

    CRP values at week 14 were lower in patients with SR when compared to those with PNR (3.1 mg/L vs 9.1 mg/L; p = 0.013) or RTA (3.1 mg/L vs 7.6 mg/L; p = 0.019), while it did not differ between patients with PNR and RTA (9.1 mg/L vs 7.6 mg/L; p = 0.363) (Fig. 4).

    Figure 4

    C-reactive protein levels (mg/L) at week 14 — median and interquartile range (Mann Whitney test). CRP, C-reactive protein; PNR, primary nonresponse; SR, sustained response; RTA, response after therapeutic adjustment.

    Lower CRP levels at week 14 (< 3 mg/L) were associated with sustained response, e.g. 78% of patients with CRP levels < 3 mg/L at week 14 compared with 57% of patients with CRP levels > 3 mg/L achieved sustained response (p = 0.053) (Table 3). CRP levels greater than 4.6 mg/L at week 14 predict PNR with 67% sensitivity and 62% specificity (Fig. 3). The NPV was 94%.

    CRP was significantly lower at week 14 in patients on combination maintenance treatment (p = 0.031).

    There was a trend to higher CRP levels at week 14 (4.7 mg/L [0.1–110.0] vs 3 mg/L [0.1–81.0], p = 0.079) predict hospitalization.

    3.3.3 CRP delta

    CRP delta between baseline and week 14 of treatment was higher in patients with SR than in those with RTA (5.2 mg/L vs 0.6 mg/L; p = 0.027), while there was no difference between PNR and SR (5.8 mg/L vs 5.2 mg/L) or PNR and RTA (5.8 mg/L vs 0.6 mg/L). The probability of RTA regarding SR decreases 0.914 for each increase of one delta unit [0.848–0.986], which means that the likelihood of SR regarding RTA increases with the increase of delta. This result is adjusted to the baseline CRP. It is possible to see that there is a linear correlation between baseline CRP and CRP delta, either in SR or in RTA, but with different Pearson's correlations (Fig. 5).

    Figure 5

    Baseline C-reactive protein and C-reactive protein delta scatter dot — response after therapeutic adjustment and sustained response. CRP, C-reactive protein.

    Considering a CRP level cut-off of 4.6 mg/dL at week 14 as predictor of non-response, a CRP delta lower than 3.1 predicts non-response with 67% sensitivity and 64% specificity.

    4 Discussion

    Our purpose was to explore the association between CRP levels and the clinical outcome of patients with CD treated with IFX, followed at a tertiary center. The applicability of clinical trials results to routine clinical practice has been frequently questioned28 and, herein, we have the opportunity to evaluate the results of current practice.

    We would like to emphasize the importance of CRP levels at week 14, once lower CRP levels at this time were associated with sustained response in patients treated with IFX, independently of baseline CRP serum levels. This has already been showed in previous reports.11,29 In the study by Reinisch et al.,29 normalization of CRP at week 14 (< 5 mg/L) predicted a higher rate of response and remission during the first year. However, no optimal cut-off at week 14 has previously been established as predictive of nonresponse. Herein, we suggest that CRP levels higher than 4.6 mg/L might be predictive of nonresponse, with 67% sensitivity and 62% specificity. This is particularly useful, when considered together with CRP deltas between beginning of treatment and week 14, once higher deltas are predictive of sustained response in relation to those with need of therapeutic adjustment. CRP delta wasn't predictive of sustained response when compared with nonresponse, however, if we define a CRP level cut-off of 4.6 mg/dL at week 14 as predictor of non-response, CRP delta lower than 3.1 predict non-response with 67% sensitivity and 64% specificity. This highlights the need to combine CRP levels at week 14 with CRP delta. This way, patients with lower CRP values at week 14 and higher CRP variations are less likely to need further therapeutic adjustments. We selected week 14 for our analysis since this is when the efficacy of the induction phase is usually assessed and maintenance treatment starts.11 Similar results were achieved with Adalimumab (ADA), as shown by Kiss LS30 where lower CRP levels at week 12 (< 5 mg/L) were associated with a higher likelihood of clinical remission at one year of treatment.31 If CRP levels do not improve at week 14, a new assessment of the patient should be done,32 considering the need for therapeutic adjustment or a combination of another drug.

    Previous studies14,29,32 suggest, in opposition to our results, that lower baseline CRP levels can predict nonresponse, as they may correspond to absence of inflammation and, thus, have no response to anti-TNF treatment. Cut-off levels in W. Reinisch29 and M. Jürgens32 were 7–25 mg/L [18.8 ± 24.1 mg/L] and 3 mg/L [IQR 1.5–35.3 mg/L], with higher values being associated with maintenance of response. E. Louis33 achieved similar results, with response rates being higher in patients with CRP levels higher than 5 mg/L [IQR 5–160 mg/L]. Similar results for response and remission were seen with ADA34 and Certolizumab35 (10 mg/L in both).

    In SONIC trial, patients who had low CRP levels and no endoscopic lesions responded worse to treatment.36 Considering this, symptoms might be attributed to causes other than active CD. In our study, primary nonresponders had higher baseline CRP, with levels greater than 15 mg/L predicting nonresponse with 67% sensitivity and 65% specificity. There was no difference regarding IFX response, when a cut-off value of 3 mg/L was established (levels below 3 mg/L were considered normal). However, all our patients were submitted to an endoscopy in the three months previous to beginning of IFX treatment, which showed, in all of them, endoscopic activity, with ulcers. Previous studies25,26,37,38 depicted that higher CRP levels were associated with more severe disease. Because of their high molecular weight, monoclonal antibodies do not undergo renal elimination or metabolism by hepatic enzymes. The catabolism occurs within the cells of the reticuloendothelial system as the primary route of elimination.39 Disease severity may influence elimination of antibodies through reticuloendothelial system-mediated mechanisms.40 It has been shown that patients with elevated CRP and serum albumin concentrations below the normal range have accelerated drug clearance.41,42 The presence of systemic inflammation may therefore increase antibody catabolism in the reticuloendothelial system. This is supported by the suboptimal IFX concentrations that have been observed in patients with severe UC undergoing IFX induction therapy.43 Based on this, we underline that patients with higher grades of inflammation (more severe disease) have a significantly lower likelihood of maintaining sustained response. These patients may benefit from earlier introduction of IFX therapy, which we support, as well as, SONIC study that emphasize the benefit of starting IFX therapy at earlier stages.36 A recent paper by Sandborn et al.,44 with results referring to adalimumab, says that in patients with moderately to severely active CD, weekly dosing of adalimumab may be most effective in the anti-TNF-experienced patients with elevated CRP at baseline, in comparison with adalimumab every other week.

    In our data, patients on combination treatment had lower CRP levels at week 14, however, that did not influence final response, need of dose adjustment nor frequency adjustment. Finally, CD patients with longer disease duration responded less well to IFX.

    In this analysis, it was possible to evaluate the results of current practice and confirm the important role displayed by CRP as a noninvasive inflammatory marker. According to our results, a higher inflammatory burden at IFX beginning is correlated with a worse response to treatment. The sensitivity and specificity achieved in the presented study isn't high enough to allow the recommendation of the CRP levels proposed by us an indication to stop treatment. However, the high NPV of CRP supports an important role for this marker, once 93% of the patients with baseline CRP levels lower than 15 mg/L and 94% of the patients with CRP levels lower than 4.6 mg/L at week 14 achieved sustained response or response after therapeutic adjustment. Therefore, a decrease in CRP levels at week 14 to levels sufficiently low may be a predictor of good response to treatment. We believe that there is a cut-off CRP from which higher levels are associated with a worse response. It may be due to a more severe disease or to a higher rate of clearance of IFX. Hence, CRP levels together with CRP delta may be a good surrogate marker of IFX response. In clinical practice we propose an evaluation of all patients at week 14 and those with high CRP and low delta should have their IFX efficacy reassessed, however, further prospective studies and external validations are required to evaluate the results achieved.

    Conflicts of interest disclosure and funding declaration

    None declared.

    References

    View Abstract