You are here

An increase in serum tumour necrosis factor-α during anti-tumour necrosis factor-α therapy for Crohn's disease – A paradox or a predictive index?

Digestive and Liver Disease, Volume 48, Issue 10, October 2016, Pages 1168 - 1171



Soluble tumour necrosis factor-α (sTNF-α) has been reported to increase in the course of anti-TNF-α therapy for rheumatoid and skin diseases.


To assess changes in sTNF-α and clinical efficacy of anti-TNF-α agents in Crohn's disease (CD).


Sixty-four patients on infliximab or adalimumab were analyzed. Clinical outcomes were assessed by using CD Activity Index after the induction therapy and at week 52. sTNF-α was measured before and after the induction therapy with high-sensitivity immunoassay.


In the majority of patients, sTNF-α increased significantly. Those with the greatest increase were more likely to experience long-term response, were more often treated with infliximab, had less frequently isolated small bowel CD, and tended to have sTNF-α levels at baseline that correlated with C-reactive protein.


Neutralization of sTNF-α does not seem to be critical for the efficacy of anti-TNF-α therapy in CD. Paradoxically – an increase in sTNF-α may reflect an ongoing process that is beneficial for the clinical outcome.

Keywords: Biological therapy, Cytokines, Gastroenterology, Inflammatory bowel disease.

1. Introduction

Tumour necrosis factor alpha (TNF-α) plays a central role in the pathogenesis of Crohn's disease (CD) [1]. Therapies based on neutralization of TNF-α have fundamentally changed treatment algorithms for CD [2]. Increasing data accumulate to suggest that binding to transmembrane form of TNF-α (tmTNF-α) on target cells rather than neutralization of soluble form of TNF-α (sTNF-α) is crucial for the anti-inflammatory effect of TNF-α antagonists [1]. Surprisingly, recent data from patients treated with anti-TNF-α agents for rheumatoid and skin diseases showed that circulating levels of sTNF-α did not change or even increased in the course of therapy [3] and [4]. However, it is not clear whether this effect is a consistent feature of anti-TNF-α regimen or it is rather related to the nature of the underlying disease. The relationship between changes in sTNF-α and the clinical efficacy of anti-TNF-α agents in the context of CD has not been investigated so far. We have therefore examined serum levels of sTNF-α in CD patients and demonstrated for the first time that the induction therapy with anti-TNF-α antibodies is associated with a significant increase in sTNF-α, especially so in patients with satisfactory long-term response to treatment.

2. Material and methods

CD patients who did not respond to a prior standard therapy with mesalamine, steroids, and thiopurines (at recommended doses and regimens) were qualified for anti-TNF-α therapy and prospectively enrolled into the study [5]. Activity of the disease was determined by parallel assessment of the CD Activity Index (CDAI) and biochemical parameters [6]. The patients were treated with either infliximab (IFX) or adalimumab (ADA) according to current therapeutic guidelines [5]. Primary response to treatment was defined as a decrease in CDAI ≥ 100 points after the induction period of 10 or 12 weeks for IFX and ADA, respectively [2] and [5]. Primary responders were then maintained on anti-TNF-α regimen until week 52. Long-term response was defined as a steroid-free remission with CDAI < 150 points at week 52. If these criteria were not met, the patients were considered to be secondary non-responders. Serum sTNF-α was measured before and after the induction phase of anti-TNF-α therapy with a high-sensitivity TNF-α Quantikine assay (R&D Systems, Minneapolis, MN, USA). Patients who required any change in treatment during the induction period were excluded from the analysis.

All statistical analyses were performed using GraphPad Prism™ 6.07 software (GraphPad Software Inc., USA) using non-parametric statistics. Paired and unpaired data were analyzed with the Wilcoxon test and the Mann–Whitney test, respectively. Multiple comparisons were performed with the Kruskal–Wallis test and the categorical data were analyzed with the chi-square test. Correlations were assessed with the Spearman's rank coefficient. The level of significance was set at p < 0.05.

The study protocol confirms to the ethical guidelines of the 1975 Declaration of Helsinki (6th revision, 2008) and it was approved by the Bioethics Committee at the Poznan University of Medical Sciences (No. 409/2013). Informed consent was obtained from each patient included in the study.

3. Results

Sixty-four patients were enrolled and analyzed. There were 28 (44%) patients with ileocolonic (L3 according to the Montreal Classification of CD), 18 (28%) patients with colonic (L2) and 18 (28%) individuals with isolated ileal (L1) disease location. Thirteen (20%) patients presented fistulizing (B3), 9 (14%) – stenosing (B2) and 42 (66%) luminal (B1) CD behaviour. In 8 (12%) cases perianal complications were present. Thirty-three (52%) patients were treated with ADA and 31 (48%) patients received IFX. Mean sTNF-α levels in the population as a whole increased over the course of the induction therapy from (mean ± SD) 2.3 ± 1.1–6.9 ± 7.3; p = 0.001 (Fig. 1A). It was clear, however, that there existed a subgroup of patients with sTNF-α levels decreasing rather than increasing. We thought it would be interesting to determine if the direction and/or the magnitude of changes in sTNF-α was associated with any particular patient characteristics or clinical outcome. To this end, we have calculated the percent change in sTNF-α over time and used this as a criterion to stratify patients into tertiles (Fig. 1B). In this way, the lowest and the highest tertiles represented patients with the most meaningful decrease and increase in sTNF-α, respectively. Then, the patients were compared according to several demographic and clinical parameters (Table 1).


Fig. 1

(A) Changes in soluble tumour necrosis factor-α (sTNF-α) levels after induction anti-TNF-α therapy. (B) Stratification of the study group into tertiles, according to the percent change in sTNF-α levels in the course of the therapy (T1: – 98% ≤ ΔsTNF-α < −10%; T2: – 10% ≤ ΔsTNF-α < 300%; T3: 300% ≤ ΔsTNF-α < 1400%).


Table 1

Baseline characteristics of the study group according to the change in soluble tumour necrosis factor-α during the induction phase of anti-tumour necrosis factor-α therapy.


Parameter % change in sTNF-α during the induction phase p (T1 vs. T2) p (T1 vs. T3) p (T2 vs. T3)
T1 – 98% ≤ ΔsTNF-α < −10% n = 22 T2 – 10% ≤ ΔsTNF-α < 300% n = 21 T3 300% ≤ ΔsTNF-α < 1400% n = 21
Age (years) 30 ± 9 28 ± 7 31 ± 9 0.5 0.6 0.5
Male/female (n) 10/12 13/8 13/8 0.2 0.5 0.3
Disease duration (years) 6 ± 4 7 ± 5 7± 6 0.1 0.3 0.2
Baseline sTNF-α (pg/ml) 2.8 ± 1.2 2.1 ± 0.8 2.1 ± 0.9 0.01 0.03 0.8
sTNF-α after treatment (pg/ml) 1.6 ± 0.9 3.2 ± 2.2 16.3 ± 5.2 0.002 <0.0001 <0.0001
C-reactive protein (mg/l) 17.2 ± 14.7 21.5 ± 33.8 9.4 ± 9.9 0.4 0.04 0.6
Crohn's Disease Activity Index (points) 309 (IQR: 291–382) 320 (IQR: 193–364) 310 (IQR: 244–375) 0.4 0.5 0.8
Haemoglobin (g/dl) 12.3 ± 1.9 12.8 ± 2.1 11.9 ± 2.1 0.4 0.4 0.1
Albumin (mg/dl) 4.1 ± 0.4 4.1 ± 0.5 4.1 ± 0.6 0.7 0.9 0.8
Platelets (103/mm3) 352 ± 97 360 ± 119 388 ± 101 0.8 0.4 0.2
Disease location according to the Montreal Classification (n)
 L3 – ileocolonic L3: 6/22 L3: 9/21 L3: 13/21 0.2 0.01 0.1
 L2 – colonic L2: 5/22 L2: 6/21 L2: 7/21 0.1 0.3 0.2
 L1 – ileal L1: 11/22 L1: 6/21 L1: 1/21 0.1 0.0009 0.001
Disease behaviour according to the Montreal Classification (n)
 B3: fistulizing B3: 5/22 B3: 3/21 B3: 5/21 0.4 0.7 0.4
 B2: stenosing B2: 4/22 B2: 2/21 B2: 3/21 0.5 0.6 0.6
 B1: luminal B1: 13/22 B1: 16/21 B1: 13/22 0.6 0.8 0.6
 p: perianal complications p: 2/22 p: 2/21 p: 4/21 0.6 0.7 0.3
Medications (n) Steroids: 9/22 Steroids: 12/21 Steroids: 8/21 0.3 0.1 0.3
Mesalamine: 21/22 Mesalamine: 20/21 Mesalamine: 19/21 0.5 0.6 0.5
Azathioprine: 14/22 Azathioprine: 13/21 Azathioprine: 11/21 0.4 0.3 0.5
Adalimumab/infliximab: 18/4 Adalimumab/infliximab: 15/6 Adalimumab/infliximab: 0/21 <0.0001 <0.0001 <0.0001

Abbreviations: sTNF-α, soluble tumour necrosis factor-α.

Patients who experienced an increase in sTNF-α were more likely to be treated with IFX rather than with ADA. Mean ± SD change in sTNF-α level was 9.8 ± 7.6 and 0.3 ± 1.1 pg/ml for IFX and ADA, respectively (p < 0.0001). Patients with an increase in sTNF-α had also less frequently the disease isolated only to the small bowel (mean ± SD change was 0.4 ± 3.2 vs. 6.2 ± 7.9 pg/ml for L1 and non-L1 location, respectively; p = 0.002), and tended to have lower levels of sTNF-α and CRP at baseline (Table 1). Moreover, these sTNF-α and CRP levels correlated significantly – an effect that was not evident in patients whose sTNF-α decreased over time (Fig. 2). There was no significant difference in the percentage of primary non-responders in each group. Accordingly, primary responders and non-responders did not differ in the magnitude of changes in sTNF-α during the induction therapy. Interestingly, however, there was a significant difference between the groups in the fraction of secondary non-responders. While there was 87% of non-responders in the group with a previous significant decrease in sTNF-α, there was none in the group that experienced an initial increase in sTNF-α (Fig. 3A). Consistently, there was a significant difference in the average change in sTNF-α during the induction therapy between long-term responders and secondary non-responders (Fig. 3B).


Fig. 2

Correlations between baseline C-reactive protein and soluble tumour necrosis factor-α (sTNF-α) in patients with decrease (black circles, T1 tertile), no significant change (grey circles, T2 tertile) and increase (white circles, T3 tertile) of sTNF-α levels in the course of anti-TNF-α therapy.



Fig. 3

(A) Differences in the fractions of primary responders vs. non-responders, and long-term responders vs. secondary non-responders to anti-tumour necrosis factor-α (anti-TNF-α) therapy according to the percent change in soluble TNF-α (sTNF-α) levels in the course of biological treatment. (B) Differences in the average change in sTNF-α levels during the induction anti-TNF-α therapy between long-term responders and secondary non-responders (SNR) to biological treatment.


4. Discussion

In the present study we showed for the first time that sTNF-α levels are increasing during anti-TNF-α therapy in CD. Moreover, this phenomenon seemed to predict long-term efficacy of maintenance treatment. Taking into account that one of the postulated mechanisms of TNF-α antagonists action is the neutralization of sTNF-α, the results of our study seem to be quite surprising [1]. However, the recent data on the molecular pathways induced by anti-TNF-α antibodies in CD point more to a role of tmTNF-α rather than sTNF-α. It has been shown that both IFX and ADA - that are used effectively in CD - bind with high affinity to sTNF-α and tmTNF-α [1] and [7]. In contrast, another anti-TNF-α agent – etanercept, which has very low affinity to tmTNF-α, is not effective in CD [1], [7], and [8]. Moreover, it has been demonstrated in a mouse model of T-cell-mediated colitis that binding to tmTNF-α is essential for remission, while neutralization limited to sTNF-α is not effective [9]. By using confocal laser endomicroscopy, Atreya et al. have shown that high numbers of intestinal tmTNF-α-positive immune cells predict higher short-term response rates to anti-TNF-α therapy, which are sustained over a period of 1 year [10]. These data show indirectly that binding to sTNF-α is not crucial for the anti-inflammatory effect of anti-TNF-α antibodies in CD and that an increase in sTNF-α (as observed in our study) is not incompatible with clinical improvement.

We also showed that baseline sTNF-α correlated with CRP only among patients with a subsequent significant increase in sTNF-α. Since these patients appeared to have better long-term prognosis for successful anti-TNF-α therapy, it could be hypothesized that inflammation in this group was mediated mainly by TNF-α. In contrast, in patients experiencing treatment failure, the lack of baseline correlation between sTNF-α and CRP may suggest that inflammation was driven by other non-TNF-α-related mechanisms [11] and [12]. This would explain different outcomes of long-term anti-TNF-α treatment.

Another possible explanation of favourable therapeutic outcomes despite increases in sTNF-α could be linked to the divergent roles of sTNF-α in inflammation. While pro-inflammatory activity of TNF-α is well recognized, there is mounting evidence for its anti-inflammatory functions [1], [2], [9], and [13]. Muller et al. showed in a mouse model that immune-stimulated T cells were sensitized for activation-induced cell death only in presence of sTNF-α [13]. The molecular mechanisms of sTNF-α-dependent enhancement of activated T cell elimination is not known, however, it appears to require both TNF receptors 1 and 2. The authors conclude that in inflammatory conditions like CD, sTNF-α plays an immunoregulatory function. Thus, hypothetically, the anti-inflammatory effect of biological therapy is related not only to tmTNF-α neutralization (which requires the availability of anti-TNF-α antibodies not bound to sTNF-α) but could also be mediated by immunoregulatory activity of sTNF-α [1] and [13].

We also observed that an increase in sTNF-α is especially pronounced in patients treated with IFX. Since IFX and ADA have similar affinities to sTNF-α, it remains to be established whether this phenomenon is related to the particular pharmacokinetics of the drugs (e.g. differences in half-life, drug clearance) or some other mechanisms [7]. Limited data suggest that IFX can increase TNF-α production by circulating CD16+ monocytes [14]. However, the exact mechanism that underlies sTNF-α increase after anti-TNF-α therapy needs to be clarified. Another interesting observation was that an increase in sTNF-α seemed to be less pronounced in patients with ileal CD than in individuals with colonic and ileocolonic involvement. One can hypothesize that it can be due to as yet undefined differences in the course of inflammatory response in L1 CD. Indeed, it is known from every day clinical practice that this location of the disease is associated with the least specific clinical symptoms, and fewer and/or smaller biochemical abnormalities [5] and [6]. Indeed, we have observed that all biochemical parameters of inflammation were less elevated in L1 patients (data not shown).

The present study shows for the first time an unexpected increase in serum sTNF-α during anti-TNF-α therapy for CD and suggests that this early increase in sTNF-α is associated with more favourable long-term clinical outcome. Whether an increase in sTNF-α mediates and/or reflects anti-inflammatory processes induced by biological agents in CD needs to be elucidated.

Conflict of interest

PE received lecture fees from Abbvie Poland and travel grants from Astellas and Abbvie Poland. LŁS received travel grants from Alvogen and Abbvie Poland. IKK received travel grants from Alvogen and Astellas. KL received travel grant from Abbvie Poland. KK, KSE, JŁ, NC and JW have nothing to disclose.


The study was financed by the Grant from the Poznan University of Medical Sciences. “A helping hand” (2014).


  • [1] S.M. Slevin, L.J. Egan. New insights into the mechanisms of action of anti-tumor necrosis factor-α monoclonal antibodies in inflammatory bowel disease. Inflammatory Bowel Diseases. 2015;12:2909-2920
  • [2] S. Danese, L. Vuitton, L. Peyrin-Biroulet. Biologic agents for IBD: practical insights. Nature Reviews Gastroenterology & Hepatology. 2015;12:537-545
  • [3] M. Schulz, H. Dotzlaw, G. Neeck. Ankylosing spondylitis and rheumatoid arthritis: serum levels of TNF-α and its soluble receptors during the course of therapy with etanercept and infliximab. BioMed Research International. 2014;2014:675108
  • [4] H.M. Walters, N. Pan, T.J.A. Lehman, et al. The impact of disease activity and tumour necrosis factor-α inhibitor therapy on cytokine levels in juvenile idiopathic arthritis. Clinical and Experimental Immunology. 2016; [Epub ahead of print]
  • [5] A. Dignass, G. Van Assche, J.O. Lindsay, et al. The second European evidence-based consensus on the diagnosis and management of Crohn's disease: current management. Journal of Crohn's and Colitis. 2010;4:28-62
  • [6] S. Vermeire, S. Schreiber, W.J. Sandborn, et al. Correlation between the Crohn's disease activity and Harvey-Bradshaw indices in assessing Crohn's disease severity. Clinical Gastroenterology and Hepatology. 2010;8:357-363
  • [7] Z. Kaymakcalan, P. Sakorafas, S. Bose, et al. Comparison of affinities, avidities, and complement activation of adalimumab, infliximab, and etanercept in binding to soluble and membrane tumor necrosis factor. Clinical Immunology. 2009;131:308-316
  • [8] A. Kaser. Not all monoclonals are created equal – lessons from failed drug trials in Crohn's disease. Best Practice and Research. Clinical Gastroenterology. 2014;28:437-449
  • [9] C. Perrier, G. de Hertogh, J. Cremer, et al. Neutralization of membrane TNF, but not soluble TNF, is crucial for the treatment of experimental colitis. Inflammatory Bowel Diseases. 2013;19:246-253
  • [10] R. Atreya, H. Neumann, C. Neufert, et al. In vivo imaging using fluorescent antibodies to tumor necrosis factor predicts therapeutic response in Crohn's disease. Nature Medicine. 2014;20:313-318
  • [11] M. Berns, D.W. Hommes. Anti-TNF-α therapies for the treatment of Crohn's disease: the past, present and future. Expert Opinion on Investigational Drugs. 2016;2:129-143
  • [12] K. Papamichael, A. Gils, P. Rutgeerts, et al. Role for therapeutic drug monitoring during induction therapy with TNF antagonists in IBD: evolution in the definition and management of primary nonresponse. Inflammatory Bowel Diseases. 2015;21:182-197
  • [13] S. Muller, S. Rihs, J.M. Schneider, et al. Soluble TNF-α but not transmembrane TNF-α sensitizes T cells for enhanced activation-induced cell death. European Journal of Immunology. 2009;39:3171-3180
  • [14] M. Nazareth, F. Magro, J. Silva, et al. Infliximab therapy increases the frequency of circulating CD16+ monocytes and modifies macrophage cytokine response to bacterial infection. Clinical and Experimental Immunology. 2014;177:703-711


a Department of Gastroenterology, Human Nutrition and Internal Diseases, Poznan University of Medical Sciences, Heliodor Swiecicki Hospital, Poznan, Poland

b Department of Pathophysiology, Poznan University of Medical Sciences, Poznan, Poland

Corresponding author at: Department of Gastroenterology, Human Nutrition and Internal Diseases, Poznan University of Medical Sciences, Heliodor Swiecicki Hospital, 49 Przybyszewskiego St., 60-355 Poznan, Poland. Tel.: +48 698050797; fax: +48 618691686.