Metabolic Profiling Predicts Response to TNF Inhibitors in RA

blood sample_TS_179244038.jpg
blood sample_TS_179244038.jpg
Although individual patients' metablome profiles may vary with disease activity, metabolic profiling can identify variations between responders and non-responders to TNF.

Plasma metabolomic analysis can successfully be used to identify clinical response to tumor necrosis factor (TNF)-inhibitors in patients with rheumatoid arthritis (RA), according to a study published in BMC Musculoskeletal Disorders.1

While TNF inhibitors are effective in many patients with RA, up to one-third do not respond to treatment.2 Because TNF inhibitors can cause adverse events, including infection and basal cell carcinoma reactivation, there is a need to identify markers that predict response to therapy in order to avoid treating patients who are not likely to receive any benefit from therapy.

The profile of the metabolome, or complete set of metabolites, has been shown to vary with disease activity levels in several inflammatory conditions. Zuzana Tatar, MD, of CHU Gabriel-Montpied in France, and colleagues sought to determine whether metabolic profiling can identify variations among TNF inhibitor responders and nonresponders.

Only those patients designated as good responders and non-responders to anti-TNF therapy were included in this study. This was done “to increase the chances of finding significant differences in metabolic profiles,” the authors explained. 

High Yield Data Summary

  • Distinct metabolic “fingerprints” distinguished patients who were adequate responders and nonresponders to TNFi therapy

 A total of 100 patients were identified as good clinical responders (Disease Activity Score 28 [DAS28] ≤3.2 and improvement by >1.2 points from baseline after anti-TNF therapy), and 40 were nonresponders (DAS28 ≥5.1 and improvement by <0.6 points from baseline after anti-TNF therapy).

Blood samples were obtained prior to and 6 months after initiation of TNF inhibitor treatment. Samples were analyzed using reverse-phase liquid chromatography-electrospray quadrupole-time-of-flight mass spectrometry.

Investigators found that both anti-TNF responders and nonresponders had different metabolic “fingerprints” after 6 months of TNF inhibitor treatment. Carbohydrate derivates, such as D-glucose and maltose, accounted for the primary differences between the 2 groups. However, no single predictive biomarker was identified.

No differences in metabolic profiles were observed when patients were grouped by rheumatoid factor or anti-citrullinated peptide status or by specific TNF inhibitor agent.

“Our metabolomic approach needs to be completed by screening larger cohorts of patients and investigating outcomes with other biologic agents in patients with severe RA,” wrote Dr Tatar and colleauges. “Metabolomic analysis remains expensive, but in case of identification of pertinent biomarkers, classic quantitative analysis could be used in clinical practice.”

Summary and Clinical Applicability

Researchers identified different metabolic profile “fingerprints” that distinguished RA patients with good clinical response to anti-TNF therapy from patients with no response to therapy. Metabolic profiles did not differ based on inflammatory marker status or type of TNF inhibitor used.

Limitations and Disclosures

  • Using blood samples, which are more complicated to analyze and interpret than urine samples, may have resulted in the failure to identify a single predictive biomarker
  • Samples from negative controls were not analyzed for comparison with the study population

This study was funded by grants from PHRC Interrégional, DGOS, and Pfizer. The authors report no competing financial interests. 

Related Articles


  1. Tatar Z, Migne C, Petera M, et al. Variations in the metabolome in response to disease activity of rheumatoid arthritis. BMC Musculoskelet Disord. 2016;17(1):353. doi: 10.1186/s12891-016-1214-5
  2. Marotte H, Miossec P. Biomarkers for prediction of TNFalpha blockers response in rheumatoid arthritis. Joint Bone Spine. 2010;77(4):297–305.

follow @RheumAdvisor