Serum levels of tumour necrosis factor-alpha predict response to recombinant human erythropoietin in patients with myelodysplastic syndrome

We measured pretreatment serum levels of tumour necrosis factor-alpha (TNF-alpha) and interleukin-1 beta (IL-1 beta) in 25 patients with myelodysplastic syndrome receiving recombinant human erythropoietin (rhEPO) at dosages up to 300 U/kg thrice weekly for 12 weeks. Both TNF-alpha and IL-1 beta levels were measured using commercially available enzyme-linked immunoassays. A complete response (CR) was defined as a rise in untransfused haemoglobin concentrations of at least 2 g/dl or a 100% decrease in RBC transfusion requirements over the treatment period; a partial response (PR) was an increase in untransfused haemoglobin values of 1-2 g/dl or a decrease in RBC transfusion requirements equal to or greater than 50%: no response (NR) was defined as a response less than a PR. After 12 weeks of rhEPO treatment, four patients showed a CR, five patients a PR, and 16 patients NR. Serum levels of both TNF-alpha (80.5 +/- 64.8 vs 8.1 +/- 4.2 ng/l, P < 0.001) and IL-1 beta (60.4 +/- 49.9 vs 8.9 +/- 4.7 ng/l, P < 0.001) were higher in MDS patients than in a group of 28 normal controls, Responders (CR + PR) showed significantly lower serum levels of TNF-alpha than non-responders (21.6 +/- 26.2 vs 106.3 +/- 60.8 ng/l, P < 0.001), whereas IL-1 beta concentrations between those who benefited from therapy and unresponsive cases were not significantly different (39.8 +/- 48.9 vs 73.4 +/- 48.2 ng/l, P = 0.120). It is noteworthy that TNF-alpha levels were within the normal range in all responsive patients but one, whereas all non-responders presented elevated cytokine concentrations. No relationship was found between TNF-alpha or IL-1 beta values and haemoglobin levels, transfusion requirement, serum EPO or ferritin concentrations. We conclude that pre-treatment TNF-alpha levels might help to select those MDS patients who are most likely to benefit from rhEPO treatment.