Variables associated with the assessment of systemic tumor necrosis factor alpha levels during hemodialysis

Conflicting results have been published concerning the systemic induction of the cytokine tumor necrosis factor alpha (TNF alpha) during hemodialysis (HD). We therefore evaluated in vitro TNF alpha production in whole blood as well as in vivo variability of TNF alpha levels in patients on long-term HD. Whole blood was incubated at room temperature (RT) with or without exogenously added endotoxin (ET), and plasma-TNF alpha was measured after 5, 30, 120, 240, and 960 min by specific enzyme immunoassay. Additionally, plasma-TNF alpha before and after 120 and 240 min HD was studied longitudinally once a week over a period of 4 weeks in 36 patients on Cuprophan (CU, n = 23) or polysulfone-F60 (PSu, n = 13) HD. Mean plasma TNF alpha levels in vitro rose from (mean) 8 pg/ml after 5 min to 12 pg/ml (120') and 32 pg/ml (960') even without ET addition, and to 18 pg/ml (after 120') and 88 pg/ml (after 960') when 0.1 microgram/ml ET were added. Pre-dialytic as well as intra-dialytic TNF alpha levels in patients showed high intra-individual variability. A substantial (> 100%) increase in plasma TNF alpha was observed during only 14 out of 84 treatments with CU and 20 out of 47 with PSu, however, the increase in TNF alpha was not statistically significant in either group. We conclude that the sampling procedure, if not carefully standardized, is a potential source of artifacts with regard to "systemic" TNF alpha levels. The high intra and inter-individual variability of plasma TNF alpha suggests that results of cross-sectional studies are questionable.