HDCN Article Review/Hyperlink

Hakim RM, Held PJ, Stannard DC, Wolfe RA, et al

Effect of the dialysis membrane on mortality of chronic hemodialysis patients

Kidney Int (Aug) 50:566-570 1996

Hakim has proposed a hypothesis that use of a complement activating membrane is detrimental to patients both in terms of morbidity and ultimate survival. In this study, the USRDS data set was used to see how type of dialysis membrane was related to mortality. Patients were randomly chosen from dialysis units, and patients using acetate dialysis were excluded. The strength of the analysis was, that the analysis was restricted to patients in whom URRs were available, controlling for the confounding effect of adequacy. Also, data on a large array of comorbid conditions was collected and controlled for in the analysis.

Membranes were divided into 3 groups: unsubstituted cellulose, modified cellulose (which includes hemophan, cellulose acetate, and cellulose triacetate), and synthetic (primarily polysulfone and PAN). The principal finding was, that the relative risk of death was about 0.75 for modified cellulose and synthetic membranes vs. for unsubstituted cellulose.

Comment: On the surface, the data appear to be support the hypothesis promulgated by Hakim, that complement activation by unsubstituted cellulose membranes has measurable adverse clinical effects. However, there are a number of confounding effects present in the data which leave the question of causality in doubt.

1) Although the data were controlled for by geographic region, a strong center effect may have been present. For example, an abstract will be presented at ASN 96 by the USRDS, that units using synthetic membranes spend more money per patient in terms of staffing than those using unsubstituted cellulose membranes. Also, use of a high flux membrane requires use of a volumetric dialysis machine, the use of which per se may be associated with better volume control and perhaps better survival.
2) There was a strong trend in the data favoring better outcome with membranes having a higher UF rate. Hakim argues that UF rate and complement activation are inversely related, which is true. However, the trend for lower mortality with higher UF rates were also seen within some membrane groups.
3) The great majority of patients in the study were reusing dialyzers. If mortality was associated with complement activation, one would expect the use of bleach to result in a higher mortality rate. This trend was not seen in the data, although this finding was not reported given the relatively small number of patients and the large number of reuse methods employed.
4) Even if a membrane effect is present, it is not at all clear if it is due to biocompatibility or to flux. Given the high percentage of reuse without bleach of unsubstituted cellulose membranes, a flux based explanation of lower mortality with synthetic membranes is equally plausible. At least the flux question will be addressed by the NIH HEMO trial, which compares low flux (cellulose acetate and low flux polysulfone) with high flux (cellulose triacetate, high flux polysulfone) membranes. Unsubstituted cellulose membranes were excluded from the NIH HEMO trial, in an attempt to separate out flux and biocompatibility issues.

Overall, the study emphasizes the difficulties of analyzing data from cross-sectional studies. I don't believe that the case against unsubstituted cellulose membranes is a strong one in 1996, certainly not in terms of mortality. Nevertheless, the USRDS data remains to be explained, and one of the explanations (not the only one) is, that the dialyzer membrane does affect mortality. (John T. Daugirdas, M.D., University of Illinois at Chicago)