To the Editor:
Drs. Yip and Jacobsen's thoughtful comments regarding our report are appreciated. We agree with their general comments that the raw data used for the analyses are not optimal. The small number of cases and small number of data points for several of the included cases lend potential error into analysis and subsequent conclusions. Limitations of this nature are unavoidable due to the infrequency of methanol poisoning. These limitations were clearly delineated and discussed in our paper.
We would take issue with Drs. Yip and Jacobsen on several points.
The first point involves statistical comparison of half-life. We also performed a secondary analysis of elimination half-lives using those patients with more optimal data (i.e.: at least three predialysis formate levels). This “subgroup” analysis did not detect any difference in the endogenous vs. hemodialysis elimination half-life for formate (see discussion section p.142). Of course, this tinkering with data creates an increased risk for type II error from a shrinking number of included subjects. Using data from only one patient as the authors of the letter suggest can only be considered less reliable.
Drs. Yip and Jacobsen commented that patients 2 and 7 had longer half-lives while on hemodialysis. We cannot comment on half-lives for individual subjects, as paired testing was not possible. We analyzed the means of the study population and found no difference.
A short endogenous elimination half-life probably decreases the additive effect of hemodialysis. Formate certainly has favorable physical properties that render it amenable to dialysis, such as small molecular weight, lack of protein binding, and small volume of distribution. Our calculated hemodialysis clearance (223 ml/min) supports dialysis. However, the endogenous elimination half-life of formate probably alters hemodialysis impact. In general, the shorter the half-life, the less likely that dialysis will be of benefit. For formate, three of the four previous case reports suggest short endogenous elimination half-lives of 2.5, 2.5, and 3.7 hr (Verstraete et al., Moore et al., Osterloh et al.). Only one case report documents a long half-life of 20 hr (Shahangian et al.). In this case, the long half-life is explained by ongoing metabolism of methanol with rising formate levels as the patient was not treated with ethanol until 21 hr post-presentation (see Fig. 1, Shahangian et al., Clin Chem 1984;30;1414). Thus this patient is a likely outlier. Our study subjects received fomepizole that eliminated further production of formate. Thus, our results reflect true elimination. Our data concurs with a rapid elimination (205 min). A relatively short endogenous elimination half-life may have minimized the impact of dialysis.
Drs. Yip and Jacobsen stated that the graph in Fig. 1 (p. 140) from the 45-year-old patient directly contradicts our conclusion that dialysis had no impact. Granted, the slope may have the appearance of decreasing during the initial hemodialysis and the half-lives with and without dialysis (146 vs. 347 min) suggest benefit. However, this patient may be an outlier for two reasons. First, while no study subject developed overt renal failure, this 45-year-old had greater than seven hours of renal dysfunction (peak BUN/creatinine of 22/2.3 mg/dl). Renal dysfunction due to hypovolemia may have contributed to relatively long predialysis elimination. Second, this patient had the highest formate of any study subject. Perhaps there is some alteration in formate metabolism/elimination at extreme concentrations. However, we want to emphasize that while there appears to be a treatment effect for this individual, no statistical comparison for individual subjects was performed. Additionally, the second dialysis session for the 45-year-old appears to greatly alter the slope of the log concentration time curve, implying an impact on formate elimination. However, this change in slope occurred at concentrations that were barely above limits of detection and late in the patient's recovery. Lastly, the slope of the concentration-time relationship does not appear to change during dialysis for the second patient depicted in the figure that had similarly elevated formate and marked acidemia.
Overall, the results of our study raise a focused question that pertains to the utility of dialysis for the elimination of formate only. The study did not address methanol elimination (nor acidemia or clinical manifestations such as blindness). Thus, the question of clinical utility of dialysis must not be extrapolated beyond our stated condition where there is only formate present and no methanol. The vast majority of significantly methanol-poisoned patients will undergo dialysis to enhance methanol elimination, and the question of the dialysis of formate is moot. For example, no patient in this trial with an elevated formate level and acidemia had a zero methanol level.
In summary, we agree (and so stated in the discussion) that the question of the benefit of dialysis for formate alone warrants further study.
William Kerns II, M.D.
Christian Tomaszewski, M.D.
Emergency Medicine and Medical ToxicologyCarolinas Medical CenterCharlotte, North Carolina, USA