Potential conflict of interest: Nothing to report.
Article first published online: 4 DEC 2012
Copyright © 2012 American Association for the Study of Liver Diseases
Volume 56, Issue 6, pages 2428–2429, December 2012
How to Cite
Remien, C. H., Adler, F. R., Box, T. D., Waddoups, L. and Sussman, N. L. (2012), Reply:. Hepatology, 56: 2428–2429. doi: 10.1002/hep.25940
- Issue published online: 4 DEC 2012
- Article first published online: 4 DEC 2012
- Accepted manuscript online: 14 JUL 2012 03:11AM EST
- Manuscript Received: 7 JUN 2012
- Manuscript Accepted: 7 JUN 2012
We appreciate the comments of Drs. Craig and Simpson and Drs. Mullins and Schwartz, as well as the challenge they raise to make Acetaminophen-Induced Liver Damage (MALD) more believable, comprehensible, and usable. Acetaminophen (APAP) toxicity is indeed a complex process involving multiple steps. Although the underlying mathematics is more involved than the statistical regression models with which physicians are more familiar, MALD could easily be reduced to a web-based or hand-held format for bedside use. The mathematical model accurately fits patient data from the University of Utah. If the modeling holds up in a prospective cohort, we will certainly convert MALD to a more user-friendly format.
The complexity of APAP metabolism and toxicity required us to consider the underlying mechanisms that contribute to end-stage organ damage. The relative values of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and international normalized ratio (INR) are key to understanding the course of injury and, of course, are central to the model. Because of differences in decay rates, AST and ALT provide complementary information about liver damage and both should always be measured in patients with acute liver injury.
Different models have different purposes. The Rumack-Matthews nomogram aids in determining the need for early N-acetylcysteine (N-Ac) treatment by relating plasma APAP level to time of overdose, as best estimated from patient history. MALD provides complementary information about the need for eventual transplant by estimating, from measured patient laboratory values, the size of overdose and elapsed time since overdose. We agree with Drs. Craig and Simpson that sequential organ failure assessment (SOFA) is a highly predictive tool, but SOFA relies on existing organ dysfunction, and its value is best realized in the context of time from overdose. Progressive organ failure is a reliable marker of poor outcome, but perhaps too late to help the patient. Our goal is early prognostication, not confirmation of events that have already occurred.
We all face the same dilemma: We wish to transplant only those patients who will die or suffer irreparable organ damage without a transplant, and we want to give ourselves maximum time to procure a suitable organ when necessary. By using only data obtained on admission, MALD serves as the earliest warning of need for transplant. Figure 3 shows how closely the model predicts actual enzyme curves after admission, indicating that data collected after treatment with N-AC provides little, if any, additional information regarding timing and size of overdose.
Drs. Mullins and Schwartz raise a number of interesting points, but have misinterpreted some of our statements. First, we state that AST is approximately twice ALT at peak, not at baseline, or in patients without acute liver injury. Second, we never discourage anyone from using N-Ac after 24 hours. Survival depends on physiological response to injury even after N-Ac, as modeled by MALD. Third, the fraction of APAP converted to N-acetyl-p-benzoquinone-imide is one of the patient-specific parameters that has no effect on the predictions of outcome because it merely rescales the lethal dose.
We disagree on the issue of arbitrary parameters—values were carefully chosen and referenced (where references existed). We also did not state that MALD was superior to King's College Criteria (KCC). We said it compared favorably and should be tested prospectively. We regret that data limitations prevented us from using the complete KCC score—this deficit will be corrected in a prospective study that compares as many methods as possible.
Like many royalists, Mullins and Schwartz vow fealty to a king who they themselves have nearly decapitated (e.g., by reducing the INR criterion in KCC by 70%). We have no doubt that MALD requires testing and refinement, but we do not understand the objection to further research on a new, promising approach to a problem. The spirit of science lies with further investigation, rather than with loud huzzahs for the past.
Christopher H. Remien*, Frederick R. Adler* , Terry D. Box, Lindsey Waddoups, Norman L. Sussman§, * Department of Mathematics, University of Utah, Salt Lake City, UT, Department of Biology, University of Utah, Salt Lake City, UT, Department of Gastroenterology, University of Utah, Salt Lake City, UT, § Department of Surgery, Baylor College of Medicine, Houston, TX.