Comment on Ammonia level as a proxy of asparaginase inactivation in children: A strategy for classification of infusion reactions

I read the article of Santos et al. ¹ They found that the ammonia levels served as a proxy of asparaginase inactivation in children with acute lymphoblastic leukemia (ALL) using native E. coli asparaginase.

Asparaginase is a non-human enzyme that catalyzes the hydrolysis of asparagine into aspartic acid and ammonia. ² The efficacy of asparaginase can be evaluated by measuring the levels of asparaginase activity.^3–6 Although the most direct way of assessing asparaginase efficacy is the measurement of asparagine from the blood. ⁷ The evaluation of asparagine depletion is, however, technically difficult.^8,9 Measuring asparagine levels in cerebrospinal fluid (CSF) is also studied to evaluate its efficacy.^10,11 In the last decade, also the role of assessing asparaginase antibodies during asparaginase therapy has been, extensively, studied. However, different results were published with sometimes controversy.^6,12–16

Previously, ammonia levels have been suggested to reflect the asparaginase activities.^17,18 It has been suggested in case reports that ammonia release could lead to encephalopathy.^19,20 Moreover in a previous prospective study, it was shown that the ammonia level was not related to central neurotoxicity. ²¹ Given the unclear role of the clinical utility of ammonia levels in daily practice, I would like to comment on the paper of Santos et al. It should be mentioned that the current standard of practice to evaluate asparaginase efficacy is therapeutic drug monitoring (TDM) with measuring asparaginase activities which is now used world-wide. ²² Also some important statistical questions can be raised, which I address below.

First, in the paper of Santos et al. the upper limit of normal of the ammonia levels is unclear. The authors had only defined the corresponding ammonia levels according to grades 1 or 2 using the Common Terminology Criteria for Adverse Events (CTCAE) 3/4.03 version. ¹ If the upper limit of normal of the ammonia levels was used according to a previous paper ¹⁸ , there would be no (cor)relation, as currently shown in Figure 3 of their paper.

Second, in my opinion, the relationship shown in panel B of this Figure 3 seems not correct. Santos et al. should not mention hyperammonemia as reflection of the ammonia levels itself. My suggestion is that by using the Fisher Exact test is the correct way to analyze this relationship. For example: low/high ammonia levels versus no hypersensitivity/or reaction in a 2-by-2 contingency tables. Also, I suggest to use a Violin plot rather than a box plot, as a Violin plot also show the probability density of the data at different values. ²³

Third, two laboratory issues. To avoid the ongoing production of ammonia by asparaginase ex vivo did the authors adhere to the following procedure: were the blood samples put in an ice bath and were these samples immediately processed at their laboratory? The authors also obtained blood samples immediately after the asparaginase courses, why did not the authors measured ammonia trough levels?

Fourth, some statistical issues. The authors studied 245 infusions in 32 patients, and 19 reactions were observed in 17 children. I was wondering if the authors noticed that given this information only two risk factors should be studied. The authors chose to use a logistic regression model. By using more than two risk factors, this model could be overfitted. More importantly, why did these authors chose to use a logistic regression model? Their study group was rather small, hence a descriptive statistical approach, to present the data, would be more appropriate.

Lastly, in their Table 1, they authors present the odds ratio of age for each year of life. Why did not the authors use the National Cancer Institute (NCI) criteria for age, for example: age less than 10 years and age at least 10 years?

To conclude, in the past decade monitoring of asparaginase efficacy has proven to be very successful, mainly by implementing asparaginase activities to monitor asparaginase pharmacokinetics. ²² Other (surrogate) measurements are available, including ammonia measurements. However, the pharmacology of asparaginase is rather difficult and some controversies do exist. Challenges herein are still to be solved.