Response to Response to Letter Regarding “Plasma Concentration of Thrombopoietin in Dogs With Immune Thrombocytopenia”

Abstract

I would like to acknowledge the hard work that Brooks et al. have put into developing their canine TPO ELISA and thank them for their response to my letter [1-3]. Unfortunately, it appears that the authors have misunderstood the concerns raised regarding the validation process of the assay, which I believe must be addressed before ineffective thrombopoiesis can be implicated in contributing to primary immune thrombocytopenia (pITP). This issue has substantial clinical importance because, as Brooks et al. conclude, findings from this study may provide (or not) “a rationale for future studies exploring the potential utility of TPO receptor agonist treatment in canine ITP patients.”

First, I did not mean to suggest that cryo-poor plasma was added to any of the canine plasma samples. Rather, my concern pertains to the use of cryo-poor plasma with an unknown TPO content in the spike-and-recovery experiment described in section 3.3 of the manuscript. Brooks et al. account for the unknown TPO concentration by subtracting the OD of blank wells containing only cryo-poor plasma in buffer from all standard and test sample wells. The problem is that this method of subtraction is only valid if the percentage recovery of recombinant and endogenous TPO behaves in an additive (linear) manner. Not all recombinant and endogenous analytes behave in such manner when measured with an ELISA; an example of nonlinear recovery is provided in Table 1 below [4].

In this example, the measured (observed) concentration of a compound measured using a ligand-binding assay is not equal to the sum of the endogenous and recombinant (target) concentrations. If the recovery of recombinant and endogenous TPO behaves in this non-linear manner, then it is not valid to subtract the OD of blank wells from the standard and test sample wells [4]. Doing so may underestimate the measured TPO at low concentrations (as seen in the “low” level in Table 1), which may have impaired the ability of the assay to detect differences in TPO concentration between healthy and pITP groups. Because the TPO content of the cryo-poor plasma was unknown, it is not possible from the data provided by Brooks et al. to determine whether recombinant and endogenous TPO behave in a linear or nonlinear manner. Accordingly, the standard curve (Figure 2) used to quantify the TPO plasma concentrations in test samples may not be valid.

Second, Brooks et al. report good spike-in recovery of recombinant TPO using their ELISA. However, this was only evaluated in pooled cryo-poor plasma, presumably obtained from healthy dogs. As stated in my previous letter, plasma from dogs with pITP may have contained interfering compounds that were not present or were present at a different concentration in the pooled healthy dog plasma. As such, the recovery of TPO in pITP dogs may have been lower than that in healthy dogs.

The author declares no conflicts of interest.