Target product profile for a test for the early assessment of treatment efficacy in Chagas disease patients: An expert consensus

Abstract

T, et al. (2020) Target product profile for a test for the early assessment of treatment efficacy in Chagas disease patients: An expert consensus. PLoS Negl Trop Dis 14(4): e0008035.

Chagas disease diagnosis and cure assessment: Getting formally hierarchical about a naturally hierarchical problem

“…In a recent Viewpoint article, Alonso-Padilla and colleagues [ 1 ] discuss the difficulties of confidently classifying a patient with Chagas disease as “cured”—with “cure” defined as the “elimination of Trypanosoma cruzi parasites from the patient’s body following treatment” [ 1 ]. This is, of course, a particular instance of a much more general problem: doctors need to confidently classify subjects as infected or not with a pathogen to make clinical decisions, to understand infectious disease epidemiology, and to measure the effects of therapeutic or preventive interventions.…”

“…Average θ is clearly <1.0 for T . cruzi parasites and their DNA in blood samples; it may approach 1.0 for certain antibody classes such as immunoglobulin G (IgG) in chronically infected subjects (often with the downside of low specificity) and possibly for some pathogen-derived biomarkers [ 1 – 6 ], and can only be 1.0 when the target was seeded in the sample by the investigators—that is, in the artificial context of laboratory studies designed to determine analytical sensitivity [ 6 , 7 ].…”

“…Clinical research teams, then, just need to take one further step ahead, as many ecologists [ 9 , 10 , 13 – 15 ] including disease ecologists [ 8 , 17 , 18 ] have done already, to get formally hierarchical about their naturally hierarchical problems. Back to Chagas disease diagnosis and cure assessment, I think that clinical or epidemiological researchers like Alonso-Padilla and colleagues [ 1 ] would greatly benefit from having reliable estimates of subject-level Ψ, sample-level θ, and test-level p ; from knowing how they vary with subject-, sample-, or test-level covariates; and from getting, for each estimate, a measure of uncertainty.…”

“…This very simple example shows how this approach can help doctors make better strategic decisions—in this case, they should choose “replicate sampling occasions, sample pooling, and limited test replication” instead of the largely useless “multiple test replication on single-occasion samples.” A “single-occasion sample, 2 test replicates” strategy (see [ 1 ]) would be all but hopeless in this scenario—about 31.6% of infections would be missed, most of them because T . cruzi DNA was simply not available for detection in the sample.…”

“…Taken together, then, results resembling those of my example would clearly suggest that efforts should be directed toward developing tests based on targets with high θ . Molecules with high θ values in easy-to-sample tissues or bodily fluids would be particularly interesting—with, e.g., saliva or urine preferable to blood, and capillary blood preferable to venous blood [ 1 ]. Among such widely bioavailable targets, parasite-derived biomarkers (e.g., excreted-secreted antigens [ 4 ], proteins or nucleic acids in extracellular vesicles [ 5 ], or small RNAs [ 19 ]) can in principle be expected to yield overall higher test specificities than host-derived molecules such as anti- T .…”