To the Editor
The imperfect diagnosis of malaria is a major impediment to the efforts to control and eliminate the disease1. The recent report by Peng et al.2 describing a novel, rapid and potentially low-cost micromagnetic resonance relaxometry technique for diagnosing malaria therefore addresses an important problem. Unfortunately, we believe that there is a major flaw in the authors' interpretation of the results of their measurements, which greatly impacts the estimates of the limits of detection of this technique.
Our most important concern relates to the interpretation of the relationship of the transverse relaxation rate (R2 index) with the time after infection (shown in Fig. 4b in the original paper2) and the parasitemia level (Figs. 2a and 4c). In these comparisons, the authors have used the interquartile range (IQR) of the measurements conducted using uninfected mice (far left scatter plot in Fig. 4b) as the 'baseline' level against which measurements of infected mice are compared. An inherent property of the IQR is that 25% of the data fall above the upper (75%) level. Any diagnostic test evaluated against the 75% level would thus, by definition, result in 25% false-positive observations in uninfected individuals. We strongly believe that reliable detection of malaria infections with a new method should be assessed using a more stringent statistic, such as the 95% confidence level.
We have re-digitized the baseline (uninfected mice) data Peng et al.2 presented in Figure 4b and have estimated the upper 95% confidence level of the background R2 index to be approximately 1.16 s−1. Assuming that reliable detection of infection is possible only if the mean signal is above this threshold, then the corresponding limit of detection in terms of 'parasitemia level' is approximately 1–2% (Fig. 4c), which is between 5,000- and 10,000-fold higher than the limit of detection claimed in the paper (10 parasites/μl of blood or 0.0002% parasitemia)2.
Visual inspection of the data shown by Peng et al.2 in Figure 4c reveals that the R2 signal does not change significantly below parasitemia levels of 1–2%, strongly indicating that these levels are in fact indistinguishable from the background.
A similar concern pertains to the results presented by Peng et al.2 in Figure 2a. Again, there is a large step between the background signal and the signal originating from the samples with the lowest parasite concentrations. If the background signal is indeed representative, then the measured signal will continuously approach the background signal at sufficiently low parasitemia levels. The observed discontinuity, rather than the expected continuous behavior, between uninfected samples and samples with extremely low parasitemia strongly indicates that an artifact (such as elevated methemoglobin levels) caused artificially high R2 indices in the samples with undetectably low parasite numbers (such as those clearly shown in Fig. 4c for parasitemia levels below 1–2%). Methemoglobin—which is formed, for example, over prolonged periods of blood sample storage or during malaria infection—contains paramagnetic high-spin Fe3+, leading to a higher R2 (ref. 3).
In conclusion, our re-analyses of the data presented by Peng et al.2 indicate that, consistent with previous findings using conventional NMR techniques, the micro-NMR technique can only reliably determine parasitemia levels above 1–2% with 95% confidence4. It may therefore not represent a viable alternative to current tests.
MalERA Consultative Group on Diagnoses and Diagnostics. PLoS Med. 8, e1000396 (2011).
Peng, W.K. et al. Nat. Med. 20, 1069–1073 (2014).
Hänscheid, T. et al. Malar. J. 13, 285 (2014).
Karl, S., Gutierrez, L., House, M.J., Davis, T.M. & St Pierre, T.G. Am. J. Trop. Med. Hyg. 85, 815–817 (2011).
The authors declare no competing financial interests.
About this article
Cite this article
Karl, S., Mueller, I. & St Pierre, T. Considerations regarding the micromagnetic resonance relaxometry technique for rapid label-free malaria diagnosis. Nat Med 21, 1387 (2015). https://doi.org/10.1038/nm.3811
Nature Communications (2021)
Scientific Reports (2017)
Reply to "Considerations regarding the micromagnetic resonance relaxometry technique for rapid label-free malaria diagnosis"
Nature Medicine (2015)