Antibody‐Dependent Enhancement in Dengue Virus Infections

To the Editor—Laoprasopwattana et al. [1] report that enhancing antibody activity in plasma does not predict subsequent disease severity or viremia in secondary dengue virus (DV) infection. They write that their “results do not support those of a previous study by Kliks et al.” (p. 516) regarding the association of antibody-dependent enhancing activity (EA) and dengue hemorrhagic fever (DHF) in dengue virus (DV) serotype 2 (DV2) infection [2, 3]. We respectfully disagree and argue that their findings simply have no relevance to our previously published conclusions. Whereas we studied asymptomatic versus symptomatic illness in DV2-infected patients and found a clear relationship between plasma EA and symptomatic illness, Laoprasopwattana et al. report 2 substudies, neither of which replicates ours. One is a controlled study of asymptomatic versus symptomatic DV3-infected patients, in which they report no relationship between preinfection plasma EA and disease severity, and the other is an uncontrolled study of only symptomatic DV2-infected patients (without asymptomatic control subjects), in which they again report no relationship between EA and disease severity. The study design used by Laoprasopwattana et al. failed to replicate our study, so it is puzzling that they should so provocatively claim its refutation. Implicit in their comments is the belief that all DVs are equally enhanced by antibodies. Unlike Laoprasopwattana et al., we think that different strains of DV may have very different enhancement properties. Indeed, we and others have reported dramatic differences between the proportions of primary versus secondary DV infections among patients with DHF who are infected with DV1, DV2, DV3, or DV4 in Thailand [4–7] and even differences in these proportions among patients infected with the same serotype (DV2) with different geographic origins [8]. Laoprasopwattana et al.'s provocative claim—based on findings with DV3—to have refuted our prior findings obtained with DV2 is flatly illogical. We were perplexed to read that Laoprasopwattana et al.'s assay of “EA” for DV3 is—paradoxically—a remarkably good surrogate marker for virus neutralization! According to results of their assay, higher “EA” correlates with lower in vivo viremia (their figure 2G and 2H), with an asymptomatic clinical presentation (their figure 4, left panel), and with higher plaque-reduction neutralization titers (as measured on LLC MK2 cells [10], the conventional measure of neutralization; their figure 5, lower panels). This, of course, raises an interesting question: for DV3, exactly why do Laoprasopwattana et al. believe that this assay is a good way to measure in vivo enhancement? We think that the aberrant results of Laoprasopwattana et al. are a consequence of their convenient high-throughput assay, which is based on K562 cells, an undifferentiated malignant cell line derived from a patient with chronic myelogenous leukemia [9]. We found that K562 cells, when infected at high multiplicities—as was done by Laoprasopwattana et al.—exhibit a bias toward EA. Instead, we quite consciously used fresh peripheral blood mononuclear cells, so as to simulate in vivo conditions in our studies, even though they were less convenient than cell lines. It is quite likely that the paradoxical results reported by Laoprasopwattana et al. reflect the peculiarities of their assay, such as the atypical properties of K562 cells or the very high MOIs required to infect these cells, rather than the true pathogenesis of dengue. Unlike their surprising and paradoxical results with DV3, an inverse correlation of higher DV2 EA with lower DV2 plaque-reduction neutralization titers (their figure 5, upper panels) is reported by Laoprasopwattana et al., as would be expected. Yet, even if we do accept the validity of their assay using DV2—with its limitations—and do accept, at face value, their main finding that for DV2 there is no correlation between EA and disease severity in symptomatic patients, we nonetheless remain unconvinced that this finding in any way refutes our own findings. As noted above, the constricted range of patient clinical presentations studied by Laoprasopwattana et al. (i.e., only symptomatic DV2-infected patients, without any DV2-infected asymptomatic control patients) makes these data oflimited value. We contend that their data simply confirm our previously reported results indicating that preillness plasma from symptomatic DV2-infected patients typically has measurable EA. Although we are critical of many aspects of their study, we are nonetheless intrigued by the strikingly different behaviors of DV2 and DV3 in the EA assay of Laoprasopwattana et al. Perhaps this convenient high-throughput assay could find productive use as a way to explore differences in enhancement properties of various DV strains.