Serologic Response to Human Cryptosporidium Infections

Okhuysen et al. (9) reported the susceptibilities and serologic responses of 19 healthy human volunteers rechallenged with Cryptosporidium oocysts. Their finding that less-severe symptoms were observed when volunteers were rechallenged supports our interpretation of the Cryptosporidium outbreak literature and may explain the difficulty in linking reported cases of cryptosporidiosis with water supplies that are chronically contaminated with oocysts (3); however, the low correlation observed between serum antibody response and infection requires clarification. The serological findings are based on an enzyme-linked immunosorbent assay (ELISA) in which the antibody responses to all antigens obtained from Cryptosporidium oocysts were measured (2). Miller (6), Voller (11), and others (10) suggested that because of the complex array of parasite antigens, ELISAs for parasite immunodiagnosis should use immunodominant antigens rather than extracts of the whole parasite. It is possible that seroconversion, based on either an ELISA which uses purified antigens or Western blotting (WB), may have more accurately estimated the rate of Cryptosporidium infection in these volunteers (7).

We compared serum antibody responses to Cryptosporidium antigens determined by an ELISA for combined immunoglobulin G (IgG), IgA, and IgM responses (4) and those by a WB assay for IgG responses to 15/17-kDa and 27-kDa antigen groups (8) using 117 serum samples collected in the spring of 1992 during a cryptosporidiosis outbreak in Jackson County, Oreg. (5). The intensity of the WB and the optical density of the ELISA responses were measured and analyzed as a ratio of the unknown sample and the positive control response (4). Low correlations were found between the ELISA and WB results: Pearson correlations for the IgG response to the 15/17- and 27-kDA antigens and for the IgA response to the 27-kDa antigen were 0.17, 0.42, and 0.35, respectively.

More than half of 34 ELISA-negative individuals had WB responses greater than 20% of the positive control in one or more bands. If the ELISA is less sensitive and specific than the WB, then by using the ELISA, Okhuysen et al. may have misclassified seroconversions as well as those volunteers who were considered seronegative at the outset of the study. Some ELISA-seronegative volunteers may have been WB seropositive. These WB-seropositive individuals may have been less susceptible to illness and, based on the definition of infection used, less likely to have become infected upon rechallenge.

Finally, the definition of infection used in the rechallenge study (being oocyst positive or having illness) differed from that (being oocyst positive) used in an earlier study by these authors (2), but neither definition considers the possibility of volunteers being asymptomatically infected but stool negative for oocysts. The same authors previously reported that oocyst stool-positive but asymptomatic volunteers excreted fewer oocysts than did stool-positive and symptomatic volunteers (1). This possibility suggests that other asymptomatically infected individuals with low numbers of oocysts in their stools may have been misclassified as uninfected. By using the occurrence of illness to identify 10 of 13 infected persons in this study, the authors may have inadvertently mingled risk factors for infection with risk factors for illness from an infection.