Eligible participants were residents of Miami–Dade County, Florida, USA, who had Zika virus disease confirmed by real-time reverse transcription PCR (rRT-PCR) and symptom onset during June–October 2016. Persons with asymptomatic infection, pregnant women, and infants with congenital infection were excluded from enrollment. We enrolled participants during October 16, 2017–February 1, 2018. We obtained written consent from study participants or their guardians.

Serum specimens were tested at the Centers for Disease Control and Prevention (Fort Collins, CO, USA) by IgM antibody capture ELISA (MAC-ELISA) for detection of Zika virus and dengue virus IgM and by plaque reduction neutralization test (PRNT) to detect Zika virus and dengue virus neutralizing antibodies (5,6,22). The PRNT endpoint titer was defined as the reciprocal of the dilution reducing the virus plaque count by 90%.

We obtained descriptive and clinical data for case-patients, including age, gender, race/ethnicity, reported symptoms, symptom onset, and origin of infection, from Merlin, the Florida Department of Health surveillance system. We used Pearson χ² and Fisher exact tests to examine associations between demographics, symptomology, and Zika virus IgM results. We performed all statistical analyses with SAS statistical software version 9.4 (https://www.sas.com/en_us/software/sas9.html). This study was approved by the Florida Department of Health Institutional Review Board.

Of 352 eligible PCR-confirmed Zika virus disease case-patients, 62 (18%) were enrolled and provided follow-up serum specimens. The 62 enrolled participants and 290 eligible case-patients who were not enrolled were similar with regard to age, sex, race/ethnicity, and clinical manifestations; however, 55% of enrolled participants acquired their infections in Florida, compared with 45% of the unenrolled cases (Table 1).

Among the enrolled participants, 8 (13%) provided a specimen at 12 months after initial symptom onset, 1 (2%) at 13 months, 13 (21%) at 14 months, 21 (34%) at 15 months, 11 (18%) at 16 months, 3 (5%) at 17 months, 3 (5%) at 18 months, and 2 (3%) at 19 months. The median age of participants was 47 years (range 8–70 years); 60 (97%) were adults >18 years of age (Table 1). Overall, 32 (52%) participants were male, and 42 (68%) were Hispanic. Two (3%) participants reported only 1 of the 4 main symptoms (fever, maculopapular rash, arthralgia, and conjunctivitis) at the time of their initial Zika virus diagnosis; 15 (24%) reported 2, 32 (52%) reported 3, and 13 (21%) reported all 4. From case investigations, we determined that 34 (55%) participants acquired their Zika virus infection locally in Miami–Dade County.

Figure

Figure. Zika virus IgM results for 62 participants in Miami–Dade County, Florida, USA, with PCR-confirmed Zika virus disease by follow-up specimen collection month.

At follow-up, 45 (73%) patients had detectable Zika virus IgM, 12 (19%) had an equivocal result, and 5 (8%) were negative (all laboratory results provided in Appendix). Results by month since Zika virus symptom onset (Figure) showed that, overall, 39 (91%) of 43 specimens collected at 12–15 months postonset were IgM positive or equivocal, and 18 (95%) of 19 specimens collected at 16–19 months were positive or equivocal. No significant differences in IgM persistence were identified by age, gender, race/ethnicity, origin of infection, or time since illness onset (Table 2).

All participants had Zika virus neutralizing antibodies at 12–19 months after their acute illness, and 39 (63%) had dengue virus neutralizing antibody titers at follow-up. Using a definition of positive or equivocal Zika virus IgM with confirmatory Zika virus neutralizing antibodies, 57 (92%) would have had a diagnosis of recent Zika virus or flavivirus infection on the basis of results from their follow-up specimens (Table 3). Overall, regardless of Zika virus IgM results, 45 (73%) of the PCR-confirmed cases had Zika virus neutralizing antibody titers that were >4-fold higher than dengue virus titers. However, substantial cross-reactivity in neutralizing antibodies was still observed; 12 (19%) patients had a <4-fold difference between Zika virus and dengue virus titers, and 5 (8%) had dengue virus titers that were >4-fold higher than Zika virus titers.

These findings demonstrate that 73% of persons with PCR-confirmed symptomatic Zika virus disease still had positive IgM test results 12–19 months after their initial illness, and another 19% had equivocal results. Because all participants had Zika virus neutralizing antibodies, a high proportion (92%) would have had a recent Zika virus or flavivirus infection diagnosis on the basis of serologic testing performed at the follow-up time point. Current Zika virus testing guidance recommends Zika virus serologic diagnosis only for symptomatic patients with a clinically compatible Zika virus illness (3). However, given the limited specificity of the clinical symptoms associated with Zika virus disease, the prolonged detection of Zika virus IgM presents a particular challenge for serologic diagnosis in pregnant women, given the importance of determining if the infection occurred during the current pregnancy, and complicates the diagnosis of new Zika virus infections in locations with known previous outbreaks.

The prolonged detection of IgM after Zika virus infection is consistent with previous findings for related flaviviruses (9–13). In a study of patients with confirmed West Nile virus encephalitis, 9 (43%) of 21 patients had detectable IgM 300–400 days after onset of their acute illness, and 5 (42%) of 12 had IgM detected >500 days after onset (11). Among asymptomatic blood donors with West Nile virus viremia detected on routine screening, IgM was still detected an average of 156 days (95% CI 70–423 days) after the viremic donation (13). A similar finding was observed in a study looking at the immune response to yellow fever vaccine in adults, in which 29 (73%) of 40 persons had detectable levels of IgM 3–4 years postvaccination (12).

Serologic cross-reactivity has been demonstrated between Zika virus and related flaviviruses, including dengue virus, but the degree to which neutralizing antibody cross-reactivity limits the ability to identify the specific virus responsible for the current infection is unclear (6,23). One published report suggests that relative levels of neutralizing antibody titers can distinguish Zika virus from dengue virus infections, especially in specimens collected months after infection (19). However, we found substantial neutralizing antibody cross-reactivity in >25% of specimens collected >1 year after symptom onset.

These findings are subject to several limitations. This report presents data from a single follow-up specimen but how long IgM may persist after this timeframe remains unknown. We cannot exclude the possibility that some participants may have been reexposed to Zika virus or another flavivirus between their initial illness and follow-up testing. The timing of the follow-up specimen varied among the participants and was limited to 12–19 months following onset of Zika virus symptoms. This, coupled with the small number of specimens at some time points, prevented us from assessing possible trends in IgM persistence over time. The small sample size and lack of specimens from the acute illness also limited our ability to detect factors that may be associated with prolonged detection of IgM, including possible differences between primary and secondary infections. Our findings of prolonged IgM seropositivity are specific to the Centers for Disease Control and Prevention MAC-ELISA, which targets Zika virus premembrane and envelope glycoproteins; other IgM serologic assays targeting other proteins are currently available and may not produce comparable findings. Finally, because we enrolled only symptomatic disease case-patients, it is uncertain whether persons with asymptomatic infections would exhibit similar IgM persistence.

These findings support data for other flaviviruses and suggest that a substantial proportion of persons with Zika virus disease will still have detectable IgM 1–2 years after their initial infection. The results highlight the complexity of using serologic diagnosis to determine the specific timing of a recent infection, which is particularly important for pregnant women and challenging for residents of areas with previous or ongoing Zika virus activity. As such, the findings further support the current recommendations to use nucleic acid amplification for screening asymptomatic pregnant women with ongoing possible Zika virus exposure (3). Further study is needed to assess IgM persistence using other approved assays, determine the full duration of Zika virus IgM after infection, and evaluate possible differences in IgM duration following primary and secondary infections.