Evidence of Ongoing Transmission of Zika Virus in Mérida, Mexico

ABSTRACT.

Since the Zika virus (ZIKV) pandemic in 2015–2017, there has been a near absence of reported cases in the Americas outside of Brazil. However, the conditions for Aedes-borne transmission persist in Latin America, and the threat of ZIKV transmission is increasing as population immunity wanes. Mexico has reported only 70 cases of laboratory-confirmed ZIKV infection since 2020, with no cases recorded in the Yucatán peninsula. Here, we provide evidence of active ZIKV transmission, despite the absence of official case reports, in the city of Mérida, Mexico, the capital of the state of Yucatán. Capitalizing on an existing cohort, we detected cases in participants with symptoms consistent with flavivirus infection from 2021 to 2022. Serum samples from suspected cases were tested for ZIKV RNA by polymerase chain reaction or ZIKV-reactive IgM by ELISA. To provide more specific evidence of exposure, focus reduction neutralization tests were performed on ELISA-positive samples. Overall, we observed 25 suspected ZIKV infections for an estimated incidence of 2.8 symptomatic cases per 1,000 persons per year. Our findings emphasize the continuing threat of ZIKV transmission in the setting of decreased surveillance and reporting.

INTRODUCTION

Zika virus (ZIKV) is an Aedes mosquito–borne virus (ABV) that was introduced to the Americas in 2015.¹ Zika virus spread rapidly through Latin America, leading to severe birth defects and the declaration of a Public Health Emergency of International Concern by the WHO in 2016.² Zika virus transmission in the Americas peaked in 2016 and decreased precipitously thereafter, with a near absence of reported cases in Central America and Mexico from 2018 to 2023.^3,4 However, because the conditions for ZIKV transmission persist and ZIKV immunity may wane over time at both the individual and population levels, resurgence remains a threat to the public health.^5,6 This threat is exacerbated by challenges in serologically differentiating between ZIKV and closely related flavivirus dengue virus (DENV) infections by routine serodiagnostic assays, the fact that the majority of ZIKV infections are asymptomatic, and the lack of resources available for surveilling ZIKV in endemic areas.^7–10

Zika virus was first detected in Mexico in January 2015, with 12,932 laboratory-confirmed (by IgM ELISA and polymerase chain reaction [PCR]) cases and 25,755 suspected cases from 2015 to 2019. These infections led to 51 cases of congenital Zika syndrome and coincided with the natal microcephaly case rate increasing from 3.7 per 100,000 births before 2015 to 11.5 per 100,000 from 2015 to 2017.¹¹ From January 2020 to June 2023, Mexico reported 70 ZIKV cases (with none reported in the state of Yucatán) (Figure 1), with a cumulative incidence of 0.01 case per 100,000 persons per year, leading to decreasing emphasis on ZIKV surveillance and countermeasures.⁴

Figure 1.

Incidence of ZIKV, DENV, and CHIKV cases in Mexico and the state of Yucatán reported to the Mexican Ministry of Health from January 2017 to June 2023.⁴ CHIKV = chikungunya virus; DENV = dengue virus; ZIKV = ZIKA virus.

Mérida, the capital of the Mexican state of Yucatán with a population of 892,000, has long been endemic for DENV and more recently experienced introductions of ZIKV and the alphavirus chikungunya virus (CHIKV).^12–14 Two cohort studies have endeavored to characterize the epidemiology of these viruses through active surveillance: Familias sin Dengue (FSD) (2015–2017, enrolling family members older than 2 years¹³) and the ongoing Targeted Indoor Residual Spraying (TIRS) trial (2020–2024, enrolling children aged 2–15 years¹⁵). Both studies identified infections by DENV and ZIKV as well as CHIKV through serological techniques. For both studies, symptomatic infections were detected during the arbovirus transmission season (epidemiological weeks 28–52) and asymptomatic cases through annual serological surveys. The FSD study observed peak ZIKV transmission in 2016 and detected 2.3 per 1,000 incident Zika infections per year.¹⁴ After 2017, detection of ZIKV cases decreased in Mérida with no reports of Zika cases in the city from 2018 to 2023 (Figure 1). Although these observations were presumably due to a decrease in ZIKV transmission and/or an increased proportion of asymptomatic ZIKV infections, decreased vigilance could also have contributed to the dearth of detected cases.

The potential for ZIKV transmission is likely increasing in the Americas as a result of waning immunity and favorable ecological factors. Indeed, ZIKV transmission has remained consistent in Brazil,³ and sporadic cases have been recorded in Bolivia and Belize.¹⁶ Our goal in this study was to determine whether ZIKV is circulating in the ABV-endemic region of Yucatán despite the absence of reported cases in the last 3 years. We capitalized on the existing TIRS cohort and performed in-depth serological analyses to quantify the prevalence of ZIKV antibodies and the incidence of ZIKV cases during 2021 and 2022 to assess the threat posed by ZIKV.

MATERIALS AND METHODS

Study site.

Mérida, the capital of the Mexican state of Yucatán, has a population of 892,000.¹⁷ The climate is tropical, with a mean annual temperature of 25.9°C and seasonal precipitation from July to December. Aedes-borne viruses (dengue, chikungunya, and Zika) in Mérida are transmitted in a highly seasonal pattern coinciding with the rainy season.^18,19 Aedes-borne virus transmission seasonality and incidence in Mérida have been previously characterized^19–21 and are consistent with high levels of transmission among the population (e.g., in 2015, the incidence of symptomatic ABVs was 3.5 cases per 1,000 person-years for DENV, 8.6 per 1,000 for chikungunya, and 2.3 per 1,000 for Zika^19–21). Emory University and the Collaborative Unit for Entomological Bioassays from the Autonomous University of Yucatán are the implementation partners for the TIRS trial. The Autonomous University of Yucatán and Yucatán State Laboratory for Public Health were implementation partners for the annual serological analyses of samples taken at yearly baselines.

Cohort design.

The TIRS trial is a two-arm, parallel, unblinded, cluster-randomized controlled trial conducted in 50 clusters of 5 × 5 city blocks each (1:1 allocation) within Mérida¹⁵. Cluster randomization was performed with an algorithm designed to maximize the average minimum pairwise distance between eligible neighborhood clusters. More information about the design of the TIRS trial¹⁵ and the covariate-constrained randomization procedure²² is described in previous publications.

Baseline sampling.

Upon informing participants of the study and giving them enough time to consider participating in the trial, epidemiological field teams consisting of a trained phlebotomist and a social worker collected whole blood samples in Vacutainer tubes (BD, Franklin Lakes, NJ). These baseline samples were taken from January to May, before ABV transmission season, in 2021 and 2022. Vacutainer tubes with blood samples were kept in coolers with cold blocks in the field for no more than 2 hours. Samples were transported to the project’s laboratory for centrifugation, and serum was aliquoted and stored at −80°C.

Active surveillance.

During July–December (epidemiological weeks 28–52), the epidemiological team performed household visits, phone calls, and cellular messaging to obtain weekly information about the health status of all participants and detect the occurrence of any ABV symptoms/signs. In addition, a 24/7 toll-free line was available for parents/tutors to report any illness. Each report was assessed by a physician, and if it was compatible with a suspected symptomatic ABV case (acute onset of fever or a nonfocal rash plus any additional symptom such as headache, conjunctivitis, arthralgia, or myalgia), peripheral blood samples were collected. Specimens obtained within 7 days of the onset of symptoms were PCR tested. If PCR was negative or the samples were obtained 8–14 days after the onset of symptoms, an IgM ELISA was performed. A convalescent sample was also collected 14–35 days after the acute sample from any PCR-negative participant. All febrile cases received medical advice and were referred to medical services according to their needs.

IgM ELISA.

The presence of anti-ZIKV IgM antibodies in human sera was determined using a commercial IgM ELISA (ZIKV Detect^™ 2.0 IgM Capture ELISA, InBios International, Seattle, WA) following manufacturer’s instructions. Briefly, each 96-well plate comes with a ready-to-use ZIKV antigen (ZIKV Ag), a cross-reactive control antigen (CCA), and a normal cell antigen (NCA) used to calculate a ZIKV immune status ratio (ZIKV ISR), which classifies a sample into three possible categories based on optical density (OD) values at 450 nm: 1) reactive for ZIKV IgM antibodies, if the specimen has a ZIKV ISR ≥1.70 and the specimen ZIKV Ag OD450 is ≥threshold ZIKV Ag OD450; 2) reactive for other flavivirus IgM antibodies, if the CCA/NCA ratio is ≥5.00; and 3) negative, if the specimen is not presumptive ZIKV positive and not presumptive other flavivirus positive (non-ZIKV). The reported sensitivity and specificity for this IgM ELISA are 90% and 96%, respectively (InBios International), whereas other studies have measured an overall sensitivity and specificity of 92.5% and 88.4%, respectively.²³

IgG ELISA.

To test yearly serological survey samples for seroconversion to ZIKV, we performed a capture ELISA with ZIKV and DENV1–4. The flavivirus cross-reactive antibody 4G2 was adsorbed to a 96-well half-area ELISA plate (Corning, Glendale, AZ) with 0.1 M carbonate buffer overnight at 4°C. Plates were washed, blocked with 3% nonfat milk in phosphate-buffered saline (PBS), and adsorbed with 10⁴ focus forming units (FFU) of ZIKV or DENV for 2 hours at room temperature. Unadsorbed virus was washed off, and plates were incubated with participant sera diluted 1:20 in blocking buffer for 1 hour at room temperature. Plates were washed and incubated with anti-human IgG conjugated to alkaline phosphatase (Thermo Fisher Scientific, Waltham, MA) for 30 minutes at room temperature. Alkaline phosphatase presence was detected with Fast p-nitrophenyl phosphate (Sigma-Aldrich, St. Louis, MO), and OD values were read at 405 nm after 20 minutes.

Real-time quantitative PCR.

To detect ZIKV genome in human sera, a Super Script III kit (Thermo Fisher Scientific) was used in a master mix containing 10.35 µL nuclease-free H₂O; 2 µL 2X Super Script III mix; 2 µL MgSO4 (50 µM); 0.15 µL deoxynucleotide triphosphates (stock 25 µM); 0.1 µL forward primer (Zika-F: CCG CTG CCC AAC ACA AG) and 0.1 µL reverse primer (Zika-R: CCA CTA ACG TTC TTT TGC AGA CAT), both at 100 µM stock concentration; 0.1 µL probe of Zika-Probe (FAM-AGC CTA CCT TGA CAA GCA GTC AGA CAC TCA A-BHQ1) (stock 100 µM); 0.2 µL of Super Script III enzyme; and 5 µL of RNA extracted using the QIAamp Viral mini RNA Kit (QIAGEN, Germantown, MD). Both primers and probes are described by Lanciotti et al.²⁴ A real-time quantitative PCR (RT-qPCR) protocol was run using a Bio-Rad (Hercules, CA) CFX96TM Real-Time System as follows: 50°C for 15 minutes (one cycle) and 95°C for 5 minutes (one cycle), followed by 40 cycles of 95°C for 30 seconds and 57.5°C for 30 seconds. Data analyses were performed using the Bio-Rad CFX master software v. 2.3. Human samples were considered positive if cycle threshold (CT) values were below 35. RNA extracted from the ZIKV prototype strain (MR-766) was used as a positive control.

Focus reduction neutralization tests.

To determine neutralizing antibody titers (focus reduction neutralization test [FRNT]₅₀ and FRNT₈₀, the serum dilution that exhibits 50% or 80% of maximum neutralization, respectively)²⁵ against DENV1–4 and Zika, 100 FFU of ZIKV (strain: MR766), DENV-1 (strain: West Pac 74), DENV-2 (strain: S-16803), DENV-3 (strain: ARB-1669/2019), or DENV-4 (strain: TVP-360) was incubated with 3-fold dilutions (dilution factors: 1:50–1:36,450) of participant serum for 1 hour at 37°C. The virus-serum mixtures were inoculated onto Vero cells, and infection was allowed to proceed for 1 hour at 37°C. Cells were overlaid with 2.5% carboxymethylcellulose (Sigma-Aldrich) in Dulbecco’s modified Eagle’s medium and incubated at 37°C for 72 hours. Plates were fixed with 2% paraformaldehyde (Sigma-Aldrich), and cells were permeabilized with 0.1% saponin in PBS with 0.1% bovine serum albumin and 1% fetal bovine serum for blocking. Viral envelope proteins were detected in infected cells with a flavivirus cross-reactive mouse monoclonal antibody (4G2 diluted at 1 µg/mL in permeabilization buffer), which was detected with anti-mouse IgG conjugated with horse radish peroxidase (The Jackson Laboratory, Bar Harbor, ME; diluted 1:2,000 in permeabilization buffer) and KBL TrueBlue substrate (SeraCare, Milford, MA). Foci were counted on a Biosys (Pasadena, CA) BioReader 7000E spot counting machine. Focus reduction neutralization test₅₀ and FRNT₈₀ titers were calculated using GraphPad Prism software (v. 6.07, Boston, MA) using nonlinear regression analysis. The FRNT titers were interpreted as below our assay’s detection limit if <50% neutralizing activity was calculated as ≤50, and titers with 50% neutralizing activity at the final dilution (36,450) were considered at the assay’s maximum and reported as FRNT₅₀ = 36,450. The identity of flaviviruses that had previously infected each participant was estimated by comparing relative FRNT values for each virus, with a  ≥4-fold difference in the FRNT considered a significant difference.²⁵ A participant with >4-fold higher FRNT against a single virus was considered previously exposed to that virus. Participants with FRNT values within a 4-fold difference were considered cross-reactive for classification.

Statistical analyses

The TIRS trial maintains a geocoded database with the locations of the houses of all children enrolled. In addition to demographic information, the data include geographic coordinates of each household. We used this information to map the distribution of cases using Quantum GIS (v. 3.28; https://qgis.org/es/site/). We applied Ripley’s K-function (also called second-order analysis) to quantify the spatial distribution of ZIKV suspected cases and identify whether they followed one of the three known distributions: randomness, clustering, or regularity.^14,15 The empirical K-function for a specified radius, d, is calculated with the following formula:

$$L\left(d\right)=\sqrt{\frac{A{\displaystyle \sum _{i=1}^N{\displaystyle \sum _{j=1}^{N}k\left(i,j\right)}}}{\pi N\left(N-1\right)}},$$

where A is the study area, N is the number of points, d is the distance, ${\displaystyle {\sum }_{i=1}^N{\displaystyle {\sum }_{j=1}^{N}k\left(i,j\right)}}$ is the number of j points within distance d of all i points, k(i,j) is the weight (calculated as k[i,j] = 1 if d[i,j] ≤ d, or k[i,j] = 0 otherwise). Basically, the K-function calculates, for a distance radius d, the average number of neighbors of a typical random point. Such value is then contrasted against the null hypothesis of complete spatial randomness (CSR). If for any distance the observed L(d) falls above or below the expected L(d), the null hypothesis of CSR can be rejected at an appropriate level of significance. The level of significance is determined by the confidence envelope, generated by Monte Carlo (MC) simulations. For this study, initial parameters for analysis were set as maximum distance = 3,000 m, number of increments to quantify L(d) = 30, number of simulations = 999. The results of the K-function are presented in a plot that shows the observed L(d), the expectation under CSR, and the 95% CI of CSR for the 999 MC simulations. All analyses were run with the spdep²⁶ package in R (v. 4.3.1).

RESULTS

Baseline FRNT-based studies of the TIRS cohort measured a ZIKV seroprevalence of ∼20% in the pediatric cohort (Earnest and Prokopec, unpublished data). We stratified these results by age and found most seropositive children (279/286, 98%) were older than 5 years, supporting the conclusion that there has been little transmission since 2018 (Table 1). We hypothesized that with 80% of the pediatric population being seronegative against ZIKV along with favorable conditions for ABV spread in Mérida, ZIKV could remain a threat to the population.

Table 1.

Flavivirus seroprevalence stratified by age in children from Mérida, Mexico, in 2020 measured from sampling of the TIRS trial

Age at Sampling (Years)	ZIKV		DENV		Cross-Reactive		Seronegative
	n	%	n	%	n	%	n	%
2–5	7	2.4	3	1.0	16	5.5	265	91.1
6–9	100	24.8	14	3.5	42	10.4	248	61.4
10–13	110	24.5	37	8.2	106	23.6	196	43.7
14–16	69	27.1	30	11.8	97	38.0	59	23.1
Total	286	20.4	84	6.0	261	18.7	768	54.9

DENV = dengue virus; TIRS = Targeted Indoor Residual Spraying; ZIKV = Zika virus.

To identify potential ZIKV cases, active surveillance was conducted during the 2021 and 2022 transmission seasons. This surveillance involved participants reporting ABV-associated symptoms (such as fever, conjunctivitis, headache, arthralgia, myalgia) to the TIRS project and epidemiological teams obtaining a weekly health status for all 4,461 participants. Symptom reports were assessed by a physician, and participants with suspected ABV infections provided blood samples (see Materials and Methods). After diagnostic criteria were established,²⁷ sera obtained within 7 days of the onset of symptoms were tested by real-time quantitative PCR (RT-qPCR) for ZIKV RNA. If the PCR was negative or the samples were obtained 8–14 days after symptom onset, an IgM ELISA was performed and a convalescent sample was collected 14–35 days later. Commercial anti-ZIKV IgM ELISA kits have a manufacturer-published sensitivity of 90% and specificity of 96%, though these assays are known to be less specific in DENV-endemic populations. Independent evaluations of this assay measured 91.5% and 94.2% sensitivities for primary and secondary infections, respectively, and specificities of 96.7% and 82.6% for primary and secondary infections, respectively.²³ All participants with potential ABV cases received medical advice and were referred to medical services according to their needs.

Of 155 participants who reported ABV symptoms in 2021, 17 (10.9%) exhibited IgM reactivity to ZIKV. In 2022, 182 participants reported symptoms consistent with acute ABV infection. Four tested positive for ZIKV RNA by RT-qPCR, and 48 (26.4%) exhibited IgM reactivity against ZIKV (Table 2). To identify potential asymptomatic cases, annual serum samples were tested for seroconversion with anti-ZIKV IgG antibodies by ELISA before and after the transmission seasons in 2021 and 2022. We identified three participants who seroconverted in an IgG ELISA against ZIKV and not DENV during this period. In total, we observed 72 potential cases that fit the criteria for suspected ZIKV infection in our cohort during active surveillance and from annual serological surveillance activities from 2021 and 2022 (Table 2).

Table 2.

PCR, ELISA, and FRNT results for potential ZIKV cases in the TIRS cohort observed by symptomatic active surveillance or from seroconversion in annual serological surveillance (passive surveillance) from 2021 and 2022

Serological test	2021 Active Surveillance	2022 Active Surveillance	2021–2022 Passive Surveillance	Total
ZIKV RT-qPCR+	0	4	NA	4
Anti-ZIKV IgM+	17	48	NA	65
FRNT
ZIKV	4	14	3	21
Cross-reactive	5	16	0	21
DENV	6	10	0	16
No exposure observed	2	8	0	10

DENV = dengue virus; FRNT = focus reduction neutralization test; NA = not applicable; RT-qPCR = real-time quantitative polymerase chain reaction; TIRS = Targeted Indoor Residual Spraying; ZIKV = Zika virus.

IgM antibodies against ZIKV can persist after ZIKV exposure for up to 2 years,^28,29 and the IgM ELISA used in this study has decreased specificity with secondary infection. Therefore, we used in-depth serological analyses to gain more evidence of active ZIKV infections in participants with suspected ZIKV cases. The 65 potential symptomatic cases identified by ZIKV IgM ELISA assays were tested by the more-specific FRNT assay; FRNT₅₀ titers were calculated against ZIKV and the four DENV serotypes for both pre- and postexposure sera (Supplemental Figures 1–4, Supplemental Table 1). Zika virus infection was determined to be likely when an increase in the ZIKV FRNT₅₀ titer was ≥4-fold the increase against all DENV serotypes. We identified four cases from 2021 and 14 cases from 2022 active surveillance that fit these criteria as well as the three suspected, asymptomatic cases from surveillance of annual serum samples obtained from all participants between 2021 and 2022 (Table 2, Figures 2 and 3). A total of five cases from 2021 and 16 cases from 2022 showed increases in both ZIKV and DENV neutralizing titers and were considered “cross-reactive,” and 16 cases from 2021 and 2022 were determined likely to be DENV using these criteria. Including the four RT-qPCR⁺ samples, we observed 25 likely ZIKV symptomatic or asymptomatic cases, resulting in an estimated incidence of 2.8 cases per 1,000 persons per year. In addition, 21 ZIKV/DENV cross-reactive cases were observed for an additional 2.4 cases per 1,000 persons per year. We performed a similar analysis using FRNT₈₀ titers and found that three likely ZIKV cases and three likely DENV cases were reclassified as cross-reactive (Supplemental Table 2). For the 22 likely ZIKV cases with symptoms data, the most common symptoms (occurring in +90% of cases) were myalgia, arthralgia, fever, and headache (Table 3).

Figure 2.

Heatmaps of the fold increase in FRNT₅₀ titers from the most recent baseline sample and either the convalescent sample (active surveillance) or the 2022 annual serological surveillance sample (passive surveillance) for each participant for ZIKV and the indicated DENV serotype (full dataset in Supplemental Figures 1–4, Supplemental Tables 1 and 2). Participants have been divided based on category of likely exposure: ZIKV, ZIKV/DENV cross-reactive, or DENV. DENV = dengue virus; FRNT = focus reduction neutralization test; PID = participant identification; ZIKV = Zika virus.

Figure 3.

(A) Map of Mérida indicating the hot spot area for dengue and other Aedes-borne viruses (light gray) and clusters of the TIRS trial cohort (dark gray), together with likely ZIKV cases (red) and ZIKV⁺/DENV⁺ cross-reactive cases (black). (B) Results from K-function analysis showing the observed value of the function (solid dark line) against the 95% CI for complete spatial randomness (dotted line). A pattern is random when the observed values are within the 95% CI band. DENV = dengue virus; TIRS = Targeted Indoor Residual Spraying; ZIKV = Zika virus.

Table 3.

Symptoms of 22 laboratory-confirmed ZIKV-infected participants reported to project physicians during 2021 and 2022

Symptom	n	%
Myalgia	21	95.5
Arthralgia	21	95.5
Fever	20	90.9
Headache	20	90.9
Cough	9	40.9
Odynophagia	9	40.9
Exanthema	6	27.3
Congestion	6	27.3
Diarrhea	4	18.2
Retro-orbital pain	3	13.6
Nausea	3	13.6
Vomiting	3	13.6
Conjunctival hemorrhage	1	4.5
Arthritis	1	4.5
Pruritus	0	0.0

ZIKV = Zika virus.

We mapped all likely (RT-qPCR⁺ or IgM⁺/FRNT⁺) and suspected (IgM⁺/FRNT cross-reactive) ZIKV cases and found that cases in the two categories displayed a similar distribution within the study site (Figure 3A). Furthermore, we applied a K-function analysis³⁰ to quantify the spatial pattern of likely cases (maximum search distance = 3,000 m, number of increments = 30, number of MC simulations = 999) and found that the distribution of likely ZIKV cases was widespread and randomly distributed throughout the surveilled regions of the city (Figure 3B).

DISCUSSION

In this report, we present robust evidence of recent ZIKV transmission dispersed throughout our study site in Mérida, Mexico, despite a lack of reported cases by the Ministry of Health. Our study also shows that the ability to monitor for ZIKV transmission in the post-2015 pandemic period may be hindered by challenges of case capture and confirmation and limitations of widely used serologic assays. Indeed, even thorough analysis of potential cases in our cohort resulted in many cross-reactive sera, which we cannot claim definitively as ZIKV cases. Along with the availability of sensitive and specific assays for measuring ZIKV seroprevalence, investment and innovation in public health surveillance strategies in ABV-endemic areas remain urgent needs.

The relatively high incidence and diffuse distribution of pediatric Zika illness in the 2020–2022 post-pandemic period, combined with potential low levels of ZIKV protective immunity in the pediatric population, represent a potential warning sign for the resurgence of ZIKV transmission in Mérida and other sites throughout the Americas. These findings are of particular relevance given that the absence of reporting of Zika illness by most countries may lead to the incorrect assumption that ZIKV has been locally eliminated. An increase in proactive surveillance for ZIKV infections and communication of existing risks to pregnant women or women planning to be pregnant who are living in or visiting endemic areas should be emphasized to reduce the likelihood of adverse pregnancy outcomes due to maternal ZIKV infection.

Our findings estimated a ZIKV incidence of 2.8–5.2 per 1,000 person-years in 2021 and 2022. These numbers represent primarily symptomatic cases within the cohort, as only three potential asymptomatic cases were identified from yearly serology surveillance of all participants. Generally, previous studies have measured variable levels of asymptomatic ZIKV infections from roughly 50% to 80%.³¹ This indicates that the incidence rates estimated in this study may be 2–3 times lower than the actual incidence. However, the fact that we observed so few asymptomatic cases suggests either that there were few such cases or that the diagnostic assays in this study were insufficient to identify them. Importantly, we only identified participants with IgG seroconversion as potential ZIKV cases and did not analyze participants who had anti-flavivirus antibodies at the time of baseline sampling. This could have led to underestimation owing to previous DENV infection and the presence of cross-reactive IgG antibodies. Another possibility is that the anti-ZIKV IgG waned in asymptomatic participants because of the length of time between the transmission season in May–October and baseline surveillance in February–April. Finally, there is a possibility that within this cohort the symptomatic-to-asymptomatic ratio is higher than in other cohorts. Further analyses and increased ZIKV surveillance within this cohort in the future may provide insight into these factors and provide a more accurate estimate of ZIKV incidence.

We observed only four participants with detectable levels of ZIKV RNA in their serum at the time of acute sampling. This result is likely due to low serum titers of ZIKV in this cohort of participants. These results highlight the challenges of accurately diagnosing ZIKV infections in populations with endemic DENV transmission. Relying on IgM ELISA and FRNT serological tests can introduce biases into this and similar studies, especially in areas in which DENV co-circulates. Zika virus IgM can persist in humans for up to 2 years after infection,^28,29 and cross-reactive antibodies are produced in people infected by either ZIKV or DENV. In the absence of PCR⁺ serum samples and without more in-depth analyses and development of improved diagnostic assays, individual cases of ZIKV cannot be diagnosed with complete certainty. However, in this study, we used a strict definition of potential ZIKV cases to increase confidence in case classification. Furthermore, previous studies have suggested that FRNTs are an accurate means of diagnosing ZIKV infections, especially in samples taken in convalescence.^24,32 Although it is clear that further development and availability of diagnostics is critical to optimally address the public health threat that ZIKV poses, the sum of our data strongly argues that ZIKV transmission is occurring in Yucatán and likely in other areas with suitable ABV transmission ecology. For these reasons, investing increased attention and resources in ZIKV surveillance is warranted.

Note: Supplemental material appears at www.ajtmh.org.