From October 1–November 4, 2016, ≈2–3 months after the peak of the Puerto Rico Zika outbreak (12), the Puerto Rico Department of Health recruited pregnant women during routine prenatal care visits at 7 clinical sites to provide serum, urine, and whole-blood specimens. Women provided verbal consent, and information was collected on demographic and clinical characteristics. Women with any laboratory evidence of ZIKV infection during pregnancy before recruitment or with any reported clinically compatible symptoms, including fever, rash, headache, eye pain, myalgia, or arthralgia, <7 days before specimen collection were excluded. This study was deemed nonresearch and exempt from institutional review board review.

At collection, specimens were refrigerated at 4°C, transported to the Centers for Disease Control and Prevention (CDC; San Juan, PR, USA) within 12 hours of collection, and stored according to Food and Drug Administration and CDC guidelines (13). For each specimen, 200 µL was tested by the Trioplex real-time reverse transcription PCR (rRT-PCR), by using previously described methods (13,14), including primer pairs specific to ZIKV, DENV, and chikungunya virus (CHIKV). According to surveillance systems in Puerto Rico, DENV and CHIKV circulation were minimal, and no confirmed cases were reported during the study period.

CDC (Fort Collins, CO, USA) performed quality-control testing by singleplex NAAT (13,14) on all ZIKV-positive specimens. In addition, whole-blood specimens were tested by the Hologic Aptima assay (Hologic, https://www.hologic.com) at Vitalant Research Institute (San Francisco, CA, USA) (8). Because results of quality-control testing were consistent and did not change findings, we report Trioplex assay results only.

We also tested serum specimens by the Zika virus IgM capture ELISA (Zika MAC-ELISA) (15). We used a positive-to-negative optical density ratio of >3 to determine sample positivity, suggesting previous ZIKV infection (15).

Among 514 pregnant women, all were asymptomatic during specimen collection; 14 were symptomatic 8–187 days before collection. The median age was 25 (range 15–43) years; specimen collection was evenly distributed by trimester of pregnancy (Table 1). Of the 1,521 specimens collected, all tested negative for DENV and CHIKV by Trioplex NAAT. Overall, 69 (13%) pregnant women tested positive for ZIKV by NAAT or IgM in >1 specimen (Table 2). A total of 24 (5%) participants tested positive for ZIKV by serum, urine, or whole-blood NAAT and had negative IgM results, whereas 41 (8%) participants had positive IgM and negative NAAT results. Only 4 (1%) participants had positive NAAT and IgM results, and 1 (<1%) woman was positive by NAAT on all specimens and IgM.

Figure

Figure. Comparison of positive NAAT results by specimen type for Zika virus infection among asymptomatic pregnant women, Puerto Rico, USA, October 1–November 4, 2016. NAAT, nucleic acid amplification testing.

Among 28 women who tested positive by NAAT, 8 were by whole blood, 10 by urine, and 20 by serum (Figure). Among the 8 women with NAAT-positive whole-blood specimens, none were positive by only whole blood; 5 tested positive by serum NAAT, 2 by urine NAAT, and 1 by urine and serum NAAT and IgM. Serum NAAT identified 13 positive results not detected by NAAT in another specimen, and urine NAAT identified 6 positive results not otherwise detected (Figure).

This study provides information about laboratory testing to maximize detection of ZIKV infection among asymptomatic pregnant women. In this small sample of ZIKV NAAT–positive asymptomatic pregnant women, no additional ZIKV-positive cases were identified by whole-blood NAAT beyond those identified through tests of other samples. This finding contrasts with other studies that note prolonged detection, higher viral load, and greater sensitivity of whole-blood NAAT versus NAAT on other specimens (6,8–10). Unlike previous studies that tested mostly symptomatic persons (6,10,11), we restricted our analysis to asymptomatic pregnant women.

All asymptomatic ZIKV-positive women had detectable ZIKV in NAAT of urine or serum samples in our study. Although previous studies detected ZIKV RNA in urine more frequently than in serum (5,7), we found that ZIKV RNA was detected in serum more frequently than in urine; 64% (18/28) tested positive by serum NAAT and negative by urine NAAT. However, serum and urine NAAT together are critical because urine alone identified only 21% (6/28) of pregnant women with a positive NAAT result.

This large study comparing NAAT for ZIKV on serum, urine, and whole-blood specimens is unique in that the study population is among asymptomatic pregnant women. Although studies have mentioned lack of overlap between different specimens tested by NAAT (7,10) and whole blood yielding fewer positive results in symptomatic persons (14), in this study NAAT on whole-blood specimens yielded fewer positive results than NAAT on serum or urine specimens among asymptomatic pregnant women. Because ZIKV-associated birth defects have been noted among asymptomatic pregnant women (1), identification of ZIKV infection is critical, especially when the virus is circulating in a community. Timely identification enables appropriate counseling and clinical management.

Our detection of acute ZIKV infections by NAAT is likely low because the study occurred 2–3 months after the peak of the Puerto Rico outbreak (12), and false-negative results in pregnant women are possible (11). Further, ZIKV RNA–positive results have been reported days or months after symptom onset or first positive test (6,9), and other cases were reported in Puerto Rico during the study period. Our analysis was among asymptomatic pregnant women, and we could not determine infection onset or whether infection occurred at all. False-positive results were also possible, but samples tested by singleplex NAAT revealed similar results, and results were independently validated in multiple laboratories.

These findings support CDC guidance to perform NAAT on asymptomatic pregnant women during outbreaks when ZIKV is widely circulating (3,4). Identification of infections among pregnant women who reside in or travel to areas at risk for ZIKV infection is critical for prenatal and postnatal counseling and clinical management (2,4). Although our understanding of viral persistence in various specimens is growing and the percentage positive in this study was small, NAAT of urine contributed to additional diagnoses, and NAAT on serum and urine combined yielded more positive results compared with whole-blood testing among asymptomatic pregnant women. Timely and accurate prenatal screening and notification of infection can optimize pregnancy and infant care during Zika outbreaks.