Partnerships Involved in Public Health Testing for Zika Virus in Florida,2016

Laboratory Tests

BPHL performed Zika virus testing in all 3 of its laboratories: Jacksonville, Miami, and Tampa. The laboratories established laboratory-developed tests in preparation for test requests because US Food and Drug Administration (FDA)–approved assays were not initially available. Laboratory-developed tests are validated according to internal procedures. These tests are often created in response to unmet testing needs and are not regulated by the FDA. ⁹ The laboratories in Jacksonville and Tampa established capacity for rRT-PCR testing in July 2015 to detect Zika virus nucleic acid in serum specimens from acute infections, which have the potential to be transmissible via the bite of a mosquito during the viremic (virus circulating in the blood) phase. During 2016, BPHL expanded the acceptable specimens to include urine, saliva, amniotic fluid, placental tissue, and semen. The laboratories used a laboratory-developed test based on a Centers for Disease Control and Prevention (CDC) method that was established in response to a Zika virus outbreak in Micronesia in 2007. This method defined 2 genome position target sequences for which rRT-PCR primers and probes were designed. ¹⁰ As suggested by CDC, BPHL modified 1 of the primer sequences to detect the Zika virus Asian genotype that was circulating in the Western hemisphere.

In January 2016, the Jacksonville and Tampa laboratories added capacity to perform Zika immunoglobulin M antibody testing to detect recent but nonviremic infections in serum and cerebrospinal fluid by using a laboratory-developed test based on a CDC method. Viral-specific immunoglobulin M antibody detection indicates a recent infection because these antibodies are produced during the first few days after symptom onset. ¹¹ The assay uses anti-immunoglobulin M antibody coated 96-well plates that capture immunoglobulin M antibody present in the specimen. A “sandwich” is then created with the addition of Zika virus noninfectious antigen that binds to Zika immunoglobulin M, followed by an enzyme-conjugated antiviral antibody that will cause a colorimetric change when a chromogenic substrate is added. The colorimetric change is detected with a spectrophotometer. This method was the basis for the FDA Emergency Use Authorization (EUA)–approved CDC immunoglobulin M antibody capture ELISA (MAC-ELISA) released in late February 2016. ¹¹ The FDA allows for emergency use of laboratory methods that are not fully approved or cleared when a diagnostic test is needed to respond to public health emergencies.

In March 2016, BPHL adopted the EUA-approved ELISA, which was used in conjunction with dengue immunoglobulin M and dengue immunoglobulin G ELISA when required diagnostically. BPHL used a laboratory-developed dengue MAC-ELISA assay based on the same CDC method as for the Zika MAC-ELISA with dengue virus noninfectious antigen. BPHL used a laboratory-developed dengue immunoglobulin G ELISA based on a CDC method. Viral-specific immunoglobulin G antibody detection indicates a past infection because these antibodies are produced later than immunoglobulin M and are lifelong but cross-reactive with other flavivirus immunoglobulin G. ¹² The assay is similar to the MAC-ELISA with sandwich of antigen, specimen antibody, and enzyme-conjugated anti-human immunoglobulin G antibody, but in a different sequence.

Also in March 2016, the FDA granted EUA approval of the Trioplex rRT-PCR assay, a multiplex assay with 1 genome position target for each of 3 viruses: dengue, chikungunya, and Zika, which makes it possible to distinguish Zika from clinically similar dengue and chikungunya infections. ¹¹ The Miami laboratory added capacity to perform the Trioplex rRT-PCR test in July 2016, using this assay to test for Zika virus nucleic acid in serum and other specimen types (ie, cerebrospinal fluid, whole blood, urine, and amniotic fluid). These other specimen types were indicated as potential other sources of Zika virus detection because of persistence of viral nucleic acid in them longer than in serum, but the FDA–EUA approval required that they be tested alongside a patient-matched serum because they had not been validated as extensively as serum for arbovirus detection. ¹³ Also, BPHL-Miami performed the Trioplex assay on urine survey specimens, a process not described in this case study. When the Zika virus testing demand began to decrease and high-capacity nucleic acid extraction instruments approved for the EUA-approved rRT-PCR assay were obtained, the laboratories in Jacksonville and Tampa verified the Trioplex assay and transitioned to it in December 2016.

CDC performed testing by using the FDA EUA–approved CDC MAC-ELISA and Trioplex rRT-PCR assay. Two to 4 shipments of specimens were sent to CDC-Atlanta weekly from August 9 through October 5, 2016, and distributed among 3 divisions for testing (Division of Health Care Quality Promotion, Division of Bacterial Diseases, and Division of Viral Diseases). CDC and BPHL established a file transfer protocol to transmit a secure Excel file with data on specimens and patient demographic characteristics. The contracted commercial laboratory performed testing by using FDA EUA–approved PCR assays on serum and urine (alongside a patient-matched serum) and immunoglobulin M antibody assays on serum.

A positive Zika virus RNA detection by PCR assay was confirmatory. For a patient with only a positive, equivocal, or inconclusive Zika virus immunoglobulin M result, CDC confirmed the presence of neutralizing antibodies by a plaque reduction neutralization test following test request and specimen referral from BPHL or the commercial laboratory. This test is a complex serologic assay requiring as long as several weeks to obtain a result; details of it are not a part of this case study and as of this writing had not been published.

Test Reports

BPHL generated specimen reports from its laboratory information management system software, used by BPHL member laboratories to manage specimens and associated data, and sent reports to non–Florida Department of Health submitters by postal mail or fax. Clinic providers in the Florida Department of Health received electronic laboratory results through messaging to the county health department health management system. The state-level Florida Department of Health Bureau of Epidemiology and county health department epidemiologists received electronic laboratory results through messaging to Merlin, Florida’s reportable disease surveillance system. ¹⁴ BPHL sent these electronic laboratory results directly from the laboratory information management system via health level 7 messaging, which is compliant with the most widely used international standard for transfer of clinical and administrative data between clinical and public health entities. ¹⁵ CDC sent specimen reports through a secure automated email attachment to BPHL, and BPHL manually entered results into the BPHL laboratory information management system and reported them as described previously. Commercial laboratory results were sent to the submitting provider and reported to Merlin through electronic laboratory reporting.

Test Volume and Turnaround Time

We obtained data on test volume from the BPHL laboratory information management system for specimens tested by BPHL and CDC and from Merlin for specimens tested in the commercial laboratory for Zika virus PCR and immunoglobulin M tests. We calculated median test turnaround times weekly from day of specimen receipt at BPHL or the commercial laboratory to day of BPHL test result release or commercial laboratory electronic laboratory results message creation, respectively. When calculating test turnaround times for specimens sent to CDC-Atlanta, we included the day of specimen receipt at BPHL to account for the time needed to prepare specimens for shipping. We also projected weekly test turnaround times for BPHL if the specimens had not been sent to CDC using turnaround times before CDC shipment, the number of specimens received, and estimated weekly test capacity.

Laboratory Triage and Testing Algorithms

BPHL developed algorithms for triaging and testing the large volume of specimens submitted to it. BPHL established a testing partnership with a large for-profit commercial laboratory through the Florida Department of Health Incident Management Team within 2 days of the free testing for pregnant women announcement on August 3, 2016. County health department clinics sent specimens from low-risk patients (asymptomatic pregnant patients with no exposure) to the commercial laboratory and specimens from high-risk patients (pregnant or not pregnant symptomatic patients and pregnant or not pregnant asymptomatic patients with exposure) to BPHL. Exposure was defined as living in or visiting the Miami-Dade transmission area, traveling to a Zika-endemic country, having sexual contact with a person with confirmed Zika virus infection, or being born to a Zika-positive or symptomatic mother. Patients were considered symptomatic if they had symptoms such as fever, rash, joint pain, conjunctivitis, Guillain-Barré syndrome, or fetal ultrasound indicating possible congenital Zika syndrome. All specimens from patients who lived or worked in Miami-Dade County were considered high risk and were sent to BPHL-Miami for testing or to a triage unit because most local Zika virus exposures occurred in Miami-Dade County. BPHL-Miami established this Zika triage unit with staff members deployed by the Florida Department of Health Incident Management Team and CDC Emergency Operations Center, a unit within CDC that provides resources in public health emergencies.

BPHL-Miami received specimens from the Florida Department of Health that were collected in Miami-Dade County’s pregnancy Zika testing clinics (special clinics set up for that purpose by the county health department), obstetrics and gynecology providers, and multiple hospitals. BPHL-Miami tested specimens from the pregnancy clinics (which were considered high risk because they came from the Miami-Dade transmission area) by Trioplex and forwarded to BPHL-Tampa for MAC-ELISA, and all other specimens went to the triage unit. Prior to receipt of the specimens in the triage unit, Florida Department of Health epidemiologists indicated that testing was warranted for public health purposes by approving testing based on patient symptom, pregnancy, and possible exposure status. When samples arrived in the laboratory, the triage unit verified prior approval and triaged the specimens based on the low-risk or high-risk criteria for shipment to the commercial laboratory, BPHL-Tampa, or BPHL-Jacksonville. BPHL did not test specimens that did not meet epidemiologic requirements. Specimens from patients who were symptomatic and pregnant with or without exposure were sent to BPHL-Tampa. Specimens from nonpregnant patients who were asymptomatic with exposure were sent to BPHL-Jacksonville. Specimens from pregnant patients without symptoms or exposure were sent to a commercial laboratory. This triage allowed BPHL to retain specimens from patients meeting high-risk criteria and distribute the specimen load between BPHL-Tampa and BPHL-Jacksonville based on staffing and capacity. BPHL-Tampa forwarded a subset of specimens from pregnant or not pregnant symptomatic patients to CDC from August 9 through October 5, 2016. BPHL-Tampa made the decision to send specimens to CDC-Atlanta based on the rapid increase in specimen submissions after the free testing announcement and an anticipated backlog (Figure 1).

OPEN IN VIEWER

Figure 1.

Triage algorithm for Zika virus specimens used by the Florida Department of Health Bureau of Public Health Laboratories (BPHL)–Miami triage unit, Florida, July–December 2016. Exposure was defined as living in or visiting the Miami-Dade transmission area, traveling to a Zika-endemic country, having sexual contact with a person with confirmed Zika virus infection, or being born to a Zika-positive or symptomatic mother. Patients were considered symptomatic if they had symptoms such as fever, rash, joint pain, conjunctivitis, Guillain-Barré syndrome, or fetal ultrasound indicating possible congenital Zika syndrome. “Has EPI stamp” means the specimen submission form had a stamp indicating testing approval by an epidemiologist (EPI). The triage unit verified the EPI stamp based on information on the specimen submission form and contacted an epidemiologist if the approval was questionable because of inaccurate forms, to confirm (or deny) the approval as appropriate. “EPI approved” means information on pregnancy status, symptoms, travel, and sexual history was required to approve or deny testing. Specimens from patients who were symptomatic and pregnant with or without exposure were sent to BPHL-Tampa. Specimens from nonpregnant patients who were asymptomatic with exposure were sent to BPHL-Jacksonville. Specimens from pregnant patients without symptoms or exposure were sent to a commercial laboratory. BPHL did not test specimens that did not meet epidemiologic requirements. BPHL-Tampa forwarded a subset of specimens from pregnant or not pregnant symptomatic patients to the Centers for Disease Control and Prevention (CDC) from August 9 through October 5, 2016.

BPHL developed testing algorithms based on guidance provided by CDC, although Florida made modifications based on local transmission and the decision to offer testing to all pregnant women in the state (Figures 2 and 3). These modifications included expanded serologic testing for suspected local cases (based on symptoms and no recent travel history outside of Florida), suspected Guillain-Barré cases (based on symptoms), and pregnant patients by Zika MAC-ELISA, dengue MAC-ELISA, and dengue immunoglobulin G ELISA testing regardless of Zika rRT-PCR result to obtain the full serologic profile for epidemiologic determination as an official case. Although immunoglobulin M levels vary, they may be detectable approximately 4 days and up to 12 weeks after symptom onset or exposure, and cross-reactivity between Zika and dengue viruses on serologic tests has been reported. ¹⁶ As another modification, BPHL performed rRT-PCR testing on all pregnant patients because the timing of a possible infection could not be established and the possibility existed of prolonged viremia during pregnancy, which was documented in other cases. ¹⁷ BPHL also performed rRT-PCR on suspected local cases because the time (or lack) of symptom onset or exposure may not have been clearly established and on suspected Guillain-Barré cases because little was known about the viremia in these patients.

OPEN IN VIEWER

Figure 2.

Testing algorithm for Zika virus specimens collected from patients who had possible Zika virus symptom onset or exposure within 14 days before specimen collection, used by the Florida Department of Health Bureau of Public Health Laboratories, Florida, 2016. The algorithm focused on real-time reverse-transcriptase polymerase chain reaction (rRT-PCR) assays using the laboratory-developed and US Food and Drug Administration (FDA) emergency use authorization (EUA)–approved tests on specimens collected <14 days after symptom onset or exposure because of a greater likelihood of patient viremia (virus circulating in the blood) and detectable Zika virus nucleic acid in specimens. For specimens collected <14 days after symptom onset, symptoms included fever, rash, joint pain, conjunctivitis, Guillain-Barré syndrome (GBS), or fetal ultrasound indicating possible congenital Zika syndrome. Exposure was defined as living in or visiting the Miami-Dade transmission area, traveling to a Zika-endemic country, having sexual contact with a person with confirmed Zika virus infection, or being born to a Zika-positive or symptomatic mother. Although specimens collected <14 days after symptom onset are most likely to be rRT-PCR positive, Zika virus detectability may vary by specimen type, specimen handling practices, and timing of the immune response. For specimens from suspected local cases (based on symptoms and no recent history of travel outside of Florida), suspected GBS cases (based on symptoms), and pregnant patients, additional serologic testing including immunoglobulin M antibody capture enzyme-linked immunosorbent assay (IgM MAC-ELISA), dengue MAC-ELISA, and dengue immunoglobulin G (IgG) ELISA regardless of Zika rRT-PCR result was performed to obtain the full serologic profile for epidemiologic determination as an official case. Specimens testing positive by Zika MAC-ELISA but negative by rRT-PCR from suspected local cases, suspected GBS cases, and pregnant patients were sent to the Centers for Disease Control and Prevention (CDC) for plaque reduction neutralization test (PRNT).

OPEN IN VIEWER

Figure 3.

Testing algorithm for Zika virus specimens collected from patients with possible Zika virus symptom onset or exposure ≥14 days before specimen collection, Florida Department of Health Bureau of Public Health Laboratories (BPHL), Florida, 2016. This algorithm focused on immunoglobulin M antibody capture enzyme-linked immunosorbent assay (IgM MAC-ELISA) testing on specimens collected ≥14 days after symptom onset or exposure due to a lower likelihood of patient viremia (virus circulating in the blood) but detectable Zika virus antibody, which develops later than detectable Zika virus nucleic acid in specimens. Symptoms included fever, rash, joint pain, conjunctivitis, Guillain-Barré syndrome (GBS), or fetal ultrasound indicating possible congenital Zika syndrome. Exposure was defined as living in or visiting the Miami-Dade transmission area, traveling to a Zika-endemic country, having sexual contact with a person with confirmed Zika virus infection, or being born to a Zika-positive or symptomatic mother. Specimens from suspected local cases (based on symptoms and no recent history of travel outside of Florida), suspected GBS cases (based on symptoms), and pregnant patients who tested positive, equivocal, or inconclusive by MAC-ELISA were also tested by real-time reverse-transcriptase polymerase chain reaction (rRT-PCR) using the laboratory-developed and US Food and Drug Administration (FDA) emergency use authorization (EUA)–approved tests for the following reasons: the timing of a possible infection could not be established and the possibility existed of prolonged viremia during pregnancy, the time (or lack) of symptom onset or exposure may not have been clearly established for the suspected local cases, little was known about the viremia in GBS patients. For these cases, BPHL also performed dengue MAC-ELISA and dengue immunoglobulin G (IgG) ELISA regardless of rRT-PCR result to obtain the full serologic profile for epidemiologic determination as an official case. Although IgM levels vary, they may be detectable approximately 4 days and up to 12 weeks after symptom onset, and cross-reactivity between Zika and dengue viruses on serologic tests has been reported. Specimens testing positive by Zika MAC-ELISA but negative or equivocal by rRT-PCR from suspected local cases, suspected GBS cases, and pregnant patients were sent to the Centers for Disease Control and Prevention (CDC) for plaque reduction neutralization test (PRNT). Abbreviation: CSF, cerebrospinal fluid.

BPHL followed 2 algorithms. The first algorithm, for specimens collected from patients with possible Zika virus symptom onset or exposure within 14 days prior to collection, focused on rRT-PCR testing using the laboratory-developed and FDA EUA–approved tests because of a greater likelihood of patient viremia and therefore detectable Zika virus nucleic acid in specimens. But Zika virus detectability may vary by specimen type, specimen handling practices, and timing of the immune response; therefore, for specimens from suspected local cases, suspected Guillain-Barré syndrome cases, and pregnant patients, BPHL performed the additional serologic testing described previously.

The second algorithm, for specimens collected from patients with possible Zika virus symptom onset or exposure ≥14 days prior to collection, focused on Zika MAC-ELISA testing by using the FDA EUA–approved test because of a lower likelihood of patient viremia at that time but a higher likelihood of the presence of detectable Zika virus antibody, which develops later in the course of infection, when Zika virus nucleic acid may no longer be detectable. Because Zika virus immunoglobulin M detection can be cross-reactive with dengue immunoglobulin M, ¹⁶ for specimens from suspected local cases, suspected Guillain-Barré syndrome cases, and pregnant patients, BPHL performed the additional dengue virus serologic testing described previously.

CDC-Atlanta performed testing following the Florida testing algorithms. The commercial laboratory performed FDA EUA–approved Zika virus PCR testing on serum and patient-matched urine specimens and FDA EUA–approved immunoglobulin M antibody testing on serum specimens unless indicated otherwise by the ordering submitter.

Test Volume

In 2016, all the laboratories combined tested 41  008 Zika specimens. In 2016, BPHL tested 17 749 Zika specimens, including 14 470 laboratory-developed PCR tests and FDA EUA–approved PCR tests and 8846 laboratory-developed immunoglobulin M antibody and FDA EUA–approved immunoglobulin M antibody tests. In 2016, the commercial laboratory tested 21 745 specimens, including 16 087 FDA EUA–approved PCR tests and 10 667 FDA EUA–approved immunoglobulin M antibody tests. In 2016, CDC-Atlanta tested 1514 specimens that BPHL sent from August 9 through October 5. During the peak specimen submission period, from July through November 2016, 82% (33 802/41 008) of all 2016 submissions were received. This period encompassed the peak of local transmission (July–October 2016), and the numbers of specimens tested by each entity during the period were 13  074 specimens (10 938 laboratory-developed PCR tests and FDA EUA–approved PCR tests and 6408 FDA EUA–approved immunoglobulin M antibody tests) at BPHL, 1514 specimens (1482 FDA EUA–approved PCR tests and 658 FDA EUA–approved immunoglobulin M antibody tests) at CDC-Atlanta, and 19  214 specimens (13 757 FDA EUA–approved PCR tests and 10 268 FDA EUA–approved immunoglobulin M antibody tests) at the commercial laboratory (Figure 4).

OPEN IN VIEWER

Figure 4.

Zika virus test volume and weekly median turnaround time (TAT) for (a) polymerase chain reaction (PCR) and (b) immunoglobulin M (IgM) antibody tests, Florida Department of Health Bureau of Public Health Laboratories (BPHL), Centers for Disease Control and Prevention (CDC), and a large for-profit commercial laboratory, Florida, July through November 2016.

Turnaround Times

From July through November 2016, median weekly turnaround times (the medians of turnaround times for all tests performed during a particular week) for Zika virus PCR testing ranged from 1 to 4 days; the total range of individual turnaround times during that entire period (hereinafter, “total range”) was 1 to 64 days at BPHL, 6 to 8 days (total range = 3-29 days) at CDC, and 3 to 5 days (total range = 2-13 days) at the commercial laboratory. The median weekly turnaround times for Zika virus immunoglobulin M antibody testing was 3 to 7 days (total range = 1-73 days) at BPHL, 7 to 19 days (total range = 3-48 days) at CDC, and 3 to 13 days (total range = 2-19 days) at the commercial laboratory (Figure 4). Peak turnaround times tended to coincide with peak testing volumes, especially in the first few weeks after the free testing announcement. CDC turnaround times were longer because of the time involved in preparing and shipping the specimens to CDC, including transcription of data on patient demographic characteristics from paper requisition forms to computer systems, resolution of missing or questionable demographic information, labeling and aliquoting of specimens into leak-proof tubes suitable for air transport and not shipping every day, and delays in the resulting communication process, including results lost in email transmission or emails missed in the BPHL result extraction process. The median turnaround time for specimens tested by CDC from time of receipt at CDC to result was 2 days for Zika virus PCR and 5 days for Zika virus immunoglobulin M. The median weekly turnaround times at BPHL were only slightly longer than pre-Zika virus testing levels, which ranged from 1 to 3 days for PCR and 2 to 5 days for immunoglobulin M antibody assays despite the laboratories performing testing beyond anticipated peak capacity. The projected weekly test turnaround time for BPHL had the specimens not been sent to CDC from August through October 2016 would have been 4 to 10 days for PCR and 5 to 15 days for immunoglobulin M antibody.