Use of FTA Cards To Transport Throat Swabs and Oral Fluid Samples for Molecular Detection and Genotyping of Measles and Rubella Viruses

A sample size of between 152 and 215 is required to estimate a binomial proportion with a desired precision (score with continuity correction, power analysis and sample size [PASS] v14) assuming a P value of 85%, alpha value of 0.05, and desired precision between 5% and 6%.

The study protocol was reviewed by the Associate Director for Science/Laboratory Science (ADS/ADLS) at the Center for Global Health, Centers for Disease Control and Prevention (CDC). The ADS/ADLS determined that, since the samples (serum, TS, and OF) were being collected during the standard measles case investigation, further review by the Institutional Review Board (IRB) of the CDC was not required. An explanation of the study was provided to every caregiver and verbal consent sought prior to sample collection.

Personnel at the Institut National de Recherche Biomédicale (INRB) in the Democratic Republic of the Congo were trained in sample collection for use of FTA cards and sent to outbreak sites to train local health facility staff and supervise sample collection. Specimens were collected prior to the implementation of measles mass vaccination campaigns using Sigma-Virocult viral collection and transportation systems (Fisher Scientific, Hampton, NH, USA) and Oracol collection devices (MMD, Malvern, UK). Tubes with 1 ml sterile saline, disposable pipettes (Globe Scientific, Paramus, NJ, USA), Whatman indicating FTA cards (Sigma-Aldrich, St. Louis, MO, USA), transport pouches for FTA cards (Sigma-Aldrich), tongue depressors (VWR, Radnor, PA, USA), and labels were provided to the health facilities. Posters describing sample collection and processing were displayed in participating health facilities (see Fig. S1 in the supplemental material). Case investigation forms were filled out for each suspected case, and data entry was performed at the INRB. From each suspected case, two TS and two OF samples were collected. The two TS samples were collected by holding the two applicators as one, and the two OF samples were collected, one from each side of the mouth. One TS and one OF sample were stored and transported at 4 to 8°C according to current measles surveillance guidelines (2). The second TS sample was replaced in its container and shaken vigorously for 1 min to elute the sample from the swab. The second OF sample was replaced in its container after the addition of 1 ml of saline and pushed down 10 times to elute the sample from the sponge. Next, 250 µl of the viral transport medium containing the TS eluate and 250 µl saline containing the OF eluate were transferred to one FTA card each using disposable pipettes. After drying for 1 h at ambient temperature, the two FTA cards were placed in a transport pouch. Pouches were transported in biohazard bags to the laboratory at INRB at ambient temperature. At INRB, pouches with FTA cards were stored at 4°C with desiccant until shipment to the CDC (Atlanta, GA). Sera were collected, stored, and transported according to current measles surveillance guidelines (2). Sera were tested at INRB for the presence of measles IgM using the Enzygnost anti-measles virus IgM kit (Siemens, Erlangen, Germany). Measles IgM-negative sera were tested for the presence of rubella IgM using the Enzygnost anti-rubella IgM kit (Siemens). TS samples were transferred to cryovials (VWR) and stored at −70°C until shipment to the CDC. OF samples were centrifuged at 2,000 rpm for 5 min at 4°C to collect the liquid from the sponge, transferred to cryovials, and stored at −70°C until shipment to the CDC. OF and TS samples were shipped to the CDC Atlanta on dry ice.

Office paper punches were used to cut five 6-mm disks from each FTA card. After punching each card, the punches were cleaned with 10% bleach and 80% ethanol. The five disks were transferred to a 1.5-ml centrifuge tube with disposable forceps. A modified version of the QIAamp Viral RNA Minikit (Qiagen, Hilden, Germany) extraction procedure was used to extract RNA from five disks per FTA card. One hundred fifty microliters phosphate-buffered saline (Fisher Scientific) was added to each tube, followed by 600 µl AVL buffer with carrier RNA. The disks were incubated for 10 min at room temperature, followed by the transfer of 700 µl of the liquid to a clean tube. The remaining steps of the procedure were performed according to the manufacturer’s recommendations. An RT-qPCR assay targeting the measles N gene was used to determine copy numbers of N gene-specific RNA (31). Sample RNA quality was assessed by RT-qPCR for the human RNase P gene (31). Eight (3.2%) of the samples that were RT-qPCR negative for measles had to be excluded because of insufficient RNA quality or quantity. Samples from suspected cases that did not have detectable measles RNA in any of the four samples and samples from suspected cases that were positive for RuV IgM were tested for the presence of rubella RNA using the CDC diagnostic rubella RT-qPCR assay. This assay used a primer and probe set located in a well-conserved region close to the 5′ terminus of the rubella virus genome and was duplexed to detect the RNase P cellular reference gene (unpublished data). Templates for sequencing and genotyping of MeV were generated as described previously (10). Templates for sequencing and genotyping of rubella virus were generated using a two-amplicon method or by nested-set amplification (32, 33). Sanger sequencing was performed on an ABI 3500 Genetic Analyzer. Phylogenetic analysis was performed using MEGA6 (rubella) or MEGA7 (measles) (34, 35). GenBank accession numbers are listed in the trees.

Microsoft Excel 2016 was used for data entry and to construct the boxplot. Analysis was completed in R 3.4 (36). Due to the absence of a gold standard test that is both 100% sensitive and specific, we present percent positive agreement and percent negative agreement with corresponding 95% Wilson (score) confidence intervals. A loess smoother, a locally weighted polynomial regression method, was plotted to compare RNA load between sample types by day since rash onset using a span value of 0.8. The gray bands in Fig. 1 represent the confidence intervals for the loess curve using a t approximation for the standard error (37, 38). For the figure, we added 0.1 to the copy numbers and then transformed to the log₁₀ scale. Time since rash onset was truncated at 25 days, and the points were jittered on the x axis to visualize points with identical values.

Sequences were deposited in GenBank under accession numbers MH752133 to MH752190 (measles) and MH654795 to MH654808 (rubella). These accession numbers are listed behind the WHO names in the phylogenetic trees.

Between March 2014 and September 2014, TS and OF samples were collected from 270 suspected measles cases. Sample collection took place at 38 collection sites in eight provinces in the Democratic Republic of the Congo. From each suspected case, two TS and two OF samples were collected. One each of the TS and OF samples were processed and transported according to the protocol recommended by the WHO (2), which specifies storage and transportation at 4 to 8°C. The second TS and OF samples were transferred to FTA cards at the collection site, followed by storage and transport at ambient temperature. These samples will be referred to as FTS and FOF. Of 270 suspected cases, 16 were excluded due to incomplete sample sets, insufficient RNA quality or quantity, missing patient information, or because they were positive for both measles and rubella. Sixteen suspected cases were excluded from the measles-specific analysis because they were positive for rubella IgM and confirmed as rubella cases. Samples from the remaining 238 patients were analyzed to evaluate the utility of FTA cards to transport samples for the detection of MeV RNA.

Of the 238 suspected cases included in the analysis, 185 (77.7%) were measles IgM positive, 8 (3.4%) were indeterminate, and 45 (18.9%) were measles IgM negative, reflecting the large measles outbreak that occurred in the Democratic Republic of the Congo in 2014. All IgM-positive and -indeterminate cases had detectable RNA either in the TS or OF sample; 96% had RNA detected in both. Therefore, the IgM-indeterminate patients were grouped with the IgM-positive cases in the following analyses. Of 112 (47.1%) suspected cases with samples collected 0 to 3 days after rash onset, 24 (21.4%) were IgM negative, but 18 (75.0%) of these had detectable MeV RNA in at least one sample. For all time points, 31 of 45 (69%) IgM-negative cases were positive by TS or OF, demonstrating the utility of molecular methods for case confirmation. Eleven suspected cases were IgM negative and had no detectable RNA in any sample. Samples from these 11 cases were collected 0 to 7 days after symptom onset. It is possible that these were not measles cases. Ninety-nine percent of the TS and 97.4% of the OF samples from IgM-positive or -indeterminate patients contained detectable MeV RNA (Table 1). The positive agreement was reduced to 79.8% and 85.5% for FTS and FOF, respectively, compared to IgM. The percentage of samples from IgM-negative suspected cases that did not contain detectable MeV RNA ranged from 42.2% to 53.3%.

For both TS and OF, 89.9% of the 238 suspected cases contained detectable MeV RNA, compared to 73.5% of FTS and 79.8% of FOF (Table 1). Both TS and OF were RNA positive in 204 (85.7%) cases and negative in 14 (5.9%) suspected cases (Table 2), indicating good agreement between these sample types. Twenty samples (8.4%) produced discordant results, resulting in a positive agreement of the OF compared to TS of 95.3% and a negative agreement of 58.3%. However, when comparing TS and FTS (Table 3), the number of discordant samples increased to 49 (20.6%). This increase was mostly due to a larger number of cases in which the TS was positive but the FTS was negative. Based on these data, the positive agreement between FTS and TS was 79.4% and the negative agreement was 79.2%. In the comparison of OF and FOF (Table 4), the number of discordant samples was 38 (16.0%), again due to the larger number of OF-positive but FOF-negative cases, with a positive agreement of FOF compared to OF of 85.5% and a negative agreement of 70.8%.

For each collection day and for all sample types, the amount of MeV RNA found in the samples varied over a wide range (Fig. 1). For the 24 cases whose samples were collected more than 7 days after rash onset, only three TS and three OF samples were RT-qPCR negative. While the FTS and FOF showed a similar broad distribution of RNA loads as the TS and OF samples, the median copy numbers per reaction were lower for every time point. Median copy numbers per reaction for TS and OF were 2.75 × 10⁵ (interquartile range, 7.70 × 10⁶) and 2.81 × 10⁵ (interquartile range, 6.26 × 10⁶), respectively. For FTS and FOF, the median copy numbers were 3.45 × 10³ (interquartile range, 1.16 × 10⁵) and 6.8 × 10³ (interquartile range, 3.65 × 10⁵), respectively, or 80- and 41-fold lower. When comparing RNA copy numbers, the volume of sample loaded on FTA cards and the fraction of the sample that is contained in the disks used for RNA extraction must be taken into account. Based on these considerations, the RNA extracted from FTA cards should have a 1.9-fold lower concentration than the corresponding TS or OF sample (data not shown). These results indicated that OF is as sensitive a sample for molecular detection as TS and that detection of MeV RNA from samples transported on FTA cards is less sensitive than from TS or OF samples.

The main advantage of using FTA cards for sample collection would be for genotyping when other methods for sample transport are unavailable. A subset of samples were selected for genotyping based on the location or time of the outbreak to obtain a more complete description of the distribution of circulating MeV in the Democratic Republic of the Congo. Therefore, some samples were chosen despite having viral RNA copy numbers below the limit of detection (LOD) for the measles genotyping RT-PCR of 1 × 10⁴ copies per reaction (Table 5) (10). Genotyping was successful for all TS and OF samples with copy numbers above the LOD but only for 86% and 80% of the FTS and FOF, respectively, with copy numbers above the LOD. While it was possible to obtain genotype information from samples with viral RNA copy numbers below the LOD, the success rate was low for all four sample types (Table 5). Using 10⁴ copies as a cutoff for successful genotyping, an estimated 184 (77.3%) of all TS, 169 OF (71.0%), 99 FTS (41.6%), and 98 FOF (41.3%) samples would provide genotyping results. These numbers account for the reduced success rate of genotyping from FTA cards and are likely slight underestimations, because suspected cases that could not be confirmed by either serology or RT-qPCR were included in the totals. These results indicate that FTA cards are useful for genotype analysis, albeit with a reduced likelihood of success compared to that for TS or OF samples.

All 58 genotyped samples belonged to genotype B3. The sequences can be divided into three separate clusters (Fig. 2). Forty-two sequences in cluster 1 were identical or very similar to the named strain, MVi/Harare.ZWE/38.09, with a maximum of four nucleotide differences. These sequences were found in five of eight provinces during epidemiologic weeks 12 to 36 of 2014, indicating a wide temporal and geographic distribution of this lineage in the Democratic Republic of the Congo. Cluster 2 comprised 14 sequences from Kasai Oriental and Katanga Provinces, and the two sequences in cluster 3 were obtained from Equateur Province. For seven cases, at least one sample from an FTA card and one TS or OF sample were genotyped. In all cases, the sequences were identical, demonstrating the reproducibility of the genotyping procedure when performed with RNA extracted from FTA cards.

Of 45 measles IgM-negative sera, 22 were positive for rubella IgM. After the exclusion of six cases for reasons described above, RNA samples from 16 rubella cases were tested by RT-qPCR to detect RuV RNA. Thirteen TS and 11 OF samples tested positive for RuV RNA compared to 8 FTS and 5 FOF. The amounts of RuV RNA varied significantly among the samples from these cases. Less RuV RNA was recovered from OF, FTS, and FOF than from TS (Fig. 3). All cases that had at least one positive RT-qPCR result had a positive RT-qPCR result for the TS. Genotyping was successful for 14 rubella cases, including one case that was excluded from the statistical analysis. Five cases occurred in May 2014 in Kasai Occidental Province in the town of Dibaya, and six cases were detected in July 2014 in Sud Kivu province in the town of Kabare, indicating rubella outbreaks in these towns. All genotyped RuVs belonged to the same lineage of genotype 2B (2BL2c) (39) and grouped with RuV genotype 2B viruses collected in the Democratic Republic of the Congo in 2012 and 2013 (Fig. 4) (33). The limited sequence variation (maximum, 1.1%) suggested continued circulation of these viruses during this time period.