Oral swab samples and DBS were evaluated as alternatives to plasma for the diagnosis of acute dengue and for screening for past exposure to dengue virus. DENV-specific antibody and NS1 antigen were detected both in oral swab samples and DBS, using standard diagnostic assays. DENV RNA was amplified from DBS but not from oral swabs. There was little loss of sensitivity or specificity in using DBS compared with plasma. Oral swabs suffered from a loss of sensitivity, but retained good specificity.
Among laboratory-confirmed dengue patients, DENV-specific IgM or IgG antibodies were detectable in enrollment plasma samples in only 6% of cases, compared with 100% of discharge plasma samples. The sensitivity of IgM detection was reduced in oral swabs compared with plasma (68.7%), and varied substantially between the two study sites. The sensitivity of IgG detection was much higher (91.9%) than IgM, but again with a large difference between the study sites. Our findings fall within the range of sensitivity and specificity estimates published previously for detection of DENV-specific IgM and IgG from oral fluid samples,7–10,20
including both those that used a swab device and those where saliva was collected directly. We did not investigate DENV-specific IgA, but others have reported a high sensitivity and specificity for diagnosis of acute dengue using serum or DBS for IgA detection (though poor results using saliva)10
and a dengue IgA rapid test is commercially available (ASSURE; MP Biomedicals, Singapore). Among patients who were NS1 antigen positive in plasma samples collected at hospital admission, NS1 antigen was also detectable in the corresponding oral swab sample in almost two-thirds of patients. To our knowledge, this is the first study to show that NS1 antigen can be detected in oral fluid collected from acute dengue patients. Although there was a substantial loss of sensitivity, this provides proof of principle that early diagnosis of dengue by NS1 detection can be done from a non-invasive oral swab sample. In patients from whom a venous blood sample cannot be collected, the availability of an oral swab sample would allow a specific, although less sensitive, diagnosis of dengue to be made.
We used standard assay procedures, using undiluted eluates from oral swabs and DBS; improvements to sensitivity may be possible with further optimization of the serological and NS1 antigen assays for oral swab samples. The performance of the oral swab sample depends on adequate saturation of the swab with fluid from the tooth-gum interface, by swabbing for 1–2 minutes. Standardized training in collection technique was provided to our study staff, but the difference in sensitivity estimates between the two sites in our study may represent differences in collection method, including inadequate location or duration of swabbing.
Our findings support existing evidence that dengue virus antibodies, NS1 antigen, and viral RNA can be detected from dried whole blood eluted from filter paper.10,15–17
Furthermore, we show a high level of sensitivity and specificity for DBS in each diagnostic assay, compared with plasma. In IgM/IgG serology, several equivocal results among a small total number of DBS samples with corresponding negative plasma serology resulted in imperfect specificity. Ideally, we would have repeated the serology on these equivocal DBS samples to classify them as positive or negative, but because the DBS eluate is used undiluted in the assay and only one blood spot was collected per patient per assay there was insufficient specimen remaining to repeat. The high sensitivity we observed for detection of DENV RNA in DBS is comparable with other reports,15–17
although we did show a lower quantitative sensitivity in DBS than plasma, using the inverse of the assay cycle threshold as a proxy measure for RNA quantity. This is to be expected because the 20 μL of whole blood in one DBS is equivalent to ~10 μL of plasma, compared with 140 μL of plasma used in the RNA extraction and RT-PCR. The implication of this is that DBS represent an adequate alternative to plasma for amplification of DENV RNA from acute patients with high viremia levels, but the sample volume they contain may be insufficient for detection of lower viral loads in studies of mild or asymptomatic DENV infections outside a hospital setting, unless RNA extraction can be performed on several blood spots combined from one patient. Furthermore, the patient population in which DBS were evaluated was enrolled on the basis of a positive NS1 rapid test and, because it is known that NS1-positive patients have on average higher viral loads than NS1-negative patients, it is conceivable that the sensitivity of RT-PCR that we report is higher than might be seen in a broader dengue patient population.
A possible limitation of our sample set was that the DBS were derived from venous blood samples spotted onto filter paper in the laboratory, rather than directly from finger prick. This was done to minimize the demand on patients of additional sampling, but means that our results may not reflect exactly the performance of capillary blood samples in these diagnostic assays. The one published comparison of venous and capillary blood samples in dengue diagnostics15
reported that DENV RNA was detectable by RT-PCR in convalescent capillary blood samples from four patients in whom the corresponding venous blood sample was negative, indicating that we might expect an equivalent or even higher sensitivity for DENV RNA detection using finger-prick blood spots than what we observed.
Seroprevalence studies of DENV exposure can provide valuable information on the spatial range and intensity of DENV transmission including, if repeated age-stratified sample sets are available, inferences on temporal changes in transmission intensity.21–24
A barrier to such studies on a large scale is the collection of blood samples from healthy individuals, particularly children. We evaluated the detection of DENV-specific IgG in oral swabs collected from healthy volunteers, as a marker of previous DENV exposure, and observed a substantial loss of sensitivity in the oral swabs, compared with plasma. Perhaps more importantly the oral swabs suffered from poor specificity, although the total number of DENV-naive volunteers was small.
In summary, we found that DBS can be used in standard serological, molecular, and antigen detection assays for diagnosis of acute dengue, with minimal loss of sensitivity or specificity compared with plasma. We found oral swabs to have inadequate sensitivity and specificity for use in population-based studies of prior DENV exposure, however their performance for detection of DENV-specific IgM, IgG, and NS1 antigen in acute patients may make them an attractive, non-invasive option where venous blood cannot be collected, or in population-based surveillance of acute or recent infection.