Results and Discussion
Of the 97 blood samples of ideal or close to ideal quality (samples from animal trials and samples H81-H91, see supporting table 1), 79 were positive by qPCR. The INGEZIM ASFV CROM Ag LFA detected 61 positives, resulting in a sensitivity of 77.2%. No false positives occurred, hence 100% specificity was observed on this dataset. The performance was therefore in line with the study published by Sastre et al. (2016) where field samples of unimpaired quality were detected with roughly 67% sensitivity when compared to an OIE listed qPCR. Specificity was also close to 100%. Sastre et al. (2016) evaluated the LFA only with fresh samples, as represented by our group of samples derived from animal trials or hunted wild boar. With the ongoing circulation of ASF in European wild boar, however, virus detection in carcasses has become an important issue. Sample quality is then usually reduced due to decomposition effects, an aspect that has not yet been elucidated for the ASFV antigen LFA. All our 80 carcass-derived blood samples were obtained from ASF-positive wild boar and confirmed by qPCR with cq values ranging from 14 to 38 (see supporting table 2). Here, significant differences were observed between the samples that were previously frozen, and those that were not: in native samples tested without any modifications (C1-C16, n=16, see supporting table 3), the LFA delivered only two positive results (sensitivity of 12.5%). After freeze-thawing, testing of the same 16 samples in the LFA yielded seven positives (sensitivity of 43.75%). Surprisingly, one of the samples that had yielded a positive result in the native context was now tested negative. The increase of overall positive results is in accordance with the sensitivity of 48.75% we observed in all of the previously freeze-thawed carcass-derived samples (C1-C80), where 39 positives were detected by the LFA (see supporting table 2). Interestingly, however, we did not observe a better sensitivity after freeze-thawing in EDTA-blood samples of high quality in a previous study by our group (Pikalo et al., 2021). The positive effect of freeze-thawing is probably due to the fact that most of the virus in blood is associated with erythrocytes (Wardley & Wilkinson, 1977), and therefore, the destruction of blood cells during freeze-thawing results in a higher antigen availability for detection in the test, a process especially effective when erythrocytes are bound to clots in samples of reduced quality. No false positive reactions occurred with any sample types.
In our study, the INGEZIM ASFV CROM Ag assay could not deliver reliable results with native blood from carcasses. Particular samples with cq values as low as 15 (C4, see supporting table 3), indicating a considerable virus load in the carcass, still delivered negative results in the LFA.
While we observed increased sensitivity after erythrocytolysis by freeze-thawing (12.5% vs 44% sensitivity, samples C1 to C16, see supporting table 3), for the practical implementation of the assay in the field, of course, freezing cannot be an option due to the technical requirements not fitting a point-of-care application. Possible alternatives to freeze-thawing for erythrocytolysis could be the dilution of blood in aqua dest . or lysis buffer. On a very limited dataset (n=4), hypotonic lysis seemed to improve the results (3 FN native, 1 FN after water lysis, no FN after freeze-thawing; data not further shown). While both methods could be feasible under field conditions, the effects of this deviation from the manufacturer’s instructions on the assay should be elucidated and could be the basis of future optimization of the assay. After all, it must be noted that even with erythrocytolysis through freeze-thawing, we could only achieve a sensitivity of roughly 50% in carcass-derived samples, a value not fit fur purpose. Considering the negative impact of immune-complexes, samples were screened for the presence of antibodies. Only seven samples were positive for ASFV-specific antibodies and three delivered doubtful results in the antibody ELISA (see supporting tables 1 and 2). Eight of these samples were positive and two were false negative in the antigen-specific LFA (see supporting tables 1 and 2). While the small number does not allow for evaluation of possible interference, the principal functionality of the test in the presence of antibodies is indicated.
In general, the LFA was more reliable using samples with cq values below 30, indicating a rather high viral load. Of those samples derived from animal trials (n=64), 56 were true positive according to the rapid test, resulting in a sensitivity of 87.5% in that group. This goes along with observations in a previous study performed in our group, when the LFA was most sensitive during the clinical phase of ASF, at the peak of viral replication (Pikalo et al., 2021). In the present study, however, it was observed that the influences of clotting and decay in the carcass-derived samples seemed be able to outweigh the effects of higher viral loads, since here no clear correlation even with very low cq values and positive results in the LFA were observed (see supporting table 2).
Taking into consideration the differences between the highly amplifying qPCR and native antigen detection by LFA, the marked lower sensitivity in the later is to be expected. Still, the possibility for point-of-care testing holds a considerable advantage and on-site assays can provide a valuable additional diagnostic tool under certain circumstances. An acceptable sensitivity of the LFA was confirmed during the clinical phase of the disease, when fresh samples can be obtained from live animals or immediately after death. Here, the application of a rapid test could be of value in domestic pig holdings, when ASF is clinically suspected and live animals can be picked for sampling (given a careful interpretation of negative results in the LFA and still immediate initiation of laboratory diagnosis). Furthermore, epidemiological investigations can benefit from antigen assays for the on-site analysis of infected populations, when weaknesses in sensitivity are considered. However, with a sensitivity of roughly 50%, or even well below when no erythrocytolytic procedure is applied as proposed by the manufacturer’s instructions, our findings imply that the LFA has only very limited use for antigen detection in blood from carcasses after extended post mortem intervals. When resources are scarce and prioritization of diagnostic workflows is needed, the high specificity may allow for positive on-site results in the LFA to surrogate a laboratory confirmation. Negative results, however, must always be rated with high caution due to the low sensitivity we observed in samples of reduced quality. OIE listed methods such as qPCR remain the only safe and proven methods for the unreserved detection of an ASFV infection. Therefore, the on-site assay should be regarded as a complimentary option rather than a substitute to laboratory diagnosis for carcass testing.