Testing, Testing, Testing…
The uncertainties about SARS-CoV-2 include critical questions such as what the death rate is for infected people, and what the infection rate is for the general public. This makes testing for infection of critical importance. There are numerous ways to establish the presence of infection. One is recovery of live virus, which in case of SARS-CoV-2 requires culturing the virus in biosafety level 3 containment. Although this would seem to be a gold standard, BSL3 facilities are not widely available, so this type of test is not routinely done. Another is to test for viral proteins, known as antigen testing. This seems to be amenable to high throughput testing, but reliable diagnostic tests have not yet become available. A third is to detect and measure viral RNA, generally done by PCR amplification following reverse transcription of viral RNA. It can be performed relatively readily in high quantity, but it is only an indirect indication of live virus. Even badly fragmented RNA can register as a positive, and positive results simply indicate that virus is being shed from somewhere within a patient, or has been recently. Thus, although recovered patients and asymptomatic people can be viral RNA positive, it is unclear as to what extent they can actually transmit virus without confirming live virus by isolation in a BSL3 facility. Fourth, serologic tests for antibodies to the virus can be performed on a massive scale, but depend critically on what the false positive and false negative rates are for a given test, and these can vary considerably among different test kits. The presence of antibodies means a person has been or is infected but may not still be infected at the time of the test. Fifth, PCR tests similar to viral RNA tests can detect spliced or subgenomic RNA. In infected cells, full length viral RNA is spliced into a variety of smaller RNAs in order to encode many of the viral proteins. Thus, the presence of spliced subgenomic RNA indicates the presence of currently infected cells somewhere within the test subject. This process is somewhat labor intensive.
- Of than these types of tests, the two that are amenable to mass testing and currently being used are viral RNA analysis and serologic tests, which are being widely and increasingly employed. To date, by far most testing that has been reported is for viral RNA. RNA is usually quantified based upon how many cycles of polymerase chain reaction are required to generate a detectable signal. These types of tests are highly specific (defined as the rate of false positives), but may lack sensitivity (defined as the rate of false negatives. This is not because of an intrinsic lack of sensitivity, but is because detection depends on the site from which the tested sample is taken. While throat and nasal swabs give reasonable sensitivity, bronchial alveolar lavage would be more sensitive, but collection is probably too onerous to be practicable. Saliva has been suggested as being conveniently obtained, but provides somewhat less sensitivity than nasopharyngeal swabs.(1) Wolfel et al.(2) performed RNA analysis of a number of tissues from moderately sick infected individuals. Swabs, saliva and stool contained viral RNA, although no live virus could be obtained from stool samples. The swabs and saliva contained live virus. Live virus could not be obtained after 8 days, although viral RNA remained positive. Urine and serum samples were never positive for viral RNA. Thus, detection of RNA in a sample is not tantamount to detection of infectious virus in the sample, or even in the tested individual.
Serologic tests for antibodies are becoming widely available and have the potential to make widespread testing much faster. As mentioned above, the presence of antibodies can indicate either that a person is infected or has been infected, and so does not necessarily indicate the presence of live virus. Following infection, IgM is the first antibody class produced, and so is more of an indicator of recent infection than is IgG, which is produced later. IgG tends to persist longer than IgM, and more sensitive to detect than past infections. Some antibody tests detect both classes; others only one. Samples collected include drawn blood or finger pricks. Widespread use of these tests has been hampered somewhat by the large number of tests of varying quality that are available; it is currently somewhat of a Wild West situation.
The specificity and sensitivity of these tests, defined respectively as the rate of false positives and false negatives, is critical for their usefulness, depending on whether they’re used for mass screening or diagnostic purposes. For diagnostic testing, sensitivity and specificity are obviously both important. For mass testing, specificity is absolutely critical. Let’s consider using a test that is 95% specific for a population of 100 people that is 5% infected. The test will detect the 5 infected people. In addition, ~5 uninfected people will test positive, leading to the conclusion that 10% of the population is infected, even though the true infection rate is only half that. For reference, the test made by Abbot, which will be used in mass testing, has been reported as 100% sensitive, 99.5% specific. The other test that is likely to be used in mass testing is one by Roche. As of this writing, the accuracy and sensitivity data are not available, but will presumably be similarly high. Consideration of specificity should be taken into account when evaluating reports of unexpectedly high seroprevalence such as the study from Santa Clara County. The article can be accessed here, along with a number of critiques. Mass testing in the future should lend a lot more clarity to the true situation. It would be of interest to follow up positive serologic tests with confirmatory tests for RNA.
- E. Williams, K. Bond, B. Zhang, M. Putland, D. A. Williamson, Saliva as a non-invasive specimen for detection of SARS-CoV-2. J Clin Microbiol, (2020).
- R. Wolfel et al., Virological assessment of hospitalized patients with COVID-2019. Nature, (2020).