SARS-CoV-2 detection methods
Let us start right in the middle with a quote from Michael Mina, MD, PhD, epidemiologist and immunologist at Harvard University:
So why are we not using it? To better understand how the two main methods work, many of you will need a quick introduction.
Let us assume that viral spread is so rapid that hospitalizations could overwhelm intensive care units in the close future, so we try to slow transmission even beyond the obvious big first step of banning large indoor gatherings. The traditional way of achieving this goal is via isolating infectious individuals, especially those who still meet many people indoors, usually due to their work settings. At the same time, it would be counterproductive to quarantine a lot of people who are unlikely to be infectious, because that causes substantial harm without meaningful benefits. This is the basic problem. A virus detection method that is ideal for these specific goals is
· sensitive enough to catch the (vast) majority of infectious people, especially those that release high viral loads, the potential superspreaders.
· ignorant to marking non-infectious people as positives.
· that provides results in a very short time, preferably on the site of measurement.
· cheap, so that it can follow through different stages of the infection, which typically present with different viral loads.
· preferably very easy to use, perhaps even at-home testing is feasible.
Now, let us see how the currently used RT-PCR fulfills those criteria:
The PCR method in general was not designed to detect live virus, rather viral nucleic acid fragments. It amplifies the mentioned segments in so-called cycles, and when it stops there is a preset number of cycles at which the color, which signals positivity, is present in the necessary strength or not. This preset number is the cycle threshold or Ct. In fact, this molecular biology tool is so good at doing the DNA amplification that picking up trace amounts of nucleic acid is a basic characteristic of the process. This eventually means that when the PCR reports a positive result that has no meaning associated to infectiousness. A positive result only means that the sample contained some matching nucleic acid sequences. Now, is there anything in the protocol that could be used for correlating to infectiousness? Yes, the cycle threshold. Unfortunately, there are some caveats: various device manufacturers have different protocols as to how many cycles should be run and what the Ct should be set to. This makes worldwide comparisons somewhat difficult. However, there are still some studies that set rather good thresholds and correlations. Still, this is rarely used in daily practice. To find out the reason read on until I discuss the typical opinion of an ‘expert’ epidemiologist, the kind that has been involved in policy making. Unfortunately.
This is where the PCR really shines, but in a bad way. Its very high sensitivity, and the high Ct values used, but not reported with the test in many countries/states make it impossible to know whether the given person is infectious or not.
As the procedure requires sophisticated lab and experienced personnel (false negatives usually come from inexperienced sampling, while false positives from poor sample handling), return time is rarely within the same day, often after 48 hours or more. Especially when a lot of tests are taken and processed in the middle of a surge. This is not at all in line with the potentially quick changes in infectiousness. There are reports of people who infected a dozen others on a long flight, even if they were non-positive 48 hours prior to boarding. What if a test is done right after going through security check and the result is there by boarding time?
The RT-PCR is not particularly cheap, a single test can cost up to 100 US dollars.
As described in #3 above, this is not something anybody can do and at any time.
Quick antigen (Ag) test concepts, alignment with criteria 1-5:
· Sensitivity seems just about the gold spot. It misses very few infectious, especially not potential superspreader people. More about this below.
· Higher quality quick antigen tests do not report all the post-infection people positive 2-12 (!) weeks later. However, they could also report presymptomatic persons negative. This can easily be fixed by applying another test 24 hours later or as it makes the most sense.
· Sampling, running the assay and evaluating results takes no more than 15-20 minutes. Running duplicates can be done in 20 minutes with some experience. Speed of getting results is simply unmatched.
· Even when 2. and 3. are combined, the four quick tests, i.e. repeat duplicates, can be of lower cost than a single PCR.
· Sampling may be the critical step, carefully phrased and visualized user manual/quick guide is sufficient to run the test in a proper way for already the very first time.
Accidentally, I bumped into this ‘rebuttal’ of Dr. Mina’s above editorial. It perfectly shows what typical governmental advisors get wrong, or perhaps they do understand the problem, just do not want to take responsibility for providing a full cost-benefit analysis. The main argument is about sensitivity. The Columbia University epidemiologist is concerned that the PCR is already not sensitive enough, producing high numbers of false negatives. An even less sensitive Ag test could eventually leave even more virus positive people running around. Where the argument fails is again the ultimate goal of infectious epidemiology during a deadly pandemic.
Are we after positive test results or infectious people? Are the two the same, and if not, what is the overlap? After a year into this epidemic, not to mention the research of previous decades, we do know that viral load in the upper respiratory tract correlates well with infectiousness. What is more, recent research suggests that saliva could be at least as good a sampling specimen, likely do to its mixing with both nasopharyngeal and lung secretions (the “mucus elevator”). Viral culture studies suggest that above a Ct barely anybody is infectious, but the zone of 25-35 is already a very mixed bag, with strong tendency of shedding meaningful amounts of “live” viruses at higher Ct positivity. Ct above 35 almost always catch (OK, describe) asymptomatic people, and a very recent study even showed that there is usually not even an overlap of presymptomatic (pre-disease) with asymptomatic (post-disease) people from a Ct perspective. The former tends to measure at 34-37 cycles, the latter always above 37. A repeat test could therefore easily distinguish between these two, but this is a totally different topic.
The message I must hammer home is that in a pandemic we are after infectious people for multiple reasons. The suggestion that catching most or even all post-infection people who keep shedding viral fragments is an important management tool while quick tests are useful for epidemiology could not be further from the truth. Testing the same people every other day with rapid Ag tests over a month still costs substantially less than testing them once a week with PCR. When do you have greater chances to check a contagious person early, if tests are done every other day or if a PCR are done once a week? Yes, that is my conclusion as well.
(If references to the claims made are required, please search my Twitter feed @gerdosi. For example, “asymptomatic” and “presymptomatic” after my handle should yield the most important papers I’ve shared so far. For more information about quick Ag tests, Dr. Mina’s Twitter feed is a good source @michaelmina_lab.
