Diagnosis of COVID19 : understanding PCR tests, their interpretation and limitations

PCR uses a unique principle: the target of the test, a fragment of viral RNA, is massively amplified to allow its detection. During the analysis, an enzymatic reaction associated with temperature variation "cycles" allows a series of successive "replications" of the targeted nucleic acid. Each cycle corresponds to a theoretical multiplication of the target by 2. Thus, the target is multiplied by 2 in one cycle, by 4 in 2 cycles, by 8 in 3 cycles, by 16 in 4 cycles, and so on exponentially. Amplification is generally practiced over 40 cycles, producing a theoretical amplification of 2^40, about a thousand billion times! As replication is not 100% efficient, the target is amplified about a million times, which allows to detect less than a dozen RNA fragments in the analyzed volume.

When viral nucleic acid is detectable after a small number of cycles, it means that the amount of viral RNA in the starting sample is large. On the contrary, when it takes many replication cycles to detect viral RNA, it means that the starting sample contains a very small amount of target RNA. The number of cycles, or Ct, which stands for cycle time, is used to define, at least semi-quantitatively, the amount of RNA present in the starting sample. Thus, a small Ct corresponds to a large number of copies, a large Ct to a small number of copies.

This spectacular sensitivity is not without drawbacks and requires special precautions. Indeed, a positive sample amplified a million times contains a very high concentration of target and the risk of it contaminating (carrying over) other samples is particularly high. Laboratory overcrowding can further increase this risk and lead to accidental false positives. Under these conditions, it is important that positive results are confirmed by a second test, especially when a positive test has significant medical, professional, or isolation-related consequences.

The second important issue concerning PCR, again because of its spectacular sensitivity, is about clinical value. A perfectly asymptomatic subject with a positive PCR test cannot be described as "sick", as we read in the media reporting the progression of the epidemic! Can we even speak of a "case" for a perfectly healthy and, probably immune subject ? Yet this is the term used in official counts. Aren't we forgetting the patient to focus on the technology? Is it an epidemic of RNA in tubes that we are monitoring, or the progression of a serious and potentially fatal disease?

Recent publications point out that the dose detectable by PCR is lower than the infectious or contagious dose: no infectious virus could be found by cell culture in asymptomatic patients with positive PCR tests with high Ct. Following these results, the question of the Ct threshold for reporting a positive specimen is being debated. Is it possible to report a negative result in an asymptomatic subject whose positivity appears beyond 35 cycles? If not, is it useful to retest these samples? As is often the case in medical diagnosis, when a threshold of positivity is determined, should the sensitivity or specificity of the test be given priority?

Moreover, a sample confirmed positive from an analytical point of view remains a false positive from the clinical point of view, if the person tested is in perfect health, sometimes ready to face a professional tennis or soccer competition ! The question becomes only that of its potential contagiousness. The possible transmission by asymptomatic subjects, who without being themselves in danger, could represent one for others requires further analysis.

To adress this question, it is important to reason quantitatively. Virology is not all or nothing! During acute viral infections, the risk of contagion and the severity of the infection vary according to the quantity of virus present in the body and its excretion into the external environment. A few copies of viruses lurking in the sinuses do not have the dangerousness of a million thrown up by repeated cough. An asymptomatic person produces fewer viruses than a symptomatic person and secretes them less infectious doses to the outside. Simply put, the amount of virus produced and therefore the risk of contagion correlates with the severity of symptoms. Even if it is not zero, the risk of transmission is therefore likely to be low for an asymptomatic person. Unfortunately, repeating over and over again that contagion from a perfectly healthy subject is possible without any precision on the level of risk obviously leads to taking measures that are disproportionate to the risk.

 Similarly, the "screen-isolate" strategy is not realistic when screening is not sufficiently reliable and especially when the virus is already widespread in the population. It is far too late to apply a method designed to stop an epidemic at birth. We cannot stop a virus that is already everywhere with a strainer that is 25% full of holes and blocked in places. Is the failure of the current strategy linked to its na?ve and inapplicable design or should it be blamed on the bad behavior of citizens? If, as we are currently observing, viral spread resumes, should we screen more massively or revise the strategy for protecting the population?

This is not a scientific question. It depends on the risks acceptable to an individual or a group. If we are in search of minimal risk, close to zero, because the risk has not been quantified, or for reasons of legal liability, we must take maximum precautions ... If we accept even a low risk, we can take back certain liberties and focus protection on those who really need it. The scientist must measure the magnitude of the risks and not simply state that an adverse event is "possible". However, it is not his or her role to decide whether these risks can be taken by others.