We have talked about polymerase chain reaction (PCR) tests before. But the ‘in the wild’ diagnostic methods are slightly different from what is used in the lab.
There are multiple ways to visualise DNA. Traditionally, DNA acts as a template for the PCR, and the result is visualised on a gel electrophoresis.
The problem with testing for COVID-19 is that viruses don’t have DNA. They have RNA. And that means the process requires an extra step.
In this case, we use a special PCR called reverse transcriptase PCR (RT-PCR).
Why is COVID-19 different from DNA?
DNA is made of a long, double-stranded chain of genetic material called nucleotides. It is relatively stable and makes it ideal to use for diagnostic testing.
On the other hand, RNA is only a single-stranded chain of nucleotides – but this degrades easily. In a PCR test, a process called Reverse Transcription converts the RNA into DNA. Like any other type of PCR, a special, unique primer identifies the target sequence – in this case the virus RNA. When it comes to COVID-19 diagnosis, special kits are available that are specific to SARS-CoV-2.
The second problem with diagnostics is that PCR tests and gel electrophoresis take a long time, potentially even a whole day. However, a machine called a real-time RT-PCR machine can show the accumulating results almost immediately, and the entire process can be finished within 2-4 hours.
The process needs to be completed to get a proper diagnosis, but seeing the results become clearer over time allows the person running the test to check that the process is working.
How does RT-PCR work?
In a real-time RT-PCR machine, the sample is tested using fluorescent light because each nucleotide that builds the reverse-transcribed DNA has a fluorescent dye.
If the virus is present in the sample, it fluoresces. The more DNA present, the more it will glow. If there is no virus in the sample, it will not fluoresce, because no fluorescent chain was built.
As the DNA accumulates over time, the machine measures how much fluorescence is in each sample and displays it on a screen. This data helps calculate whether the positive result is strong or weak.
This image shows three real-time RT-PCR samples, comparing levels of fluorescence against the number of PCR cycles (AKA, time in the machine). The horizontal line represents the level of fluorescence needed to accurately determine unique fluorescence.
In this example, the three samples reach that threshold level at different times. The longer it takes the reach the threshold, the less template there was to begin with. In this case, we could consider sample three a weak positive, because there was not very much virus to begin with.
The whole process of real-time-RT-PCR is very reliable, but false positives sometimes happen. Laboratory error – bad samples, cross-contamination, clerical error – is the main cause of false positives. However, sometimes the special primer amplifies the wrong DNA, but this is uncommon.
In the latter case, this sometimes occurs when people were once infected but got better. This is because they still have traces of the viral RNA in their system. Don’t worry, they aren’t infectious anymore!
Sometimes, diagnostics use a special probe that recognises the correct DNA instead of fluorescence, but it is still useful form of detection.
What is the rate of false-postivies?
Overall, false-positives occur in 0.8%-4% of samples, with a median occurrence of 2.3%. The diagnostic teas retests any positive sample, especially ones that look like a weak positive. This lowers the chance of a laboratory error or contamination causing the positive result.
The combination of high accuracy PCR tests, and resampling of positive tests means the real-time RT-PCR gives very reliable results.
The likelihood of a false negative is higher because anything that causes the PCR to ‘fail’ looks negative. This might happen because of bad sample or laboratory errors and occur in up to 10% of samples, according to a review.
However, a negative control accompanies each PCR – a sample known to be negative. In this case, the results don’t threshold but might look different to the control. This means they could be retested to see if it is a false negative.