True crime series like Serial and Making a Murderer invite us to sharpen our abilities and to outperform the detectives assigned to the case in figuring out what really happened. We hear the official narrative and we then inspect various clues provided to us by the storyteller, turning them around in our mind鈥檚 eye, looking for that smoking gun.
As the COVID-19 pandemic raged on, some people decided that the disease was nowhere near as severe as we had been told and looked for ways to discredit public health efforts. One of their alleged smoking guns? The laboratory test for COVID-19, called a PCR test, was not being done properly, they told us. The Ct values were apparently too high and the test was spitting out random noise. The COVID-19 PCR test, they decided, was a broken machine churning out false positives and fear.
None of this is true. In my laboratory days, I did countless PCRs, and I want to explain what a PCR is, what those Ct values are and what they mean, and why the scientists interpreting these tests actually know what they are doing
The Adventure of the Polymerase Chain Reaction
Imagine you have a short story collection in front of you and you want to know if it contains the Sherlock Holmes story 鈥淎 Scandal in Bohemia.鈥� Obviously, you could consult the index or perhaps even the dust jacket, but I want to use this example to explain how scientists go about hunting for the coronavirus鈥� genes in the samples they get from the people lining up to get tested.
We can focus on one of the last sentences in this story: 鈥淎nd that was how a great scandal threatened to affect the kingdom of Bohemia, and how the best plans of Mr. Sherlock Holmes were beaten by a woman鈥檚 wit.鈥� This sentence does not occur in any other story: therefore, if we can find it in this book, it will indicate that 鈥淎 Scandal in Bohemia鈥� is present within its pages.
What scientists would do is they would choose two short phrases called 鈥減rimers鈥� that will frame this sentence. For example, 鈥淎nd that was how鈥� for the beginning and 鈥渁 woman鈥檚 wit鈥� for the end. What will happen in the lab is that everything between these two phrases or primers will get amplified: copies will be made. This sentence will be copied over and over and over again in a process known as PCR or the polymerase chain reaction, named after the enzyme that does the copying.
PCR is a workhorse in molecular biology laboratories. It is used to detect infectious microorganisms; to see if a patient has a certain disease-causing mutation in their DNA; and to compare a specimen found at a crime scene with samples in a database or with the DNA of a suspect. It functions by amplifying a specific part of the genetic material so many times that it becomes detectable. But when it comes to detecting the coronavirus, time is of the essence, so the type of PCR that is used is called a real-time PCR assay, and it contains an additional twist: detecting these copied sentences is done as they are being copied.
Returning to our short story, we would design a probe that would bind to the central phrase 鈥渢he kingdom of Bohemia.鈥� Imagine we start with a single copy of our whole sentence. After one cycle of amplification, we now have two. After a second cycle, we have four. Then 8, 16, 32, 64, 128 copies. The number of copies increases exponentially because each copy becomes a template for the next round of amplification. And during each round of amplification, our 鈥渢he kingdom of Bohemia鈥� probe binds to all available copies of this phrase. Attached to this probe is a fluorescent molecule that, when excited by a laser, emits light like a distress beacon. What happens to this light as the probe is allowed to bind 8 copies of the phrase, then 16, then 32? More and more light gets detected. This light emission is our clue that the sentence from 鈥淎 Scandal in Bohemia鈥� was in there all along and is being amplified.
The last element I need to go over, and the one on which COVID contrarians have jumped on, is the limit of detection. You see, if you gave me a stack of papers filled with the last page from 鈥淎 Scandal in Bohemia鈥� and asked me to run a real-time PCR on it, the light signal I would get from the probe would be loud and clear since I would be introducing a large amount of starting material. But if we are starting with a single page, it will take many cycles of amplification before enough probes can bind to 鈥渢he kingdom of Bohemia鈥� and emit enough light to be detectable.
So this real-time PCR test has a limit of detection. When there is a lot of virus present, the answer it gives is pretty unequivocal. But when there are few viral particles in the sample (because, for example, we just got infected with the virus and it hasn鈥檛 had time to make many copies of itself), it will take many, many cycles of amplification before we can detect enough light to see that the virus is there.
Some people claim that the coronavirus PCR test is run for too many cycles and is thus mostly unreliable. They are asking for a thing called the Ct value to be disclosed for each test so they can judge for themselves if the test is accurate or not. There is a grain of truth to the allegation but, as you will see, not much more.
Ct values are not elementary
This infamous Ct value, or cycle threshold, is pretty simple. It is the number of cycles of amplification needed to get enough light. By 鈥渆nough light,鈥� we mean that it crosses the threshold where it is clearly above what would be called background noise. In our fictional example above, we amplify the sentence once and get two copies: a tiny bit of light is emitted and detected. With another cycle, we have four copies: more light. Then eight copies, 16 copies, 32 copies, and with each cycle we detect more and more light as more and more probes get to bind to the phrase 鈥渢he kingdom of Bohemia.鈥� At some point, let鈥檚 say at cycle 25, enough light will be detected that we will have cleared this threshold. The Ct value will be 25.
When the Ct value is low, it means that there was a lot of starting material (many pages with the sentence we were interested in, or many copies of the coronavirus). When the Ct value is high, it means there was little starting material, so it takes more time to have enough copies so that you can see them.
The danger when seeing high Ct values (e.g. 38-40) is that the signal you are eventually getting may not be specific: to go back to Mr. Holmes, it could be amplifying a somewhat similar sentence from a different story. This is the kernel of truth that COVID contrarians have jumped on. But the reality of PCR technology is much more complicated: we can鈥檛 simply set a universal Ct value beyond which we declare all tests to be negative. The Ct value is, in a way, relative.
Different laboratories have set up different PCR tests to look for the coronavirus, using different probe-and-primer combinations to look for different genes in the coronavirus鈥� genome on different PCR machines. Unsurprisingly, when participated in a proficiency test, they saw a variability of Ct values of up to eight cycles across them when testing the same specimen. Samples that are known to be positive and negative for the coronavirus are run alongside the unknown samples, and their behaviour during the run also affects interpretation of the results. This is why reporting the Ct value is not recommended in Canada: on its own, it does not mean much.
In a way, it鈥檚 not unlike chicken soup. Many families have their own recipe. As long as it鈥檚 been internally validated, meaning that it looks like chicken soup and tastes like chicken soup and the people eating it are happy with it, it鈥檚 a perfectly functional chicken soup. PCR tests come in many different flavours, but as long as they are validated (by using a known quantity of virus, diluting it many times and running these samples to see what Ct values they generate), they are reliable. They are not perfect, because no test is perfect, but they are absolutely not the futile garbage some folks on the Internet would have you believe.
The pandemic saw a rise in armchair experts, people who had never stepped foot in a laboratory suddenly learning about PCR and thinking, as in true crime dramas, that they had cracked the case wide open. But the interpretation of PCR tests for the coronavirus relies on a lot more than a single Ct value: it depends on all of the above 鈥渃hicken soup鈥� variability, plus the type of specimen collected, whether or not samples are pooled in a single well to save on reagents (with positive pools being tested individually afterwards), and on pre-test probability, meaning whether or not the person being tested has symptoms and whether or not they were potentially exposed to the virus. The blind reliance on Ct values unfortunately shows a misunderstanding of the complexities of molecular diagnostics. Ct values are not elementary; they require expertise to interpret.
Take-home message:
- Some people claim, wrongly, that the PCR test for the coronavirus is useless because the so-called Ct values are too high and the test is picking up things that are not the coronavirus
- The PCR tests for the coronavirus that have been internally validated by public health agencies are actually very reliable
- Scientists cannot declare any result above a certain Ct value to be unreliable because Ct values are somewhat relative and must be interpreted by taking into account a variety of factors
Photo credit:聽James Gathany
