What was the deal with the CDC Covid-19 diagnostic tests? It’s taken pretty long, but we now know more about what went wrong. The US got a really slow start to testing, partly because the original tests the CDC put out were faulty. Over the weekend (4/18/20) the Washington Post put out an article describing what went wrong at the administrative levels, and today I want to tell you more about what they expose went wrong at the biochemical level.
The CDC tests, like most current tests for SARS-Cov-2 (the virus that causes the disease Covid19) are based on using a technique called PCR to look for genetic information specific to that virus, such as stretches of the genes it has for making proteins that it needs but we don’t have. There’s not much of this viral info, so scientists have to make lots of copies of it and use sensitive methods to detect it – because you’re making lots of copies, if a sample gets “cross-contaminated” with some viral RNA that isn’t supposed to be there, the sample will produce a positive result – regardless of whether there was actually viral RNA in the sample. It’s kinda like if a crime scene detective didn’t wear gloves when he was testing objects for fingerprints, and then his fingerprints were found all over the evidence. Is he the culprit or not? We can’t know. And this was what happened with the CDC tests – they got viral RNA in one of their “detective mixes.”
Biochemically-speaking, these “detective mixes” are solutions containing short pieces of DNA called primers that bind to a region of genetic info you want copied and fluorescent probes that bind to the copies and let off light when copies get made. The copying happens through a technique called Polymerase Chain Reaction (PCR), which, as the name suggests, is a chain reaction – thanks to the ability of DNA letters to form specific base pairs (A to T & C to G) one strand of DNA is used to make a copy of the complementary strand which is made to make a copy of the original strand which is used to make a copy of the complementary strand… get it going and it’ll keep going and going and going until it runs out of ingredients.
PCR copies DNA, but SARS-Cov-2 has an RNA genome, so the first step, after you extract RNA from a patient sample (such as a nose or throat swab) is to “reverse transcribe” that RNA into a DNA copy. Then you can test that DNA (potentially containing DNA versions of the viral RNA) for “anything” – different primer/probe sets can be used to look for different genetic sequences – either sequences from different places within the same genetic blueprint (genome) (i.e. different genes or different parts of the same gene) or the genomes of different things.
So the basic premise of the test is – doctor collects patient sample (often taken by swabbing the nose and/or throat) and sends it to a lab and then the lab scientists:⠀
1. extract the RNA⠀
2. reverse transcribe that RNA into DNA form in preparation for copying⠀
3. copy it and copy it and copy it… and detect the copies as they’re made⠀
Like strong passwords, the primer & probe sequences can be designed so that they’re very specific, allowing you to differentiate between different viruses, bacteria, etc.
The Covid-19 PCR tests use primer/probe sets designed to match the SARS-Cov-2 genome, and copies should only be able to be made if there was that original matching viral info to copy. So you should only see light if a sample contains viral RNA – this is taken as a positive result – and if you don’t see light (at least not above a threshold determined by background “noise”), it’s a negative result. If the controls are okay, that is… But what if you see light when you KNOW there isn’t any RNA, like when you test ultra pure water as a negative control? Then you have some issues… And this was what raised red flags in the labs the CDC sent their original test kits to back in January.
To better get a sense of what happened, I think it helps to understand the test format. The tests are usually carried out in trays containing lots of rows of really little tubes (typically 96 (12 x 8) – so you can run many reactions at once. These reactions can be testing different samples for the same target (i.e. using the same primer/probe set) and/or testing the same sample for different targets (ie. using different primer/probe sets). So different reaction tubes will have different combinations of sample and primer/probe set.
Additionally, ALL the reactions will need some essential components, including a Reverse Transcriptase (RTase) to do the initial RNA -> DNA conversion, a DNA Polymerase (DNA Pol) to do the DNA copying, DNA letters (nucleotides) to be added, and salts, pH-stabilizers (buffers), etc. to keep everything happy. Since all the reactions will need these same ingredients, instead of having to add them separately, they come pre-mixed in a “Master Mix.”
Each reaction tube contains 1 test sample + 1 primer/probe set + 1 serving of Master Mix (MM)
As I will tell you more about in a sec, the original CDC test kits contained 4 primer/probe sets. 3 were designed to copy regions of the virus’ “N” gene (which has genetic instructions for making the nucleocapsid protein that surrounds the viral RNA): N1 & N2 were designed to specifically detect SARS-Cov-2 (and ONLY SARS-Cov-2) – they included 2 different regions to help be super sure of a result. N3 was designed to detect SARS-Cov-2 AND related SARS viruses (like the “original” SARS virus and related bat viruses) – I’m still not quite clear why, other than to show that their SARS-Cov-2 primers really were super specific?
And then there was a 4th primer/probe set (RP) to detect a human gene, RNaseP, to serve as a control that there was enough cells in the sample to have a good chance of finding the virus, that the genetic material extraction went okay, and that the PCR machine works, etc.
So, for each patient sample you had to run 4 reactions:
- sample X + N1 + MM
- sample X + N2 + MM
- sample X + N3 + MM
- sample X + RP + MM
Additionally, you had to (and still have to because it’s really important!) run a positive control (PC) and a negative control (NC) for each primer/probe set. These are to make sure that each set can detect its target when that target is known to be there (the test gives a positive result for the positive control) and they don’t detect anything when the target is known NOT to be there (the test gives a negative result for the negative control).
- NC + N1 + MM
- NC + N2 + MM
- NC + N3 + MM
- NC + RP + MM
- PC + N1 + MM
- PC + N2 + MM
- PC + N3 + MM
- PC + RP + MM
The negative control is just really pure water (user supplied) so, when multiple labs started reporting false positives in their negative controls you can’t blame the water! Instead, it’s probably the positive control to blame (or more like lack of proper control of the handling of the positive control)
In the case of these tests, the positive control consists of non-infectious “contrived viral RNA” – RNA that’s identical to the region of RNA containing all of the sought-for sequences, but, instead of being made by the virus, it was “in vitro transcribed” (made in a test tube). Since it’s only part of the viral RNA and it doesn’t have any of the many other components the virus needs to infect anything, it’s safe to work with from the standpoint of it’s not going to hurt you – but it still needs to be worked with with great caution when you’re designing tests to detect it!
And it seems like there wasn’t enough of this caution by the CDC’s test-makers. Instead of having specialized manufacturers make the tests, the CDC was initially insisting that all of their tests be made “in-house” – in research labs that, unlike the big factories designed for making lots of super super pure test components, were designed for designing the tests.
Not only was the CDC slow to roll out testing, but when they sent out the first test kits to 26 public labs in the US, 24 of them reported that the tests “didn’t work”. Or, at least a part of them didn’t work. The N1, N2, & RP primer/probe sets seemed fine, but the N3 set was giving false positives with its negative control – and this was invalidating the whole test. To make things worse, the CDC wouldn’t let other labs develop their own tests – labs did – and made tests that worked – but these results were only considered “presumptive positives” until the samples were confirmed with the CDC tests… When the CDC outsourced their test-making to one of those companies that’s that routinely makes this sort of thing (in this case IDT) the tests worked fine. And they’ve since authorized additional tests.
But the long delay really set the US back in its coronavirus response. And one of the worst parts about it all was that the N3 primer/probe set wasn’t even needed. The WHO tests, which were available but the US didn’t want to use, don’t include such a pan-corona test and the US might have thought it would make their test better – but it turned out to make the test worse…When the CDC realized it was just the N3 set with issues, they told the labs that had received test kits that they could just throw out that part and use the rest of the test. And the new tests don’t contain N3.
link to Washington Post article: https://wapo.st/2VEhK6W
more on how coronavirus tests work as well as other covid-19 resources: https://bit.ly/covid19bbresources