Thread by @MackayIM

I know the technical PCR nerdery is hard for the general public. I feel this whenever I read a hardcore immunology paper!
A few thoughts on the concerns about PCR...
PCR threshold cycles are test-/machine-/reagent-/kit-/lab-/handler-/sample site-/sample storage/sample handling/..

..cold chain-/extraction method-/primer pair-/probe-/disease stage-specific.
To compare one lab's results to another lab's results (or even another run within the same lab), needs both labs to be using the same PCR setup & to have the same controls or calibrators.

This is easier when using commercial kits because they all use the same reagents and they've been optimised and (hopefully) well validated. But that doesn't mean they've done the work to correlate result with infectiousness, disease severity etc-it would be helpful if they did.

When using lab-developed (in-house) real-time RT-PCRs (i.e. published/bespoke primer & probe sequences, applying your own favourite regents on your fave instrument & with fave conditions, *maybe* optimising & validating the test yourself), we really need quality assurance panels

These could contain controls of known amounts to see how lab A is performing compared to lab B, C. D, E etc. These sorts of programs go on around the world.
The control amounts *could* reflect the cut-off beyond which a person is much less likely (it'll never be 100%, because

biology) to be shedding enough virus to infect another person, even if they are in close and prolonged contact (the goal of such a control & result).
But again, with everyone using different in-house methods, this won't be a directly comparable result.

If anyone has ever looked over a few PCR Quality Assurance Panel (QAP) results, they'll know how variable different lab's results can be from each other's.

Another point is the one that says "X cycles is too many". Just to clarify, we all run our PCRs for X, Y or Z cycles (some basic info on 'cycles' here virologydownunder.blogspot.com/2015/05/the-me…) usually 35-50 cycles). That's their *end point*. What we call positive can vary.

We can also see (+) sample results that come up anytime *before* (because we can watch in real-time), or we can walk away, come back at the end & see which are positive (=sample result has produced enough fluorescence to pass arbitrary fluorescence threshold setting-dashed line)

This is a real-world example of an optimised & validated in-house test for a virus. In this, the threshold is the red horizontal line set at 0.05 normalised fluorescence units (1 of the things that differs in terminology between instruments & their software). (+) control in pink.

Non-template (water) controls in black & some negative samples also staying flat (=they did not entered the exponential amplification we'd expect to see of a positive result; S-shaped curve in this view). Yes, even at 50 cycles (😲).

On the subject of "everything comes up positive after 40 cycles". Bullshit. It will if you have a poorly designed or optimised test. Otherwise, anyone who says that, has an agenda or is parroting someone who does. There's a lot of that on twitter. Stay clear.

"PCR is too sensitive". It is what it is. But a PCR result's limitations are known by those who order these tests (they should be anyway🙄), those creating the tests & those who release the results from these tests to the Doctors caring for the patients/cases.

A PCR positive is usually a key factor in the case definition for an acute virus infection because we've spent decades understanding that PCR (I mean RT-PCR for RNA viruses, PCR for DNA viruses) are a fantastic surrogate for infection which used to be defined by growth...

..using human, animal, organ, tissue or cell culture. These methods relied on the sample still having enough infectious virus in it to show cell damage in culture or to be detected in other ways. A note here: as soon as a newly assembled virion leaves it host cell its on a...

path to obliteration. If it doesn't find a new cell it will decay beyond the ability to infect and replicate in hours to days depending on what trials it faces once its left that cell. Culture can be 100s to 1000s of times less sensitive than PCR-based methods because..

..PCR can detect genetic material of even non-infectious viruses. Usually not a problem because we just want to say, yes, that sick person is/was infected with *that* virus so we can act accordingly for their care. This is also the case for flu using PCR...

..which can also spread before symptoms start, just like SARS-CoV-2. And undoubtedly other viruses as well (we haven't looked at others as intensely as we do for flu [vaccine & drugs] or SARS-CoV-2 [first real-time RT-PCR tested pandemic].

But keep a few other things in mind here. PCR results are *part* of a patient's diagnosis & care. Yes, a lot of things are probably being skipped over during the pandemic, but a test or a lab alone shouldn't be making a _diagnosis_ on whether a well person is a transmission risk

That comes down to context and consultation. I hope so anyway, but pandemic🤷‍♂️. Another thing to remember is that there is a massive amount of interpretation to be done in diagnosing a patient-& that requires expertise. I'm not seeing a lot of this discussed in the media.

The "PCR is causing a casedemic" narrative doesn't even touch on the processes above. It also skips the importance of a positive result - on a sick, well, infectious or not contact (not sure why else a well person needs to be tested) - for contact *tracing*.

We heard this a lot early on, but testing, tracing & quarantine/isolation are awesome ways to stop spread (physical distancing is prime but we've seen how that's going around the world). PCR testing, with its high sensitivity, specificity & speed... (virologydownunder.com/yes-pcr-tests-…)

...identifies those who *have* been infectious, even if they may not be when eventually sampled (get in earlier, not test's fault your load dropped!) & eventually get a result (not test's fault labs weren't pandemic-planning for predictably high-throughput needs during pandemic!)

This sensitivity is essential, *if* you're seriously trying to reduce spread through contact tracing & the measures that follow it to prevent more transmission, disruption, illness, sequelae & death. If not (for whatever reason including TAT too slow to be able to)..

..then yes, define a method (not just a global Ct value) which labs can adopt which allows some reliable definition of likelihood that a person is still infectious. PCR is a very useful tool. Clearly it's not a widely understood one, but we can easily do better at using it