Two updates.

1)  This week’s Thursday night update will be short(er).  I have a mandatory test this week for my “every 10 years” board recertification.

2)  Since it is hitting the headlines today as well, I figured I would send out a quick update on the “game changing” new coronavirus test from Yale, sponsored by the NBA.

The test, called SalivaDirect, got emergency use authorization from the FDA yesterday.  Reading the articles about the test today, you will hear things about how the test is “open source” (meaning any lab can take the procedure and do it), is faster than existing tests, uses fewer reagents to thus avoid shortages, and is accurate enough.  Rapid testing at this scale, we are told, could “end the pandemic even without a vaccine.”  Yale is not making money marketing the test (they couldn’t if they tried–more on that in a second) and neither is the NBA.

So what does this test have right, and what does it have wrong?
Well, fortunately, the test is NOT antigen based, at home on a paper strip.  Instead, the test is still RT-PCR, which is the gold standard for diagnosis still.  Like the assay we are running here (you can also find our emergency use authorization on the FDA’s website), SalivaDirect uses the CDC’s primer and probe set.  Unlike our assay, they use the primer and probes for only one viral gene target (we use two–we find we get better sensitivity that way). 

<redacted commentary on hyperbole in lay reporting on the SalivaDirect test>

So minor snarks aside on some of the marketing in these articles, SalivaDirect does a few things right.  For one, the test will be highly specific for SARS-CoV-2.  The gene they test for, and the CDC’s primers and probes for it, are VERY specific for that virus.  So no real chance that a cousin coronavirus trips it up, as seems to happen sometimes with some antigen based methods.  And it’s probably sensitive enough, although it was not tested in asymptomatic patients.

Whom the articles out today clearly intend on testing, when they are talking about getting yourself screened in three hours for about $10.

To get some of the other advantages and disadvantages, let’s briefly go through the general steps in testing coronavirus by PCR.
1)  Collect the sample from the patient

2)  Ship the sample to the lab, and check it in (you need to make sure the right patient and physician gets the result!)

3)  Extract the viral genome

4)  Do RT-PCR on the viral genome

5)  Report the result of RT-PCR.

Got it?

Your major time sink in those steps is actually step 2), shipping and check in.  Step 3 and 4 can be done in about 3.5-4.5 hours depending on how exactly you do those steps and how many samples you have. 

***The rate limiting steps are the number of people and machines you have to do steps 3) and 4).  How many people and how many and how big the thermocyclers, the device that does the RT-PCR itself, control how many tests you can do per day***

Keep that in mind.

Now, back to SalivaDirect.  By using saliva, sample collection is a little bit nicer for everyone than using a nasopharyngeal swab.  If you have not had that done in the pandemic yet, just YouTube videos for patient experience of those.  In March and April, my company was making its own swab collection kits because those were, in fairness, hard to come by for everyone.  We were providing those to the state and hospitals in the area too.  From my understanding, there is now less pressure on getting those swab kits in sufficient number.  But the articles about SalivaDirect are correct in that you avoid the possible bottleneck of not having enough swab kits to go around.  Self collected saliva is cheaper.  Otherwise, they are using many of the same reagents for the actual RT-PCR that everyone else is.  There is no saving on availability for those (and there is no major problem with availability of those right now either). 

In addition, SalivaDirect skips a full extraction of the viral genome.  The basic idea is that you can damage enough virus, if present, while avoiding a lot of the snot in a nasopharyngeal swab that you have to clear out chemically in an extraction step.  This saves about 45 minutes to 2 hours depending on how you do extraction and how many extractions you do, because step 3) above gets a LOT shorter.

This is how you get down to “results in as little as three hours!”

BUT, you are ONLY getting results in three hours if you are standing in the lab, spit into the tube there, and immediately hand it to the person checking samples in.  From there, your sample must be lucky enough in that it’s the last one in a batch of them, since you run as many patients as you can at once on each thermocycler.  If all of that is true, yeah, results in three hours.

The problem, and why I doubt this will be as “rapid” as the articles today claim, is that you still have to send this to a lab for the diagnostic part, the RT-PCR, to be done.  That is Step 2), and remember, Step 2) is the biggest time sink.  If you are in rural Idaho, for example, your test is going to be on the road for a bit before it gets to the actual lab that can do the PCR part.  3 hour turnaround is NOT happening for you.  Even for large labs FedEx’ing samples overnight all around the country.  Your sample still has to get picked up by FedEx, sorted, put on the right plane, sorted, and finally dropped off at the lab doing the testing.  All of that takes time–more often than not, that takes more time than the actual test!

Further, the biggest reason you are hearing about quotes of 7 day turnaround for some of these tests is just the sheer volume of tests.  Again, the rate limiting factor is how many (and how big) your thermocyclers and how many people to run them.  We’ll keep the math easy, and assume you have a thermocycler that can do 10 patient samples per run.  If you staff it to run 24 hours a day, and each run takes 2 hours, you can do 120 tests per day.  If you have 240 samples show up one Monday, 120 of them are not going to be run until tomorrow.  You can’t fit 240 samples onto your thermocycler in a single day.  Your turnaround time is already stretched by a day.  Now add tomorrow’s new sample volumes, and you see the problem developing.

Every lab will vary a bit in how many samples it can handle per day, and that’s before your staff gets sick or has car breakdowns, or the equipment itself needs maintenance.

***This is why testing turnaround is out to 7 days or better at some labs (not all–others can still do 48 hours or less, or even less than 24 hours)***

It’s not for lack of reagents.   It’s not even the PCR technique being used.  Some labs, even large national reference labs, are getting more samples than they can handle in a given amount of time.  Testing is so popular in some places that it is creating a backlog of samples to be tested.

Removing/shortening step 3, extraction, means you can run step 4) more often per day, and help clear that backlog. 

But as we have said, testing numbers may be going down because people who don’t –need– the test are not getting it because they don’t want to wait as long as the quote from some particularly slammed labs.  You get into an interesting dichotomy where if the test gets faster and cheaper, more people want it (even those who probably don’t need it), and more tests get ordered until these, too, create a backlog because there are only so many molecular lab techs and thermocycles available. 

So, long summary made short.  Will SalivaDirect help?  Yes, in some ways, it will definitely help.  The lab process gets shorter and a little cheaper (not sure about $10 per test cheap though), and labs with the equipment Yale used already in their lab can adopt this pretty quickly.  They can, in theory, do more runs of PCR per day with it, reducing any log jams they may have.  But is it a “game changer,” where everyone is going to be getting tested now, supa’ fast, and we can instantly track where all the ‘rona is and quarantine it to death?

I have some doubts.

We’ll see as it all rolls out though.

<Paladin>