Thursday, 27 June 2013

What is PCR and what does a positive virus detection mean..or not mean?

There have been a few comments of late about the finding of MERS-CoV in asymptomatic or mild cases of disease. For example, from ProMED
As to my knowledge "infection" is defined as the proof of an infectious agent AND the proof of multiplication of this agent inside the body (OR associated inflammatory response that can be linked to this agent). Therefore my question is: where was the specimen taken from the "asymptomatic" nurses? PCR, as we all know, just detects nucleic acids. So if multiplication cannot be proven and no local or systemic inflammatory response is given, how was "contamination" (no matter whether due to inactivated parts of MERS-CoV or even complete virus particles) ruled out? This is not a semantic question only but should impact the reported number of "cases". [Prof Martin Haditsch]
The feeling I get is that some don't trust the PCR results (or result-getters?) alone because they do not fulfil a need to see replicating virus in a host before assigning causality to that virus. Culturing an isolate of course, wouldn't assign causality to a pedant either it just proves it was in the patient. Similarly, serology would only prove it was in the patient and was recognized by their immune system. The sorts of extremely rare and poorly define studies needed to actually prove causality (human infection - but unethical) are usually well down the priority list in a public health management phase such as we are in with MERS.

So there is no easy answer to address this and a lot revolves around pedantry. As we know, Koch's postulates (apart from being a guide only and applying to bacteria) don't hold up too well in the increasingly molecular world of new and emerging virology. One certainly can't test a flu or CoV which kills 20-50% of the cases its found in, using human volunteers in order to reproduce disease and prove the agent was responsible. There are modifications of the postulates though. I've even had a go in a book chapter once.

I've added some musings below. I'll be working on this throughout the week day-check back for final version:
  1. PCR is an enzyme-driven cyclical cooking and cooling process that amplifies a pre-defined stretch of DNA using a pair of equally pre-defined and commercially made oligonucleotide (short DNA stretches) primers. Add a reverse transcription step beforehand and you can use PCR to amplify DNA made from a pre-defined RNA template (the thing being copied).

    Viruses come with DNA or RNA genomes. A PCR POS result does not prove active replication of a virus. It does not prove infectious virus is present. This is why we prefer to talk about virus being detected in patients. Some still use the term "viral isolate" - but this should be reserved for when a virus is grown using cell/tissue/organ culture.
  2. PCR doubles the number of pieces of target DNA (copies or amplicon-meaning DNA copies of a single species) every cycle (repeating series of heatings and coolings). The more DNA/RNA that you start (usually equates with the viral load) within your patient sample, the sooner you get enough copies to be measured using your detection method of choice (agarose gel or, most commonly, fluorescence signal). I have a page on the mechanics of PCR here
  3. Of late, with increased use of PCR methods in respiratory virus investigations, research studies are more often finding detections among people with no signs or symptoms of disease at the time of sampling...also not immediately before or after if the people are followed up carefully.

    Side issue #1 - if we're looking to declare such a person "asymptomatic" - we had better be completely sure they have absolutely no symptoms of an infection. Disease, a deviation from normal body function, may be as mild as a headache or just feeling crook - and still, pedantically, be a disease that ascribed to the virus detected by PCR. Let's assume that level of care has been taken (often it hasn't)

    Side issue #2 - we are walking platforms for viruses, bacteria and fungi. Take a look at this paper from Eric Delwart; in 2 kids-92% of 72 samples collected weekly over more than 250-days were virus-positive (no or mild signs of disease developed during this time)! It is very likely, as we discover more diverse agents, better characterize them and develop (and employ) better testing capabilities, that we will learn we're hardly ever without a passenger or 3. And yet we are not always sick - why is that?
  4. The immune system to the rescue! It won't come as surprise to the experts, but we have very capable, active and multifunctional immune system. It is constantly being challenged by..things - not just viruses, bacteria and fungi but chemicals, various proteins, carbohydrates, allergens, dirt and dust (composed of bits of all that other stuff) and it does a great job of sorting through it all, responding to what we need to, to prevent disease, and usually squashing the response to the rest in such a way that we don't under-perform 24/7 because of allergy and illness. It is due to this system that we shouldn't be surprised that as technology lets us find lower levels of infectious agents, it is completely reasonable to be "infected" but not diseased. At least in some proportion our encounters with viruses this should be no surprise.
  5. I think its worth highlighting that we breath in up to 60L of air per minute, with 0.1-7200 submicron particles (aersolised; 0.3-0.5μm) per litre. In one study, influenza RNA was found more often in particles ≤5μM than in those greater than 5μM. These small particles can be inhaled and exhaled fairly easily and commonly.
  6. Sample site and transmission in asymptomatic cases. I don't have any data for this but its seems to me that if you swab the upper airways, you have swabbed the site from which virus is easily coughed, sneezed and therefore transmitted in larger particles. If virus is detected there, then there would seem to be a good chance of transmission being possible. Does it happen? Those studies are not easily visible...or perhaps they have not been done.
  7. The big grey peanut-loving amplicon in the room is contamination. That is, amplicon from previous PCR reactions that accidentally gets into your new mix/common PCR reagents/pipettes/lab and gives us a positive result when the patient is actually not infected. It's what some think of when a PCR positive occurs in an asymptomatic person. It's easy to point at it as a reason. However, if the lab is expert in PCR, and those testing in the WHO network are, then the necessary controls will be in place to give warning of contamination. Plus, for some viruses and in some labs, more than one PCR assay, each targeting different genetic regions and using separate primers, are employed. This is the case with the recommended work-flow for MERS-CoV screening. It does not mean these are the assays used of course. And more detail on testing would be useful.
Other useful steps include separate labs for specimen preparation, reaction mix preparation and cycling; labs with directional air- and work-flows; disposable everything; commercial reagents; pre-made, target-specific kits ("analytes"); lots of negative controls and so on. These provide considerable confidence in a PCR result and are standard working practice for a professional diagnostic PCR lab. Taking more than one sample from a suspected case is also helpful.

PCR has been used to detect viruses since the 1980s and there are many very knowledgeable experts in its use who know how to get useful results and avoid contamination. Three are also many publications that link infectious virus with a PCR result - PCR positivity has meaning. Also, those of us who have conducted PCR experiments on human samples over many years will know, PCR is not so practically sensitive that it will regularly detects virus that is just sitting around in your nose because it was inhaled, but is not replicating. Dilution factors see to that.

Still, problems happen and non-professional labs exist. There is no easy answer to the questions posed by Prof Martin Haditsch on ProMED. As with any test, reliability is about reducing the risk of failures in the process. But even expert labs have bad days.