Re: [DIYbio] Re: Anybody working on Lyme Disease?

One problem is that the data that comes out of these tests is not "positive" or "negative", but (for example) stain densities in seven different lanes.  I have ambiguous results; what does that mean?  Another is that the tests are less reliable for someone who's been taking antibiotics - they generally kill off the most-accessible organisms, but may leave a reservoir hidden inside cells which will not show up on PCR.  Sampling synovial fluid is supposed to be more reliable, but doctors don't like to do that.  They prefer using blood or urine because these are accessible, not because they work.  No one knows how long you need to go off antibiotics before a PCR test, and it may be quite painful to do so.  Additionally, if you go off antibiotics for a PCR test, you may find that those antibiotics no longer work for you when you resume them.

Possibly a bigger problem is that there are a variety of bacterial infections communicated by ticks, and there's little reason to think that we've enumerated them all. This should change in the next few years, if someone does a metagenomic study of wood ticks. Someone did one metagenomic study of wood ticks, but they were looking for known organisms rather than doing assemblies and looking for unknown infectious bacteria IIRC.

A lot of patients have mysterious diseases that clear up under antibiotic administration.  These often get classified as Lyme, because they co-infect with Lyme and so the patient has Lyme antibodies, or just because bacterial infections generally cause fatigue and chronic pain.  Then someone will argue that these patients don't have Lyme, and therefore should not be on antibiotics.  This is moronic.  The question is not what disease they have; the question is how to make them better.  This is one reason why double-blind tests are kind of pointless in this case.  First, I don't care whether the effect is a placebo effect.  If it works, it works.  Second, double-blind tests assume that you can accurately categorize your  patients into those who have the disease, and those who don't, and that EVERY SINGLE PATIENT who has the disease responds in the same way to treatment.  Even those with Lyme are going to all respond differently; and your set of "Lyme patients" is certainly going to contain a lot of people with other infections, and a lot of hypochondriacs.

I could write an article on why the standard double-blind test is so often misleading.  If you're testing a treatment, or a drug toxicity, typically some subset of patients has a positive or negative response, based on environment or genetics.  If 10% of your population has a set of SNPs that dictates a negative response, a standard double-blind test will be cited as "proof" that no such response exists, because the hypothesis that all patients with this condition have a negative response will most likely be rejected with 95% certainty.  This is the case for hyperactivity in response to certain food colorings, which the medical establishment vociferously proclaimed to be a myth for 40 years, until recently it was shown that the effect is strong if you restrict yourself to looking at certain types of food dyes and patients with certain SNPs.  (You have to account for multiple hypothesis testing when dealing with restrictions, of course.)