running wild said...
Garzie - Igenex has said its immunoblots are more than 93% sensitive. Not arguing what you are saying, but would you disagree with their number? Reason I ask is because I always put a lot into the immunoblot because of what I have read. /
testing is a complex and nuanced subject because we don't have very good tests that give a simple black and white result - and so we could discuss the ins and outs of it all day
essentially, because the tests are not ideal - it all comes down to interpretation of less than ideal tests - and this is where the complexity creeps in...
to make sense of this complexity we have to ask more questions about
what is really happening in the real world - vs for instance the validation studies.
so, yes, all manufacturers that produce serological Lyme tests do quote high numbers for sensitivity - bear in mind this is pretty much a given - since they would not be able to sell the test otherwise - and it would therefore be pointless making it.
however - independent attempts to validate these tests demonstrate that they do not in fact achieve the sensitivities quoted by the manufacturers - for instance by comparing serological results like immunoblot with indisputable methods such as culture, this discrepancy occurs pretty much across the board.
(the independent studies i have read average out at something like an actual 40-50% sensitivity for commercial blots)
so how can this be ?
well, when you dig into it - and ask question like :
1,what strain of the Lyme spirochete was used to arrive at the 90% + sensitivity number ( was it 1000 samples from infected humans in the wild - or some simulated sample with a lab grown strain of borrelia like B31)
Hint - its almost always its lab grown strains or samples
2,what gold standard test did the manufacturer use to define which samples were truly infected and truly uninfected - so that they could generate sensitivity numbers ( sensitivity = percentage of test positive vs truly infected)
hint - often they do this using other serologic tests as reference - in which case the logic is deeply flawed (as the reference sample is not actually a "golden sample") and the sensitivity numbers can come out very high if there is simply agreement between the data sets ( which might happen if for instance you just used a similar method to the reference test ....which is common)
3,what was the mathematical method used by the manufacturer to generate the sensitivity number ?
here again there is significant room to impact the claimed number - and especially if the gold standard test chosen as reference also misses some positive samples ( ie is not a true "Gold standard" test - that this method of validation relies upon) - then the mathematical calculation for sensitivity and specificity starts to become highly problematic.
on top of these procedural issues with almost every validation study i have ever read - you have the very general issue with serologic testing for lyme as a whole - in that lyme is known to damage the germinal centres in the lymph nodes very early in the course of infection.
this is a key point as - it is in these "germinal centres" that the immune system is supposed to learn what antibodies to make - and when this process is interrupted - the host may not make the expected antibodies at all(* there are studies showing that 20% of Lyme patients NEVER produce the expected antibodies )
in this case no serologic test could possibly deliver 90% + sensitivity!
and in fact
using a serologic test as the basis to detect an infection, when that same infection damages the hosts ability to make the very thing the test looks for, is deeply flawed logic.ask yourself - why every person with chronic lyme does not light up every band on the western / immuno blot, every time they are tested ......
bottom line - serologic testing is cheap and easy to administer and as such has become the mainstay of testing for infectious disease - but for lyme all serologic testing is deeply flawed - and there are good reasons to be very cautious about
manufacturer's claims - and especially cautious about
negative results
however,
sensitivity tells you absolutely nothing about the reliability of a positive result -
which is i think the subject that causes you most stress - ie
"The test says positive - but do i really have Lyme???"
this is covered by the SPECIFICITY of the test - not its sensitivity.
the reality is that, although there are issues with false positives for Lyme - they are typically much smaller in size than false negatives - due to the issues above driving high false negative rates not driving high false positive rates in the same way - thus serologic tests are much more likely to tell a person they don't have lyme - when in fact they do - and if it tells them they do have lyme - its much less likely that this is an error
its therefore critical to understand the meanings of SENSITIVITY vs SPECITIVITY - when discussing test interpretations - and understand that when one is high the other may be low and vice-versa - as apposed to the extremely ambiguous term "Accuracy" - which implies a kind of symmetry(which does not exist)
running wild said...
As far as PCR, I've seen positive results are hard to argue against, but negative ones cannot be counted on. Is that where you are going with /
yep with PCR - the specificity is typically very good indeed - and its raw sensitivity is also excellent.
but sensitivity issues exist due mainly to the real world physics of what is going on when you sample.
for instance, a test that is 100% sensitive is useless if there is no DNA in the medium you have chosen to sample - and this is known to be the case for blood samples - a lot of the time ( approx. 50% as far as i can gather) in the case of chronic Lyme infections