Re: Some Random Comments 29 October 2004
Previous Rapid Response Next Rapid Response Top
Nicholas Bennett,
Infectious Disease Postdoc/Clinician
Department of Pediatrics, University Hospital, Syracuse NY

Send response to journal:
Re: Re: Some Random Comments

As usual, Mr Bartlett makes some interesting points: even fewer I can answer but I can respond with a few random comments of my own.

At what time should we stop questioning? As a scientist, I would say never! As a clinician, I suppose we have to draw the line somewhere or else we'd never make any decisions. The problem then is that the line occasionally gets shifted a bit as new knowledge comes to light. In my first years of med school I was taught facts, then in my final preclinical year they gave the complicated version. During the clinical training we learn that medicine really is an art, and not a science, and during the PhD I learnt we really knew nothing at all! There's a kind of Heisenberg uncertainty about everything: the harder you look, the less things make sense.

The comments on interpreting a hypothetical 6% false positive HIV rate in SA are quite right. But without any more information, and based on current evidence, one has to assume the results are real if surprising in some areas.

The points about HAART have to be borne alongside the fact that therapy is changed as needs dictate, most often because viral load starts to rise. This usually signifies an escape mutation in the virus. Certainly other studies have shown long-term continual rises in CD4 count up to 6 years, and I think the paper in question actually states that the _rate_ of rise drops off, not the actual numbers [1]. They were looking at proviral load, not viral load (but I can't get access to the fulltext either right now, so I'm only going by the abstract), so the findings make perfect sense as regards our current understanding of the HIV proviral reservoir. I do however agree that to be at the stage where patients are surviving long enough with HIV to suffer from the side effects of the drugs is hardly something to shout about. The problem of course is that initiating HAART at low CD4 counts (<200) is associated with a much poorer outcome. How does research like this affect things? The treatment -initiation point came down (from 500 to 350 CD4 per ul). Not due to this paper of course, but other work.

The US may use the CD4< 200 as AIDS-defining, but that's simply because it is associated with a greatly increased risk of OIs. Additionally, HAART studies have shown worse prognosis when treatment is initiated at or below 200 (so, for example, current guidelines recommend treatment at a count of 350 even without OIs). The problem with AIDS is that it is an invisible disease process, that of immune-deficiency. HIV itself causes some direct effects, but largely the problem is the secondary infections. Definitions therefore vary depending on the location and need for accuracy versus sensitivity: this doesn't really affect treatment initiation (which in Canada is with the same criteria as elsewhere). This highlights the difference between a surveillance case definition and clinical practise. Another example is that in a trial I'm involved with, diagnosis of a certain viral infection is made by a central lab. Even if the local hospital tests come back as positive, if the central sample is negative then it'll be a negative datapoint on the clinical trial dataset. Unlikely to happen, but strangly true.

This is also perhaps a cautionary tale in getting info "passed on by a friend"! It doesn't take much to misread an article and suddenly it's all over the dissident discussion groups, and then it's dogma..! A quick look around suggests that didn't happen with this article however, but I've seen it before.

As regards Giraldo's work, I have no real doubt that it's real, but irrelevant. If you have to dilute a sample prior to running it or not, it doesn't really matter so long as the test still means something! The fact that you have to dilute it doesn't mean much above the fact that the antibodies are at a high level and cross-reactions are then seen more frequently. It's nothing astounding: in laboratory research antibodies are frequently diluted 1:100 or 1:10000 or whatever is needed to get a balance of sensitivity and specificity. Other serological tests also dilute the sample: a 5 minute google search brings up leishmania, RSV, influenza, measles... It's a standard laboratory technique.

Nick Bennett njb35@cantab.net

1. Viard et al. "Impact of 5 years of maximally successful highly active antiretroviral therapy on CD4 cell count and HIV-1 DNA level." AIDS. 18(1):45-49, January 2, 2004.

Competing interests: None declared