Hard to believe, but we have to get rid of the HIV Western blot -- at least as our HIV confirmatory test.
Here's why (case adapted from several seen the past few years; I'm sure most of you have seen similar):
30-year-old man, high risk for HIV. He's worried he might have become infected due to recent exposures, so he requests testing at a community health center.
The clinician does an oral rapid test, which is positive, then sends a blood sample to the lab for confirmation.
The ELISA test is also positive, so the lab sends this sample off-site for confirmation by Western blot.
One week later (the Western blots always take longer than you think they should), the result comes back:
Does that worry you? If not, it should -- the guy's HIV viral load was > 1 million copies/mL. Not only that, but the clinician who did the testing told him that he should return in 6 weeks for a repeat test to see if he fully seroconverts. The patient suspected something might be wrong with that advice (you think?) and sought a second opinion from one of my colleagues, who sent the viral load.
The case highlights what Bernie Branson from the CDC has been telling us for years, which is that the Western blot is lousy at detecting recently acquired HIV. (Great CROI 2012 presentation by Bernie here, when you have a moment.) The Western blot is barely more effective as a "confirmatory" test in this setting than if the sample had been sent for Tropheryma whipplei antibody -- and we all know how often those are helpful.
(Little inside ID joke there.)
The fact is that there is a 40-day delay from when "4th Generation" antigen/antibody combination test turns positive to when the Western blot does so. That's a long time for a person not to know whether he/she has HIV, especially since this is the most contagious period in all of HIV infection.
So what should we do? Chaired by Eric Rosenberg, The Clinical and Laboratory Standards Institute (CLSI) issued a new testing algorithm (M53-A) in July, and the CDC will soon follow with similar recommendations. In essence, the most important algorithm will look like this (figure from Bernie's excellent 2010 paper):
One immediate problem is that most labs don't have an "HIV-1/HIV-2 differentiation immunoassay". Only one such test -- the Bio-Rad Multispot HIV-1/HIV-2 Rapid test -- is FDA approved, although apparently another test is in the late stages of development.
But until such assays are widely available, here's some practical advice we can give clinicians today:
Current HIV screening tests (ELISA IgM or ELISA/Ag combinations) are much more sensitive than the Western blot.
Every case with a positive screening test but a negative Western blot must have an HIV viral load.
Every case in which HIV acquisition might have been recent -- symptoms of acute HIV, or HIV testing in the context of a recent STI (especially syphilis), or known recent negative HIV test -- should have an HIV viral load sent along with HIV antibody testing (or at the very least, be tested using a 4th Generation combined antigen/antibody test).
Check out this article from the March 2012 issue of CAP Today for a complete discussion.
And to the HIV Western blot, hey, it's been great -- I thank you for your 20+ years of excellent service! Now it's time for us to move on.
Paul Sax is Clinical Director of Infectious Diseases at Brigham and Women's Hospital. His blog HIV and ID Observations is part of Journal Watch, where he is Editor-in-Chief of Journal Watch AIDS Clinical Care.