The recent paper by Owens and colleagues [1] presents a meta-analysis of the performance of polymerase chain reaction (PCR) for the diagnosis of human immunodeficiency virus type 1 (HIV-1) infection. The authors do not comment on either the genetic sequences to which the PCR primers are targeted or the genetic differences documented among the subtypes of HIV-1. Variability in PCR across subtypes was the subject of a recent UNAIDS meeting, at which the need for further assay development was identified [2].

At the Walter Reed Army Institute of Research, a DNA PCR assay is used to supplement serologic testing for HIV diagnosis. The gag primer pairs used, both of which must be reactive for a positive result, are commercially available and widely used. We have investigated the assay's sensitivity for non-B-subtype viruses. Cell pellets from co-cultures of genetically typed viruses (n = 28) were selected to represent the major HIV-1 subtypes. In a blinded manner, DNA was extracted and serial dilutions were assayed by PCR. One pair of primers (SK462/SK431) amplified DNA from all specimens. The other pair (SK38/SK39) amplified only 3 of 5 preparations from subtype A, 4 of 5 from subtype B, 3 of 5 from subtype C, 2 of 5 from subtype D, 4 of 5 from subtype E, and 0 of 3 from subtype F.

Thus, although we agree with the recommendations about study design by Owens and colleagues [1], the level of PCR sensitivity derived from their analysis should not be assumed to be valid for non-subtype-B viruses. Using available nucleic acid sequences, it is now possible to select more "universal" primers for all known subtypes of HIV-1. Modification of amplification conditions may also improve sensitivity across multiple clades but, to our knowledge, no DNA PCR methods have been verified to amplify all subtypes equally. It would serve a useful surveillance function to have both "universal" and subtype-B-specific primers as part of a diagnostic PCR algorithm.