Oral lesions as markers of severe immunosuppression in HIV-infected homosexual men and injection drug users.

OBJECTIVES

We examined the diagnostic utility of the presence of oral lesions, individually and in combination, in identifying severe immunosuppression, defined as CD4 cell count under 200.

STUDY DESIGN

Data were collected on 82 HIV-seropositive homosexual men and 82 HIV-seropositive injection drug users who volunteered to participate in a longitudinal study of HIV infection. CD4 cell counts were measured within 24 hours of oral examination.

METHODS

Sensitivity, specificity, positive predictive value, negative predictive value, and the odds ratio were computed to assess the association between oral lesions and CD4 less than 200. In addition to the individual lesions, we studied the diagnostic properties of sets of three to six lesions. For each set of lesions, a patient was classified as positive for the set if he or she had one or more lesions in that set.

RESULTS

In homosexual men and injection drug users, individual lesions had low sensitivity, high specificity, and moderate positive and negative predictive values. Odds ratios reflected weak correlation to immunosuppression. When lesion sets were considered in homosexual men, sensitivity rose dramatically with only modest decreases in specificity. The positive and negative predictive values remained almost the same. Similar results for lesion sets were obtained in injection drug users, with greater reduction in specificity but stable positive and negative predictive values. Odds ratios indicated that for homosexual men, the more lesions included in the set, the stronger the correlation with immunosuppression. For injection drug users, strong correlations were observed for all lesion sets.

CONCLUSIONS

Analysis of sensitivities and odds ratios in homosexual men suggest that it may be valid to note the occurrence of a greater number of oral lesions than is currently done in staging patients with HIV infection. Among injection drug users, monitoring a larger number of lesions neither improves nor reduces the correlation to severe immunosuppression.