Immunologic marker paths for seroconversion: single determinations of immunoglobulin A and beta 2-microglobulin are not adequate to estimate time of HIV infection. Multicohort Analysis Project Workshop. Part II

Abstract

Objective: To investigate the usefulness of single determinations of serum immunoglobulin (Ig) A and beta 2-microglobulin (beta 2M) levels in estimating the time from HIV seroconversion.

Subjects: Five cohorts were represented in the workshop. The Multicohort Analysis Project (MAP) workshop database comprised 1744 HIV-infected patients with documented HIV seroconversion times, 363 of whom had AIDS. Overall, 1430 patients had two or more pre-AIDS CD4+ cell counts (13056 counts); 896 patients had two or more pre-AIDS IgA measurements (6081 observations); but only 2964 beta 2M measurements were available.

Main outcome measures: Marker paths for loge IgA and loge beta 2M and for square root of CD4+ cell count. Dependence on cofactors (age, sex, mode of HIV transmission) and attention to inter-individual variation in seroconversion level and annual decline in square root of CD4+ cell count. Logistic discrimination was performed using cofactor-adjusted paired loge IgA and loge beta 2M to date HIV infections as recent (within 3 years of seroconversion), intermediate, or distant (> or = 6 years after seroconversion). Transition intensities between immunologically defined states (using IgA or beta 2M) and to clinical AIDS.

Results: Linear functional form described the decline in square root of CD4+ cell count, except for the Italian cohort where a steeper annual loss of square root of CD4+ cell count occurred in the first year than thereafter. Square root of CD4+ cell count at seroconversion depended on age and mode of transmission. Annual loss of square root of CD4+ cell count was less severe in those infected by sexual transmission. Non-monotone functional form emerged for loge IgA with an initial decrease in the first year after seroconversion followed by an increase thereafter. Loge IgA levels were higher in older subjects and in those infected by sexual transmission, but lower in women. A quadratic growth curve described the marker path for loge beta 2M, which increased for 5-6 years after seroconversion but declined thereafter. Loge beta 2M values were significantly higher in older patients and injecting drug users, but lower in women. The considerable heterogeneity of marker paths between individuals affected all three markers, but marker values appeared to track within an individual. Discrimination based on paired IgA and beta 2M measurements from single blood samples performed poorly in classifying HIV infections as recent, intermediate or distant. High intensities of backward as well as forward transitions between immunologically defined states explained the poor discrimination based on single sample per individual.

Conclusions: Further study of how serum IgA reflects HIV infection and careful clinical and statistical assessment of reduced beta 2M from 5 or 6 years after HIV infection are needed. The dependence of marker paths on mode of HIV transmission suggests that sexual transmission may have implications for HIV disease progression. The feasibility of using serum IgA and beta 2M, evaluable in single stored blood samples, to estimate time of HIV infection has been set back by our results.