CD4+T Cell Surface CCR5 Density and Virus Load in Persons Infected with Human Immunodeficiency Virus Type 1
Year: 2000
Type: article
Abstract: Although CCR5 density may play a role in human immunodeficiency virus (HIV) disease progression, we suggest that the relationship between CCR5 density and virus load is more complex than was proposed recently by Reynes et al. [1]. The authors' conclusion that this relationship is independent of cellular activation was based on measurements of a single activation marker, HLA-DR, which has not been strongly associated with virus load or disease progression. Other factors of importance to HIV pathogenesis (e.g., cellular immune responses to HIV, HLA types, CCR2b, and viral characteristics) might have influenced the study's findings. Furthermore, the conclusion of Reynes et al.—that CCR5 density is independent of cellular activation—is in conflict with that of Husman et al. [2], who argued that CCR5 expression increased with HIV disease progression as a result of immune activation. The third figure and the text on page 929 in the article by Reynes et al. [1] support Husman et al.'s view, in that they report a significantly higher density of CCR5 on DR+ than on DR− CD4+ cells. The small number of HIV-infected patients analyzed by Reynes et al. [1] (14 not undergoing highly active antiretroviral therapy [HAART] and 6 undergoing HAART) prompts caution in the extrapolation of their conclusions to the general population. Also, analysis of figure 1A of this article suggests that an “outlier,” the right-most data point in the plot, may have skewed the statistical analysis. Without this observation, the Spearman rank correlation coefficient between plasma virus load and CCR5 density on all CD4+ T cells increases from r = .666 (see page 929) to r = .769 (P = .004), a value similar to the correlation coefficient, r = .0749 (P = .002), between virus load and CCR5 density on DR− CD4+ T cells. (The latter correlation coefficient also increases, albeit slightly, to r = .77 with deletion of the same data point.) Thus, if this 1 data point is a true outlier, is in error, or represents a patient who differed from others in some important, unanalyzed way, then the correlation from the remaining data suggests that CCR5 expression is influenced by immune activation, and it would contradict the authors' assertion (page 930), “We found a correlation between virus load and CCR5 density on DR− CD4+ T cells that was higher (r = .749, P = .002) than the correlation between virus load and CCR5 density for the whole CD4+ T cell population (DR+ and DR−)….” Without the outlying observation (see their second figure, panel B), the correlation coefficient between virus load and CCR5 density on DR+ cells (r = 75) is similar to that between virus load and CCR5 density on DR− cells (r = .77). A, Complete data from figure 1A of Reynes et al. [1], with fitted regression line and 95% confidence interval; B, the same data, fit, and 95% confidence interval, less 1 outlying observation. The elimination of the outlier observation from figure 1A also is appealing, because it results in an improved log—linear regression fit to the data with a lesser y-intercept: virus load p exp5.2 + 4.80 × 10−4] × CCR5 density)] (compare figures 1A and 1B, with and without the outlier observation, respectively). This equation explains 69% of the variance in virus load, in contrast to 35% of the variance when all 14 observations are included, and predicts a virus load in plasma of 182 copies/mL (95% confidence interval [CI], 23–1448 copies/mL) among persons with zero CCR5 density, that is, those who are homozygous for the 32 bp deletion, in contrast to 1408 copies/mL (95% CI, 131–15,087 copies/mL). The log-linear modeling prevents a prediction of zero virus load at zero CCR5 density. The regression model without the outlier observation, thus, is more in keeping with the unobservable virus load, and presumably unsustainable HIV-1 infection, among persons homozygous for the 32 bp deletion. Finally, we question the conclusion of Reynes et al., from their fifth figure, that CCR5 density decreases transiently with immune activation. First, this experiment appears to be based on only 1 sample of peripheral blood lymphocytes, infected with 1 R5-tropic HIV isolate. Second, their use of relative densities (i.e., “percentages of densities expressed on control un-infected cells cultured concurrently”), rather than an absolute scale, on the left y-axis in this figure is troublesome. Last, their results in this experiment may be due to their choice of phytohemagglutinin (PHA) to stimulate cells. Bleul et al. [3] have shown that interleukin-2 is a much more efficient and perhaps a more physiological stimulator of CCR5 density than PHA. In sum, we applaud Reynes et al.'s efforts and complete reporting of data, which allowed us to further analyze their results. We question, however, their analyses and interpretations, and we urge further research on the relationships among CCR5 density, immune activation, and HIV-1 virus load.
DOI: 10.1086/315845
Institution: New York University
Language: English
Topic: HIV Research and Treatment
Subfield: Virology
Field: Immunology and Microbiology
Domain: Life Sciences
Open Access status: bronze
OpenAlex ID: W2076501786