IVD Technology
Magazine
IVDT Article Index
Soluble markers of immune system activation
Part 2: AIDS prognosis and the clinical utility of ß2-microglobulin
David A. George
Studies suggest that serial measurements of immune system activation markers could have value for monitoring the progress of HIV disease.
To help determine the best course of treatment for individuals infected with the human immunodeficiency virus (HIV), researchers and clinicians are increasingly looking to soluble markers of immune system activation. The hope is that such markers will provide useful prognostic data to supplement the markers of disease progression already in routine use.
The first installment of this article outlined the association between ß2-microglobulin (ß2M) and HIV-1 infection, noting that ß2M levels in serum generally increase as the disease progresses. This second installment further investigates the association between ß2M and HIV-1 infection, as well as the relationships among ß2M, the general immune-system response to HIV-1 infection, and HIV-1 viral load.
ß2M versus CD4 Count
Two recent studies investigated ß2M and CD4 T-cell lymphocyte (CD4) count, and explored the relationship between the two and how this relationship may relate to the onset of acquired immunodeficiency syndrome (AIDS) and death.
In a prospective study involving 539 HIV-infected individuals without AIDS, the researchers measured ß2M and CD4 count approximately every 1½ months for a period of 21 months.1 Data were separately analyzed with respect to disease progression to AIDS or death (see sidebar). Contrary to some of the reports discussed in the first installment of this article, univariate analysis indicated that both ß2M and CD4 count were significantly associated with an increased risk of developing AIDS (see Table I).
Rendering of the class I major histocompatibility complex H-2DD heavy chain (blue), complexed with 2-microglobulin (green) and an immunodominant peptide, P18-110 (red), from the HIV-1 virus envelope glycoprotein 120.
Furthermore, multivariate analysis—adjusting for ß2M and CD4 count as well as for treatment and demographic factors—demonstrated that both ß2M and CD4 count were independent markers of progression to AIDS. With death as the endpoint, instead of AIDS onset, multivariate analysis adjusted for the same parameters again indicated that both ß2M and CD4 count were strong prognostic markers (see Table I).
When the data were analyzed using only the baseline measurements rather than serial measurements (simulating the availability of only a single patient sample), univariate analysis showed a significant reduction in the predictive value of both ß2M and CD4 count for both progression to AIDS and death. Similarly, multivariate analysis using only baseline values resulted in altered predictive ability of both ß2M and CD4 count.
Analyzing their data further, the researchers noted that it is the most recent determination of ß2M and CD4 count that provides the most accurate prognostic data with respect to AIDS onset or death, and that, among patients with similar CD4 counts, those with higher ß2M serum levels are at increased risk of disease progression. The authors concluded that ß2M supplied information supplemental to and independent of CD4 count and that serial measurements of ß2M were more valuable than just a single measurement.
In another in-depth study, researchers evaluated CD4 count, ß2M, neopterin, and the relationship among these markers in 198 HIV-infected individuals for a period of approximately 10 years.2 This study investigated the relationship between the onset of AIDS and the levels of these markers at various time periods:
- Pre-seroconversion.
- Post-seroconversion 0–12 months.
- Post-seroconversion 13–30 months.
| Marker | Univariate | Multivariate |
| Relative Hazard of AIDS | ||
| ß2M (per g/L increase) | 1.79 (p < 0.0001) | 1.37 (p = 0.0012) |
| CD4 count (per log increase) | 2.5 (p < 0.0001) | 2.17 (p < 0.0001) |
| Relative Hazard of Death | ||
| ß2M (per g/L increase) | 2.06 (p < 0.0001) | 1.34 (p = 0.0004) |
| CD4 count (per log increase) | 2.99 (p < 0.0001) | 1.91 (p < 0.0001) |
Pre-seroconversion. The results showed that the general trends for the pre-seroconversion levels of these markers were consistent throughout the follow-up period. If pre-seroconversion CD4 count was high, CD4 count generally remained high. Likewise, if pre-seroconversion ß2M or neopterin levels were low, they remained low throughout. As in the first study discussed above, the markers were independent of one another.
Post-seroconversion 0–12 Months. During the first year post-seroconversion, CD4 count inversely correlated with both ß2M and neopterin levels. If first-year ß2M or neopterin was low, future CD4 count was high. Similarly, if first-year ß2M or neopterin was high, future CD4 count was generally low. As above, general trends were consistent for each marker: low first-year levels remained low, and high first-year levels remained high, throughout the follow-up period.
| Subject Groups | Pre- seroconversion | 0–12 Months Post- seroconversion | 13–30 Months Post- seroconversion |
| CD4 count—high | 1.0 | 1.0 | 1.0 |
| CD4 count—medium | 1.204 | 1.656 | 0.848 |
| CD4 count—low | 1.369 | 2.931* | 2.407* |
| ß2M—low | 1.0 | 1.0 | 1.0 |
| ß2M—medium | 1.014 | 1.628 | 1.794 |
| ß2M—high | 1.109 | 2.510* | 3.552* |
| Neopterin—low | 1.0 | 1.0 | 1.0 |
| Neopterin—medium | 0.898 | 3.645* | 1.565 |
| Neopterin—high | 1.319 | 4.235* | 2.261* |
| *Denotes statistically significant increased relative hazard (p < 0.05). | |||
Post-seroconversion 13–30 Months. During this period of monitoring, those patients with the greatest ß2M or neopterin increases had the lowest future CD4 count, and those with the smallest ß2M or neopterin increases had the highest future CD4 count.
Throughout this study, ß2M and neopterin levels correlated closely with one another. Although CD4 count correlated inversely to ß2M and neopterin levels, this association was much weaker. CD4 count had little effect on future ß2M and neopterin levels, but the reverse associations were quite strong: Changes in ß2M and neopterin levels correlated strongly with future CD4 count. In particular, ß2M and neopterin levels in the 0–12-month post-seroconversion period were the best predictors of future CD4 count. The authors thus proposed that immune system activation, marked by significant increases in ß2M and neopterin, is indicative of future CD4 depletion.
Cox proportional hazards models were used to estimate the relative hazard of progression to AIDS for the subjects studied (see Table II). Individuals with the lowest CD4 counts and highest ß2M and neopterin levels during the 0–12-month postseroconversion period were at the greatest risk of developing AIDS, while those with the highest CD4 counts and lowest ß2M and neopterin levels had the least risk of developing AIDS. No statistically significant correlation could be made between the onset of AIDS and any of the pre-seroconversion marker levels.
| Marker | p Value | Correlation Coefficient |
| ß2M | 0.0001 | 0.68 |
| TNFR-II | 0.0002 | 0.66 |
| IL-2R | 0.0008 | 0.62 |
| Neopterin | 0.001 | 0.61 |
TNF- | 0.01 | 0.54 |
The authors noted that although individuals with higher pre-seroconversion CD4 counts maintained higher CD4 counts throughout the course of infection, those higher CD4 counts did not correlate to a longer period before progression to AIDS. It simply resulted in these patients developing AIDS at a higher CD4 count than most others.
From their data, the investigators concluded that the response of the immune system to HIV-1 infection during the first year post-seroconversion has a significant impact on the course of disease and progression to AIDS, while the state of the immune system at the time of infection has less influence. It appears that during the first year after infection a relationship is established between the virus and the infected individual, and the patients with the best prognosis are those with the least-active immune systems, as indicated by low levels of ß2M and neopterin.
These results were corroborated in a similar study of the hazard of progression to AIDS, in which the researchers reported a similar link between CD4 count and both ß2M and neopterin.4 Although the correlation was stronger with ß2M than with neopterin, the investigators reported that hazard estimates made using CD4 count were enhanced by the additional information provided by ß2M and neopterin. This was especially true in the later stages of disease, when the CD4 count was low (100 cells/µl or less). The authors concluded that when CD4 count is known, ß2M determinations significantly augment the prediction of the hazard of progression to AIDS in the late stages of disease.
ß2M and HIV-1 Viral Load
In a study designed to determine the relationship between cytokine expression and HIV-1 viral load (the number of copies of HIV-1 RNA found per milliliter of plasma or serum), another group of researchers evaluated HIV-1 viral load, tumor necrosis factor- (TNF-), TNF receptor type II (TNFR-II), interleukin type 2 receptor (IL-2R), ß2M, and neopterin in 34 asymptomatic HIV-infected individuals with CD4 counts greater than 100 cells/µl.5 Of all the markers studied, ß2M had the highest correlation with HIV-1 viral load (p = 0.0001; see Table III).
In addition, when patients were staged into high and low HIV-1 viral load groups (above and below 40,000 copies/ml), statistically significant differences were noted between the two groups for all markers. On the other hand, when patients were staged according to their CD4 count (above and below 300 cells/µl), statistically significant differences were noted only for ß2M, neopterin, and TNFR-II; no significant differences were observed for HIV-1 viral load, TNF- and IL-2R.
HIV-1 viral loadA patient's HIV-1 viral load—the number of copies of HIV-1 RNA found per milliliter of plasma or serum—is closely associated with progression to AIDS. Several studies have asserted the excellent prognostic value of HIV-1 viral load determinations, noting that high HIV-1 viral load after primary infection is predictive of rapid progression to AIDS, while low HIV-1 viral load is associated with long-term asymptomatic disease.5,7–12 Measurements of viral load have rapidly come into routine clinical use for monitoring both disease progression and a patient's response to antiretroviral therapy.8,13,14 Although HIV-1 viral load is currently considered the best single predictor of progression to AIDS and AIDS-related death, its predictive ability can be improved by adding measurements of CD4 count and soluble markers of immune system activation. Refinement of the assay methods for HIV-1 viral load (including reverse transcriptase polymerase chain reaction, branched DNA signal amplification, and nucleic acid sequence-based amplification) make it useful not only to clinicians but also to researchers investigating the dynamics of HIV-1 replication in vivo.15–17 Using such techniques, Perelson et al. report that a high level of virus replication and CD4 T-cell lymphocyte destruction occurs at all stages of HIV-1 infection.18 Their results are summarized below.
Length of virus generation includes the entire cycle of virus release, infection of a new cell, cell death, and subsequent release of new virus particles. |
The researchers concluded that HIV-1 viral load correlated well with levels of both ß2M and neopterin in asymptomatic HIV-infected patients, strengthening the link between these immune system activation markers and HIV-1 disease progression. They also demonstrated the absence of a significant correlation between CD4 count and HIV-1 viral load.
In a recent prospective study, another group of investigators evaluated the ability of HIV-1 viral load, CD4 count, CD8 count, ß2M, and immune complex dissociated (ICD) HIV-1 p24 antigen to predict disease progression in 67 long-term nonprogressors (LTPNs; asymptomatic HIV-infected patients with CD4 count of at least 500 cells/µl).6 Each marker was measured at study entry and at six-month intervals for a period of two years.
| HIV-1 Viral Load (RNA copies/ml) | ||||
| Marker | < 200 (n = 11) | 201–10,000 (n = 30) | > 10,000 (n = 26) | p Value |
| CD4 count (cells/µl) median range | 819 561–1375 | 699 504–1195 | 616 507–1258 | 0.077 |
| CD8 count (cells/µl) median range | 861 580–1350 | 1188 432–2368 | 1091 476–2080 | 0.425 |
| ß2M (mg/ml) median range | 2.0 1.4–2.5 | 2.5 1.5–3.5 | 3.0 1.3–5.5 | 0.0005* |
| ICD p24 (pg/ml) median range | 5.0 5.0–5.0 | 5.09 5.0–13.9 | 5.0 5.0–100 | 0.0014* |
| *Denotes statistically significant correlation with HIV-1 viral load (p < 0.05). | ||||
When study subjects were staged by their baseline HIV-1 viral load levels, a statistically significant and strong correlation was evident between HIV-1 viral load and ß2M (p = 0.0005; see Table IV). A significant correlation was also noted between HIV-1 viral load and ICD p24, but it was less strong (p = 0.0014). When staged according to the rate of CD4-count decline, baseline elevations of HIV-1 viral load, ß2M, and ICD p24 all correlated with a more rapid decline in CD4 count, but only ß2M significantly predicted CD4 decline in both univariate and multivariate analyses (see Table V). This was true both for those patients receiving and not receiving antiretroviral therapy.
The researchers concluded that in LTNPs, ß2M is a stronger predictor of CD4 decline than is HIV-1 viral load. They also noted that elevated ß2M may indicate the occurrence of an immunological event before the appearance of clinical disease. For this reason, when used in conjunction with HIV-1 viral load and CD4 measurements, markers of immune system activation such as ß2M provide valuable prognostic information for determining the risk of disease progression in LTNPs.
| Marker | Rate of CD4 Decline Per Month | Univariate | Multivariate | ||||
| >1.5 | <1.5 | Odds Ratio | p Value | Odds Ratio | 95% Confidence Interval | p Value | |
| HIV-1 RNA (copies/ml) < 200 200–10,000 > 10,000 | 3 9 13 | 7 10 8 | 1.0 1.91 4.33 | 0.432 0.079 | 1.0 1.24 1.56 | (0.22, 7.04) (0.25, 9.68) | 0.811 0.636 |
CD4 count (cells/µl) 650> 650 | 13 12 | 12 13 | 1.0 0.85 | 0.777 | 1.0 1.47 | (0.41, 5.31) | 0.559 |
| CD8 count (cells/µl) 1000 > 1000 | 15 10 | 10 15 | 1.0 0.44 | 0.160 | 1.0 0.30 | (0.08, 1.12) | 0.073 |
| ß2M (mg/ml) 2.2 2.3–2.8  2.9 | 5 8 12 | 13 8 4 | 1.0 2.60 7.80 | 0.188 0.009* | 1.0 2.42 6.51 | (0.56, 10.39) (1.21, 32.07) | 0.234 0.029* |
| ICD p24 (pg/ml) Undetectable > 5 | 19 6 | 22 3 | 1.0 2.32 | 0.278 | 1.0 1.2 | (0.22, 6.40) | 0.832 |
| *Denotes statistically significant prediction of CD4 decline (p < 0.05). | |||||||
In studying the relationship between percent change in CD4 count over the course of a one-year period and the onset of AIDS, another group evaluated the associations among HIV-1 viral load, ß2M, neopterin, and TNF receptor 75 (TNFR-75).7 Cited as the most significant finding of their work, the researchers reported that soluble immune system activation markers correlated better with both single-year CD4-count decline and disease progression than did HIV-1 viral load. Neopterin was the best predictor of CD4-count decline and death, while in the subgroup of patients with CD4 counts greater than 200 cells/µl, TNFR-75 was the strongest predictor of the onset of AIDS. Of note, in patients with a baseline CD4 count of less than 200 cells/µl, no correlation was found for HIV-1 viral load. The investigators concluded that HIV-1 viral load determinations and CD4 counts should be combined with a marker of immune system activation to develop an optimal predictive model for the progression of HIV-1 disease (see box, page 44).
Conclusion
It is clear from the studies discussed here that an undeniable link exists between ß2M and HIV-1 disease progression. During the course of HIV-1 infection, ß2M serum levels correlate with CD4 count, AIDS onset, and death. In addition, the link observed between ß2M and HIV-1 RNA plasma load, as well as the apparent effect of immune system response during the first year of infection on subsequent disease progression, suggest that immune system activation may be involved in the amount of viral replication that occurs and, hence, may influence the clinical progression of disease.
Although the prognostic value of a single, early measurement is limited, serial measurements of immune activation markers such as ß2M have been shown to be very useful in monitoring the course of disease in HIV-infected individuals. Compared to HIV-1 viral load, soluble markers of immune system activation show a better correlation with CD4-count decline, disease progression, and death in LTNPs and in late-stage disease.
Future studies should include more in-depth evaluations of serial measurements of ß2M. Although immunoassays for neopterin, TNFR-II, and TNFR-75 are not available as readily as those for ß2M, similar time-course studies of those markers should also be considered.
Study methods and statisticsCorrelation coefficient. The correlation coefficient is the linear association between two random variables, x and y. Correlation coefficients range from –1 to +1. A positive value indicates a positive association: as x increases, y also increases; and as x decreases, y also decreases. A negative value indicates a negative association: as x increases, y decreases; and as x decreases, y increases. The larger the correlation coefficient, the stronger the association between x and y. Cox proportional hazards model. This statistical model is a logarithmic presentation of a set of data, such as ß2M serum levels, and its relationship to a progression from a specific starting point (e.g., the diagnosis of HIV-1 infection) to a defined endpoint (e.g., the onset of AIDS or death). In effect, this is a numerical presentation of the relative risk or likelihood of an event, with larger numbers indicating a higher risk of the event being evaluated. Friedman's test, Kruskal-Wallis test, Two-sided Mann-Whitney test, and Wilcoxon's test. These are examples of nonparametric rank tests. In rank tests, obtained data are replaced by values representing their relative rank. Nonparametric tests do not make distributional assumptions, meaning they do not involve population parameters. The tests do not assume that the data come from normally distributed populations, nor does their validity rely on the population distribution from which the data have been sampled. They do make some assumptions, however, such as equality of population variances. Tests such as these are typically used to compare data from two unpaired groups, with the data being ranked for analysis from low to high, regardless of the population from which they are derived. Least squares linear regression. A statistical method used to fit a straight line to a series of data. Multivariate longitudinal studies. In multivariate longitudinal studies, serial measurements of two or more variables are obtained from an individual over a period of time (e.g., CD4+ T-cell lymphocyte count, ß2M, and neopterin measured at three-month intervals). In analyzing the relationships among the variables, researchers must consider both the correlations among the variables taken at the same time and at different times. 95% confidence interval. For a marker under investigation, such as ß2M, the 95% confidence interval is that range of values within which 95% of the study population will fall. For instance, a 95% confidence interval of 1.21–32.07 for ß2M indicates that 95% of the study group will have ß2M values between 1.21 and 32.07 mg/ml. To avoid inadvertently biased results, confidence interval calculations should be used only in a random sampling of a population. Odds ratio. In an effort to identify factors that may cause harm to a study population, epidemiologists calculate the odds ratio for each of the factors involved. Values for odds ratios are between zero and infinity. An odds ratio of 1 is the neutral value, indicating that the factor being studied shows no difference between the control group and the study group. Harmful factors have odds ratio values greater than 1; higher values are associated with factors that are more harmful. Odds ratios are typically used in studies in which disease prevalence is not known. p value. The outcome of statistical testing is often expressed as a p value. This value is an indicator of the risk of a Type I statistical error, or the rejection of the null hypothesis. In effect, the p value describes the probability or risk of false-positive statistical error. As this type of error is undesirable, statistical significance is said to be achieved for a set of data when the risk of a false-positive error is low, at the 5% level or less (p ¾ 0.05). Univariate longitudinal studies. In univariate longitudinal studies, a single variable from one individual is measured over a period of time (e.g., CD4+ T-cell lymphocyte count measured at three-month intervals). In analyzing such data, correlations among the serial measurements are considered. |
References
1. AJ Mocroft et al., "The Relationship between ß2-Microglobulin, CD4 Lymphocyte Count, AIDS, and Death in HIV-Positive Individuals," Epidemiology and Infection 118 (1997): 259–266.
2. M Shi et al., "Early Levels of CD4, Neopterin, and ß2-Microglobulin Indicate Future Disease Progression," Journal of Clinical Immunology 17, no. 1 (1997): 43–52.
3. V Henne, P Frei, and P Bürgisser, "ß2-Microglobulin—A Rapid and Automated Determination for a Broad Range of Clinical Applications," Anticancer Research 17 (1997): 2915–2918.
4. F Yong et al., "Dependence of the Hazard of AIDS on Markers," AIDS 11, no. 2 (1997): 217–228.
5. J Salazar-Gonzales et al., "Relationship of Plasma HIV-RNA Levels and Levels of TNF-* and Immune Activation Products in HIV Infection," Clinical Immunology and Immunopathology 84, no. 1 (1997): 3645.
06. L Ashton et al., "Predictors of Progression in Long-Term Nonprogressors," AIDS Research and Human Retroviruses 14, no. 2 (1998): 117–121.
07. R Zangerle et al., "Serum HIV-1 RNA Levels Compared to Soluble Markers of Immune Activation to Predict Disease Progression in HIV-1-Infected Individuals," International Archives of Allergy and Immunology 116 (1998): 228–239.
08. S Riddler and J Mellors, "HIV-1 Viral Load and Clinical Outcome: Review of Recent Articles," AIDS 11, supp. A (1997): S141–S148.
09. JW Mellors et al., "Quantitation of HIV-1 RNA in Plasma Predicts Outcome after Seroconversion," Annals of Internal Medicine 122 (1995): 573-579.
10. S Jurriaans et al., "The Natural History of HIV-1 Infection: Virus Load and Virus Phenotype Independent Determinants of Clinical Course?" Virology 204 (1994): 223–233.
11. R Iuliano et al., "Correlation between Plasma HIV-1 RNA Levels and the Rate of Immunologic Decline," Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology 14 (1997): 408–414.
12. J Mellors et al., "Plasma Viral Load and CD4+ Lymphocytes as Prognostic Markers of HIV-1 Infection," Annals of Internal Medicine 126 (1997): 946–954.
13. DS Stein et al., "Predicting Clinical Progression or Death in Subjects with Early-Stage Human Immunodeficiency Virus (HIV) Infection: A Comparative Analysis of Quantification of HIV RNA, Soluble Tumor Necrosis Factor Type II Receptors, Neopterin, and ß2-Microglobulin," Journal of Infectious Disease 176 (1997): 1161–1167.
14. SM Bruisten et al., "Prospective Longitudinal Analysis of Viral Load and Surrogate Markers in Relation to Clinical Progression in HIV Type 1Infected Persons," AIDS Research and Human Retroviruses 13, no. 4 (1997): 327–335.
15. C Pachl et al., "Rapid and Precise Quantification of HIV-1 RNA in Plasma Using a Branched DNA Signal Amplification Assay," Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology 8 (1995): 446–454.
16. J Mulder et al., "Rapid and Simple PCR Assay for Quantitation of Human Immunodeficiency Virus Type 1 RNA in Plasma: Application to Acute Retroviral Infection," Journal of Clinical Microbiology 32 (1994): 292–300.
17. T Kievits et al., "NASBA Isothermal Enzymatic In Vitro Nucleic Acid Amplification Optimized for the Diagnosis of HIV-1 Infection," Journal of Virological Methods 36 (1991): 273–286.
18. A Perelson et al., "HIV-1 Dynamics In Vivo: Virion Clearance Rate, Infected Cell Life-Span, and Viral Generation Time," Science 271 (1996): 1582–1586.
David A. George is a product manager at Scripps Laboratories (San Diego).
