|Home | About | Journals | Submit | Contact Us | Français|
A decrease in circulating albumin levels after seroconversion has been reported as a predictor of disease progression in HIV-infected adults. We hypothesized that a similar decrease would be seen in pig-tailed macaques in early SIV infection, and that the degree of this decrease would be related to the pathogenicity of the infecting viral strain. Ten juvenile pig-tailed macaques were previously inoculated with virus derived from molecular clones representing different stages of infection: early (SIVMneCL8, n=2), intermediate (SIVMne35wkSU, n=2), late blood (SIVMne170, n=3), or late lymph node (SIVMne027, n=3). Albumin was measured in stored samples. Changes from baseline were evaluated by paired sample t tests and by linear regression with generalized estimating equations (GEE). Albumin levels decreased in the week after SIV inoculation (p=0.02), increased above baseline at week 5, then fell, returning below baseline by week 16 (p=0.03). In GEE modeling, albumin decreased significantly in both early and chronic infection (weeks 0–3, p<0.001; weeks 5–16, p=0.004) and this change differed significantly between infections caused by late versus early or intermediate virus variants (weeks 0–3, p=0.002; weeks 5–16, p=0.001). A decrease in albumin levels occurs in both early and chronic SIV infection, and is more marked in macaques infected with more pathogenic virus variants. These results suggest that both early and late events in SIV pathogenesis are influenced by properties of the infecting viral strain.
Since the advent of the HIV pandemic, biomarkers that could serve as predictors of disease progression or severity have been sought for clinical use. A number of biomarkers related to nutrition and gastrointestinal function, including serum albumin and microbial translocation of gastrointestinal flora, are related to disease progression in chronic HIV or SIV infection.1–5 Serum albumin was reported as a strong independent predictor of mortality in HIV-1-infected women after adjustment for known disease markers (including CD4 counts) and has been proposed as a useful tool for clinical monitoring.6 Moreover, the association between albumin and disease progression is not limited to chronic infection. Albumin levels decrease significantly during early HIV infection,7,8 and HIV-1 seroconverters in whom albumin levels decreased by >10% during the early infection period have been shown to have more rapid disease progression.8
Disease progression is at least partially determined by the pathogenicity of the infecting virus, which varies at different stages of infection. Late virus variants are phenotypically and antigenically distinct from infecting strains, replicate to higher levels, cause cytopathic effects in T cells, and are resistant to neutralizing antibodies, becoming increasingly fit for replication in the host.9–17 Epidemiologic studies have shown this phenotype to be associated with disease progression.18,19 In macaques infected with cloned SIV strains representing prototype variants from early-, intermediate-, and late-stage infection, late-emerging variants have had increased pathogenicity and accelerated disease progression.20 The experimental SIV/macaque model provides a setting in which to explore how early biomarkers predict AIDS pathogenesis, given a known time of infection and known characteristics of the infecting viral strain. Our hypothesis for this study was that infection with the more pathogenic SIV strains that led to accelerated disease progression would be associated with a significant, early decrease in albumin levels.
Banked plasma samples were used from a previous study of juvenile pig-tailed macaques (Macaca nemestrina), each infected intravenously with one of the four highly related SIV variants presented in Table 1.20 SIVMneCL8 is a macrophage-tropic, minimally cytopathic clone derived from a pig-tailed macaque that establishes a low viral set point (<2.4×103RNA copies per ml) more characteristic of HIV infection in humans than many SIV variants. The other three viruses are highly related: two evolved directly from SIVMneCL8 in infected animals (SIVMne35kSU and SIVMne170) and one evolved from an animal infected with a related isolate (SIVMne027). The intermediate-stage variant chosen was a viral chimera encoding four amino acid changes in the V1 region of env (SIVMne35wkSU) compared with SIVMneCL8, which encoded the dominant env sequences present at that time in the animal.21 SIVMne35wkSU is a neutralization-escape variant that induces a higher viral set point (2.4×105RNA copies per ml) than SIVMneCL8.20,22,23 SIVMne170 is a full-length molecular clone obtained 170 weeks after inoculation of a macaque with SIVMneCL824; it is T cell tropic, cytopathic, and induces a high level of viral replication in macaques (viral set point, 6.0×106RNA copies per ml).20 SIVMne027 was cloned directly from the lymph node,25 and is distinguished by its ability to be induced from resting T cells and replicate efficiently both in unstimulated macaque peripheral blood mononuclear cells and in dendritic cell–T cell cocultures.24–27 Infection with SIVMne027 also results in a high viral load set point (8.3×106RNA copies per ml).20
As described previously,20 juvenile pig-tailed macaques (M. nemestrina) 1–2 years old were inoculated intravenously with equal amounts (1×105 tissue culture infectious doses) of SIVMneCL8 (n=2), SIVMne35wkSU (n=2), SIVMne170 (n=3), or SIVMne027 (n=3). Inoculations were performed in April 1997 for late virus variants and in February–March 1998 for early and intermediate virus variants. From each inoculated macaque, serial samples of peripheral blood, sera, and plasma were obtained weekly through the first 8 weeks after inoculation and monthly thereafter. Each macaque was also carefully monitored for the duration of the study for clinical signs of disease. All animals were maintained and cared for in accordance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care and the Animal Care and Use Committee of the University of Washington.
Measurements of viral RNA (vRNA) in plasma were done by the Chiron branch DNA (bDNA) assay and expressed as RNA copies per ml plasma. CD4 lymphocyte counts were performed by standard flow cytometry. Stored serum or plasma samples were aliquoted and sent to the University of Washington Research Testing Service in May or October 2007 for albumin testing by chemical dye-binding.28 Time points assessed included baseline, weeks 1–3, weeks 5 and 6, and weeks 8, 12, and 16 from the time of infection. The normal range of albumin in macaques is 3.0–5.0g/dl.5
Fisher's exact test was used to evaluate the association between morbidity or mortality and virus type. Independent sample t tests were used to compare albumin values in different groups of macaques. A paired sample t test was used to compare albumin values at different time points to the value for a given macaque at baseline. Because the relationship of albumin with time was nonlinear, time since infection was divided into two periods: early infection, defined as week 0–week 3 (when the viral set point is established in macaques), and chronic infection, defined as week 5–week 16. The change over week 3–week 5 was excluded, as albumin increased in this time period for several monkeys. Simple linear regression was used to explore the relationship between change in albumin, virus type, and change in viral load in the first week of infection. To evaluate the relationship between infecting virus type and changes in albumin over all time points in the two periods, two separate regression analyses were performed using generalized estimating equations (GEE) with an exchangeable correlation matrix. Analysis for the chronic infection period was run with and without the week 5 data, to determine the effect of the week 5 increase on results.
Both animals infected with early virus had CD4 cell counts similar to baseline through 90 weeks postinoculation, when the experiment concluded. Those infected with intermediate virus survived to 90 weeks, although one suffered diarrhea and weight loss (>10% body weight), and both had CD4 cell counts that declined to ~430 cells/μl by 90 weeks after inoculation. Of the three macaques infected with late lymph node virus, one died at 24 weeks, one was euthanized at 84 weeks due to symptomatic disease, and a third was alive at 90 weeks but had very low CD4 counts (<200 cells/ml3) from week 69. The three animals infected with late blood virus all died or were euthanized due to symptomatic disease at 31 weeks, 71 weeks, and 90 weeks. Late virus variants were associated with both morbidity (p=0.03) and mortality (p=0.05).
Macaques infected with late virus variants had higher mean albumin levels at baseline than did macaques infected by early or intermediate variants (3.67mg/dl vs. 3.15mg/dl, p=0.02), but only one macaque infected with early virus had a baseline albumin level below the normal range (2.90mg/dl). Overall, mean albumin levels decreased in the week after SIV inoculation, from 3.46mg/dl to 2.94mg/dl (p=0.02). Levels remained low relative to baseline at week 2 (mean 3.13mg/dl, p=0.04) and week 3 (mean 2.91mg/dl, p=0.008), then increased significantly above baseline at week 5 (mean 3.87mg/dl, p=0.04). Despite this increase, albumin levels again fell, and were significantly lower than baseline by week 16 (mean 3.04mg/dl, p=0.03). The decrease that occurred after infection relative to baseline was more striking among macaques infected by late virus variants, with highly significant differences at weeks 1, 2, 3, 12, and 16 (p=0.002, 0.004, 0.004, 0.04, and 0.02). The increase above baseline seen at week 5 was of borderline significance (p=0.07) in these macaques. Figure 1 shows the changes in mean albumin levels over time both overall (Fig. 1A) and for the macaques infected by each virus type (Fig. 1B).
Overall, the percent decrease in albumin levels in early infection from baseline to week 4 was larger in macaques infected with late virus variants than in those infected with early or intermediate virus variants, but this difference was not significant (23.0% vs. 10.0% decrease, p=0.07). Albumin decreases were also more notable among macaques that became ill (19.7% vs. 13.4% decrease, p=0.44) or were euthanized (22.0% vs. 14.6% decrease, p=0.26) during the study period. In a linear regression analysis with viral load at week 1 as outcome (i.e., the rate of increase of virus through the first week of infection), there was a significant interaction between the decrease in albumin levels seen at week 1 (i.e., the rate of decay of albumin through the first week of infection) and the type of virus (p=0.04), with a greater decrease seen in macaques infected with late virus variants. In other words, there is a significant difference in the relationship in the first week of infection between viral replication and the decay of albumin, based on the type of infecting virus (early versus late).
In linear regression using GEE, albumin decreased significantly over time in early infection (weeks 0–3, p<0.001). In multivariate modeling, changes in albumin differed significantly between late virus variants and early or intermediate virus variants, as shown by a significant virus-by-time interaction term (also known as the “treatment effect,” p=0.002). Figure 2 presents the changes in albumin during this period for each virus type and results of the multivariate GEE analysis.
In the chronic period after the viral set point had been reached, there was also a significant decrease in albumin over time (p=0.004). Again, changes in albumin over time differed significantly by virus type (p=0.001 for the virus-by-time interaction). Figure 3 presents the changes in albumin during this period for each virus type and results of the multivariate GEE analysis. If the peak in albumin levels we observed at week 5 is excluded, the virus-by-time interaction is still significant for late virus variants compared to early and intermediate virus variants (p=0.03).
The highly related cloned viruses used in this study lead to well-documented differences in outcomes after the infection of the macaques.20 Stored samples from this earlier study allowed us to examine the relationship between changes in albumin after SIV infection and viral pathogenicity. In summary, we found that a decrease in albumin levels occurs in early SIV infection, and is more marked in macaques infected with more pathogenic virus variants. The degree of change in albumin levels was significantly related to the infecting virus variant and to the rate of growth in viral load in the first week after infection. Despite a compensatory peak at week 5, albumin levels continued to decrease in chronic infection, especially in macaques with more rapid disease progression. Although variability in response was evident among hosts infected with the same virus variant, these results suggest that both early and late events in SIV pathogenesis are defined in large part by properties of the infecting viral strain. We were unable to establish whether the more rapid decay in albumin in macaques infected with late virus variants is due to the more rapid rate of viral replication we observed or to some other factor associated with pathogenic infection.
Luetscher observed in 1947 that “the common denominator of almost every pathological state is a relative or absolute decrease in the serum albumin.”29 Albumin has prognostic value in a number of conditions, and early studies investigating markers for advanced HIV infection found decreased albumin to be a marker for disease progression.2,3,6 Studies of early HIV infection have examined the association of albumin with clinical outcomes,7,8 and found that its prognostic significance was independent of viral load. This is the first study to investigate the relationship between infection with virus strains of differing pathogenicity and albumin followed at multiple time points, and we have found a significant decrease that differs by virus strain. Indirect evidence that albumin levels increase as viral replication is suppressed is provided by a Nigerian study demonstrating that albumin increases after antiretroviral therapy initiation30; however, viral strains were not characterized in this study and differences between patients in terms of treatment response were not reported.
The cause of the decrease in albumin levels occurring in early SIV or HIV infection is not clear. Albumin levels are known to decline as part of the acute phase response,31 and this is a possible mechanism.29 However, we found significant decreases in albumin in our study of Kenyan women after HIV-1 seroconversion in the absence of elevated C-reactive protein levels.8 Renal loss is a possibility, but proteinuria is not common in acute HIV infection.32 Hepatic dysfunction is also possible, although the increase in albumin at week 5 argues against impaired synthetic capacity. The albumin synthesis rate can increase by over 2-fold with severe albumin loss, and the highest synthesis rate reported in humans occurs during loss from the gastrointestinal tract.29 An intriguing possibility is that the changes we observed may be a result of SIV-induced gastrointestinal enteropathy. Indeed, the SIV macaque model provides the opportunity to explore the question of why albumin levels have been predictive of disease progression by taking advantage of controlled experiments such as this, in which macaques are monitored closely after inoculation with viral isolates of known pathogenicity.
By as early as 28 days after SIV infection, multiple microabscesses have been observed in the small bowel of infected macaques.11 Massive apoptosis of intestinal epithelial cells has been reported in the first 1–4 weeks after SIV infection, triggered by established mechanisms of gut epithelial cell apoptosis, and possibly by direct virus interactions with GPR15/Bob, an intestinal epithelial cell-associated alternative coreceptor for SIV and HIV-1.33 This apoptosis of intestinal epithelial cells occurs at the same time that CD4+ memory T lymphocytes in gut-associated lymphoid tissue undergo massive destruction.34–38 As this happens, a vigorous proinflammatory cytokine cascade is initiated in the colonic mucosa,39–42 with collagen deposition and disruption of the lymphatic tissue architecture.43 This inflammation contributes directly to lymphocyte destruction,44 but lymphocyte depletion is not sufficient to induce AIDS.45,46 Further research is needed to investigate whether albumin loss is related to destruction of the mucosal epithelium and whether the pathogenicity of the infecting viral strain is an important determinant of gastrointestinal injury in early SIV infection.
Our study has several limitations. First, only 10 macaques were included in the study, leading to a very small dataset. Despite this limitation, highly significant associations were found between albumin, time, and virus type in early SIV infection. Second, samples had been in storage for several years when assayed. However, albumin levels change minimally with storage (approximately 0.5% increase per year),47 and because all macaques were inoculated within 1 year, differences in storage time are unlikely to have affected our results. Finally, the macaques in this study were not monitored for potential causes of albumin loss and no gastrointestinal biopsies were taken, limiting our ability to investigate potential mechanisms for the observed decrease in albumin levels.
In conclusion, albumin levels decrease in early SIV infection in macaques, as in early HIV infection in humans, further validating albumin as a biomarker for AIDS progression. The cause of this decrease is not clear, but may be linked to damage to the gut in early SIV infection. Given the significant differences we detected with only a small number of animals, albumin could be a useful marker to evaluate disease progression in vaccine studies and other pathogenesis research using the nonhuman primate model. In our study, a greater decrease in albumin levels was associated with infection with more pathogenic virus variants. This may explain, in part, the association with more rapid disease progression seen in earlier studies.7,8
This work was supported by National Institutes of Health (NIH) Grant AI34251. Dr. Graham is supported by a Clinician Scientist Award from the University of Toronto and by an NIH Mentored Patient-Oriented Career Development Award (K23 AI069990). We would like to thank LaRene Kuller of the Washington Regional Primate Center in Seattle for her assistance with verifying macaque outcome data.
No competing financial interests exist.