HIV-1 is a progressive infection accompanied by destruction of the immune system largely through a depletion of CD4 cell lymphocytes. Acute phase reactants, nonspecific indices of infection and inflammation, may also be affected as part of this process (1) and may, thus, be useful markers in predicting prognosis and effectiveness of intervention. For example, we have shown that the concentration of serum albumin (a “negative” acute phase reactant), even in ranges considered “normal,” is a powerful predictor of survival in HIV-1–infected women independent of the specific disease markers, CD4 lymphocyte cell count, and HIV-1 RNA level (2). C-reactive protein (CRP), another inflammatory marker, has been found to be of independent prognostic value in a variety of disease processes, such as atherosclerosis, end-stage renal disease, and cancer, but has not yet been examined in depth as a prognostic factor in HIV infection (3–5). Given the relatively low cost of CRP, if proved to be useful as a prognostic tool, it may have a role in monitoring HIV-infected patients in regions where antiretroviral therapies are becoming available but where frequent monitoring with viral loads and CD4 counts is still cost prohibitive. In the current study, we examine the association of CRP with survival in HIV-1–infected women separately and in conjunction with other predictors.
Enrollment and Collection of Baseline and Outcome Data
From October 1994 to November 1995, 277 HIV-positive women and 82 HIV-uninfected control volunteers were enrolled at the Brooklyn, New York site of the Women's Interagency HIV Study (WIHS), a multicenter study of the natural history of HIV-1 infection. The women were recruited through HIV-1 primary care sites, drug treatment facilities, and community-based outreach organizations (6), and every 6 months, the women underwent an interview using a structured questionnaire and received a physical examination. Multiple gynecologic and blood specimens were collected at each visit. Notification of death of a participant was obtained from friends, relatives, and medical providers of the participants and through national and local death registries. Data on survival were collected up to March 15, 2000 and excluded four deaths from trauma.
Informed consent was obtained from all study subjects, and human experimentation guidelines of the US Department of Health and Human Services were followed. IRB approval was obtained at all participating institutions.
Of the 359 women enrolled, a sufficient volume for CRP assay was available for 75 HIV-uninfected and 209 HIV-infected women at baseline. Accordingly, for the purpose of correlating CRP with serum albumin, we used specimens from visit 3 (12 months after enrollment, which was the next available specimen) for the remaining 75 women. The survival analysis was limited to the 204 HIV-infected women with adequate specimens at baseline and at least one follow-up visit.
Acute bacterial infections in the cohort were assessed by several mechanisms, all of which suggested that such infections were uncommon. Women were screened by physical examination for acute illness at the time of the CRP measurement. Only 3 women were considered to be acutely ill, and none had an oral temperature >100°F. The levels of CRP among these 3 women were 0.40, 0.41, and 2.11 mg/dL. No woman had a white blood cell count that exceeded 11,000, and, in fact, there was no correlation in the cohort between white blood cell count and CRP level (Spearman r = .01, p = .92). The prevalence of sexually transmitted bacterial infections, such as chlamydia, gonorrhea, or pelvic inflammatory disease, in the cohort used for survival analysis was <1% (2 of 204). Thus, it is unlikely that coexisting bacterial infection influenced the CRP results.
HIV-1 RNA determinations were performed in a laboratory that participated in the National Institutes of Health, Virology Quality Assurance Laboratory Proficiency Testing Program. T-cell subsets were determined using standard flow cytometry performed in a laboratory certified by the AIDS Clinical Trials Group. Albumin was measured as a part of a standard clinical chemistry panel performed on a Vitros (Johnson & Johnson). Serum specimens were stored for up to 5 years at −70°C. C-reactive protein was assayed by rate nephelometry using a Beckman immunochemistry system (P/N 449760,465315). The system measures the rate of increase in light scattered from particles suspended in solutions as a result of complexes formed during an antigen–antibody reaction and is said to measure concentrations within a range of 0.4 to 12 mg/dL. However, for many patients, the laboratory was able to record measures <0.4. Therefore, to most completely use the available data, CRP was categorized in three ways. In the most conservative scheme (“manufacturer-recommended”), we ignored apparent variation <0.4, the nominal lowest level of detection (LLD), setting all such values ≤0.4 (the reference group), and established the following three groups: 0.41 to 0.6, 0.61 to 1.35, and ≥1.35 based on an equal number of women in the groups >0.4. In a second categorization, we used the Downstate Medical Center clinical laboratory's recommended reference range grouping, which was <0.2 (the reference group = negative), with the other groups being 0.2 to 0.9 (low positive), 0.91 to 1.49 (positive), and ≥1.5 (high positive). Finally, we ranked the values of CPR according to quartile, which produced a reference grouping that was <0.25 with the following groups: 0.25 to 0.34, 0.35 to 0.59, and ≥0.6.
Either logarithmic (log) transformations of the actual data or categorical groupings were used in the analysis of CRP, HIV-1 RNA, or CD4 cell count, since these distributions were highly skewed. The Pearson product moment correlation coefficients of the transformed data were used to examine bivariate correlations. The overall impact of CRP on survival was assessed using the Kaplan-Meier approach, whereas Cox proportional hazard models were used to estimate the independent contribution of CRP in multivariate analyses. The impact of a variable on survival was assessed based on the relative hazard (RH) and on its contribution to a change in the log likelihood ratio. All analyses were performed using SPSS 11.0.
The distributions of actual CRP levels using values registered below the manufacturer's recommended LLD for 277 HIV-infected and for 82 uninfected women are similar in shape and central tendency (t-test for log of means p = .73), and both are highly skewed to the right. At enrollment, the 204 HIV-1–infected women with at least one follow-up visit had the following median values: age, 35 years; CRP, 0.35 mg/dL; serum albumin, 43 g/L; CD4 cell count, 290; and HIV-1 RNA, 21,000. They were also characterized by prevalence rates of 32% for an AIDS-defining condition, 15.8% for current use of hard drugs (i.e., cocaine, heroin), and a median family income of $10,300.
For the 277 HIV-infected women with a specimen either at baseline or at 1-year follow-up evaluation, we computed the Pearson product moment correlation coefficient of serum albumin, and the log (CRP) with each other and with log (CD4 lymphocytes) and log (HIV-1 RNA), and BMI and age. The correlation coefficients for most pairs, although statistically significant (p < .05), are not very strong. The highest correlation coefficient was between serum albumin and the log (HIV-1 RNA); it equaled 0.2, indicating that variations in serum albumin accounted for only 4% of the explained variability in HIV-1 RNA. CRP level did not vary significantly by smoking status (p < .51).
The 204 HIV-1–infected women included in the survival analysis were observed for a median of 44.7 months (range, 0.7–56.6). During the study follow-up period, 49 of the women died (24%). Kaplan-Meier curves for categories of CRP value at enrollment and survival are shown in Figure 1, using the manufacturer's recommended LLD (i.e., ≤0.4 mg/dL) as the reference category. Patients who had higher CRP values experienced much shorter survival (p < .001). Table 1 shows the results of the Cox proportional hazards model incorporating serum albumin, CD4 cell count, HIV-1 RNA, CRP, BMI, and age, which were all statistically significant (p < .05) independent predictors. This analysis employed the most conservative method of ranking CRP (“manufacturer-recommended”), combining all values of CRP of ≤0.4 into the reference category. Nonetheless, CRP values >0.4 were associated with a threefold to fourfold increase in the RH of death independent of age, serum albumin, and the disease-specific measures CD4 cell count and HIV-1 RNA level. Even when the analysis was restricted to patients with levels <0.4, there was a significant log linear relationship between the actual CRP value and the risk of mortality (p < .04).
In addition to analysis above, which employed the manufacturer-suggested range for CRP, we further examined the relationship of survival to CRP using two other grouping schemes: 1) by quartile, and 2) laboratory recommended. Using these schemes and adjusting for age, serum albumin, CD4 cell lymphocyte count, and HIV RNA level, the adjusted RHs of mortality associated with CRP level in the highest category versus the reference (lowest) category were 1) 4.5 for the quartile grouping, and 2) 13.6 for the laboratory grouping. The change in the likelihood ratio for CRP level, not surprisingly, was significantly greater when the laboratory-recommended grouping of CRP values was used in the model since the range between the highest and reference categories was wider in this grouping than it was for the others.
The acute phase reaction includes a programmed increase in the liver's synthesis of certain proteins (acute phase reactants) and a decrease in the synthesis of others (negative acute phase reactants) in response to acute infection and other stimuli. Evidence of chronic stimulation of the “acute” phase reaction seems to indicate a worsened prognosis in a variety of chronic disease settings (7–9). We have previously shown in a larger series of HIV-infected women with apparently normal kidney and liver function that serum albumin, a negative acute phase reactant, is a powerful predictor of mortality in HIV-1–infected women independent of disease markers CD4 cell count and HIV-1 RNA (2). It has been reported that weight loss reduces CRP levels in obese postmenopausal women and weight loss is a common manifestation of HIV disease and HIV disease progression (10). Despite that, the current study shows that CRP predicts survival in HIV-1–infected women independent of serum albumin, BMI, and other well-established markers of HIV-1 disease progression. This is true even using the most conservative categorization of CRP values, i.e., risk more than tripled in those women whose CRP was >0.4 mg/dL (a relatively high and consequently insensitive cutoff value, which defined a reference group that included two thirds of the women).
Despite a modest correlation of serum albumin and CRP, serum albumin remained an independent predictor of mortality in the current study, after adjusting for CRP and disease-specific markers. This is important from economic and biologic perspectives. Because serum albumin and CRP are only weakly associated and each appears to be a strong predictor of survival, they may be especially useful when used in combination in underdeveloped countries, where there is a pressing need for relatively inexpensive and easy to obtain markers of prognosis that can be used to determine when to initiate or change antiretroviral therapy.
Among the 204 patients in the survival analysis, the mean and median values of CRP were 0.99 and 0.35, respectively, which are two to three times higher than levels reported in the cardiovascular literature (11,12). Because HIV-1 is a chronic progressive infection, it is not surprising that CRP values are higher in infected patients than in the general population. But when comparing CRP levels in HIV-1–infected and comparable uninfected women who were selected to be similar for demographic characteristics, drug use, and sexual behavior, no differences in CRP levels were found, suggesting that factors other than HIV-1 infection may play a role.
Several studies have shown that high sensitivity CRP assay, capable of quantifying CRP in the very low concentration range, improves prediction of cardiovascular events independently of lipid levels (13,14). For example, in assessing cardiovascular risk in women using 12 plasma measures, high sensitivity CRP was the most significant predictor of the risk of cardiovascular events (15). The comparatively young women in our study (median age, 35 years) are obviously at low risk of cardiovascular mortality or morbidity, and we did not use a high sensitivity CRP assay. Nevertheless, in the present cohort, quantification of concentrations of CRP reported <0.4 LLD level did improve risk stratification, despite our use of a low sensitivity assay. For example, compared with the lowest quartile (<0.25), women in the highest quartile (>0.6) had a fivefold increase in risk independent of other predictors. Using a normal cutoff of 0.2 for the reference group (recommended by the Downstate Medical Center Clinical Laboratory), the adjusted RH was >13 for women with CRP >1.5. Finally, even when the group was restricted to women with CRP levels <0.4 there was a statistically significant association of CRP level with mortality. The steeper risk gradient obtained when using the “laboratory-recommended” levels of CRP suggests that a high sensitivity assay would be a much better predictor and could have important clinical applications in the HIV setting. It would also be important to assess the increase in predictive ability that a high sensitivity assay produces beyond that of the low sensitivity assay.
In summary, we have shown that even low sensitivity CRP is an independent predictor of mortality in HIV-1–infected women. If confirmed and extended, using a high sensitivity assay, this finding will improve our ability to quantify risk in the clinical and research settings.
Data in this manuscript were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group with centers (principal investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium (Ruth Greenblatt, Herminia Palacio); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Alvaro Muñoz, Stephen J. Gange ). The WIHS is funded by the National Institute of Allergy and Infectious Diseases, and the National Institute of Child Health and Human Development, with supplemental funding from the National Cancer Institute, the National Institute on Drug Abuse, and the National Institute of Dental Research. U01-AI-35004, U01-AI-31834, U01-AI-34994, U01-AI-34989, U01-HD-32632, U01-AI-34993, U01-AI-42590, MO1_RR00079, M01-RR00083. The C-reactive protein testing for this study was performed by the Immunology/Immunochemistry laboratory staff at the SUNY Downstate Medical Center.
1. Fauci AS. The human immunodeficiency virus: infectivity and mechanism of pathogenesis. Science 1988; 238:617–22.
2. Feldman JG, Burns DN, Gange SJ, et al. Serum albumin as a predictor of survival
in HIV-infected women in the Women's Interagency HIV Study. AIDS 2000; 14:863–70.
3. Rifai N, Buring JE, I-Min L, et al. Is C-reactive protein
specific for vascular disease in women? Ann Intern Med 2002; 136:529–33.
4. Harris TB, Ferrucci L, Tracey RP, et al. Associations of elevated interlukin-6 and C-reactive protein
levels with mortality in the elderly. Am J Med 1999; 106:506–12.
5. Melchior JC, Niyongabo T, Henzel D, et al. Malnutrition and wasting immunodepression, and chronic inflammation as independent predictors of survival
in HIV infected patients. Nutrition 1999; 15:865–9.
6. Barken SE, Melnick SL, Preston-Martin S, et al. The women's interagency study (WIHS)-design, methods, cohort characteristics and comparison with reported AIDS cases in U.S. women. Epidemiology 1998; 9:117–25.
7. Gaby C, Kushner I. Acute phase proteins and other systemic responses to inflammation. N Engl J Med 1999; 340:448–54.
8. Gitlin JD, Colten HR. Molecular biology of acute phase plasma proteins. In Pick E, Landy M. Lymphokines, vol 14.
San Diego: Academic Press, 1987:123–53.
9. Munford RS. Statins and acute phase response. N Engl J Med 2001; 344:2016–8.
10. Tchernof A, Nolan A, Sites Ck, et al. Weight loss reduces C-reactive protein
levels in obese postmenopausal women. Circulation 2002; 105:564–9.
11. Ridker PM, Rifai N, Clearfield M, et al. Measurement of C-reactive protein
for the targeting of statin therapy in the primary prevention of acute coronary events. N Engl J Med 2001; 344:1959–65.
12. Danesh J, Whincup P, Walker M, et al. Low grade inflammation and cornary disease: prospective study and updated meta analyses. BMJ 2000; 321:199–204.
13. Rifai N, Ridker PM. C-reactive protein
: a new strong predictor of cardiovascular disease. Clinical Laboratory News Oct. 2001;12–4.
14. Rafai N, Tracy RP, Ridker PM. Clinical efficacy of an automated high-sensitivity C-reactive protein
assay. Clin Chem 1999; 45:2136–41.
15. Ridker PM, Hennekens CH, Buring JE, et al. C-reactive protein
and other markers of inflammation in the prediction of cardiovascular disease in women. N Engl J Med 2000; 342:836–43.