Gebo, Kelly A. MD, MPH; Gallant, Joel E. MD, MPH; Keruly, Jeanne C. CRNP, MS; Moore, Richard D. MD, MHS
The CD4 lymphocyte is the biologic target of HIV, and the depletion of these cells is recognized as an important measure of the clinical stage of HIV infection. 1,2 As early as 1988, it was determined that an absolute CD4 threshold of 200 cells/mm3 could define when prophylactic treatment should be initiated to reduce the risk of Pneumocystis jiroveci (formerly carinii) pneumonia. 3,4 Other early studies identified the importance of an absolute CD4 threshold of 50–100 cells/mm3 for an increased risk of Mycobacterium avium bacteremia, toxoplasmosis, and cytomegalovirus infection. 5–7 In 1993, the US Centers for Disease Control and Prevention expanded the AIDS surveillance case definition to include all HIV-infected persons who have an absolute CD4 count <200 cells/mm3 or a CD4% (CD4%) <14%, because of the increased risk of opportunistic disease. 8 The absolute CD4 count has also been an important surrogate marker used to determine the need for antiretroviral therapy and to measure response to therapy. Early guidelines recommended the initiation of antiretroviral therapy when the absolute CD4 count declined below 500 cell/mm3.9 More recent guidelines suggest that antiretroviral therapy should be started when the CD4 count is between 200–350 cells/mm3.10
The absolute CD4 count is a calculated product of the total lymphocyte count and the percentage of lymphocytes that are CD4 lymphocytes determined by a flow cytometry. 11 There are numerous sources of variability among the measured components used to calculate the absolute CD4 count, including day-to-day and diurnal variability. 12,13 In contrast, because it is directly measured, the CD4% is less variable on repeated measurement than the absolute CD4 count. 14,15 Early studies suggested that the CD4% has greater prognostic significance than the absolute CD4 count. 15,16 Discordance between the absolute CD4 count and the CD4% can occur when the measured CD4% is unexpectedly low or high compared with the calculated absolute CD4 count. Examples of discordant pairs would include an absolute CD4 count of 200 cells/mm3 with a CD4% of 22%, or an absolute CD4 count of 350 cell/mm3 with a CD4% of 10%. A study of nearly 10,000 patients over a 10-year period indicated that there is an 8% discordant rate in men with early infection. 17 The rate of discordance in later infection is not well documented. When a discordant CD4-CD4% pair is encountered in clinical practice, the question arises as to whether prophylactic and antiretroviral therapy should be guided by the absolute CD4 count or the CD4%. The current study was designed to address that question by assessing the risk of clinical disease across a wide range of CD4-CD4% pair measures in a cohort of HIV-infected patients.
The Johns Hopkins AIDS Service provides care for a large proportion of the HIV-infected patients in the Baltimore metropolitan area. Longitudinal primary and subspecialty care are integrated in a hospital-based HIV clinic. An observational, longitudinal, clinical database has been maintained on patients receiving primary HIV care in our HIV/AIDS Clinic since 1990. 18 Consenting patients presenting for care at the clinic are simultaneously enrolled into the cohort. Comprehensive demographic, clinical, therapeutic, and laboratory data are collected at baseline (enrollment) and are updated at 6-month intervals using structured data collection forms and coding criteria. Professionally trained abstractors update these data using the records from clinic and inpatient visits (at Johns Hopkins and elsewhere), laboratory testing, pharmacy, social services, and all other available clinical sources. A 4% random sample of abstracted data are validated by independent, trained individuals, with correction of data and retraining of primary abstractors as necessary. Dropouts from our cohort are relatively uncommon, averaging only 9% (defined as >18 months out of care) since 1990. Laboratory data are collected from all locations at which they are obtained. Over 90% of the laboratory data are obtained in the Johns Hopkins Hospital Department of Pathology. Two other major commercial laboratories are used depending upon the patient’s insurer, and these data are also collected. This observational cohort collects all CD4 count and CD4% pairs that are obtained by the patients’ providers. The timing of laboratory testing generally follows practice guidelines for the management of patients with HIV infection 19 but is not imposed by this study. This cohort study was approved by the Johns Hopkins Institutional Review Board.
Prior to January 1996, CD4 measurement was calculated as a product of the total lymphocyte count and the CD4%. At Johns Hopkins, CD4 measurement was changed to a direct flow cytometry method in 1996. This method allows for direct measurement of the CD4 count, as opposed to a calculation. All CD4-CD4% pairs that were measured on the patients in our cohort since January 1996 were used in this analysis. For each CD4-CD4% pair, we assessed whether an AIDS-defining opportunistic illness (ADI) occurred after the measurement of the subsequent CD4-CD4% pair. The time after the pair was obtained was limited to a maximum of 180 days if no subsequent pair was obtained, otherwise the next CD4-CD4% pair was used for the next interval. If the next CD4 count was >350 cells/mm3, the analysis was truncated at the date of the new CD4 count. The median time of risk assessment for development of an ADI was 70 days (range: 14–180 days). For this analysis, we stratified the absolute CD4 count as follows: <50, 50–100, 101–200, and 200–50 cells/mm3. CD4% was stratified as follows: <7, 7–14, 15–21, and >21%. The analysis was limited to patients with an absolute CD4 count of ≤350 cells/mm3.
The incidence (number of ADI events/person-years of time) was calculated for development of any ADI for all 16 possible absolute CD4-CD4% categories using the 4 absolute CD4 count strata and the 4 CD4% strata. In addition, we used negative binomial regression with a generalized estimating equation to adjust for repeated measures to compute the incidence rate ratio (IRR) for developing an ADI by all of the possible absolute CD4 count and CD4% strata. Independent correlation was used for the generalized estimating equation. Negative binomial regression is more robust than Poisson regression for variation in the assumption that the variance is equivalent to the mean of the distribution. 20
The IRR is interpreted as the relative increase in the rate of developing an ADI for each category, compared with a selected reference category. In a multivariate analysis, we further adjusted the computation of the IRR by age (≥40 years vs. <40 years), sex, race (African American vs. all others), HIV transmission risk factor (injecting drug use vs. others), the HIV-1 RNA level (≥100,000 vs. <100,000 copies/mL), and use of highly active antiretroviral therapy (HAART). HAART was defined as the use of one or more protease inhibitors or a non-nucleoside reverse transcriptase inhibitor with nucleoside reverse transcriptase inhibitors, or the use of ≥3 reverse transcriptase inhibitors or a protease inhibitor/nonnucleoside reverse transcriptase inhibitor combination. We also examined interaction terms in the multivariate analysis as the product of the CD4-CD4% and each of the other above-listed covariates. A significant interaction would indicate that the association between the CD4-CD4% categories was different for each of the categories of the covariate (e.g., in patients receiving HAART, the association between the CD4-CD4% and ADI was different than the association in patients who were not receiving HAART).
We analyzed 15,736 absolute CD4-CD4% pairs from 2185 patients who were followed in our cohort after January 1996, with follow-up extending to July 2002 (Table 1). Our study population was predominately male (68.6%), African American (77.5%), and had injection drug use (45.8%) or heterosexual sex (45.8%) as an HIV risk factor. The mean age of our sample was 37.6 years, with a range between 17–77 years, and 59% received HAART. Of those with an initial CD4 cell count of <200 cells/mm3, 76.9% received P. jiroveci prophylaxis (PCP), and of those with an initial CD4 cell count of <50 cells/mm3, 73.5% received M. avium complex (MAC) prophylaxis.
During the study interval, there were 608 ADIs. Three hundred thirty-seven patients (15.4%) had one ADI, 70 (3.2%) had 2 ADIs, and 38 (1.7%) had ≥3 ADIs. The distribution of each ADI is demonstrated in Table 2. The most common diagnoses were Candida esophagitis, herpes zoster, and P. jiroveci pneumonia.
The distributions of absolute CD4 count and CD4% are shown in Table 3. As would be expected, most patients were concordant, and there were relatively few pairs of extreme discordance. There was no difference in rates of discordance by HAART usage.
The incidence of ADIs for each of the 16 absolute CD4-CD4% categories is shown in Figure 1. The incidence rate of ADIs is highest in patients with an absolute CD4 cell count of <50 cells/mm3, with a progressive decrease for CD4 counts of 51–100, 101–200, and 201–350 cells/mm3. The ADI incidence rate is also highest in patients with a CD4% <7%, with a progressive decreases for CD4% of 7–14%, 15–21%, and >21%. The IRR for developing an ADI by absolute CD4 category was 17.9 (95% CI: 13.2, 24.4) events/100 person-years for <50 cells/mm3, 6.2 (95% CI: 4.4, 7.9) for 50–100 cells/mm3, and 2.7 (95% CI: 1.9, 4.0) for 100–200 cells/mm3, compared with the referent stratum of 200–350 cells/mm3. Without adjustment for absolute CD4, the IRR for developing an ADI by CD4% was 14.4 (95% CI: 9.3, 22.6) for <7%, 3.7 (2.4, 5.9) for 7–14%, 1.9 (1.1, 3.1) for 15–21%, compared with the reference group of >21%.
The adjusted IRR of ADI is shown in Figure 2 for each of the 16 categories. The IRR is the relative increase in the rate of ADI for each CD4-CD4% category compared with the reference category of absolute CD4 250–350 cells/mm3 with CD4% >21%. The IRR was adjusted for patient age, sex, race, HIV transmission risk group, HIV-1 RNA measurement, and use of HAART. In multivariate analysis, the absolute CD4 count was associated with developing an ADI (Table 4). After adjusting for the absolute CD4 count, however, the CD4% was not associated with ADI development. In addition to the absolute CD4, other significant variables associated with development of an ADI were female gender, HIV-1 RNA >100,000 copies/mL, and no use of HAART. There were no significant interactions between any of these variables and either the CD4 or the CD4% variables. This indicates that the association found between absolute CD4 count and risk of ADI does not change based on the values of these variables.
A sensitivity analysis was performed repeating this same regression and stratifying by HAART usage. The trends for those on and off HAART were unchanged (data not shown).
Discordance between the absolute CD4 count and the CD4% frequently complicates treatment decisions regarding the need for opportunistic illness prophylaxis, use of antiretroviral therapy, and determination of prognosis. Although CD4-CD4% pairs are frequently concordant, an unexpectedly high or low CD4% can occur with a particular absolute CD4, making it difficult to assess the subsequent risk of clinical disease. Our analysis, of CD4-CD4% pairs obtained during the clinical care of adult HIV-infected patients followed in a large observational clinical cohort indicates that the absolute CD4 count is the surrogate measure that is most predictive of subsequent short-term risk of developing an ADI, and that the CD4% adds relatively little additional information that would change a treatment decision based on the absolute CD4 count. It is important to note that these data are based on patients followed in an observational cohort who were receiving treatment according to current guidelines for HIV-1–infected individuals, which include consideration of the CD4%. Thus, these data must be interpreted cautiously and not used to alter treatment recommendations.
Earlier studies done before the availability of HAART suggested that CD4% was a more useful prognostic measure. 15,16 These studies in homosexual men assessed the risk of ADI from 1–3 years after the initial absolute CD4 and CD4% measurement. Of note, these patients had higher CD4 counts than in our study. 15,16 Consistent with the current HIV epidemic, we had a higher proportion of women and IDUs in our cohort than in previous work. In addition, we included only patients with evidence of immunosuppression with CD4 counts of <350 cells/mm3, and our study assessed the short-term risk of developing an ADI over a median of 90 days, after the CD4-CD4% was measured. We believe that this is a clinically relevant result because decisions regarding prophylactic and antiretroviral therapy must be based upon the most recent laboratory data obtained. Another possible reason for the discordance between our study and previous work is the increased accuracy of the measurement of CD4 cells in the HAART era compared with the pre-HAART era. CD4 counts are now direct measurements and no longer calculated from the total lymphocyte count.
In contrast to studies conducted prior to HAART, the majority of our patients received HAART. 15 Use of HAART alters the natural history of the immunosuppression that accompanies HIV infection. However, our analysis did not find a significant interaction between HAART use and CD4–CD4%, suggesting that the association of CD4 with ADI was not significantly modified by use of HAART. We would point out that our results do not necessarily extend to children with HIV infection, 21,22 to splenectomized patients, 23 or to pregnant women, in whom the CD4% may have greater prognostic significance.
Despite the large sample size of absolute CD4–CD4% pairs from our cohort, there were relatively few pairs with extreme discordance. Therefore, our estimates at these extremes are less robust, and we are unable to be as confident that both measures may not provide significant predictive information. However, it is notable that the proportion of extremely discordant pairs is very low, and the need to make a prognostic or treatment decision based on this degree of discordance will be uncommon.
The relative importance of the absolute CD4 count was not altered by the concomitant use of HAART. We specifically used CD4–CD4% pairs obtained after January 1996 to assess whether use of HAART affected subsequent therapeutic decision-making regarding opportunistic illness prophylaxis and prognosis. HAART use had no significant effect on the IRRs, nor were there differences in the relationship with ADI development by sex, race/ethnicity, HIV transmission risk group, or age. In addition, the HIV-1 RNA level failed to influence the associations we found.
In summary, our results suggest that the absolute CD4 count may be the more useful measure for decisions regarding use of opportunistic infection prophylaxis and antiretroviral therapy, as well as assessment of prognosis. Although it was previously obtained because it was necessary to calculate the absolute CD4, the CD4% adds little additional independent information. It is possible, however, that the CD4% may be useful in patients with extremely discordant CD4–CD4% values.
1. Farizo KM, Buehler JW, Chamberland ME, et al. Spectrum of disease in persons with human immunodeficiency virus infection in the United States. JAMA
2. Moore RD, Chaisson RE. Natural history of HIV infection in the era of combination antiretroviral therapy. AIDS
3. Phair J. Munoz, Detels R, et al. The risk of pneumocystis carinii pneumonia among men infected with human immunodeficiency virus type 1. N Engl J Med
4. Fischl MA, Dickinson GM, La Voie L. Safety and efficacy of sulfamethoxazole and trimethoprim chemoprophylaxis for Pneumocystis carinii
pneumonia in AIDS. JAMA
5. Porter SB, Sande MA. Toxoplasmosis of the central nervous system in the acquired immunodeficiency syndrome. N Engl J Med
6. Horsburgh CR Jr. Mycobacterium avium complex infection in the acquired immunodeficiency syndrome. N Engl J Med
7. Masur H, Ognibene FP, Yarchoan R, et al. CD4 counts as predictors of opportunistic pneumonias in human immunodeficiency virus (HIV) infection. Ann Intern Med
8. Centers for Disease Control. 1993 Revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR Morb Mortal Wkly Rep
9. Carpenter CCJ, Fischl MA, Hammer SM, et al. Antiretroviral therapy for HIV infection in 1998. JAMA
10. February 4 2002 Guidelines for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents. Department of Health and Human Services (DHHS) and the Henry J. Kaiser Family Foundation. (http://aidsinfo.nih.gov
). Accessed April 1, 2004.
11. Centers for Disease Control. Guidelines for the performance of CD4+ T-cell determinations in persons with human immunodeficiency virus infection. MMWR Morb Mortal Wkly Rep
12. Simson E, Groner W. Variability in absolute lymphocyte counts obtained by automated cell counters. Cytometry
13. Stein DS, Korvick JA, Vermund SH. CD4+ lymphocyte cell enumeration for prediction of clinical course of human immunodeficiency virus disease: a review. J Infect Dis
14. Hughes MD, Stein DS, Gundacker HM, et al. Within-subject variation in CD4 lymphocyte count in asymptomatic human immunodeficiency virus infection: implications for patient monitoring. J Infect Dis
15. Taylor JM, Fahey JL, Detels R, et al. CD4 percentage, CD4 number, and CD4:CD8 ratio in HIV infection: which to choose and how to use. J Acquir Immune Defic Syndr
16. Burcham J, Marmor M, Dubin N, et al. CD4% is the best predictor of development of AIDS in a cohort of HIV-infected homosexual men. AIDS
17. Le TP, Tribble DR, Zhou SYJ, et al. Clinical significance of discordant CD4 count and CD4 percentage in HIV-infected individuals. AIDS
18. Moore RD. Understanding the clinical and economic outcomes of HIV therapy: the Johns Hopkins HIV Clinical Practice Cohort. J Acquir Immune Defic Syndr
. 1998;17(suppl 1):S38–S41.
19. Yeni PG, Hammer SM, Carpenter CC, et al. Antiretroviral treatment for adult HIV infection in 2002: updated recommendations of the International AIDS Society-USA Panel. JAMA
20. Long JS. Regression Models for Categorical and Linear Dependent Variables
. Thousand Oaks, CA: Sage Publications; 1997.
21. Mofenson LM, Korelitz J, Meyer WA III, et al. The relationship between serum human immunodeficiency virus type 1 (HIV-1) RNA level, CD4 lymphocyte percent, and long-term mortality risk in HIV-1-infected children. J Infect Dis
22. Valentine ME, Jackson CR, Vavro C, et al. Evaluation of surrogate markers and clinical outcomes in two-year follow-up of eighty-six human immunodeficiency virus-infected pediatric patients. Pediatr Infect Dis J
23. Zurlo JJ, Wood L, Mackrell Gaglione M, et al. Effect of splenectomy on T lymphocyte subsets in patients infected with the human immunodeficiency virus. Clin Infect Dis
© 2004 Lippincott Williams & Wilkins, Inc.