Quantitative HIV-1 RNA, CD4+ lymphocyte counts, and the presence of HIV-related clinical disease prior to initiation of highly active antiretroviral therapy (HAART) have been demonstrated to predict AIDS-defining illness (ADI) and death after HAART initiation, in both clinical trials and cohort studies. 1–3 CD4+ cell counts and HIV-1 RNA are used in resource-replete settings as primary indicators of when to initiate HAART. 4,5 As HAART becomes less expensive and more readily available in resource-limited settings, CD4+ lymphocyte and HIV-1 RNA determinations remain prohibitively costly, and less expensive predictors of outcome after HAART initiation are needed.
Total lymphocyte count (TLC) has been demonstrated to correlate with and have a similar tempo of decline as CD4+ cell count, to predict HIV disease progression in untreated individuals, 6–8 and to predict virologic and immunologic response after HAART initiation. 9 TLC has been suggested as a measure of when to initiate both HAART and prophylaxis against opportunistic illness in resource-limited settings. 7–12 Guidelines provided by the World Health Organization (WHO) recommend starting HAART in those with WHO stage 2 or 3 disease with TLC <1200 cells/μL. 13 However, pre-HAART TLC as a predictor of clinical outcomes after HAART initiation has not been well characterized.
Progression to AIDS and death are also predicted by the presence of anemia both in HAART-treated and -untreated individuals. 14–16 Anemia has been demonstrated to improve after initiation of HAART 17 and has been suggested as a parameter for initiation and monitoring of HAART. 10,12 Delayed-type hypersensitivity (DTH) testing through intradermal antigen administration has also been demonstrated in both developed and developing countries 18–21 to predict disease progression in HIV-1–infected individuals not receiving HAART, independent of CD4+ cell count. TLC, hemoglobin (HGB) levels, and DTH testing may thus be feasible in developing countries as substitutes for more expensive monitoring of CD4+ cell counts and quantitative HIV-1 RNA.
However, data from resource-limited settings assessing predictors of clinical outcomes in HIV-infected persons treated with HAART with long-term follow-up are not available. We therefore investigated the association of TLC, HGB level, and DTH responses measured prior to initiation of HAART with incidence of new ADIs and death in a cohort of HIV-infected women in the United States.
Overall Study Population and Data Collection
The Women’s Interagency HIV Study (WIHS) is a multicenter prospective study of HIV-1 infection in women, conducted in New York City (2 sites), Washington, DC, Chicago, southern California, and the San Francisco Bay area. The WIHS methods and baseline cohort characteristics have been described previously. 22 Briefly, from October 1994 through November 1995, 2628 women (2059 HIV-1 seropositive and 569 seronegative) were enrolled. Informed consent was obtained from the participants and consent materials reviewed and approved by the committee on human experimentation at each of the collaborating institutions. At 6-month visits, WIHS participants are interviewed with a structured questionnaire and receive a physical examination. Data collected include clinical, demographic, and behavioral information. Multiple gynecologic and blood specimens are collected at each visit.
Response to DTH
Skin testing for DTH was conducted annually from study entry through visit 11 (March 31, 2000) by separate intradermal injections on the volar aspect of the forearm of 0.1 mL each of Candida, mumps, and tetanus antigens. Candida and mumps antigen consisted of premixed preparations provided by Miles Pharmaceuticals (Elkhart, IN) (Hollister/Stier) and Connaught Pharmaceuticals, (Swiftwater, PA) respectively. Tetanus antigen consisted of a 1:10 dilution of adsorbed toxoid (Connaught) and serum. For a small portion (<20%) of the women, Candida antigen was unavailable during the baseline visit, and these visits are not included in this analysis. Multiple batches of each antigen were used in the different study sites. Standardization in the administration and reading of the intradermal skin tests was achieved by individual training and through the use of a training video from the Centers for Disease Control. Reactivity to each of the administered antigens was measured in millimeters of induration. A response to any antigen was defined as ≥2 mm of induration. Only skin test interpretations performed by trained study personnel 2–3 days after implantation of the antigens were analyzed. A DTH test was classified as “responsive” at a visit if a person responded to any one of the 3 antigens and “nonresponsive” or “anergic” at a visit if the person did not respond to any of the antigens.
Specific Study Populations
The specific subgroups used in the analyses here were determined based on use of HAART. HAART was defined to include the use of ≥2 nucleoside analogue reverse transcriptase inhibitors (NRTIs) in combination with at least one protease inhibitor (PI) or nonnucleoside reverse transcriptase inhibitor (NNRTI); one NRTI in combination with at least one PI and at least one NNRTI; a regimen containing ritonavir and saquinavir in combination with one NRTI; an abacavir-containing regimen of ≥3 NRTIs in the absence of PIs and NNRTIs. Combinations of zidovudine and stavudine with a PI or NNRTI were not considered HAART.
All WIHS participants who reported initiation of HAART after July 1995, for whom the date of HAART initiation could be estimated to within a 1-year interval and who had DTH testing prior to HAART initiation, were included in this analysis. The date of HAART initiation was defined as the midpoint between the last visit reporting not using HAART and the first visit reporting HAART use and the date of analyses for which we had complete data on subjects was March 31, 2001.
TLC and HGB were determined with automated complete blood counts at state-certified laboratories at each of the participating institutions. HIV-1 RNA in plasma was quantified for all participants using Nuclisens, the more sensitive method of the isothermal nucleic acid sequence–based amplification (NASBA) method (Organon Teknika Corp., Durham, NC) with a lower limit of detection of 80 copies/mL, in laboratories that were certified by the National Institutes of Health, Virology Quality Assurance Laboratory proficiency testing program. Previous analyses have found NASBA to be statistically equivalent to reverse transcription polymerase chain reaction values among WIHS samples. 23 T-cell subsets were measured by flow cytometry in laboratories certified by the Division of AIDS of the National Institute of Allergy and Infectious Diseases. 24
The primary outcome variables were death and self-report of a new clinical ADI following initiation of HAART. For this study ADIs conform to the class C clinical conditions in the 1993 case definition of AIDS 25; i.e., they do not include the immunologic criterion of CD4+ cell count <200 cells/μL. Deaths were ascertained by notification of participant death from participant friends, relatives, and medical providers and periodically through national and local death registries. Death certificates were requested for all women believed to have died. Date of death was ascertained in descending order of priority from the death certificate, medical records, medical provider, and family/friends. We determined event-free and survival times from the date participants reported initiating HAART through March 31, 2001. Those event-free or alive at March 31, 2001 contributed with censored observations to the survival analyses of time to development of an AIDS-defining event and time to death, respectively. Participants seen from October 1, 1999 to March 31, 2001 with no report of outcome of interest (i.e., ADI or death) were considered censored at date of analysis (March 31, 2001). For analysis of time to death, we considered all deaths as events.
We considered primary laboratory variables from the closest visit to HAART initiation within 1 year prior to HAART initiation that all 3 measures were available. The primary laboratory variables considered were TLC, HGB, response to DTH, CD4+ cell count, and HIV-1 RNA level. Self-reported clinical ADI prior to HAART initiation was also considered as a primary exposure variable. Secondary exposure variables included ethnicity, HIV transmission category, age, and employment status. In practice, clinicians base treatment decisions on laboratory measures reaching thresholds (e.g., to treat when CD4+ <200/μL) and so we attempted to find relevant thresholds for the laboratory variables used here. Unfortunately, no single metric or approach can be guaranteed to determine the “optimal threshold” for measuring prognosis or need to treat, especially as we are using 2 outcomes (time to new ADI and time to death).
We therefore used a semiquantitative approach to determine thresholds that mediated good separation of participants in terms of post-HAART prognosis. We determined threshold cutpoints for TLC, HIV-1 RNA, HGB, and CD4+ in this study based on maximizing log-likelihood χ2 in models of time to death and time to new ADI under systematic search of possible cutpoints throughout the range of the variable. For example, with HGB, and time to new ADI, we first fit a predictive model of time from HAART to new ADI with (HGB <8.0 vs. HGB ≥8.0) as the only predictor; the likelihood ratio χ2 from this model was 2.8. We modified this model by sequential increments of 0.2, i.e., first (HGB <8.2 vs. HGB ≥8.2), then (HGB <8.4 vs. HGB ≥8.4), …, up to (HGB <15.0 vs. HGB ≥15.0). The maximum likelihood ratio χ2 of 21.8 in this sequence occurred for (HGB <10.6 vs. HGB ≥10.6). (Note that larger likelihood ratio ≥ 2 indicates greater statistical prediction by the model). We fit the same sequence of models to time to death. The optimal likelihood ratio χ2 of 42.8 occurred for HGB <10.6 vs. HGB ≥10.6, and so we chose HGB <10.6 vs. HGB ≥10.6 as the cutpoint for this analysis.
A similar approach was taken with TLC using <700, 750, 800, …, 2000 as cutpoints. For time to death, the optimal model likelihood χ2 of 26.4 occurred at 800/μL, whereas for time to new ADI, the optimal model χ2 occurred at 850/μL. Because only 15% of the participants had pre-HAART TLC <850/μL and even fewer had TLC <800/μL, we chose the larger value of 850 as a cutpoint. In addition, the distribution of
χ2 values for the TLC analysis was bimodal, with a second peak (χ2 of 17.0) at 1250 cells/μL. This value of 1250 cells/μL also closely approximates the WHO guidelines, has been found to be optimal in predicting disease progression (without HAART) in prior literature, 6 and represents a similar proportion of our participants (39.1%) as those with CD4+ <200 cells/μL (35.7%). We therefore also fit models using TLC <1250 cells/μL as the cutpoint.
For CD4+, we considered cutpoints of 25, 50, …, 800. The model likelihood χ2 for time to death was maximized with 125 cells/μL as a cutpoint while that for time to death was maximized using 225 cells/μL as a cutpoint. We thus compromised at the commonly used threshold of CD4 <200 vs. CD4 ≥200 cells/μL. Finally, for log viral load, we considered cut-points of < (vs. ≥) 1.91, 2.11, …, 5.71. The model likelihood χ2 for time to death was optimized at the log viral load cutpoint of 4.71, whereas the models for time to new ADI had peaks of similar optimal likelihood χ2 values at 3.91 and 4.71 log viral load. We therefore chose 4.71, which corresponds to a viral load of 51,286, and then rounded this threshold to the commonly used viral load cutpoint of <55,000, which (on the log scale) is close to 51,286.
For most analyses, DTH response was dichotomized as responding to at least 1 antigen vs. responding to no antigens (anergy). However, to examine qualitatively if there were threshold or dose-response patterns, we also fit models with the number of DTH antigens responded to (0, 1, 2, or 3) as a categorical variable, with response to 3 antigens as the reference category.
Medians were reported for continuous variables and comparisons are based on rank tests. Proportions are compared with exact tests. Kaplan-Meier and proportional hazards models were fit to assess pre-HAART predictors of survival and time to new AIDS illness, with the following predictor variables: report of an ADI prior to HAART initiation, TLC, HGB, CD4+ cell count, and DTH measured within 1 year prior to HAART initiation; and secondary variables.
Of the 2059 HIV-1-seropositive and 11 HIV-incident women in the cohort, 1107 (53.8%) initiated HAART prior to March 31, 2001. Of these, 873 (79%) initiated HAART on or after July 1995, had a HAART initiation date known within 1 year, had known DTH status within 1 year prior to HAART, and were followed up after HAART initiation. The median follow-up after HAART initiation was 3.6 years. Of the included women, 341 (39.1%), 143 (16.4%), 312 (35.7%), 374 (42.8%), and 103 (11.8%) had TLC <1250 cells/μL, TLC <850 cells/μL, CD4+ cell count <200 cells/μL, DTH response to 0 of 3 antigens, and HGB <10.6 g/dL, respectively (Table 1). TLC was correlated with CD4+ cell count (Spearman correlation coefficient 0.68, P < 0.001), as was response to DTH testing (at least 1 antigen vs. no antigens) (Spearman correlation coefficient 0.34, P < 0.0001).
Among the 873 included women, there were 106 deaths and 345 (post-HAART initiation) diagnoses of AIDS-defining events. In univariate models (Table 2), TLC <850 cells/μL and TLC <1250 cells/μL, HGB <10.6 g/dL, HIV-1 RNA <55,000 copies/mL, pre-HAART anergy to DTH testing, CD4+ cell count <200 cells/μL, and self-report of an ADI prior to HAART initiation all predicted progression to death (Fig. 1, A–E) and to self-report of a new ADI (Fig. 2, A–E). The results of 3 different multivariate analyses are shown in Table 2: 2 models that exclude CD4+ cell count and assess TLC at either <850 or <1250 cells/μL, and 1 model that excludes TLC and includes CD4+ <200 cells/μL. In these models, all exposure variables remained independently associated with death, except TLC <1250 cells/μL (Hazard ratio = 1.40, P = 0.114).
The largest χ2 values were associated with HGB, indicating that in multivariate models of the laboratory exposures considered here, HGB maintained the greatest independent predictive power: partial log likelihood χ2 (partial LLCS) = 22.56 and 24.88 for HGB in models including CD4+ <200 cells/μL (partial LLCS for CD4 = 7.71) or TLC <850 cells/μL (partial LLCS for TLC = 5.11), respectively. In addition, the LLCS values for the 3 models were similar, suggesting similar power in predicting death after HAART initiation.
All 7 clinical measures except HIV-1 RNA level were also independently associated with report of an incident ADI in all models. HIV-1 RNA level did not independently predict incident ADI in any of the models (P > 0.05 in all models). Self-report of an AIDS-defining event prior to HAART initiation was the strongest independent predictor of an incident ADI after HAART initiation, with the highest hazard ratios (HR) and with partial LLCS >46 in all models. In addition, both thresholds for TLC had HRs and partial LLCSs that were similar to those of CD4+ cell count (HR = 1.64, 1.46, 1.56; and partial LLCS = 14.12, 11.25, 13.50, for TLC <850, TLC <1250 and CD4+ <200 cells/μL, respectively). HGB was a less powerful independent predictor of incident ADI than of death and was less powerful than either TLC or CD4+ cell count as a predictor of incident ADI (partial LLCS = 8.39, 8.12, 7.40 for HGB in the 3 models, respectively). All 3 models were more powerful predictors of ADI than of death, as indicated by the higher partial LLCS values of 111.30–113.46 for models predicting ADI, compared with 81.30-86.69 for models predicting death.
Because quantitation of HIV-1 RNA at a cost affordable in resource-limited settings ($10) may not be available until a few years from now, we also fit multivariate models without viral load. The removal of HIV-1 RNA <55,000 copies/mL from the multivariate models with TLC <850 or CD4 <200 cells/μL had minimal impact on the relative hazards, P values, and full (i.e., total model) log likelihood ratio χ2 values and did not alter the relationships of the other predictor variables with each other. When HIV-1 RNA was removed the third model, TLC <1250 cells/μL, became significantly associated with death (RH = 1.52, P = 0.045).
Pre-HAART Response to Number of DTH Antigens as Predictor of Death
Because anergy represents late-stage immune suppression, a marker of response to therapy prior to complete anergy could be useful. We therefore performed a sensitivity analysis of associations of clinical response to HAART and the number of administered DTH antigens to which a participant responded (Table 3 and Fig. 1E). Compared with those who were anergic to all 3 administered antigens, there was a monotonic decrease in the risk of death by increasing number of antigen responses with HRs for death (P value) of 0.51 (0.005), 0.32 (<0.001), and 0.27 (0.011), in participants responding to 1, 2, and 3 antigens, respectively. This relationship remained significant after adjustment for TLC <1250 cells/μL, HGB, and pre-HAART AIDS illness (Table 3).
In this large US cohort of HIV-1–infected women who initiated HAART and were followed for up to 5.5 years, TLC, HGB level, anergy to DTH testing, and report of an ADI prior to initiation of HAART each independently predicted both death and AIDS-related morbidity. These are all markers of disease severity that are of relatively low cost and could be performed in resource-limited settings. Although in the setting of a cohort study, and therefore not randomized and subject to lead-time bias, these findings provide some validation for a threshold approximating current WHO guidelines of TLC <1200 cells/μL for HAART initiation in settings in which CD4+ cell counts are not available, and for treating in the presence of class IV clinical disease (comparable to class C AIDS-defining conditions in the 1993 Centers for Disease Control classification of HIV infection). In addition, our findings suggest that HGB <10.6 g/dL and responses to DTH skin testing may also provide guidance on when to recommend HAART.
Although the strongest predictive value of TLC occurred at <850 cells/μL, the significantly greater occurrence of an incident ADI (HR = 1.46, P = 0.0008) and death (1.52, P = 0.045 when HIV-1 RNA was extended from model) when HAART was initiated with TLC <1250 cells/μL suggests that the earlier threshold of 1250 cells/μL may be the better threshold for treatment because of the clinical benefit that can be obtained. The lower relative hazards and P values for HGB’s association with incident ADI than with death suggest that it may be a better marker of non–HIV-specific morbidity and mortality than TLC, CD4+ cell count, or prior report of an ADI. However, by improving the immune system, HAART may also indirectly reduce non-HIV mortality.
TLCs have been shown to predict HIV disease progression in the absence of treatment, 6 to correlate highly with CD4+ cell counts (r = 0.69–0.79), 7,9,10 and to have patterns of decline similar to that of CD4+ cells. 8 TLC has also been shown to rise after HAART in correlation with increases in CD4+ cell counts. 9,10 To our knowledge, our findings are the first to describe the association of TLC measured before HAART with clinical disease progression after initiating HAART.
HGB has also been demonstrated to be strongly associated with progression of HIV infection both in untreated and treated populations 14–16 and to improve with HAART. 17 However, in our study, only 11.8% of the participants had pre-HAART HGB <10.6 mg/dL, the value determined in our study to be the strongest predictor of adverse outcomes. In Africa the prevalence of anemia is significantly higher, and a threshold of 10.6 g/dL might include many individuals whose anemia was unrelated to their HIV disease and carried different prognosis, thus attenuating the predictive value of HGB. It seems prudent to be cautious when extrapolating from our population to a resource-limited setting in determining a threshold for HGB. This should be studied further in such settings, as others have suggested. 10
Previous studies in resource-limited settings have indicated that DTH responses predicted both HIV serostatus and prognosis before the availability of HAART 18,20 and that DTH response improved after initiation of antiretroviral treatment, both with HAART 26,27 and even with 3–6 months of zidovudine monotherapy. 28 To our knowledge, ours is the first study to demonstrate that the DTH status prior to HAART initiation predicts clinical outcomes on treatment and to define a possible threshold in DTH responsiveness for initiation of HAART. Our data suggest that DTH responsiveness to only 2 of 3 intradermally administered antigens may provide a minimal threshold at which to initiate HAART. Women who responded to 2 antigens survived nearly as long as those responding to 3 antigens (RH = 0.32 vs. RH = 0.27) and significantly longer than anergic women. Further study of this, preferably in a clinical trial, may be useful in resource-limited settings.
The possible use of DTH testing to assess functional immune status as an indication for HAART would likely be limited to specific settings in which conditions make such testing feasible. The cost and logistic challenges are greater for DTH testing than for TLC and HGB determinations. DTH testing is labor intensive and requires reading by trained personnel 2–3 days after antigen injection. In addition, the specific antigens used, especially tetanus, for which vaccination is spotty in Africa, may need to be replaced with antigens specific to universal exposures in Africa. The Multitest CMI (Institut Merieux, Lyons, France) device, a self-contained multiple-puncture device delivering 7 antigens and a glycerol control (tuberculin, candidin, diphtheria, tetanus, trichophytin, Proteus, and Streptococcus), proved predictive of clinical disease in Tanzania. 21 Thus, it may provide some refinement of clinical decision-making and may be worthwhile in some settings, e.g., those providing directly observed therapy, which provide village-based care with a single worker who sees each patient daily, or when there is an already established infrastructure providing care with good access by patients.
Additional limitations in extrapolating from our data regarding DTH responsiveness to resource-limited settings are factors other than HIV-1 infection that can impair DTH response 29–33 and the variability in DTH itself over short periods (1 year). 34–37 However, in resource-limited settings, anergy has been demonstrated to predict poor clinical outcomes in HIV-1–infected people not treated with HAART 19–21 —despite these other factors influencing DTH responsiveness—and thus may also prove valuable in predicting response to HAART.
In summary, our findings support suggestions 9,12 that TLC and HGB may be useful as indicators of when to initiate HAART if CD4+ cell counts are unavailable and suggest specific thresholds of TLC <1250 cells/μL and HGB <10.6 g/dL. Less expensive technology for CD4+ and HIV-1 RNA determinations is expected, with a cost of approximately US $8–10 per test for CD4+ cell count and US $7 per test for viral load. This compares to current costs of US $30 for flow cytometry and US $0.85 for a complete blood count, which would provide both TLC and HGB. Thus, although a very substantial cost advantage for TLC exists, it may still be feasible and advisable to incorporate some viral load determinations (at a cost of US $7–10) into care in resource-poor settings, e.g., immediately prior to HAART initiation, if CD4+ cell count fails to respond, or if CD4+ declines after an initial improvement. In addition, our findings suggest that in settings in which DTH is performed, anergy, or responding to only 1 or 2 of 3 administered antigens, may indicate a need for HAART.
1. Anastos K, Barron Y, Greenblatt R, et al. Risk of disease progression
and death in HIV
infected women initiating highly active antiretroviral therapy
at different stages of disease. Arch Intern Med
2. Phillips AN, Staszewski S, Weber R, et al. HIV
viral load response to antiretroviral therapy according to the baseline CD4 cell count and viral load. JAMA
3. Hogg RS, Yip B, Chan KJ, et al. Rates of disease progression
by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA
4. Panel on clinical practices for treatment of HIV
. Guidelines for using antiretroviral agents among HIV
-infected adults and adolescents. Ann Intern Med
5. British HIV
Association (BHIVA) guidelines for the treatment of HIV
-infected adults with antiretroviral therapy. HIV Med
6. Post FA, Wood R, Maartens G. CD4 and total lymphocyte counts as predictors of HIV disease progression
. Q J Med
7. Kumarasamy N, Mahajan AP, Flanigan TP, et al. Total lymphocyte count
(TLC) is a useful tool for the timing of opportunistic infection prophylaxis in India and other resource-constrained countries. J Acquir Immune Defic Syndr
8. Gange SJ, Lau B, Phair J, et al. Rapid declines in total lymphocyte count
and hemoglobin in HIV
infection begin at CD4 lymphocyte counts that justify antiretroviral therapy. AIDS
9. Kumarasamy N, Flanigan TP, Mahajan AP, et al. Monitoring HIV
treatment in the developing world. Lancet Infect Dis
10. Badri M, Wood R. Usefulness of total lymphocyte count
in monitoring highly active antiretroviral therapy
in resource-limited settings. AIDS
11. Flanigan T, Mahajan A, Kumarasamy N, et al. Total lymphocyte count
as a surrogate for CD4 count to initiate and monitor HAART in resource-limited countries. Paper presented at: Ninth Conference on Retroviruses and Opportunistic Infections; February 2002; Seattle, WA.
12. Spacek LA, Griswold M, Quinn TC, et al. Total lymphocyte count
and hemoglobin combined in an algorithm to initiate the use of highly active antiretroviral therapy
in resource-limited settings. AIDS
14. Levine AM, Berhane K, Masri-Lavine L, et al. Prevalence and correlates of anemia in a large cohort of HIV
-infected women. Women’s Interagency HIV
Study. J Acquir Immune Defic Syndr.
15. Berhane K, Gange SJ, Masri-Lavine L, et al. Impact of highly active anti-retroviral therapy on anemia, and relationship between anemia and survival
in a large cohort of HIV
infected women. Women’s Interagency HIV
16. Moore RD, Keruly JC, Chaisson RE. Anemia and survival
infection. J Acquir Immune Defic Syndr
17. Semba RD, Shah N, Klein RS, et al. Highly active antiretroviral therapy
associated with improved anemia among HIV
-infected women. Aids Patient Care STDS
18. Birx DL, Brundage J, Larson K, et al. The prognostic utility of delayed type hypersensitivity skin testing in the evaluation of HIV
infected patients. J Acquir Immune Defic Syndr
19. Gordin FM, Hartigan PM, Klimas NG, et al. Delayed type hypersensitivity skin tests are an independent predictor of HIV disease progression
. J Infect Dis
20. Blatt SP, Hendrix CW, Butzin CA, et al. Delayed type hypersensitivity testing predicts progression to AIDS in HIV
infected patients. Ann Intern Med
21. Miller WC, Thielman NM, Swai N, et al. Delayed type hypersensitivity testing in Tanzanian adults with HIV
infection. J Acquir Immune Defic Syndr
22. Barkan SE, Melnick SL, Preston-Martin S, et al. The Women’s Interagency HIV
Study. WIHS Collaborative Study Group. Epidemiology
23. Lew J, Reichelderfer P, Fowler M, et al. Determinations of levels of human immunodeficiency virus type 1 RNA in plasma: reassessment of parameters affecting assay outcome. J Clin Microbiol
24. Calvelli T, Denny T, Paxton H, et al. Guidelines for flow cytometric immunophenotyping. Cytometry
25. Centers for Disease Control and Prevention. 1993 Revised classification system for HIV
infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR Recomm Rep
26. Carr A, Chuah J, Hudson J, et al. A randomised, open-label comparison of three highly active antiretroviral therapy
regimens including two nucleoside analogues and indinavir for previously untreated HIV
-1 infection: the OzCombo1 study. AIDS
27. Valdez H, Smith KY, Landay A, et al. Response to immunization with recall and neoantigens after prolonged administration of an HIV
-1 protease inhibitor-containing regimen. ACTG 375 team. AIDS Clinical Trials Group . AIDS
28. French MA, Cameron PU, Grimsley G, et al. Correction of human immunodeficiency virus-associated depression of delayed-type hypersensitivity (DTH) after zidovudine therapy: DTH, CD4+ T-cell numbers, and epidermal Langerhans cell density are independent variables. Clin Immunol Immunopathol
29. Shell-Duncan B, Wood JW. The evaluation of delayed-type hypersensitivity responsiveness and nutritional status as predictors of gastrointestinal and acute respiratory infection: a prospective field study among traditional nomadic Kenyan children. J Trop Pediatr
30. Smith SM, Levy NS, Hayes CE. Impaired immunity in vitamin A-deficient mice. J Nutr
31. Rall LC, Meydani SN. Vitamin B6 and immune competence. Nutr Rev
32. Kukreja R, Khan A. Effect of selenium deficiency and its supplementation on DTH response, antibody forming cells and antibody titre. Indian J Exp Biol
33. Matarese G. Leptin and the immune system: how nutritional status influences the immune response. Eur Cytokine Netw
34. Chin DP, Osmond D, Page-Shafer K, et al. Reliability of anergy skin testing in the person with HIV
infection. Am J Respir Crit Care Med
35. Caiaffa WT, Graham NM, Galai N, et al. Instability of delayed-type hypersensitivity skin test anergy in human immunodeficiency virus infection. Arch Intern Med
36. Klein RS, Sobel J, Flanigan T, et al. Stability of cutaneous anergy in women with or at risk for HIV
Epidemiology Research Study Group. J Acquir Immune Defic Syndr Hum Retrovirol
37. Pouchot J, Grasland A, Collet C, et al. Reliability of tuberculin skin test measurement. Ann Intern Med
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, Phyllis Tien); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); and Data Coordinating Center (Alvaro Muñoz). 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, N01-AI-35161, MO1-RR-00079, MO1-RR-00083.