905

Equal Plasma Viral Loads Predict a Similar Rate of CD4+ T Cell Decline in Human Immunodeficiency Virus (HIV) Type 1– and HIV-2– Infected Individuals from Senegal, West Africa Geoffrey S. Gottlieb,1 Papa Salif Sow,7 Stephen E. Hawes,2,5 Ibra Ndoye,8 Mary Redman,4 Awa M. Coll-Seck,7 Mame A. Faye-Niang,7 Aissatou Diop,7 Jane M. Kuypers,2 Cathy W. Critchlow,5 Richard Respess,6 James I. Mullins,1,3 and Nancy B. Kiviat1,2

1

Division of Allergy and Infectious Diseases, Department of Medicine, and Departments of 2Pathology and 3Microbiology, School of Medicine, and Departments of 4Biostatistics and 5Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle; 6Roche Molecular Diagnostics, Alameda, California; 7Department of Infectious Diseases, University of Dakar, and 8Institute of Public Health, Dakar, Senegal, West Africa

Human immunodeficiency virus (HIV) type 2 infection is characterized by slower disease progression to acquired immunodeficiency syndrome than results from HIV-1 infection. To better understand the biological factors underlying the different natural histories of infection with these 2 retroviruses, we examined the relationship between HIV RNA and DNA levels and the rate of CD4+ T cell decline among 472 HIV-1– and 114 HIV-2– infected individuals from Senegal. The annual rate of CD4+ T cell decline in the HIV-2 cohort was approximately one-fourth that seen in the HIV-1 cohort. However, when the analysis was adjusted for baseline plasma HIV RNA level, the rates of CD4+ T cell decline per year for the HIV-1 and HIV-2 cohorts were similar (a rate increase of
4% per year for each increase in viral load of 1 log10 copies/mL). Therefore, plasma HIV load is predictive of the rate of CD4+ T cell decline over time, and the correlation between viral load and the rate of decline appears to be similar among all HIV-infected individuals, regardless of whether they harbor HIV-1 or HIV-2.

AIDS in humans is caused by 2 related retroviruses: human immunodeficiency virus (HIV) type 1 and HIV-2. Although both viruses cause AIDS via progressive CD4+ T cell depletion, HIV-2 infection is characterized by a much longer asymptomatic stage, lower plasma viral loads, slower decline in CD4+ T cell count, lower mortality rate, and lower rates of mother-tochild transmission [1–7]. In addition, although HIV-1 is responsible for the global AIDS pandemic, HIV-2 is endemic in West Africa and has been found only sporadically elsewhere [8, 9]. The basis for the markedly different natural histories of these 2 viral infections is not completely understood. Gaining insight into the relationship between plasma HIV RNA levels, peripheral blood mononuclear cell (PBMC) HIV DNA levels, and the decline in levels of CD4+ T cells that is associated Received 13 July 2001; revised 28 October 2001; electronically published 19 March 2002.

Informed consent was obtained from all study participants, and the study protocol was approved by the University of Washington Human Subjects Committee and the Senegalese AIDS National Committee, in accordance with the human experimentation guidelines of the US Department of Health and Human Services. Financial support: National Institutes of Health (CA62801 and CA75920 to N.B.K.; AI49755 to G.S.G.); University of Washington Center for AIDS Research Training Grant (AI07140 to G.S.G.). Reprints or correspondence: Dr. Nancy B. Kiviat, Box 359933, University of Washington, HPV Research Group, 1914 N. 34th St., Ste. 300, Seattle, WA 98103 ([email protected]). The Journal of Infectious Diseases 2002;185:905–14 q 2002 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2002/18507-0008$02.00

with HIV-1 and HIV-2 infections is essential to understanding the different natural histories of these 2 infections. It is widely believed that an HIV-1 plasma RNA level, or set point, is established early during HIV-1 infection and maintained during a large part of the asymptomatic stage of infection. This RNA set point has been shown to be an important predictor of the rate of CD4+ T cell decline, the time to development of AIDS, and the risk of death from HIV-1 infection [10– 14]. The PBMC HIV-1 DNA level, which is highly correlated with the plasma HIV-1 RNA level, is also thought to be a strong predictor of the rate of CD4+ T cell decline and the occurrence of AIDS-defining events [15–19]. However, the relatively narrow range of PBMC HIV-1 DNA levels makes this measure less clinically useful than are plasma HIV-1 RNA levels. In contrast to what is known about HIV-1, only a few studies have examined the role of plasma HIV-2 RNA levels and CD4+ T cell decline [20–22], and these studies were limited by small sample size, cross-sectional design, or lack of direct comparison to a matched HIV-1– infected cohort. Available data suggest that HIV-2 infection is also characterized by the presence of a viral RNA set point in plasma during the long asymptomatic phase, although the set points of HIV-2– infected individuals have generally been found to be severalfold lower than those of HIV-1– infected individuals [20]. As is true for HIV-1– infected individuals, plasma viral load predicts the rate of CD4+ T cell decline in individuals infected with HIV-2 [21]. However, conflicting data exist with regard to whether plasma HIV-2 RNA level and PBMC HIV-2 DNA level are correlated [23–25].

906

Gottlieb et al.

Previously published studies suggest that although levels of viral DNA in PBMC appear to be similar for individuals infected with HIV-1 and those infected with HIV-2, in general, plasma HIV-2 RNA levels are significantly lower than plasma HIV-1 RNA levels [2, 20, 23–26]. However, whether the slower rate of disease progression that is characteristic of most HIV-2 infections is solely the result of lower viral loads or whether HIV-1 and HIV-2 are characterized by fundamentally different rates of CD4+ T cell decline at any plasma HIV RNA level is unknown. Thus, whether HIV-2– infected persons with plasma viral loads similar to those seen in HIV-1– infected persons would progress at a rate characteristic of HIV-1 infection remains unclear. This question has important clinical treatment implications, and insight into the issue will contribute to a greater understanding of HIV pathogenesis. Therefore, we undertook the present study to compare viral loads and immunologic characteristics among 472 HIV-1– and 114 HIV-2– infected individuals in Senegal, West Africa.

Subjects, Materials, and Methods Study population. Between October 1994 and November 1998, all male and female patients aged >16 years who presented to the University of Dakar Infectious Disease Clinic (Fann Hospital, Dakar, Senegal, West Africa), as well as commercial sex workers (CSWs) who presented to the sexually transmitted disease (STD) clinics in Dakar and M’Bour (a city
83 km from Dakar), were offered serologic testing for HIV-1 and HIV-2. HIV-seropositive persons, along with a subset of HIV-seronegative persons, were invited to participate in longitudinal studies. During this period, 6788 adults were screened for the presence of antibodies to HIV-1 and HIV-2; 864 were found to be seropositive for HIV-1 (450 women and 414 men), and 168 were found to be seropositive for HIV-2 (115 women and 53 men). Seventy subjects who were seropositive for both HIV-1 and HIV-2 were excluded from the study. Five hundred forty-four individuals who were seropositive for HIV-1 and 129 individuals who were seropositive for HIV-2 were screened for the presence of HIV RNA and DNA; subjects in whom we failed to detect any HIV-1 (6 subjects) or HIV-2 (5 subjects) RNA or DNA were excluded from our analyses. The study group was further limited to the 472 subjects with HIV-1 and 114 subjects with HIV-2 for whom data on CD4+ T cell count and plasma HIV RNA levels from at least 1 visit to a clinic were available. Patients who had opportunistic infections were treated with appropriate therapy; no patients received antiretroviral therapy. Collection of specimens and study procedures. At the screening visit, blood was collected for HIV-1 and HIV-2 serologic assays. Subjects completed a short standardized screening interview that included questions about medical and sexual history and underwent a general physical examination, as described elsewhere [27]. All subjects enrolled in longitudinal studies returned 1 month later and every 4– 6 months subsequently, at which time more-detailed questionnaires were completed and blood was obtained for qualitative and quantitative assays for HIV-1 and HIV-2 RNA and DNA and for determination of T cell counts. Whole blood collected in

JID 2002;185 (1 April)

EDTA tubes was analyzed using the FACSCount analyzer (Becton Dickinson) to determine the number of CD4+, CD8+, and CD3+ T cells per microliter of blood. Classification of subjects. Subjects were classified as having an HIV infection if the results of both a microwell plate HIV-1/HIV2 EIA (Genetic Systems) and a confirmatory rapid synthetic peptide– based membrane immunoassay, which also distinguishes HIV-1 from HIV-2 (Multispot; Sanofi Diagnostic Pasteur), were positive. All positive results of these assays were further confirmed by HIV-1– and HIV-2–specific Western blot assays (Genetic Systems) performed according to the manufacturer’s specifications. A further criterion for study inclusion was demonstration of the matching type of HIV RNA and/or DNA in at least 1 blood sample obtained during the study from persons classified as having HIV-1 or HIV-2 infection. Quantitative and qualitative detection of HIV-1 RNA. RNA was extracted and quantified from blood mixed with the anticoagulant EDTA, using the Amplicor HIV-1 Monitor test (Roche Molecular Systems) according to the manufacturer’s recommendations, with the addition of 2 primers, SK145 and SK151, to the master mix. These primers are of the same length and hybridize to the same regions as the first-generation primers (SK462 and SK431) but were designed to have fewer mismatches with non– subtype B HIV-1 sequences and, thus, to provide more accurate quantification of HIV-1 RNA in plasma from individuals infected with different genetic subtypes. The resulting assay (using the equivalent of 25 mL of plasma) provided reliable HIV-1 RNA quantification at levels .400 copies/ mL, with a minimum level of detection of 80 copies/mL, a linear range of 400– 750,000 copies/mL, and an interassay coefficient of variation of 12.4% [28, 29]. Qualitative detection of HIV RNA in plasma was performed on all samples that had negative results of the quantitative assay. This assay was conducted as described for the quantitative assay, except that plasma was suspended in 200 mL of specimen diluent, amplifications were performed for 35 cycles, and the amplified products were analyzed undiluted in microwell plates coated with the respective HIV and quantitation standard (QS) probes. A positive reaction had an absorbance at 450 nm of >0.350. This assay had a reproducible sensitivity of 200 HIV RNA copies/mL and a lower limit of detection of 40 HIV RNA copies/mL. Detection and quantification of HIV-1 DNA. DNA was prepared from a white blood cell pellet from 0.5 mL of whole blood, using the Amplicor HIV-1 Monitor test as recommended by the manufacturer. Approximately 1 mg of total cellular DNA was coamplified with a known number of copies of the DNA QS. The total genomic DNA concentration was determined using Hoechst dye (Aventis Pharma) [30]. The master mix for DNA amplification was the same as that for RNA amplification, except that Taq DNA polymerase and Mg2+ were added instead of rTth DNA polymerase and Mn2+, respectively. The thermal cycling parameters were also the same, with the exception of elimination of the 30-min reverse transcription (RT). Amplified DNA was diluted and detected as for the RNA reaction. The number of HIV-1 DNA copies was calculated from the known number of DNA QS copies added and then normalized to the total number of cellular DNA copies added. The quantitative assay for HIV-1 DNA can detect as little as 1 HIV-1 DNA copy/mg of PBMC DNA and had a linear range of

JID 2002;185 (1 April)

HIV RNA as a Predictor of CD4+ T Cell Decline

10–25,000 copies per polymerase chain reaction (PCR). The interassay coefficient of variation was 3.0%. Qualitative detection of HIV-1 DNA in PBMC was performed on all samples that yielded negative results of the quantitative assay. The qualitative assay was conducted as described for the quantitative assay, except that twice the volume of input DNA was used. The assay had a lower limit of detection of 1 HIV-1 DNA copy/mg of PBMC DNA, and it reliably detected .5 HIV-1 DNA copies/mg of PBMC DNA [31]. Qualitative and quantitative detection of HIV-2 RNA and DNA. HIV-2 RNA and DNA were detected with PCR-based assays developed at Roche Molecular Systems [20], using primers RAR1000 (50 -GCTGGCAGATTGAGCCCTGGGAGGTTCTCT-30 ) and RAR04 (50 -GAATGACCAGGCGGCGACTAGGAGAGAT-30 ) to amplify a 201-bp fragment of the HIV-2 long terminal repeat. The target region is highly conserved among known HIV-2 and simian immunodeficiency virus isolates; the upstream primer has no more than a 2-bp mismatch, and the downstream primer has no mismatches with known sequences. These primers did not amplify HIV-1. Qualitative and quantitative detection of HIV-2 RNA and DNA was conducted as described for HIV-1, with some exceptions. The QS probe was composed of the HIV-2 primer binding sites but had different internal sequences to allow differentiation between the probe and the HIV target. RT-PCR reactions were performed using the following thermal cycling parameters: 1 cycle at 60 C for 30 min for RT; 5 cycles at 94 C for 20 s, 60 C for 20 s, and 72 C for 10 s; 35 cycles at 90 C for 10 s, 65 C for 20 s, and 72 C for 20 s; and 1 cycle of 10 min at 72 C. The RT step was omitted during amplification of DNA targets. Amplified products were detected in microwells coated with the oligonucleotide probe RAR05 (50 -TGGCTGTTCCTGCTAGACTCTCACCAGTACT-30 ). The quantitative HIV-2 RNA assay reliably detected 200 HIV-2 RNA copies/mL, with a lower limit of detection of as few as 40 copies/mL. The linear range of the HIV-2 RNA assay was between 400 and 750,000 copies/ mL. The interassay coefficient of variability was 20.3%. Qualitative detection of HIV RNA in plasma was performed on all samples that yielded negative results of the quantitative assay. Using the equivalent of 50 mL of plasma, the qualitative HIV-2 RNA assay reliably detected 100 HIV-2 RNA copies/mL, with a lower limit of detection of as few as 20 copies/mL. The quantitative HIV-2 DNA assay reliably quantified .10 HIV-2 DNA copies/mg of PBMC DNA, with a linear range of 10– 25,000 copies per PCR and a lower limit of detection of 1 HIV-2 DNA copy/mg of PBMC DNA. The interassay coefficient of variability was 4.8%. Qualitative detection of HIV-2 DNA in PBMC was performed on all samples that yielded negative results on quantitative assay. The qualitative assay reliably detected .5 HIV-2 DNA copies/mg of PBMC DNA, with a lower limit of detection of 1 copy/mg. Statistical methods and data analysis. Groups were compared using x2 tests for categorical data, Mantel-Haenszel tests for trend for ordinal data, and Student’s t test for continuous data. Linear regression models were fit for cross-sectional data, and linear mixed models with autocorrelated errors were fit for longitudinal data (SPLUS 2000; Mathsoft). To correct for nonnormality, plasma HIV RNA and PBMC HIV DNA data were log-transformed (base 10), when levels were detectable, or set to 0, when levels were undetectable. The natural logarithms of the CD4+ T lymphocyte counts were used to normalize error terms. To evaluate the sensitivity and robust-

907

ness of our conclusions to samples that had negative results on qualitative assay for HIV RNA or positive results on qualitative assay but negative results on quantitative assay, we performed a series of analyses, setting the values for those samples to a range of values between 0 and the limit of detection. None of these analyses gave qualitatively different results. The level of statistical significance used in all analyses was P , :05.

Results We enrolled 472 subjects infected with HIV-1 and 114 subjects infected with HIV-2 in our study (table 1). Study participants were drawn from 3 cohorts: (1) male and female patients Table 1. Demographic characteristics and stage of disease of Senegalese individuals infected with human immunodeficiency virus (HIV) type 1 or HIV-2.

Characteristic Site of recruitment Fann Hospital Infectious Disease Clinic Male Female Commercial sex worker Dakar sexually transmitted diseases clinica M’Bour sexually transmitted diseases clinica Sex, female Commercial sex workerb Age, yearsc ,30 30–39 40–49 >50 Mean ^ SD Born in Senegal Marital statusd Monogamous Polygamous Widowed Divorced/separated Never married HIV disease stagee 1 2 3 4

Subjects infected with HIV-1 (n = 472)

Subjects infected with HIV-2 (n = 114)

392 (83)

79 (69)

191 201 17 52

31 48 4 16

(40) (43) (4) (11)

(27) (42) (4) (14)

28 (6)

19 (17)

281 (60) 97 (35)

83 (73) 39 (47)

149 (32) 209 (44) 91 (19) 22 (5) 33.7 ^ 8.4 404 (86)

18 (16) 58 (51) 31 (27) 7 (6) 36.9 ^ 7.9 105 (92)

171 52 58 97 90

(37) (11) (12) (21) (19)

40 14 9 34 14

(36) (13) (8) (31) (13)

79 142 188 19

(19) (33) (44) (4)

39 31 34 3

(36) (29) (32) (3)

NOTE. Data are no. (%) of subjects, unless otherwise indicated. a All study subjects from the Dakar and M’Bour sexually transmitted diseases clinics are female commercial sex workers. b All commercial sex workers were women. c Age of 1 patient was unknown. d This information was available for 468 subjects infected with HIV-1 and 111 subjects infected with HIV-2. e Based on the World Health Organization staging system [32]. This information was available for 428 subjects infected with HIV-1 and 107 subjects infected with HIV-2.

908

Gottlieb et al.

(aged, >16 years) attending the University of Dakar Infectious Disease Clinic, (2) female CSWs from an STD clinic in Dakar, and (3) female CSWs from an STD clinic in M’Bour. Subjects infected with HIV-2, compared with subjects infected with HIV-1, were older (mean age, 36.9 vs. 33.7 years; P ¼ :0002) and were more likely to be female (73% vs. 60%; P ¼ :009) and to be employed as CSWs (47% vs. 35%; P ¼ :04). Male subjects in our study population tended to present with later-stage HIV disease, lower baseline CD4+ T cell counts, and higher viral loads than did female subjects, regardless of whether they were infected with HIV-1 or HIV-2 (data not shown). This difference most likely resulted from the fact that female subjects often presented for non–AIDS-associated medical care, such as family planning, gynecologic examinations, and routine STD screening (CSWs), whereas male subjects tended to present to the infectious disease clinic with symptomatic HIV disease. Furthermore, female CSWs who visited the STD clinics for government-required STD screening tended to have earlier-stage disease, lower plasma viral loads, and higher CD4+ T cell counts than did female subjects who presented to the infectious disease clinic (data not shown). Subjects infected with HIV-1 presented with lower CD4+ T cell counts, higher CD8+ T cell counts, and higher plasma RNA

JID 2002;185 (1 April)

viral loads (table 2), in addition to later-stage HIV disease, than did subjects infected with HIV-2 (table 1). At baseline, plasma viral RNA was detected in 99% of subjects infected with HIV-1 but in only 75% of subjects infected with HIV-2. Mean log10 plasma HIV RNA levels were greater in HIV-1–infected subjects than in HIV-2– infected subjects (4:92 ^ 1:24 vs. 2:18 ^ 1:86 log10 copies/mL, respectively; P , :0001; interquartile range [IQR], 25,914– 543,320 HIV-1 RNA copies/mL vs. 0–5469 HIV-2 RNA copies/mL). Although the mean log10 PBMC HIV DNA loads among HIV-1– infected subjects and HIV-2–infected subjects were not significantly different (1:42^ 0:76 vs. 1:54 ^ 0:92 log10 copies/mg, respectively; P ¼ :17; IQR, 8–95.5 HIV-1 DNA copies/mg vs. 11– 168 HIV-2 DNA copies/mg), the distribution of values was shifted toward higher levels in subjects with HIV-2 infections (table 2). Among subjects with HIV-1 and subjects with HIV-2, lower CD4+ T cell counts were strongly associated with higher levels of viral RNA (figure 1A) and viral DNA (figure 1B). Persons infected with HIV-1 were found to have, on average, plasma RNA levels higher than those found in persons with HIV-2 who had similar CD4+ T cell counts (P , :0001) (figure 1A); however, the relationship between CD4+ T cell count and levels of DNA in PBMC appeared to be more complex (figure 1B). Although

Table 2. Immunologic and virologic characteristics of study subjects infected with human immunodeficiency virus (HIV) type 1 or HIV-2. Characteristic

Subjects infected with HIV-1 (n = 472)

Subjects infected with HIV-2 (n = 114)

P

+

CD4 T cell count, cells/mL ,100 100–199 200–499 >500 Mean ^ SD CD8+ T cell count, mean cells/mL ^ SD CD3+ T cell count, mean cells/mL ^ SD HIV RNA level in plasma, mean log10 copies/mL ^ SD HIV RNA level in plasma, copies/mL Undetectable: ,40 40–999 1000–9999 10,000–99,999 100,000–999,999 >1,000,000 Median (range) HIV DNA level in PBMC,b mean log10 copies/mg of PBMC DNA HIV DNA level in PBMC,b copies/mg of PBMC DNA Undetectable: ,1 1–9 10–49 50–99 100–499 >500 Median (range)

148 (31) 103 (22) 147 (31) 74 (16) 259 ^ 241 945 ^ 541 1296 ^ 699 4.92 ^ 1.24 4 (1) 20 (4) 59 (13) 124 (26) 192 (41) 73 (16) 149,533 (0–17,585,106) 1.42 ^ 0.76 9 67 115 48 73 11 29

(3) (21) (36) (15) (23) (3) (0–6636)

20 (18) 12 (11) 34 (30) 48 (42) 461 ^ 337 692 ^ 531 1224 ^ 723 2.18 ^ 1.86

,.0001a

29 (25) 45 (40) 17 (15) 16 (14) 7 (6) 0 221 (0–548,969) 1.54 ^ 0.92

,.0001a

1 12 27 12 24 10 38

(1) (14) (31) (14) (28) (12) (0–3450)

NOTE. Data are no. (%) of subjects, unless otherwise indicated. PBMC, peripheral blood mononuclear cells. a Mantel-Haenszel x2 test for trend. b PBMC DNA data were available for 323 subjects infected with HIV-1 and 86 subjects infected with HIV-2.

,.0001 .0002 .17 ,.0001

.17 .003a

JID 2002;185 (1 April)

HIV RNA as a Predictor of CD4+ T Cell Decline

909

Figure 1. Cross-sectional analysis of plasma RNA and peripheral blood mononuclear cell (PBMC) DNA viral load trends vs. CD4+ T cell count of patients with human immunodeficiency virus (HIV) type 1 and HIV-2 infection at baseline. A, Plasma RNA log10 copies/mL vs. CD4+ T cell count. B, HIV DNA log10 copies/mg of PBMC DNA vs. CD4+ T cell count. Linear regression lines for HIV-1 and HIV-2 are also shown.

lower CD4+ T cell counts were strongly correlated with higher PBMC HIV DNA levels among both HIV-1– and HIV-2– infected subjects, the slope of the regression line for HIV-2 is steeper than that of the regression line for HIV-1 (P ¼ :01), which suggests that HIV-2 DNA levels are higher in subjects with low CD4+ T cell counts and lower in subjects with high CD4+ T cell counts than in HIV-1–infected subjects with comparable CD4+ T cell counts. We also investigated relationships between plasma HIV RNA and PBMC HIV DNA levels (figure 2). For subjects infected with HIV-1 and subjects infected with HIV-2, higher PBMC HIV DNA loads were associated with higher plasma HIV RNA loads (P , :0001); however, this positive correlation was more robust for HIV-2 (P , :0001). Furthermore, these positive correlations remained after adjustment for CD4+ T cell count for both HIV-1 and HIV-2 infection (P , :0001). We examined relationships between type of HIV infection (HIV-1 vs. HIV-2) and rate of CD4+ T cell decline among the subset of persons who were seen on at least 3 occasions during a time period of >1 year and who had baseline CD4+ T cell counts .200 cells/mL. This group, which included 120 subjects with HIV-1 and 49 subjects with HIV-2, had a slightly higher

proportion of female subjects and subjects with earlier-stage disease than did the study population as a whole (data not shown). In this subgroup, the mean ^ SD baseline CD4+ T cell count among subjects with HIV-1 was 458 ^ 203 cells/mL, and the mean ^ SD log10 plasma HIV RNA load was 4:3 ^ 1:3 log10 copies/mL, compared with values of 618 ^ 203 cells/mL and 1:3 ^ 1:5 log10 copies/mL among subjects with HIV-2. The 120 subjects infected with HIV-1 contributed 861 samples during a mean ^ SD follow-up period of 2:73 ^ 1:25 years, and the 49 subjects infected with HIV-2 contributed 372 samples during a mean ^ SD follow-up period of 3:07 ^ 1:22 years. The median annual rate of CD4+ T cell decline among HIV-1–infected subjects was 15.9% (95% confidence interval [CI], 11.1%–20.3%), compared with 4.1% (95% CI, 20.5 to 9.2%) among subjects infected with HIV-2 (table 3). The annual rate of CD4+ T cell decline in subjects infected with HIV-1 or HIV-2 appeared to be greater among those with higher plasma viral loads (table 3 and figure 3A and 3B). However, similar rates of CD4+ T cell decline were observed for HIV-1– infected subjects and HIV2–infected subjects with similar plasma HIV RNA levels (table 3 and figure 3A and 3B). In addition, after adjustment for HIV RNA levels, the estimated median annual rate of CD4+ T cell

910

Gottlieb et al.

JID 2002;185 (1 April)

Figure 2. Cross-sectional analysis of the association between plasma human immunodeficiency virus (HIV) RNA levels and peripheral blood mononuclear cell (PBMC) HIV DNA levels in subjects infected with HIV-1 or HIV-2 at baseline. Linear regression lines for HIV-1 and HIV-2 are also shown.

decline among HIV-1–infected subjects (4.1%; 95% CI, 2.7%– 5.4%) was similar to that for HIV-2– infected subjects (3.3%; 95% CI, 0.9%– 5.5%; P ¼ :66). Likewise, after adjustment for HIV type, the estimated median annual rate of CD4+ T cell decline per 1 log10 plasma HIV RNA level was 3.9% (95% CI, 2.8%–5.0%). We also analyzed the relationship between PBMC HIV DNA levels and rate of CD4+ T cell decline. Annual rates of CD4+ T cell decline per log10 increase in baseline PBMC HIV DNA

level did not differ for subjects infected with HIV-1 (10.2%; 95% CI, 1.4%–18.2%; n ¼ 97) and subjects infected with HIV-2 (10.6%; 95% CI, 4.3%– 16.4%; n ¼ 41). To further investigate whether HIV RNA and DNA level were independently associated with the rate of CD4+ T cell decline, we performed an analysis among all subjects, using CD4+ T cell decline as the dependent variable and plasma HIV RNA and PBMC HIV DNA levels as independent variables. We found an average annual CD4+ T cell decline of 8.7% (95% CI, 2.1%–14.8%) for

Table 3. Association between the annual rate of CD4+ T cell decline and baseline plasma RNA viral load among subjects with baseline CD4+ T cell counts .200 cells/mL. Subjects infected with HIV-1 (n = 120)

Subjects infected with HIV-2 (n = 49)

+

Variable

No. (%)

Baseline plasma HIV RNA level, copies/mL ,100 100–999 1000–9999 10,000–99,999 >100,000 Median rate Median rate per log10 RNA viral load

6 5 31 42 36

NOTE.

CI, confidence interval.

(5) (4) (26) (35) (30)

Yearly decline in CD4 T cell levels, % (95% CI)

3.8 (219.3 to 22.5) 14.9 (29.9 to 34) 9.9 (0.8–18.1) 13.3 (5.6–20.4) 26.2 (18.8–32.9) 15.9 (11.1–20.3) 4.1 (2.7–5.4 )

No. (%)

28 14 5 2 0

(57) (29) (10) (4)

Yearly decline in CD4+ T cell levels, % (95% CI)

20.1 12.4 8.1 22.7

(28 to 7) (0–23) (26 to 20) (6–36) — 4.1 (20.5 to 9.2) 3.3 (0.9–5.5)

JID 2002;185 (1 April)

HIV RNA as a Predictor of CD4+ T Cell Decline

911

Figure 3. Relationship between baseline human immunodeficiency virus (HIV) RNA levels and median annual rate (%) of CD4+ T cell decline. A, HIV-1– infected subjects (n ¼ 120); the median annual rate of CD4+ T cell decline for each increase in HIV RNA levels of 1 log10 copies/mL was 4.1% (95% confidence interval, 2.7%– 5.4%). B, HIV-2–infected subjects (n ¼ 49); the median annual rate of CD4+ T cell decline for each increase in HIV RNA levels of 1 log10 copies/mL was 3.3% (95% confidence interval, 0.9%– 5.5%). P ¼ :66 for difference between median annual rate of CD4+ T cell decline for each increase in HIV RNA levels of 1 log10 copies/mL in HIV-1– infected subjects and that in HIV-2– infected subjects. Open circles, median rates of CD4+ T cell decline for individual subjects; solid lines, population median rates of CD4+ T cell decline.

each increase in HIV DNA levels of 1 log10 copies/mg of PBMC DNA and a 2.9% (95% CI, 0.2%–5.6%) annual rate of CD4+ T cell decline for each increase in plasma HIV RNA levels of 1 log10 copies/mL, regardless of HIV type. Therefore, despite the strong correlations seen between HIV RNA and DNA levels, our data suggest that plasma and PBMC viral loads each independently contribute to the rate of CD4+ T cell decline. Baseline CD4+ T cell counts among HIV-1–infected subjects and HIV-2– infected subjects were not associated with the rate of CD4+ T cell decline (data not shown). Although male subjects, compared with female subjects, had slightly higher plasma levels of HIV-1 RNA (mean ^ SD, 5:3 ^ 0:9 vs. 4:6 ^ 1:4 log10 copies/mL, respectively; P , :0001) and HIV-2 (mean ^ SD, 3:2 ^ 1:5 vs. 2:0 ^ 1:7 log10 copies/mL, respectively; P ¼ :0008), rates of CD4+ T cell decline did not significantly differ among these groups (data not shown). Furthermore, we

found no significant differences in the relationship between viral load and rate of CD4+ T cell decline when we compared data by site of recruitment or CSW status. Discussion In this study, we compared the virologic, immunologic, and clinical characteristics of 472 HIV-1– and 114 HIV-2–infected subjects from Senegal. Plasma HIV RNA levels were significantly higher among subjects infected with HIV-1 than among subjects infected with HIV-2, despite similar levels of HIV DNA in PBMC. This phenomenon has been observed in other cohorts in Senegal, The Gambia, and Guinea Bissau [2, 7, 20, 23–26]. The reasons for this distinction are unclear. It may be that, for a given infected target cell, HIV-1 proviruses are more transcriptionally active than HIV-2 proviruses and therefore are capable

912

Gottlieb et al.

of producing more plasma virus copies per infected target cell [33, 34]. Because our assay does not differentiate between integrated and unintegrated viral DNA in PBMC, it is also possible that HIV-1 has more integrated, replication-competent proviruses per infected target cell. Interestingly, in a recent study by Ibanez et al. [35] of 10 HIV-1– infected subjects, a 5-fold decrease in unintegrated HIV-1 DNA was found after 48 weeks of highly active antiretroviral therapy (HAART), despite maintenance of the integrated viral DNA copy number. Because infected PBMC represent only a small proportion of the potentially infected target cells in tissues, increased plasma levels of HIV-1 may reflect increased proviral DNA levels in compartments other than the blood. Jobe et al. [36] found that proviral DNA levels in lymph-node mononuclear cells in HIV-1– infected persons were
5 times higher than levels in HIV-2– infected persons. However, interpretation of that study is complicated by the fact that PBMC HIV-1 DNA levels were also found to be higher than PBMC HIV-2 DNA levels, which is contrary to the results of the present study and of other studies [23– 26]. Further studies to assess HIV-1 and HIV-2 RNA and DNA levels in other compartments may help to explain this phenomenon. Another factor may be host immune responses, which are likely to be important in viral clearance from the blood; it may be that anti– HIV-2 immune responses are more efficient than anti– HIV-1 responses at clearing virus from plasma. Previous studies have shown a correlation between baseline HIV-1 proviral DNA copy numbers, plasma HIV RNA levels, and HIV disease progression [15–18]. It has also been suggested that monitoring PBMC proviral DNA levels in subjects with undetectable plasma HIV-1 RNA who are receiving HAART may provide data relevant to disease progression [19]. The relationship between PBMC proviral DNA levels, plasma RNA levels, and disease progression in persons infected with HIV-2 is less clear. Several studies have examined HIV-2 DNA levels in PBMC and found an inverse correlation with CD4+ T cell count and a direct correlation with disease stage [2, 23, 25, 26, 37, 38]. Our findings confirm these studies. However, interpretation of correlations across studies should be made cautiously, because such comparisons may be confounded by differences in the expression of DNA measurements (as either copies per PBMC or copies per CD4+ T cells and not in volume of blood) [39]. Other studies investigating the relationship between plasma HIV-2 RNA levels and PBMC HIV-2 DNA levels are less consistent. Popper et al. [24] did not find a correlation between viral RNA and viral DNA levels or between viral DNA levels and CD4+ T cell count in 34 subjects infected with HIV-2. This contrasts with our findings and with a recent study by Ariyoshi et al. [21] that showed a strong correlation between HIV-2 RNA and DNA levels. Comparison of the rates of CD4+ T cell decline in the HIV-1 and HIV-2 cohorts shows some interesting differences. CD4+ T cell counts in the HIV-1 group are declining
4-fold faster (15.9% per year vs. 4.1% per year) than are CD4+ T cell counts

JID 2002;185 (1 April)

in the HIV-2 group. These rates are consistent with those observed by Jaffar et al. [40] in untreated subjects in The Gambia and correlate with the natural history of immune decline and disease progression observed in other cohorts of individuals infected with HIV-1 and HIV-2 in Africa [3]. The most intriguing finding of the present study was that plasma viral load, regardless of HIV type, predicts the rate of CD4+ T cell decline; consequently, the same plasma HIV RNA level in an HIV-1– or HIV-2–infected individual results in a similar rate of CD4+ T cell decline (
4% per year for each increase in plasma viral load of 1 log10 copies/mL). In addition, PBMC viral DNA levels were independently and strongly correlated with the rate of CD4+ T cell decline. We suspect that this is due, in part, to the strong positive correlation we found between HIV RNA and DNA levels in both HIV-1– and HIV-2–infected subjects. Our results are in contrast to those of Ariyoshi et al. [21], who found that, although plasma RNA levels and PBMC DNA levels are both predictive of CD4+ T cell decline in HIV2– infected subjects, PBMC DNA levels were not independently associated with the rate of CD4+ T cell decline. To our knowledge, this is the first study to demonstrate that the rate of CD4+ T cell decline and, therefore, progression to AIDS is a function of viral load, regardless of whether subjects are infected with HIV-1 or HIV-2. The implications of these findings are noteworthy. First, differences in the pathologic features and natural histories of HIV-1 and HIV-2 disease likely are the result of host-virus relationships that, on average, maintain HIV-1 RNA load at a higher level than HIV-2 RNA load. Of note, Popper et al. [7], as well as Andersson et al. [20], have speculated that differences in pathogenicity and the time from seroconversion to development of AIDS that were observed after HIV-1 infection and after HIV-2 infection resulted from the lower plasma viral RNA loads seen in the HIV-2 cohorts. However, our study provides the first direct evidence to support this hypothesis. Second, subjects with HIV-2 infection who have high viral loads are at significant risk for CD4+ T cell depletion and, consequently, for AIDS-defining events; these occur at rates comparable to the rate among subjects infected with HIV-1. Further studies aimed at confirming these findings and correlating them with rates of mortality and AIDS-defining events are warranted. This study had several potential limitations. Of the 864 HIV1– and 168 HIV-2– seropositive individuals we screened, only 54.6% (n ¼ 472) of the HIV-1– and 67.9% (n ¼ 114) of the HIV-2– infected individuals had available data on CD4+ T cell count and viral load and, consequently, could be included in the study. How this selection bias may have affected (i.e., how inclusion of excluded individuals would have affected) the reported results is unclear. Additionally, the initial date of HIV infection was unknown for the vast majority of subjects, which precluded assessment of the true rates of clinical disease progression. Subjects in our cohort who were infected with HIV-2 probably were infected longer than were those with HIV-1, be-

JID 2002;185 (1 April)

HIV RNA as a Predictor of CD4+ T Cell Decline

cause HIV-2 has been circulating in Senegal for a much longer period of time than has HIV-1 [41– 44] and because the course of HIV-2 disease is typically longer than that of HIV-1 disease. Clinical follow-up in our cohorts has been difficult, because many patients, once they develop end-stage disease, do not present again for medical care and are lost to follow-up. As a result, no comprehensive attempt was made to analyze AIDS-defining events or mortality longitudinally. Rather, infected individuals were stratified to discrete disease stages (based on the World Health Organization staging system [32] and CD4+ T cell count) to generate the comparisons described. The validity of our results relies on our ability to accurately quantify viral RNA and DNA levels. The HIV-1 PCR assay used here was based on the Roche Amplicor test and has been shown to accurately quantitate the non–subtype B HIV-1 strains found in Senegal [29] (authors’ unpublished observations). The HIV-2 PCR assay we used is also based on the Roche Amplicor test, using primers specific for HIV-2, and has been shown to accurately quantitate HIV-2 strains, including the subtype A strain that is pervasive in Senegal [20] (authors’ unpublished observations).The PCR assays used in this study have also produced measurements similar in range and distribution to those reported in numerous other studies in Africa. Finally, although we observed rates of CD4+ T cell decline that were similar among individuals with similar plasma RNA levels, regardless of whether the infection was caused by HIV-1 or HIV-2, the distributions of plasma HIV-1 and HIV-2 RNA levels did not overlap in the majority of subjects. Therefore, overlapping data points had to be extrapolated from a small number of HIV-1–infected subjects (n ¼ 11) with low RNA levels (,1000 copies/mL) and a small number of HIV-2– infected subjects (n ¼ 7) with high RNA levels (>1000 copies/mL). In summary, we have demonstrated that, even after the analysis has been adjusted for CD4+ T cell count, plasma viral RNA levels are significantly greater among individuals infected with HIV-1 than among individuals infected with HIV-2. This occurs even though HIV-1– and HIV-2–infected subjects have similar PBMC viral DNA levels. We also have shown a strong correlation between HIV-1 and HIV-2 plasma RNA and PBMC DNA levels. Finally, we have presented evidence that plasma HIV load is predictive of the rate of CD4+ T cell decline over time and that the correlation between viral load and rate of CD4+ T cell decline is similar among all HIV-infected individuals, regardless of whether they harbor HIV-1 or HIV-2. We hope that other studies will rapidly confirm this finding, because it suggests that appropriate antiretroviral therapy should be considered for HIV-2–infected patients who have viral loads comparable to loads that warrant treatment in patients with HIV-1 infection. Future studies to address the underlying mechanisms of these observations should provide more insight into the clinical and pathogenic features that distinguish infections with each of these viruses and, thus, provide insight into ways to combat them.

913

Acknowledgments

We thank Deana Rich, Elise Reay-Ellers, and Macoumba Toure, for their invaluable coordination of all study procedures in Senegal; Mame Dieumbe Mbengue-Ly, Marie Pierre Sy, and Dr. Pierre Ndiaye, for patient care; Diouana Ba and Haby Agne, for their laboratory work; Fatou Faye-Diop, for data entry; and Alison Starling, for forms development and data management. We thank Cindy Christopherson, Kelly LeGassic, and Shirley Kwok, from Roche Molecular Diagnostics, for human immunodeficiency virus load assay development and performance. We thank Marcel Curlin and Mark Jensen for helpful discussions.

References 1. Kanki PJ, Travers KU, MBoup S, et al. Slower heterosexual spread of HIV-2 than HIV-1. Lancet 1994; 343:943–6. 2. Simon F, Matheron S, Tamalet C, et al. Cellular and plasma viral load in patients infected with HIV-2. AIDS 1993; 7:1411– 7. 3. Marlink R, Kanki P, Thior I, et al. Reduced rate of disease development after HIV-2 infection as compared to HIV-1. Science 1994; 265: 1587– 90. 4. Comparison of vertical human immunodeficiency virus type 2 and human immunodeficiency virus type 1 transmission in the French prospective cohort. The HIV Infection in Newborns French Collaborative Study Group. Pediatr Infect Dis J 1994; 13:502–6. 5. Adjorlolo-Johnson G, De Cock KM, Ekpini E, et al. Prospective comparison of mother-to-child transmission of HIV-1 and HIV-2 in Abidjan, Ivory Coast. JAMA 1994; 272:462– 6 (erratum: JAMA 1994; 272: 1482). 6. Prazuck T, Yameogo JM, Heylinck B, et al. Mother-to-child transmission of human immunodeficiency virus type 1 and type 2 and dual infection: a cohort study in Banfora, Burkina Faso. Pediatr Infect Dis J 1995; 14: 940– 7. 7. Popper SJ, Sarr AD, Travers KU, et al. Lower human immunodeficiency virus (HIV) type 2 viral load reflects the difference in pathogenicity of HIV-1 and HIV-2. J Infect Dis 1999; 180:1116–21. 8. Los Alamos National Laboratory. HIV sequence database. Available at http://hiv-web.lanl.gov/seq-db.html. Accessed 2001. 9. Remy G. HIV-2 infection throughout the world: a geographical perspective [in French]. Sante 1998; 8:440–6. 10. O’Brien TR, Blattner WA, Waters D, et al. Serum HIV-1 RNA levels and time to development of AIDS in the Multicenter Hemophilia Cohort Study. JAMA 1996; 276:105–10. 11. Schacker TW, Hughes JP, Shea T, Coombs RW, Corey L. Biological and virologic characteristics of primary HIV infection. Ann Intern Med 1998; 128:613– 20. 12. Mellors JW, Kingsley LA, Rinaldo CRJ, et al. Quantitation of HIV-1 RNA in plasma predicts outcome after seroconversion. Ann Intern Med 1995; 122:573–9. 13. Mellors JW, Mun˜oz A, Giorgi JV, et al. Plasma viral load and CD4+ lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med 1997; 126:946– 54. 14. Mellors JW, Rinaldo CR Jr, Gupta P, White RM, Todd JA, Kingsley LA. Prognosis in HIV-1 infection predicted by the quantity of virus in plasma. Science 1996; 272:1167–70. 15. Verhofstede C, Reniers S, Van Wanzeele F, Plum J. Evaluation of proviral copy number and plasma RNA level as early indicators of progression in HIV-1 infection: correlation with virological and immunological markers of disease. AIDS 1994; 8:1421–7. 16. Escaich S, Ritter J, Rougier P, et al. Relevance of the quantitative detection

914

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

Gottlieb et al.

of HIV proviral sequences in PBMC of infected individuals. AIDS Res Hum Retroviruses 1992; 8:1833– 7. Lefrere JJ, Mariotti M, Wattel E, et al. Towards a new predictor of AIDS progression through the quantitation of HIV-1 DNA copies by PCR in HIV-infected individuals. Br J Haematol 1992; 82:467– 71. Chevret S, Kirstetter M, Mariotti M, Lefrere F, Frottier J, Lefrere JJ. Provirus copy number to predict disease progression in asymptomatic human immunodeficiency virus type 1 infection. J Infect Dis 1994; 169: 882–5. Ferre F, Moss RB, Daigle A, Richieri SP, Jensen F, Carlo DJ. Viral load in peripheral blood mononuclear cells as surrogate for clinical progression. J Acquir Immune Defic Syndr Hum Retrovirol 1995; 10:S51– 6. Andersson S, Norrgren H, da Silva Z, et al. Plasma viral load in HIV-1 and HIV-2 singly and dually infected individuals in Guinea-Bissau, West Africa: significantly lower plasma virus set point in HIV-2 infection than in HIV-1 infection. Arch Intern Med 2000; 160:3286–93. Ariyoshi K, Jaffar S, Alabi AS, et al. Plasma RNA viral load predicts the rate of CD4 T cell decline and death in HIV-2–infected patients in West Africa. AIDS 2000; 14:339–44. Berry N, Ariyoshi K, Jaffar S, et al. Low peripheral blood viral HIV-2 RNA in individuals with high CD4 percentage differentiates HIV-2 from HIV-1 infection. J Hum Virol 1998; 1:457–68. Berry N, Ariyoshi K, Jobe O, et al. HIV type 2 proviral load measured by quantitative polymerase chain reaction correlates with CD4+ lymphopenia in HIV type 2– infected individuals. AIDS Res Hum Retroviruses 1994; 10:1031–7. Popper SJ, Sarr AD, Gueye-Ndiaye A, Mboup S, Essex ME, Kanki PJ. Low plasma human immunodeficiency virus type 2 viral load is independent of proviral load: low virus production in vivo. J Virol 2000; 74:1554–7. Ariyoshi K, Berry N, Wilkins A, et al. A community-based study of human immunodeficiency virus type 2 provirus load in rural village in West Africa. J Infect Dis 1996; 173:245–8. Norrgren H, Marquina S, Leitner T, et al. HIV-2 genetic variation and DNA load in asymptomatic carriers and AIDS cases in Guinea-Bissau. J Acquir Immune Defic Syndr Hum Retrovirol 1997; 16:31– 8. Langley CL, Benga-De E, Critchlow CW, et al. HIV-1, HIV-2, human papillomavirus infection and cervical neoplasia in high-risk African women. AIDS 1996; 10:413– 7. Triques K, Coste J, Perret JL, et al. Efficiencies of four versions of the AMPLICOR HIV-1 MONITOR test for quantification of different subtypes of human immunodeficiency virus type 1. J Clin Microbiol 1999; 37:110– 6. Michael NL, Herman SA, Kwok S, et al. Development of calibrated viral load standards for group M subtypes of human immunodeficiency virus type 1 and performance of an improved AMPLICOR HIV-1 MONITOR

30. 31.

32. 33.

34.

35.

36.

37.

38.

39.

40.

41. 42. 43. 44.

JID 2002;185 (1 April)

test with isolates of diverse subtypes. J Clin Microbiol 1999; 37: 2557–63. Christopherson C, Lu S, Kwok S. Laboratory markers of antiviral activity. Antivir Ther 1998; 3:247– 50. Christopherson C, Kidane Y, Conway B, Krowka J, Sheppard H, Kwok S. PCR-based assay to quantify human immunodeficiency virus type 1 DNA in peripheral blood mononuclear cells. J Clin Microbiol 2000; 38: 630–4. Interim proposal for a WHO staging system for HIV infection and disease. Wkly Epidemiol Rec 1990; 65:221–4. Hannibal MC, Markovitz DM, Clark N, Nabel GJ. Differential activation of human immunodeficiency virus type 1 and 2 transcription by specific T-cell activation signals. J Virol 1993; 67:5035–40. Tong-Starksen SE, Welsh TM, Peterlin BM. Differences in transcriptional enhancers of HIV-1 and HIV-2: response to T cell activation signals. J Immunol 1990; 145:4348–54. Ibanez A, Puig T, Elias J, Clotet B, Ruiz L, Martinez MA. Quantification of integrated and total HIV-1 DNA after long-term highly active antiretroviral therapy in HIV-1–infected patients. AIDS 1999; 13:1045–9. Jobe O, Ariyoshi K, Marchant A, et al. Proviral load and immune function in blood and lymph node during HIV-1 and HIV-2 infection. Clin Exp Immunol 1999; 116:474–8. Sarr AD, Popper SJ, Thior I, et al. Relation between HIV-2 proviral load and CD4+ lymphocyte count differs in monotypic and dual HIV infections. J Hum Virol 1999; 2:45–51. Gomes P, Taveira NC, Pereira JM, Antunes F, Ferreira MO, Lourenco MH. Quantitation of human immunodeficiency virus type 2 DNA in peripheral blood mononuclear cells by using a quantitative-competitive PCR assay. J Clin Microbiol 1999; 37:453–6. Cone RW, Gowland P, Opravil M, Grob P, Ledergerber B. Levels of HIVinfected peripheral blood cells remain stable throughout the natural history of HIV-1 infection. Swiss HIV Cohort Study. AIDS 1998; 12: 2253–60. Jaffar S, Wilkins A, Ngom PT, et al. Rate of decline of percentage CD4+ cells is faster in HIV-1 than in HIV-2 infection. J Acquir Immune Defic Syndr Hum Retrovirol 1997; 16:327– 32. Kawamura M, Yamazaki S, Ishikawa K, Kwofie TB, Tsujimoto H, Hayami M. HIV-2 in West Africa in 1966. Lancet 1989; 1:385. De Cock KM, Brun-Vezinet F, Soro B. HIV-1 and HIV-2 infections and AIDS in West Africa. AIDS 1991; 5(Suppl 1):S21–8. Marlink R. Lessons from the second AIDS virus, HIV-2. AIDS 1996; 10: 689–99. Ndour M, Sow PS, Coll-Seck AM, et al. AIDS caused by HIV1 and HIV2 infection: are there clinical differences? Results of AIDS surveillance 1986– 97 at Fann Hospital in Dakar, Senegal. Trop Med Int Health 2000; 5:687–91.