Biological and Virologic Characteristics of Primary HIV Infection

Methods

Participants

In September 1993, with support from the University of Washington Center for AIDS Research, we established a clinical facility to enroll and follow persons with acute or very recent HIV-1 infection. Persons were eligible for the study if they were 1) positive for plasma HIV-1 RNA and negative for HIV-1 on enzyme immunoassay at study entry, 2) negative for HIV-1 on enzyme immunoassay within 6 months before study entry, or 3) negative for HIV-1 on enzyme immunoassay within 12 months before study entry with a documented retroviral syndrome in the 90 days before study entry [19]. Written documentation of previous negative HIV-1 test results was obtained from all participants.

Study Design

Enrollment Procedures

After we obtained informed consent, each participant completed a standardized interview designed to document recent medical illnesses and history of recent HIV-1 testing. A physical examination was done, and the results were recorded on standardized case report forms. Medical records and documentation of HIV-1 testing in the previous 12 months were reviewed. Peripheral blood mononuclear cells and blood were drawn for measurement of CD4 cell count and HIV-1 culture, HIV-1 RNA quantitation in plasma, and determination of HIV-1 antibody status.

Follow-up Protocol

Participants were seen every other week for 1 month, every month for 2 months, and quarterly thereafter. At each follow-up visit, a standardized interview was completed to document interim illnesses, use of medications, and visits to medical providers. Blood was drawn for measurement of CD4 cell count, HIV-1 cell culture, and measurement of HIV-1 RNA in plasma.

Participant Management and Primary Care

To ensure appropriate medical management during the study, decisions about clinical care and the initiation of antiretroviral therapy were made by physicians who were experienced in the care of patients with HIV-1 but were not associated with the study. At the time of the study, therapy for acute HIV-1 infection was uncommon. Laboratory data, including CD4 cell counts and plasma HIV-1 RNA levels, were supplied to primary care providers.

Data Analysis and Definitions

Participants were seen by two of the authors. Data were collected on standardized forms, were entered into a computer database by using a double-entry system, and were subsequently verified by the data manager. The date of acquisition of HIV-1 infection was defined as the earlier of the following two events: 1) the date on which the participant was found to be positive for p24 antigen [or HIV-1 RNA] and negative for HIV-1 on enzyme immunoassay or 2) the date of symptom onset, if an acute illness consistent with HIV-1 seroconversion was identified as temporally related to a suspected sexual exposure. Participants who had asymptomatic seroconversion were assigned a date midway between the date of their previous negative test result and the date of the new positive result on Western blot.

On the basis of the interview and review of the medical records at study entry, symptoms associated with HIV-1 seroconversion were graded on a five-point scale (1 = no symptoms, 2 = mild symptoms possibly associated with HIV-1 seroconversion but no medical evaluation of those symptoms, 3 = mild or moderate symptoms consistent with HIV-1 seroconversion and medical evaluation of those symptoms, 4 = moderate or severe symptoms and medical evaluation of those symptoms, and 5 = hospitalization for symptoms of an acute retroviral syndrome).

Laboratory Methods

HIV-1 Serology and T-Cell Subset Analysis

Testing for HIV-1 antibody was done with enzyme immunoassay and Western blot according to the manufacturers' instructions. T-cell subset analysis was done by using flow cytometry in the hematopathology laboratory of the University of Washington, which has been accredited by the AIDS Clinical Trials Group since 1986.

Viral Load

Isolation of HIV-1, quantitative microculture, and HIV-1 RNA testing were done as described elsewhere [20-24]. Plasma HIV-1 RNA levels were measured by using the branched-chain DNA assay (Chiron, Emeryville, California), which has a lower limit of detection of 10 000 copies of HIV-1 RNA per mL of plasma [22]. Plasma samples near the limit of detection (<12 000 copies/mL) were subsequently retested with the Amplicor method (Roche, Somerville, New Jersey), which has a lower limit of detection of 400 copies/mL [24, 25]. Of the 655 HIV-1 RNA samples in this report, 92 were tested by using both assays; 75 of the 92 (82%) also had fewer than 12 000 copies/mL according to testing with the Amplicor method.

Statistical Analysis

All analyses were done at the University of Washington Center for AIDS Research with the Splus statistical package, version 3.1 (StatSci, Seattle, Washington). The data were summarized by using means, medians, SDs, and percentages. On plots depicting log₁₀HIV-1 RNA and CD4 values versus time from seroconversion, the overall trend in the data was characterized by a line generated by using the Loess procedure [26] (based on all data), which is a procedure for drawing a representative smooth curve through data by using local regressions. We used this line to estimate the point at which the slope of the curve changed. The slope of the curve before and after the inflection point was then estimated by using a random-effects model [27].

A covariate of particular interest was defined by the average log₁₀HIV-1 RNA levels of each participant within specified time intervals (for example, 0 to 4 months after acquisition of HIV). However, because participants were seen at different times relative to acquisition and because mean plasma HIV-1 RNA levels varied over time, participants were classified as having “high” or “low” RNA levels within a particular period if their log₁₀HIV-1 RNA levels during that period were above or below the solid line shown in Figure 1 (the Loess value was subtracted from the individual RNA values and averaged within a subject; positive averages were classified as “high” levels, and negative averages were classified as “low” levels). Mixed-effects models [28] were used to compare the effect of various covariates on CD4 counts 6, 12, 18, and 24 months after acquisition of HIV. Each covariate was added individually to a base model, which included an intercept, effect of antiretroviral therapy, and separate slopes for the period before and after the change point (117 days for the log₁₀HIV-1 RNA level and 157 days for the CD4 cell count). An interaction term between the covariate and time was included because we expected that the effects of the covariates might be transitory. In addition to these fixed effects, random effects corresponding to the intercept (which captures possible participant-to-participant variation in initial CD4 cell count) and slope (which captures possible participant-to-participant variation in rate of decline in CD4 cell count) were included.

Time-to-event analysis of CD4 cell counts (for example, time to a CD4 cell count ≤ 300 cells/mm³) was done by using the Turnbull method [29, 30] to compute the survival curve. This approach corrects for the interval-censored nature of the end point. A Wald test was used to evaluate the significance of covariates in this analysis.

Results

Demographic Characteristics and Clinical Presentation of the Cohort

Between September 1993 and March 1996, 155 persons were referred for evaluation of acute or recent HIV-1 seroconversion. Seventy-four met the entry criteria and were enrolled within 6 months of seroconversion. Reasons for nonenrollment were depression about HIV-1 seroconversion (n = 2), inadequate documentation of a previous negative result on HIV-1 testing (n = 13), unwillingness to participate in the protocol (n = 3), and lack of HIV-1 seroconversion (n = 63). The clinical and serologic presentation of these participants is shown in Table 1. The demographic characteristics of the enrollees are similar to those described elsewhere [19].

Longitudinal Follow-up

Participants completed a total of 810 visits for 87.0 person-years of follow-up. The mean follow-up time was 14.1 months (range, 1 to 42 months); 43 participants (58%) completed 10 to 12 months of follow-up, and 31 (42%) completed more than 18 months of follow-up. Overall, 17 of the 74 enrollees (23%) discontinued participation in the protocol, and 9 of the 17 (53%) dropped out within 3 months of seroconversion. Reasons for discontinuation were depression about HIV-1 seroconversion (2 participants [12%]), moving to another area (4 participants [24%]), or loss to follow-up (11 participants [65%]). Review of medical records and laboratory data obtained before discontinuation indicated that the frequency, duration, and severity of HIV-1 seroconversion symptoms; CD4 cell counts; and plasma HIV-1 RNA levels at study entry were similar in the 17 participants who dropped out of the study and the 57 participants who did not.

Antiretroviral Therapy

Fourteen of 74 participants (19%) began receiving antiretroviral therapy during follow-up. The earliest initiation of antiretroviral therapy was 5 months after seroconversion. All 14 participants had antiretroviral therapy prescribed by their primary care providers for clinical progression. Initial antiretroviral regimens were reverse transcriptase inhibitor monotherapy in 8 participants and combinations of reverse transcriptase inhibitors in 6 participants.

Plasma HIV-1 RNA Levels

Figure 1 shows the rapid decrease in plasma levels of HIV-1 RNA over time. The median plasma HIV-1 RNA level within the first 30 days of infection was 235 000 copies/mL (range; 1 600 000 to 27 200 copies/mL). This decreased to 46 000 copies/mL (range, 543 000 to <200 copies/mL) 60 days after infection, 52 000 copies/mL (range, 1 074 000 to 200 copies/mL) 90 days after infection, and 36 000 copies/mL (range, 717 000 to 200 copies/mL) 120 days after infection. After 117 days after infection, an inflection point was reached at which HIV-1 levels stopped decreasing and gradually increased. Before day 117, plasma HIV-1 RNA levels decreased by 6.5% per week (95% CI, 3.2% to 9.6%). In comparison, the estimated decline in plasma RNA levels immediately after initiation of antiretroviral therapy among the 14 treated participants was 45% per week (CI, 2% to 69%). After day 120, plasma HIV-1 RNA levels increased by an average of 0.15% per week (CI, −0.3% to 0.6%).

Many participants with acute HIV-1 infection had relatively low plasma HIV-1 RNA levels in the first 30 days of infection. Figure 2 shows the sequential findings in two such participants. Of 14 participants seen within 30 days of onset of seroconversion symptoms, 3 (21%) had plasma HIV-1 RNA levels between 20 000 and 50 000 copies/mL, 1 (7%) had levels between 50 000 and 100 000 copies/mL, 7 (50%) had levels between 100 000 and 500 000 copies/mL, and 3 (21%) had levels of more than 500 000 copies/mL at study entry.

Considerable interparticipant variability was seen in plasma HIV-1 RNA levels at all time points. The interquartile range of values in plasma HIV-1 RNA was 473 000 copies/mL in the first month after seroconversion, 95 700 copies/mL in the second month, and 98 800 copies/mL after the inflection point at 4 months. We detected no significant change in the coefficient of variation for plasma HIV-1 RNA levels before and after the 4-month inflection point.

We isolated HIV-1 from peripheral blood mononuclear cells in all 74 participants. The median numbers of infectious units per million cells (IUPM) in peripheral blood mononuclear cells were 34.4, 15.2, 19.6, and 28.3 at 0 to 3, 4 to 6, 7 to 9, and 10 to 12 months after acquisition, respectively. A significant correlation (r = 0.52 [CI, 0.45 to 0.58]; P < 0.001) was seen between simultaneous measurements of infectious units in peripheral blood mononuclear cells and plasma HIV-1 RNA levels; however, plasma HIV-1 RNA levels were more sensitive for predicting subsequent rates of CD4 cell loss and disease progression.

CD4 T-Cell Measurements

In the 74 participants, a total of 703 T-cell subset measurements was made (mean, 8.6 samples per participant). Like plasma HIV RNA levels, CD4 cell counts were highest in the month after HIV-1 seroconversion. This peak was followed by a steady decline over several months and subsequent stabilization. Median CD4 cell counts were 700 cells/mm³ in the first month of infection and 608, 538, 461, 441, and 384 cells/mm³ at 0 to 3, 4 to 6, 7 to 9, 10 to 12, and 13 to 18 months after seroconversion, respectively. The rate of decrease in CD4 cell counts was highest in the first 160 days after acquisition of HIV-1, when CD4 cell counts decreased by a mean of 5.3 cells/mm³ per week (CI, 3.0 to 7.6 cells/mm³ per week). After day 160, the mean decrease in CD4 cell count was 1.9 cells/mm³ per week (CI, 1.4 to 2.5 cells/mm³ per week) (P = 0.007).

Clinical Course

Overall, 7 of the 74 participants (9%) progressed to a diagnosis of AIDS during follow-up (Table 2). One participant had biopsy-confirmed cytomegalovirus esophagitis and the wasting syndrome, and 6 participants progressed to a CD4 count less than 200 cells/mm³. For these 7 participants, the median time from acquisition of HIV-1 to diagnosis of AIDS was 18 months, and the median plasma HIV-1 RNA levels were 113 000 copies/mL at study entry, 81 000 copies/mL 6 months after symptom onset, and 62 000 copies/mL 12 months after symptom onset. Twelve other participants progressed to a CD4 cell count of 300 cells/mm³ or less without clinically definable AIDS-associated illnesses.

Association between Seroconversion Events and Rate of Progression

We evaluated virologic and clinical variables seen during the acute retroviral syndrome for their association with the subsequent rate of decline in CD4 cell count. Because we had seen an inflection point in the slope of plasma HIV-1 RNA levels at 120 days after seroconversion, we analyzed viral levels in plasma before and after this point for their association with decline in CD4 cell count and AIDS. We also analyzed the relation between the rate of decline in CD4 cell count and the clinical, diagnostic, and demographic characteristics outlined in Table 1.

Participants consulting a physician for symptoms associated with HIV seroconversion had a significantly faster progression to a CD4 cell count of 300 cells/mm³ or less over the course of follow-up than did participants who had subclinical seroconversion (or mild symptoms that did not prompt medical evaluation) (Table 3, Figure 3). However, the survival analysis shown in Figure 3 censors all events that occurred after a participant had a single CD4 cell count of 300 cells/mm³ or less; this ignores the possibility that a participant may have had a subsequent measurement greater than 300 cells/mm³. In fact, of the 19 participants who reached an end point of 300 cells/mm³ or less before starting antiretroviral therapy, 7 had a subsequent measurement greater than 300 cells/mm³ before the initiation of therapy and 5 had a subsequent measurement greater than 300 cells/mm³ after the initiation of antiretroviral therapy. To account for this variability, we fitted random-effects models to all data on CD4 count, comparing participants who were seen by their physicians for symptoms of HIV-1 seroconversion with those who were not. At 6 months after seroconversion, participants who consulted a medical caregiver during their acute retroviral syndrome had 155 fewer CD4 cells/mm³ than those who did not (P = 0.01). The difference was 150 cells/mm³ at 12 months (P = 0.016), 145 cells/mm³ at 18 months (P = 0.02), and 140 cells/mm³ at 24 months (P = 0.05) (Table 3).

High plasma HIV-1 RNA levels between days 120 and 365 were associated with a significantly faster rate of CD4 cell decline, but high plasma HIV-1 RNA levels in the first 120 days of infection were not. For example, the difference in CD4 cell count at 6, 12, 18, and 24 months between the group with high RNA levels and the group with low RNA levels in the first 4 months of infection varied from −98 to −119 cells/mm³ (P > 0.07) (Table 3). In contrast, comparison of persons with high (above the mean) and low (below the mean) RNA titers after 120 days of infection showed a difference of 230 cells/mm³ at 6 months after acquisition, 311 cells/mm³ at 12 months, 191 cells/mm³ at 18 months, and 172 cells/mm³ at 24 months (P < 0.01 at each time point). Figure 2 shows the findings in two participants with low plasma HIV-1 RNA levels at study entry who subsequently developed high plasma HIV-1 RNA levels between 6 and 12 months after acquisition and also had rapid progression of infection. Overall, nine participants (12%) followed this pattern.

No correlation was seen between plasma HIV RNA levels at study entry and subsequent rate of decline in CD4 cell count, and no difference was seen in rate of progression between participants enrolled with known dates of seroconversion (for example, those who were positive for p24 antigen and negative on Western blot) and those with estimated dates of seroconversion derived by using paired serologic test results (one negative and the follow-up positive).

Discussion

Our study provides a detailed picture of the virologic events surrounding untreated acute HIV-1 infection and subsequent rates of disease progression. We saw participants frequently and repeatedly during follow-up, especially during the first 2 months after enrollment, to develop a more detailed description of the virologic natural history of HIV-1 than previous studies have allowed. As others have described [7-10], plasma HIV-1 RNA levels are highest immediately after acquisition and then rapidly decrease. We observed an inflection point in the rate of change in the plasma HIV-1 RNA level at 120 days after acquisition and in the rate of change in the CD4 cell count at 160 days. During the first 4 months after acquisition, plasma HIV-1 RNA levels declined at a steady rate, averaging 6.5% per week, from a median of 235 000 copies/mL in the first 30 days after seroconversion to a median of 36 000 copies/mL at about day 120. After day 120, plasma RNA levels increased by 0.15% per week; this rate is similar to the rate of increase associated with the “chronic” stage of viral replication, suggesting that viral replication is increasing at a rate of about 1000 copies/mL per month. These data fit well with data from studies of the natural history of HIV that indicate that the copy number increases over the course of infection and that differences of 4000 to 8000 copies/mL are associated with major differences in the rate of outcome over a 3- to 5-year interval [31, 32].

Neither plasma HIV-1 RNA levels at study entry nor the average HIV-1 RNA plasma level during the first 120 days of infection was associated with the subsequent rate of disease progression. This lack of association may be due to the variability in plasma HIV-1 RNA levels that we saw during the acute and very early phases of infection, which was greater than previously reported [8-10, 33]. Possible explanations for the greater variability seen in our series compared with others include a larger cohort overall; greater diversity of referral sources; and more variability in the period between HIV-1 acquisition and study entry. Because one of our goals was to more thoroughly define the diversity of primary HIV-1 infection, these findings may not be surprising. Our data suggest that levels of virus in plasma measured shortly after HIV-1 acquisition are not likely to be accurate surrogate markers of host immunity or disease activity in trials of HIV-1 vaccines or immunotherapy.

Viral levels at the inflection point were associated with subsequent rates of CD4 cell decline and progression. Persons with high plasma HIV-1 RNA levels after day 120 after acquisition were more likely to have a rapid decline in CD4 cell count than were those with plasma HIV-1 RNA levels below the median. Mellors and colleagues [31, 32] reported that in the Multicenter AIDS Cohort Study (MACS) cohort, the first plasma HIV RNA level after seroconversion was significantly correlated with the rate of disease progression. We found no association of plasma HIV-1 RNA levels at study entry with subsequent rate of CD4 cell loss or AIDS progression (Table 3). However, the date of acquisition of HIV was not as precisely defined in the cohort reported on by Mellors and colleagues as it was in our cohort. The discrepancy between our results and those of Mellors and colleagues is probably due to the fact that their cohort was enrolled at or after the inflection point in HIV-1 plasma RNA levels (after day 120). This interpretation is supported by the observation that the median plasma HIV-1 RNA levels in our cohort in the 120- to 365-day period are similar to those reported by Mellors and colleagues at the time of enrollment of their cohort [31, 32].

Although the virologic titers of our participants seem to be similar to those described by Mellors and colleagues, our participants seemed to progress more rapidly than anticipated. In MACS, the median time from the estimated date of seroconversion to a CD4 count of 500 cells/mm³ was 48 months [31, 32, 34]; in our group, it was 6 months. In addition, 7 of 74 of our participants (9%) progressed to a diagnosis of AIDS within 31 months of follow-up; this rate is two to three times higher than that previously reported [34]. Alterations in the definition of AIDS make direct comparisons between the two data sets problematic. The high frequency of symptomatic seroconversion in our cohort may be one explanation for the high progression rate. Several retrospective studies [35-38] have suggested that symptomatic seroconversion may be associated with more rapid progression of disease. We found a marked difference in the rate of progression between those who were symptomatic enough to seek medical care during the acute retroviral illness and those who were not (Figure 3, Table 3). It is not clear why some persons have severe symptoms (resulting in a physician encounter) and some do not. Severe symptoms may be a surrogate marker of dissemination or an inadequate immune response or may be secondary to expression of cytokines. Prospective studies of patients at risk for HIV-1 are needed to evaluate the true rate of subclinical as opposed to clinically symptomatic seroconversion to determine which factors (for example, immunologic or epidemiologic factors) are associated with a more severe seroconversion syndrome.

There is growing interest in using antiretroviral therapy during primary HIV-1 infection to see whether the infection can be “aborted” or whether the rate of disease progression can be changed [15, 39, 40]. We have found that the magnitude of the HIV-1 envelope precursor frequency is an important factor in establishing the rate of subsequent disease progression [41]. Whether early aggressive therapy of acute HIV-1 infection will affect host responses and whether such therapy will make a difference in long-term virologic or clinical outcomes remains to be determined.

In summary, the early virologic and clinical course of HIV-1 infection varies widely. Perhaps more intriguing is that within 4 to 6 months after acquisition of HIV-1, viral host interactions are established that are predictive of subsequent disease. Study of these early virus-host interactions may provide important opportunities to increase our understanding of the pathogenesis of HIV-1.

Dr. Hughes: 1001 Broadway, Suite 215, University of Washington, Seattle, WA 98122.

Dr. Shea: Cabrini Medical Tower, Suite 1320, 901 Boren Avenue, Seattle, WA 98104.

Dr. Coombs: Department of Laboratory Medicine (357110), University of Washington, 1959 Pacific Avenue, Seattle, WA 98195.

Dr. Corey: Fred Hutchinson Cancer Research Center, 1124 Columbia Street (M-115), Seattle, WA 98104.