HIV-1 and T cell dynamics after interruption of highly
active antiretroviral therapy (HAART) in patients with
a history of sustained viral suppression
Richard T. Davey, Jr.*†, Niranjan Bhat*, Christian Yoder‡, Tae-Wook Chun*, Julia A. Metcalf*, Robin Dewar§, Ven Natarajan§,
Richard A. Lempicki§, Joseph W. Adelsberger§, Kirk D. Miller‡, Joseph A. Kovacs‡, Michael A. Polis*, Robert E. Walker*, Judith Falloon*,
Henry Masur‡, Dennis Gee¶, Michael Baseler§, Dimiter S. Dimitrov¶, Anthony S. Fauci*, and H. Clifford Lane*
*National Institute of Allergy and Infectious Diseases, (NIAID) and‡Critical Care Medicine Department, National Institutes of Health, Bethesda, MD 20892;
§Science Applications International Corporation (SAIC), Frederick, MD 21702; and¶National Cancer Institute?Frederick Cancer Research and Development
Center, National Institutes of Health, Frederick, MD 21702
Contributed by Anthony S. Fauci, October 20, 1999
Identifying the immunologic and virologic consequences of dis-
continuing antiretroviral therapy in HIV-infected patients is of
patients with HIV-1 infection. We designed a trial to characterize
these parameters after interruption of highly active antiretroviral
therapy (HAART) in patients who had maintained prolonged viral
suppression on antiretroviral drugs. Eighteen patients with CD4?
T cell counts > 350 cells??l and viral load below the limits of
detection for >1 year while on HAART were enrolled prospectively
in a trial in which HAART was discontinued. Twelve of these
resting, latently infected CD4 cells. Viral load relapse to >50
copies?ml occurred in all 18 patients independent of prior IL-2
treatment, beginning most commonly during weeks 2–3 after
cessation of HAART. The mean relapse rate constant was 0.45 (0.20
log10 copies) day?1, which was very similar to the mean viral
clearance rate constant after drug resumption of 0.35 (0.15 log10
copies) day?1(P ? 0.28). One patient experienced a relapse delay
to week 7. All patients except one experienced a relapse burden to
>5,000 RNA copies?ml. Ex vivo labeling with BrdUrd showed that
CD4 and CD8 cell turnover increased after withdrawal of HAART
and correlated with viral load whereas lymphocyte turnover de-
creased after reinitiation of drug treatment. Virologic relapse
occurs rapidly in patients who discontinue suppressive drug ther-
apy, even in patients with a markedly diminished pool of resting,
latently infected CD4?T cells.
HIV-1 infection ? antiretroviral drugs ? viral load ? relapse ? CD4
of HIV-1 replication to very low levels for an extended period of
time in patients with HIV-1 infection. Sustained control over
viral propagation has translated into a dramatic reduction in the
incidence of AIDS-associated opportunistic infections and
deaths in those with access to these medications (1, 2). However,
the benefits of HAART are often accompanied by a substantial
degree of drug-related toxicity, the full spectrum of which
continues to be defined (3–6). The considerable expense, in-
convenience, and rigid requirement for adherence associated
with the use of HAART regimens has further weakened the
enthusiasm of many patients for their long-term use (7).
One of the implicit hopes accompanying strategies designed to
suppress viral activity is that, at some point, the host’s immune
system might be able to maintain this suppression either in the
absence of, or at least after a de-intensification of, the current
medications (8–12). Unfortunately, with the exception of occa-
sional anecdotal reports of sustained suppression, fairly prompt
virologic relapse after discontinuation of HAART has been the
predominant pattern observed thus far (13–16). Given that the
process of lymphocyte recovery after institution of HAART has
he advent of highly active antiretroviral therapy (HAART)
been projected to occur on the scale of months to years, it may
be necessary to continue HAART without interruption for an
extended period of time before embarking on any strategy in
which direct viral suppression is withdrawn (17,18). In addition,
the success rate of any withdrawal strategy may be critically
resting, latently infected CD4?T cells, and possibly other viral
reservoirs, have already been substantially depleted by using a
variety of strategies (19–23).
With these precepts in mind, we instituted a prospective trial
in which long-term responders to HAART were offered partic-
ipation in a study in which all antiretroviral drug therapy was
stopped. Included in this trial were several recipients of therapy
with intermittent cycles of IL-2 plus continuous HAART, a
treatment shown in a previous study to be associated with a low
frequency of resting, latently infected CD4?T cells (24). We
sought to establish whether a subset of individuals could be
identified in whom viral rebound either did not occur or did so
to comparatively low levels or with a delayed onset. Our goal was
also to determine whether baseline immunologic and?or viro-
logic parameters were predictive of the pattern of relapse.
Study Population. Eligible patients were asymptomatic HIV-
infected adults with baseline CD4?T cell counts ?350 cells??l
who had been on continuous HAART for a minimum period of
1 year and who had documentation of viral loads consistently
below the limits of detection for at least that period of time.
HAART was defined as a minimum of a triple drug combina-
tion, including at least two licensed reverse transcriptase inhib-
itors and at least one licensed protease inhibitor. Viral loads
below the limits of detection were defined as being those below
a maximum of 500 HIV-1 RNA copies?ml on every determi-
nation for a period of at least 1 year before screening and a level
below 50 copies?ml on at least two determinations immediately
before enrollment. All patients provided written informed con-
sent as approved by the National Institute of Allergy and
Infectious Diseases Institutional Review Board.
Laboratory Tests. Peripheral blood specimens were collected at
baseline for determination of viral load by bDNA methodology
(Bayer Diagnostics, Norwood, MA), HIV-1 gag RNA and pro-
viral DNA quantitation by PCR (25), viral co-cultivation for
peripheral blood mononuclear cell.
†To whom reprint requests should be addressed at: Building 10, Room 11C-103, National
Institutes of Health, Bethesda, MD 20892-1880. E-mail: email@example.com.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
December 21, 1999 ?
vol. 96 ?
no. 26 ?
frequency of resting, latently infected CD4?cells (24, 26, 27),
flow cytometry for lymphocyte phenotyping, and antigen-
stimulated lymphocyte blastogenesis. Incorporation of 5-bro-
mo-2? deoxyuridine (BrdUrd) into lymphocytes ex vivo was
measured after a 4-hour incubation of BrdUrd with whole blood,
cell surface staining for CD3, CD4, and CD8 phenotypes,
paraformaldehyde fixation, DNase treatment, staining with anti-
Lymph node biopsies for in situ studies and viral quantitation,
and lumbar punctures for viral quantitation (reverse transcrip-
tion–PCR) in cerebrospinal fluid (CSF) were also performed in
a subset of patients.
On day 0, patients discontinued all of their antiretroviral
medications simultaneously and then underwent 6 months of
intensive monitoring. Repeat lymph node biopsies and lumbar
punctures were performed after viral relapse in a subset of
Study Intervention. Patients were instructed to resume their
three conditions were met: (i) the CD4?T cell count declined at
least 25% from the mean of three baseline determinations, (ii)
viral load increased at least two log10-fold (i.e., reached 5000
HIV-1 RNA copies?ml) above baseline, or (iii) the patient
elected to resume drug treatment independent of these thresh-
olds. Patients who declined to resume antiretroviral medications
were counseled, but remained eligible for continued monitoring.
Data Analysis. Linear regression analysis of the first measurable
data points was used to calculate rate constants of viral rebound
and to extrapolate to the plasma virus concentration, V0, at the
time of withdrawing therapy. The empirical constants, k, char-
acterizing the rate of virus rebound, were calculated assuming an
exponential dependence of the plasma virus concentration V on
time t: V ? Voexp(kt). Similarly, V ? Vmexp(?ct) defined the
concentration at time of reinitiation of therapy, VM. Depen-
dence of log10V on t was computer-fitted (Scientist, MicroMath,
Salt Lake City) by using duplicated data points.
Similar equations were used to fit the data for CD4?and
CD8?T cell changes. For example, the number of CD4 cells, T4,
was assumed to depend exponentially on time [T4 ?
T40exp(kT4?t)], and, therefore, log10T4 ? 0.434?kT4?t ? const,
where T40is the number of cells at t ? 0, the coefficient 0.434
is a conversion factor from rate constants to log10cell number,
and const does not depend on time. In all cases, the empirical
equations described above were used in this study to describe the
dynamics of HIV-1 RNA and T cells as a means of comparison
between the periods after stopping HAART and its reinitiation,
as well as between IL-2-treated and untreated patients. Accord-
ingly, the constants in these equations were determined for the
stated (limited) time intervals and may not accurately reflect
complex viral and T cell dynamics over a longer time interval.
Enrollment commenced in January 1999. Baseline characteris-
tics of the first 18 patients are shown in Table 1. All were
CD4?T cell count of 921 ? 85 cells??l (range 454–1,868), and
whose major risk factor for exposure to HIV-1 was unprotected
sex with other men. Experience of this cohort with any of the
licensed protease inhibitors extended back for a mean period of
137 ? 8.1 weeks (range 83–248 weeks) before study entry. The
most current regimen of HAART medications had been admin-
istered for a mean period of 117 ? 12.0 weeks (range 15–248
weeks) before stopping drugs. The numbers of prestudy recip-
ients of three-drug, four-drug, or five-drug regimens were 14, 2,
and 2, respectively. Four patients included an non-nucleoside
reverse transcriptase inhibitor in their current HAART regi-
mens for a mean duration of 84 ? 4.3 weeks (range 76–96
plasma viremia could be documented to have been consistently
under 500 RNA copies?ml was 108 ? 13.3 weeks. PCR ampli-
fication of proviral DNA in peripheral blood mononuclear cells
(PBMCs) was below the quantitation threshold of one copy per
106cells in one participant. In three patients, including the one
individual with proviral DNA signal below the limits of detec-
tion, gag RNA in PBMCs was not found despite using a sensitive
nested PCR technique. In three of the five patients who under-
went inguinal lymph node biopsies at study entry, virus levels in
the excised nodes were below the limits of detection of 50 copies
Table 1. Demographics and baseline characteristics of the study population
Most recent HAART
gag RNA PCR
Lymph node virus,
Not detected Not detected
www.pnas.org Davey et al.
per 107cells. In all 10 patients who underwent a lumbar puncture
for viral quantitation at baseline, the level of CSF HIV-1 was
Twelve (67%) participants had received prior therapy with
either intermittent intravenous or subcutaneous IL-2 in combi-
nation with HAART as part of ongoing clinical trials. Of these
12, lymphocytes from two [patients 3 and 15 (Table 2)] had
recently been reported as having a frequency of resting, latently
infected CD4?cells below the limits of detection (28). Three
additional recipients of HAART ? IL-2 in the present cohort
also fulfilled this criterion, as did two recipients of HAART
alone (Table 2).
Viral Dynamics. Plasma virus rose above 50 RNA copies?ml in all
participants, generally within 2–3 weeks of stopping antiretro-
viral medications (Figs. 1 and 2). It is noteworthy that, in some
patients, plasma viremia appeared to peak quickly and then
spontaneously fall to an intermediate level even before resuming
drug therapy. Prior therapy with IL-2 did not influence the rate
a baseline proviral DNA copy number below detection, absent
gag RNA PCR signal, and a baseline frequency of resting,
rebound was delayed by several more weeks relative to the other
patients. In an additional patient (patient 18), plasma virus
RNA copies?ml despite the patient’s continued abstinence from
Reinstitution of HAART after relapse was associated with a
prompt reduction in plasma viral load in all patients who elected
to resume treatment (Fig. 4). The mean (median) time to
resuppression of plasma viremia to ?50 copies?ml in the first 16
patients to resume HAART therapy was 79 (84) days (range
when HAART was restarted.
Three representative graphs of viral relapse after discontinuation of HAART, with linear regression lines (red) depicting back extrapolation of viral load
Table 2. Parameters of viral dynamics after stopping and reinitiating therapy
Patient k, day?1
260,000 (4.7) 0.36
0.225.4 (1.6)9 11 760,000 (5.9)0.06
k and c are the viral relapse and viral clearance rate constants, respectively, as calculated based on the extrapolated plasma virus concentration at baseline,
Voand the measured concentration at drug resumption, VM. t50and t500are the time to return of plasma virus above 50 and 500 copies/ml, respectively. N0and
U, below the limits of detection.
*CSF viral quantitation by reverse transcription–PCR at the time of restarting drug.
Davey et al.
December 21, 1999 ?
vol. 96 ?
no. 26 ?
Virus relapse kinetics were rapid, with rebound rate constants
(k) ranging from 0.12 to 0.91 day?1(Table 2). Viral clearance
constants (c) after restarting drugs did not vary significantly,
their mean value being approximately the same as the mean
rebound rate constant (0.36 ? 0.06 vs. 0.45 ? 0.22 day?1, P ?
0.28). Correlation between k and c was strong (r ? 0.77, P ?
0.008). The rebound and clearance rate constants also weakly
correlated with the logarithm of the estimated HIV RNA
concentrations before therapy was withdrawn (P ? 0.046 and
P ? 0.084, respectively). However, no correlation was found
between k and the baseline frequency of resting, latently infected
CD4 cells (Table 2, column N0) in the peripheral blood. Re-
bound and clearance rate constants could also be calculated by
fitting time points for all patients normalized to a common data
point ?1 week after the first detectable HIV RNA value or the
data point at the time when therapy was reinitiated. By using this
approach, the mean values of the rebound and clearance rate
constants, 0.43 and 0.30 day?1, respectively, were quite similar to
those presented above.
After reappearance of plasma virus, the frequency of resting,
latently infected CD4 cells (Table 2, column N1) quickly in-
creased in all participants. The logarithm of N1correlated with
the log10plasma HIV-1 RNA concentration (P ? 0.001). Eight
of ten patients with baseline CSF viral quantitation also under-
went a repeat lumbar puncture just before restarting drug: CSF
HIV-1 in six had risen to ?50 copies?ml (Table 2).
CD4 T Cell Dynamics. The number and percentage of CD4?T cells
decreased in most patients after withdrawal of therapy and
increased after reinstitution of treatment (Fig. 5). Compared
with the HIV-1 RNA rebound and clearance rate constants, the
absolute values of the CD4?T cell initial decrease and increase
rate constants, kT4 and cT4, respectively, were on average more
than one log10smaller (Table 3). There were statistically signif-
icant differences between kT4 and cT4 (P ? 0.001). Differences
was discontinued, color-coded as indicated.
The viral relapse patterns observed in 18 patients in whom HAART
separated into the 12 previous IL-2-recipients (red line) and the 6 non-IL-2
recipients (blue line).
The mean viral load results (? SE) for the 18 patients (black line),
in each patient), color-coded as in Fig. 2.
The individual patterns of viral load suppression in patients after
HAART was discontinued. (Upper) CD4 cell changes in three representative
patients over time after discontinuation and then resumption of HAART.
Vertical black arrows indicate when HAART was restarted in each patient.
(Lower) The mean CD4 cell counts (? SE) for the 18 patients (black line) after
discontinuation of HAART, separated into the 12 previous IL-2-recipients (red
line) and the 6 non-IL-2 recipients (blue line).
www.pnas.orgDavey et al.
between the linear rate constants characterizing the rates of
change in CD4?percentage after withdrawal and then resump-
tion of therapy were also statistically significant (P ? 0.007).
Similar results were obtained when the number of CD4?cells
was normalized either to day 2 after drug cessation or at the
resumption of therapy—the mean rate constants being equal to
?0.015 and 0.011 day?1, respectively. No significant correlations
existed between parameters characterizing the changes in CD4?
T cell numbers and the viral dynamics parameters except be-
tween cT4 and c (P ? 0.002). There were no significant changes
in the number of CD8?T cells.
Both CD4?and CD8?T cell turnover rates, as measured by
BrdUrd incorporation ex vivo by fresh PBMCs, increased after
termination of therapy and decreased after resumption of
HAART (Table 3; Fig. 6). BrdUrd incorporation into CD8?
cells was approximately twice that observed in CD4?cells. The
respective rate constants for the percentage of labeled CD4?
cells, kBT4 and cBT4, differed significantly (P ? 0.001). Inter-
estingly, there was a correlation (P ? 0.011) between the rate
constant of cell proliferation after reinitiation of therapy, cBT4,
and the log10number of resting, latently infected CD4?T cells
at the time drug therapy was resumed, N1. A strong correlation
(P ? 0.001) also existed between the logarithmic percentage
BrdUrd incorporation into both CD4?and CD8?T cells and
log10plasma viral RNA levels (Fig. 7).
Comparing week 8 to baseline values, four patients displayed
a five-fold or greater increase in lymphocyte proliferation ex vivo
to p24 antigen, five displayed increases between 1- and 5-fold,
and, in the remaining nine, these values actually declined over
that time period. The largest increase (28.6-fold) occurred in the
patient (patient 18) whose plasma viremia rebounded only to
very low copy numbers.
We report here the results of a prospective trial of cessation of
HAART in individuals chosen for their success at achieving
long-term virologic control. Despite a mean period on effective
suppressive therapy exceeding 2 years, varying degrees of viro-
logic relapse occurred promptly in all individuals. These findings
are especially notable not only for the consistency of HIV-1
rebound within only a few weeks of stopping drugs, but also for
the apparent lack of a significant treatment effect on the kinetics
of viral relapse and, by implication, viral reservoirs.
This cohort included a number of individuals in whom sensi-
tive methods of viral quantitation identified little or no virus at
baseline in the peripheral blood, lymph node tissue, CSF, or pool
of resting CD4?T cells. Some of the IL-2-treated patients, in
particular, had very low, sometimes unmeasurable, burdens of
infectious virus in their resting CD4?T cell pools, a trait already
distinguishing them from most recipients of HAART alone (24).
These favorable conditions notwithstanding, the kinetics of
ration of BrdUrd, in relationship to plasma viral load before and after inter-
ruption of HAART. (Upper) Mean percent BrdUrd incorporation (?SE) into
stopping HAART in comparison to plasma viral load (green dashed line). The
mean viral relapse rate constant k is indicated. (Lower) Similar mean changes
in CD4?(solid blue line) and CD8?(solid red line) T cells after reinitiating
HAART therapy, normalized to time zero (the day HAART was restarted in
line). The mean viral clearance rate constant c is indicated.
Changes in lymphocyte turnover, as measured by percent incorpo-
logarithmic percent BrdUrd incorporation into both CD4?and CD8?T cells
range after withdrawal of therapy are plotted.
The correlation between log10plasma RNA HIV-1 RNA levels and
Table 3. Rate constants for CD4?T Cell dynamics after stopping and reinitiating therapy
Mean (?s.d.) 0.015 (?0.011) 0.026 (?0.016) 0.018 (?0.015) 0.020 (?0.016) 0.047 (?0.039) 0.043 (?0.048)
was used for fitting the data after resuming therapy but with constants cBT4 and cBT8 for CD4 and CD8 cells, respectively.
Davey et al.
December 21, 1999 ?
vol. 96 ?
no. 26 ?
HIV-1 rebound in the IL-2-treated patients generally appeared
no different from those who had received HAART alone. The
lack of correlation between the number of latently infected cells
and the HIV-1 dynamics after stopping HAART suggests the
existence of a small reservoir of actively replicating virus that
could be of major importance in rekindling viral replication.
These findings also underscore the results of earlier studies
demonstrating persistence of HIV-1 RNA signal by PCR in
PBMCs and lymph node tissue from treated patients, as well as
more recent reports confirming that replication-competent
HIV-1 can persist for long periods at low levels in the majority
of treated individuals despite the use of maximally suppressive
antiretroviral regimens (19, 25, 29, 30).
One patient (patient 18) in this study appeared capable of
maintaining significant control of viral replication after discon-
tinuation of drug therapy. Lymphocytes from this particular
individual also demonstrated the greatest increase in prolifera-
a key element in achieving sustained virologic control is to either
preserve or to reconstitute a virus-specific CTL response in the
setting of CD4 helper function, such as by inducing regeneration
of CTL responses through antigenic priming in the setting of
limited viral rebound (31).
The kinetics of viral relapse after withdrawal of drug therapy
were similar to those reported by Harrigan et al. in a smaller
cohort of patients (15). By using two separate methods, the
calculated rebound rate constants (k) ranged from 0.12 to 0.91
day?1. The rate constants for viral clearance (c) on reinstitution
of drug therapy were much more uniform with a mean value of
0.36 day?1. These two constants correlated not only with each
other but also with the baseline log10viral load at time zero.
On stopping drug therapy, CD4 cell numbers fell at a similar
rate in both IL-2- and non-IL-2 recipients, and then began
recovering shortly after reinstitution of suppressive antiviral
BrdUrd, we were able to demonstrate that both CD4?and CD8?
T cells consistently increased their rate of BrdUrd incorporation
after drug cessation, and then decreased their level of incorpo-
ration back to baseline beginning shortly after resumption of
therapy. These data strongly support the hypothesis that lym-
phocyte turnover is increased in the setting of untreated HIV-1
infection and decreased after treatment and suppression of viral
replication. It is likely that active viral replication leads to
increased lymphocyte activation, destruction, and turnover and
that restoration of control over viral replication through resump-
tion of HAART leads to a return to more normal levels of
In summary, we have observed that long-term suppression of
HIV-1 by HAART does not confer on the host the ability to
ultimately control viral replication once drug therapy is with-
drawn. Laboratory surrogates indicating low or absent levels of
measurable virus may provide useful comparative data on the
relative potency of various therapeutic interventions, but do not
as yet appear predictive of the ultimate success in maintaining
control of viral replication once therapy is withdrawn. Greater
attention needs to be focused on developing not only more
potent antiretroviral drug regimens but also immune-enhancing
strategies with greater potential for depleting virus from pro-
tected or sequestered sites throughout the body. Until such
strategies are successful, it remains unclear whether the goal of
being able to discontinue antiretroviral therapy permanently will
ever be realized.
1. Chiasson, M. A., Berenson, L., Li, W. H., Schwartz, S., Singh, T., Forlenza, S.,
Mojica, B. A. & Hamburg, M. A. (1999) J. Acquired Immune Defic. Syndr. Hum.
Retrovirol. 21, 59–64.
2. Powderly, W. G., Landay, A. & Lederman, M. M. (1998) J. Am. Med. Assoc.
3. Carr, A., Samaras, K., Burton, S., Law, M., Freund, J., Chisholm, D. J. &
Cooper, D. A. (1998) AIDS 12, F51–F58.
4. Miller, K. D., Jones, E., Yanovski, J. A., Shankar, R., Feuerstein, I. & Falloon,
J. (1998) Lancet 351, 871–875.
5. Carr, A., Samaras, K., Chisholm, D. J. & Cooper, D. A. (1998) Lancet 351,
6. Notermans, D. W., van Leeuwen, R. & Lange, J. M. (1996) Drug Saf. 15,
7. Gallant, J. E. & Block, D. S. (1998) J. Int. Assoc. Physicians AIDS Care 4, 32–35.
8. Hall, D. B., Montaner, J. G. S., Reiss, P., Cooper, D., Vella, S., Dohnanyi, C.,
Myers, M., Lange, J. & Conway, B. (1998) AIDS 12, F41–F44.
9. Havlir, D. V., Marschner, I. C., Hirsch, M. S., Collier, A. C., Tebas, P., Bassett,
R. L., Ioannidis, J. P. A., Holohan, M. K., Leavitt, R., Boone, G., et al. (1998)
N. Engl. J. Med. 339, 1261–1268.
10. Reijers, M. H. E., Weverling, G. J., Jurriaans, S., Wit, F. W. N. M., Weigel,
P., et al. (1998) Lancet 352, 185–190.
11. Pialoux, G., Raffi, F., Brun-Vezinet, F., Meiffredy, V., Flandre, P., Gastaut,
J. A., Dellamonica, P., Yeni, P., Delfraissy, J. F. & Aboulker, J. P. (1998)
N. Engl. J. Med. 339, 1269–1276.
12. Cooper, D. A. & Emery, S. (1998) N. Engl. J. Med. 339, 1319–1321.
13. Garcia, F., Plana, M., Vidal, C., Cruceta, A., O’Brien, W. A., Panteleo, G.,
Pumarola, T., Gallart, T., Miro, J. M. & Gatell, J. M. (1999) AIDS 13, F79–F86.
14. Lisziewicz, J., Rosenberg, E., Lieberman, J., Jessen, H., Lopalco, L., Siliciano,
R., Walker, B. & Lori, F. (1999) N. Engl. J. Med. 340, 1683–1684.
15. Harrigan, P. R., Whaley, M. & Montaner, J. S. G. (1999) AIDS 13, F59–F62.
16. Neumann, A. U., Tubiana, R., Calvez, V., Robert, C., Li, T. S., Agut, H.,
Autran, B. & Katlama, C. (1999) AIDS 13, 677–683.
C., Debre, P. & Leibowitch, J. (1997) Science 277, 112–116.
18. Pakker, N. G., Notermans, D. W., de Boer, R. J., Roos, M. T., de Wolf, F., Hill,
A., Leonard, J. M., Danner, S. A., Miedema, F. & Schellekens, P. T. (1998) Nat.
Med. 4, 208–214.
19. Wong, J. K., Hezareh, M., Gunthard, H. F., Havlir, D. V., Ignacio, C. C., Spina,
C. A. & Richman, D. D. (1997) Science 278, 1291–1294.
20. Finzi, D., Hermankova, M., Pierson, T., Carruth, L. M., Buck, C., Chaisson,
R. E., Quinn, T. C., Chadwick, K., Margolick, J., Brookmeyer, R., et al. (1997)
Science 278, 1295–1300.
21. Chun, T. W., Stuyver, L., Mizell, S. B., Ehler, L. A., Mican, J. A., Baseler, M.,
Lloyd, A. L., Nowak, M. A. & Fauci, A. S. (1997) Proc. Natl. Acad. Sci. USA
22. Chun, T. W., Engel, D., Mizell, S. B., Ehler, L. A. & Fauci, A. S. (1998) J. Exp.
Med. 188, 83–91.
23. Ho, D. D. (1998) Science 280, 1866–1867.
24. Chun, T. W., Engel, D., Mizell, S. B., Hallahan, C. W., Fischette, M., Park, S.,
Davey, R. T., Jr., Dybul, M., Kovacs, J. A., Metcalf, J. A., et al. (1999) Nat. Med
25. Natarajan, V., Bosche, M., Metcalf, J. A., Ward, D. J., Lane, H. C. & Kovacs,
J. A. (1999) Lancet 353, 119–120.
26. Chun, T. W., Engel, D., Berrey, M. M., Shea, T., Corey, L. & Fauci, A. S. (1998)
Proc. Natl. Acad. Sci. USA 95, 8869–8873.
27. Chun, T. W., Carruth, L., Finzi, D., Shen, X., DiGiuseppe, J. A., Taylor, H.,
(London) 387, 183–188.
28. Chun, T. W., Davey, R. T. Jr., Engel, D., Lane, H. C. & Fauci, A. S. (1999)
Nature (London) 401, 874–875.
29. Zhang, L., Ramratnam, B., Tenner-Racz, K., He, Y., Vesanen, M., Lewin, S.,
Talal, A., Racz, P., Perelson, A. S., Korber, B. T., et al. (1999) N. Engl. J. Med.
30. Furtado, M. R., Callaway, D. C., Phair, J. P., Kunstman, K. J., Stanton, J. L.,
Macken, C. A., Perelson, A. S. & Wolinsky, S. M. (1999) N. Engl. J. Med. 340,
31. Rosenberg, E. S., Billingsley, J. M., Caliendo, A. M., Boswell, S. L., Sax, P. E.,
Kalams, S. A. & Walker, B. D. (1997) Science 278, 1447–1450.
www.pnas.org Davey et al.