HIV/AIDS: retroviral biology, immune pathogenesis, antiretroviral therapy, and the 40-year pandemic

Intuition Beginner

HIV is a virus that attacks the immune system itself — specifically CD4 T cells, the "generals" that coordinate immune defense. Over years, the virus gradually destroys these cells. When CD4 counts fall below 200 cells per microliter, the body can no longer fight off opportunistic infections (Pneumocystis pneumonia, Kaposi sarcoma, toxoplasmosis); this late stage is AIDS. The virus is hard to kill because it integrates into the host genome, becoming part of the patient's DNA, and mutates rapidly.

The 1996 breakthrough was to hit the virus with three drugs at once, a regime called HAART. Resistance to all three would require three simultaneous mutations, which is statistically impossible at the scale of a single patient. HIV went from a near-certain death sentence to a chronic, manageable condition in wealthy countries.

HIV/AIDS is the defining pandemic of the late twentieth and early twenty-first century. Since 1981, an estimated 88 million people have been infected and 42 million have died. In 2023, about 39 million people were living with HIV, 65 percent of them in sub-Saharan Africa. Understanding HIV is central to public health, immunology, virology, and health-equity work.

Visual Beginner

The HIV lifecycle is a chain of seven steps, each of which is the target of at least one drug class. The virus first attaches to the CD4 receptor and a co-receptor (CCR5 or CXCR4) on the surface of a T cell, fuses with the membrane, and releases its RNA. Reverse transcriptase converts RNA into DNA, integrase inserts the DNA into the host genome, and the host cell then transcribes and translates viral proteins. New virions assemble at the cell surface, bud off, and mature when protease cleaves the Gag-Pol polyprotein into functional units.

The picture shows where each drug class acts: entry inhibitors block the first step, nucleoside and non-nucleoside reverse transcriptase inhibitors block RNA-to-DNA conversion, integrase inhibitors block insertion, and protease inhibitors block maturation.

Worked example Beginner

Consider two 25-year-olds diagnosed with HIV, one in 1995 and one in 2024.

1995 diagnosis. No effective therapy exists yet. The viral load settles at a set-point around 100,000 copies per milliliter. CD4 T cell counts decline by about 50 cells per microliter per year. Within a median of 10 years, CD4 falls below 200, opportunistic infections appear, and the patient progresses to AIDS. Life expectancy from diagnosis is about 12 years.

2024 diagnosis. Within weeks of diagnosis, the patient starts a single-pill regimen: an integrase inhibitor (dolutegravir or bictegravir) plus two nucleoside reverse transcriptase inhibitors (tenofovir and emtricitabine). The viral load drops below the detection limit of 50 copies per milliliter within 3 to 6 months. CD4 counts recover toward the normal range of 500 to 1,500 cells per microliter. Life expectancy on therapy is about 78 years, near normal.

Step 1: viral suppression removes the selective pressure that drives CD4 depletion, so the immune system partially rebuilds itself.

Step 2: once viral load is undetectable in blood and genital fluids, sexual transmission does not occur. This is the formula "Undetectable = Untransmittable," or U=U.

What this tells us: HIV is now a chronic manageable condition in countries with drug access, but the reservoir of latent virus inside resting memory T cells persists for decades and rebounds within weeks if therapy is stopped.

Check your understanding Beginner

Formal definition Intermediate+

Definition (HIV and AIDS). Human immunodeficiency virus (HIV) is an enveloped positive-sense single-stranded RNA retrovirus of the genus Lentivirus that infects CD4+ T lymphocytes, monocytes, macrophages, and dendritic cells. Acquired immunodeficiency syndrome (AIDS) is the clinical syndrome defined by a CD4+ T cell count below 200 cells per microliter, or by the presence of an AIDS-defining opportunistic infection or malignancy (Pneumocystis jirovecii pneumonia, Kaposi sarcoma, cerebral toxoplasmosis, disseminated Mycobacterium avium complex, HIV wasting syndrome, and others).

Definition (retroviral lifecycle). The HIV replication cycle proceeds in seven steps, each druggable:

1. Attachment — the viral envelope glycoprotein gp120 binds the CD4 receptor on the host T cell, then engages a co-receptor (CCR5 in early infection, CXCR4 in about half of late infections).
2. Fusion — gp41 undergoes a conformational change that merges the viral envelope with the host membrane, releasing the capsid into the cytoplasm.
3. Reverse transcription — viral reverse transcriptase (RT), an RNA-dependent DNA polymerase without 3' to 5' proof-reading exonuclease activity, synthesizes double-stranded DNA from the viral RNA genome. Per-base per-replication mutation rate $\mu \approx 3.4\times 10^{-5}$.
4. Integration — viral integrase cleaves the host genome and ligates the viral DNA into a host chromosome, forming the provirus. Once integrated, the provirus is indistinguishable from a host gene and persists for the lifetime of the cell.
5. Transcription and translation — host RNA polymerase II transcribes the provirus; viral mRNAs are translated into the Gag, Gag-Pol, and Env polyproteins.
6. Assembly — Gag and Gag-Pol polyproteins accumulate at the plasma membrane with two copies of viral RNA; new virions bud outward.
7. Maturation — viral protease cleaves the Gag-Pol polyprotein into the mature structural proteins (matrix, capsid, nucleocapsid) and enzymes (protease, RT, integrase). Only after maturation is the virion infectious.

Definition (clinical stages). Untreated HIV infection progresses in three stages. Acute infection (weeks 2 to 4 after exposure) presents with high viremia (peak plasma viral load around $10^7$ copies/mL), flu-like illness, and a sweeping but transient CD8 T cell response. Clinical latency (chronic infection) lasts a median of about 10 years; viral load settles to a set-point whose height predicts the rate of progression, and CD4 counts decline at an average rate of about 50 cells per microliter per year. AIDS is the late clinical syndrome with CD4 below 200, opportunistic infections, and untreated median survival of 12 to 18 months.

Definition (antiretroviral drug classes). Modern antiretroviral therapy (ART, formerly HAART) combines drugs from at least two classes against two distinct viral targets. The five classes and their targets:

• Nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) — chain terminators that compete with natural dNTPs at the RT active site (tenofovir, emtricitabine, abacavir, lamivudine, zidovudine).
• Non-nucleoside reverse transcriptase inhibitors (NNRTIs) — allosteric RT inhibitors (efavirenz, rilpivirine, nevirapine, doravirine).
• Integrase strand transfer inhibitors (INSTIs) — block the strand-transfer step of integration (dolutegravir, bictegravir, raltegravir, cabotegravir).
• Protease inhibitors (PIs) — block Gag-Pol cleavage, preventing maturation (darunavir, atazanavir, lopinavir/ritonavir).
• Entry and fusion inhibitors — maraviroc blocks CCR5, enfuvirtide blocks gp41.

The 2024 WHO first-line regimen is an INSTI (dolutegravir or bictegravir) plus two NRTIs (tenofovir plus emtricitabine, or abacavir plus lamivudine). Long-acting injectable cabotegravir plus rilpivirine, dosed every two months, is an alternative for patients with sustained virologic suppression.

Definition (prevention modalities). PrEP (pre-exposure prophylaxis) is daily oral tenofovir and emtricitabine taken by HIV-negative people at high risk, reducing sexual acquisition by about 99 percent when adherence is high. PEP (post-exposure prophylaxis) is a 28-day three-drug regimen started within 72 hours of a possible exposure. Treatment as prevention (TasP) means that an HIV-positive partner on suppressive ART does not transmit the virus sexually, the U=U finding.

Counterexamples to common slips Intermediate+

• "HIV and AIDS are the same thing." No. HIV is the virus; AIDS is the late clinical syndrome. A person can be HIV-positive for a decade without having AIDS, and effective ART prevents AIDS from ever developing.
• "HIV always progresses to AIDS without treatment." Median time is about 10 years, but a small fraction (about 5 percent) are long-term non-progressors who maintain normal CD4 counts for more than 15 years without therapy, and a smaller group of elite controllers keep viral load below 50 copies/mL without ART.
• "ART cures HIV." It does not. The latent reservoir in resting memory CD4 T cells persists with a half-life of about 44 months. Stopping ART triggers viral rebound within 2 to 4 weeks.
• "HIV is highly contagious." Per-act transmission probability is about 0.1 percent for receptive vaginal sex and about 1.4 percent for receptive anal sex, far lower than measles (about 90 percent) or influenza. Transmission risk rises sharply with viral load, which is why early ART and U=U are the central prevention tools.
• "PrEP causes resistance." Rare when adherence is good. Resistance arises primarily when PrEP is started during undiagnosed acute HIV infection, which is why HIV testing before PrEP initiation is mandatory.
• "You can tell someone has HIV by looking." Acute infection resembles mononucleosis; clinical latency is asymptomatic; only late AIDS has visible stigmata. Testing, not appearance, diagnoses HIV.

Key theorem with proof Intermediate+

Theorem (Ho-Shaw-Perelson viral dynamics, 1995 to 1996). Let $V(t)$ be the concentration of free plasma virions and $T^{\ast }(t)$ the concentration of productively infected cells. Assume the donor-recipient model

$$\frac{dV}{dt}=P+\pi T^{\ast }-cV,\frac{d{T}^{\ast }}{dt}=kVT-\delta T^{\ast },$$

where $P$ is the influx from the long-lived reservoir, $\pi$ is the per-cell virion production rate, $c$ is the virion clearance rate, $k$ is the infection rate, $T$ is the uninfected CD4 count, and $\delta$ is the death rate of productively infected cells. Then at untreated steady state with plasma viral load $V_{ss}\approx 10^5$ copies/mL and virion half-life $\ln 2/c\approx 0.5$ days, the total body production rate is approximately $10^{10}$ virions per day, every possible single-point mutation arises at least $10^4$ times per day somewhere in the patient, and monotherapy therefore fails within weeks while triple therapy is durable.

Proof.

Step 1 — steady state. At equilibrium $dV/dt=d{T}^{\ast }/dt=0$, so $c{V}_{ss}=P+\pi T_{ss}^{\ast }$ and $T_{ss}^{\ast }=k{V}_{ss}T/\delta$. Substituting gives $c{V}_{ss}=P+(\pi kT/\delta )V_{ss}$, hence

$$V_{ss}=\frac{P}{c-\pi kT/\delta }.$$

Step 2 — parameter identification from post-treatment decay. Initiation of a protease inhibitor renders newly produced virions non-infectious without immediately changing $V$. Under therapy the production term $\pi T^{\ast }$ is unchanged for a few days but the new virions cannot replenish $T^{\ast }$, so $T^{\ast }$ decays exponentially as $T^{\ast }(t)=T_0^{\ast }{e}^{-\delta t}$. Substituting back into the $V$ equation with the source term now removed:

$$V(t)=V_0{e}^{-ct}+\frac{\pi T_0^{\ast }}{c-\delta }(e^{-\delta t}-e^{-ct}).$$

Fitting this biphasic curve to patient data (Perelson 1996) gives the fast first-phase slope $c\approx 3$ day${}^{-1}$ (virion half-life about 0.5 days) and the slow second-phase slope $\delta \approx 1$ day${}^{-1}$ (productively infected cell half-life about 1 day).

Step 3 — daily production. The total free-virion population at any moment is approximately $N_V\approx 10^5\text{ copies/mL}\times 3\times 10^3\text{ mL plasma}\approx 3\times 10^8$ virions, of order $10^9$ when extracellular fluid equilibration is included. Steady-state turnover replaces the entire pool every $1/c$ days, so daily production is $P_{\text{tot}}=c{N}_V\approx 3\times 10^9\approx 10^{10}$ virions per day (Ho 1995, Perelson 1996).

Step 4 — mutational coverage. With reverse transcriptase mutation rate $\mu \approx 3.4\times 10^{-5}$ per base per cycle and genome length $L\approx 9.7\times 10^3$ bases, the expected number of new point mutants produced per day at every site is

$$P_{\text{tot}}\cdot \mu \cdot L\approx 10^{10}\cdot 3.4\times 10^{-5}\cdot 9.7\times 10^3\approx 3.3\times 10^9\text{ mutations/day},$$

or about $3.4\times 10^5$ copies of every specific single-point mutant per day. The entire single-mutation landscape is explored many times per day.

Step 5 — therapeutic implication. Under monotherapy, resistance mutants at any single site already exist in the reservoir before treatment starts and sweep within weeks. For an $n$-drug regimen with independent resistance sites at per-cycle frequency $\mu$, the probability that a single replication event produces a fully resistant virion is ${\mu }^n$. For $n=1$ this is about $10^{-5}$ per cycle, easily cleared against $10^{10}$ daily cycles. For $n=3$ it is about $10^{-15}$ per cycle, requiring more than $10^5$ years to expect a single triple mutant. Triple therapy suppresses the virus indefinitely provided adherence maintains drug levels above the resistance-selection threshold. $\square$

Bridge. This result builds toward the modern first-line INSTI-plus-two-NRTI regimen standardized by WHO, and appears again in the design of the HPTN 052 PrEP trial 35.02.01, where a single drug pair suffices because transmission is a bottleneck event rather than a within-host evolutionary contest. The foundational reason combination therapy works is that resistance is a per-replication lottery, and this is exactly the evolutionary logic that identifies the latent reservoir in resting memory CD4 T cells as the dual obstacle to a cure — the same calculation that proves suppression while on therapy also proves rebound within weeks of stopping.

Exercises Intermediate+

Advanced results Master

Result 1 (CDC, 5 June 1981). Between October 1980 and May 1981, five previously healthy young men in Los Angeles were diagnosed with Pneumocystis carinii pneumonia, an illness previously seen only in severely immunocompromised patients. The CDC MMWR report on 5 June 1981 ^{[CDCMMWR1981]} is the first published description of the epidemic. Within months, additional clusters of Kaposi sarcoma in previously healthy gay men, and of opportunistic infections in people with hemophilia and in injection-drug users, established that the underlying cause was transmissible and blood-borne.

Result 2 (Montagnier 1983 — viral isolation). Barre-Sinoussi, Chermann, and Montagnier at the Institut Pasteur isolated a novel retrovirus, which they called LAV (lymphadenopathy-associated virus), from the lymph node of a patient with persistent lymphadenopathy ^{[Montagnier1983]}. The 1983 paper in Science 220:868 demonstrated that LAV killed CD4 T cells in culture, had a magnesium-dependent reverse transcriptase, and was morphologically a lentivirus. Montagnier shared the 2008 Nobel Prize in Physiology or Medicine for this work.

Result 3 (Gallo 1984 — definitive etiology). Gallo and colleagues at the National Cancer Institute isolated and characterized HTLV-III (a name later reconciled with LAV as HIV-1), establishing it as the etiologic agent of AIDS through four independent lines of evidence: presence in patients, isolation in culture, serologic response, and prospective validation ^[Gallo1984]. The 1984 Science 224:497 series also yielded the reagents for the first commercial HIV antibody test, enabling blood-supply screening.

Result 4 (Ho-Shaw 1995 — viral dynamics). Ho, Shaw, and colleagues measured plasma viral load every few days after initiation of protease inhibitor therapy and observed exponential decay with two phases ^[HoShaw1995]. The fast phase yielded virion half-life of about 0.5 days and the slow phase yielded productively-infected-cell half-life of about 1 day. Combined with the steady-state viral load, this implied a daily production of about $10^{10}$ virions per day and overturned the prevailing picture of clinical latency as a dormant phase — latency is in fact a dynamic equilibrium between massive viral production and equally massive immune clearance, with a complete turnover of the viral population every few days.

Result 5 (Perelson 1996 — mathematical modeling). Perelson, Neumann, Ho, and colleagues formulated the biphasic decay model

$$V(t)=V_0{e}^{-ct}+\frac{\pi T_0^{\ast }}{c-\delta }(e^{-\delta t}-e^{-ct})$$

and fit it to patient data, yielding $c\approx 3$ day${}^{-1}$, $\delta \approx 1$ day${}^{-1}$, and an estimated viral generation time of about 2 days ^{[Perelson1996]}. The model also produced the estimate that the minimum duration of infection in a typical patient was 7 to 10 years, matching the observed clinical latency. Perelson 1996 is the canonical reference for HIV within-host dynamics.

Result 6 (Palella 1998 — HAART survival benefit). The HIV Outpatient Study (HOPS) cohort, reported by Palella and colleagues ^{[Palella1998]}, compared mortality and opportunistic-infection incidence before and after the introduction of HAART in 1996. The result was an age-adjusted mortality decline of about 70 percent and an opportunistic-infection incidence decline of about 75 percent within two years of triple-therapy becoming standard. This is the empirical anchor for the claim that HAART transformed HIV from a death sentence into a chronic manageable condition.

Result 7 (Cohen 2011, HPTN 052 — treatment as prevention). The HPTN 052 trial randomized 1,763 serodiscordant couples to immediate versus delayed ART for the HIV-positive partner ^[Cohen2011]. The trial was stopped 4 years early by the Data and Safety Monitoring Board because the immediate-ART arm showed a 96 percent reduction in genetically linked HIV transmission to the negative partner. HPTN 052 is the empirical foundation of the U=U consensus and of WHO's 2015 "treat all" recommendation.

Result 8 (Hütter 2009, Berlin patient — proof of concept for cure). Timothy Brown received two allogeneic stem-cell transplants from a CCR5-Δ32-homozygous donor to treat acute myeloid leukemia; ART was interrupted and the viral load never rebounded in 12 years of follow-up ^[Hutter2009]. Subsequent confirmed cures (the "London patient" Gupta 2019, the "New York patient" Hsu 2022, the "City of Hope" patient, the "Düsseldorf" patient) all used CCR5-Δ32 donors, confirming that abrogating CCR5 entry is the load-bearing mechanism. These cases are not scalable — stem-cell transplantation has substantial mortality — but they proved that a sterile cure is possible and renewed interest in CCR5-targeted gene therapy.

Synthesis. The 40-year arc from the 1981 CDC report to modern injectable cabotegravir plus rilpivirine is the foundational reason that HIV/AIDS is the canonical case study in modern biomedicine. The central insight — that viral dynamics, not viral dormancy, defines clinical latency — built toward the Ho-Shaw-Perelson dynamics that explained why triple therapy works, and the same mathematical framework appears again in the HPTN 052 result, where treatment-as-prevention identifies within-host suppression with population-level transmission block. Putting these together with the reservoir characterization (half-life about 44 months) and the CCR5-Δ32 cures, the bridge is between the molecular biology of entry and the population biology of pandemic spread, and the pattern generalises through every major antiviral strategy of the past three decades, from HCV direct-acting antivirals to COVID-19 antivirals.

Full proof set Master

Proposition 1 (latency despite high turnover). Under the Ho-Shaw-Perelson model, the half-life of productively infected cells (about 1 day) is inconsistent with the multi-year duration of clinical latency unless the infected-cell pool is continuously replenished from a longer-lived compartment.

Proof. Define $N(t)$ as the total number of productively infected cells at time $t$. Under ART that blocks new infection, $dN/dt=-\delta N$ with $\delta \approx 1$ day${}^{-1}$, so $N(t)=N_0{e}^{-t}$ and after 30 days $N=N_0{e}^{-30}\approx 10^{-13}{N}_0$, far below one cell. Observed viral load after 14 days of ART is well above this prediction (typically near the detection limit, not zero), which forces the conclusion that a second, longer-lived compartment — the latently infected resting memory CD4 T cell pool, half-life about 44 months — continues to reactivate and replenish productively infected cells. Latency is therefore not viral dormancy but a slow reseeding dynamic. $\square$

Proposition 2 (reservoir extinction under ART is impractical). With latent-reservoir half-life $t_{1/2}\approx 44$ months and initial reservoir $N_0\approx 10^6$ cells, the expected time to reduce the reservoir below one cell on suppressive ART alone exceeds 50 years.

Proof. Reservoir size after $n$ half-lives is $N_0/2^n$. Setting $N_0/2^n<1$ gives $n>{\log }_2{N}_0={\log }_{2}10^6\approx 20$ half-lives, or about $20\times 44=880$ months, approximately 73 years. This calculation is the original basis of the Pinkevych 2015 estimate and is the motivation for "shock and kill" strategies (latent-reactivating agents such as HDAC inhibitors combined with immune clearance) and for CCR5 gene-editing approaches that protect new T cells from infection regardless of whether the reservoir is cleared. $\square$

Proposition 3 (combination therapy durability). Under a regimen of $n$ drugs with independent resistance sites at per-cycle mutation rate $\mu =10^{-5}$, and daily production $P_{\text{tot}}=10^{10}$ virions, the expected number of fully resistant virions produced per day is $P_{\text{tot}}{\mu }^n$. For $n=3$ this is below the per-day stochastic threshold for establishment, but for $n=2$ resistance emerges within months.

Proof. For $n=1$: $P_{\text{tot}}\mu =10^5$ resistant virions per day; resistance sweeps in days to weeks. For $n=2$: $P_{\text{tot}}{\mu }^2=1$ resistant virion per day, with a 50 percent chance that at least one virion establishes a productive resistant lineage per month; resistance emerges within months. For $n=3$: $P_{\text{tot}}{\mu }^3=10^{-5}$ per day, expected waiting time 270 years. The threshold for durable suppression lies between $n=2$ and $n=3$, which is why three-drug regimens are the minimum standard of care. $\square$

Connections Master

• Infectious disease, immunity, and vaccines 35.02.01. This unit supplies the depth slice for one of the canonical case studies named in the chapter survey, building on the survey's overview of pathogens, immune system architecture, and vaccination. The SIR compartmental model and basic reproduction number $R_0$ introduced there generalize here to the within-host Ho-Shaw-Perelson dynamics, and the survey's account of how vaccines exploit immunological memory connects to the HIV vaccine-development challenge, where the virus's rapid mutation and glycan shield have so far defeated the classical vaccine paradigm.

• Innate immunity at the molecular level 17.10.01. The complement system, restriction factors, and innate-sensing pathways introduced there are the substrate HIV must evade. HIV's accessory proteins Nef, Vif, Vpu, and Vpr each target a specific innate restriction factor: Nef downregulates MHC-I (evading CD8 T cell recognition) and CD4 (preventing superinfection), Vif degrades APOBEC3G (which would otherwise hypermutate the viral genome during reverse transcription), and Vpu degrades tetherin/BST-2 (which would otherwise retain budding virions at the cell surface). The molecular arms race described in 17.10.01 is the substrate of HIV immune evasion.

• Mutation and repair 17.06.01. This peer established the principle that polymerase fidelity is determined by 3' to 5' proof-reading exonuclease activity. HIV reverse transcriptase lacks this activity, which is the molecular source of the per-base per-cycle mutation rate $\mu \approx 3.4\times 10^{-5}$ that drives the viral dynamics calculation in the Key theorem. The DNA repair pathway deficiencies catalogued in 17.06.01 are also why host cells rarely proof-read integrated proviruses back to wild-type, making integration a one-way molecular ratchet.

Historical & philosophical context Master

The epidemic was first recognized in a five-case CDC report on 5 June 1981 ^{[CDCMMWR1981]}; the viral etiology was established by Montagnier's group at the Institut Pasteur in 1983 ^{[Montagnier1983]} and definitively characterized by Gallo's group at the National Cancer Institute in 1984 ^[Gallo1984]. The 1986 International Committee on Taxonomy of Viruses standardized the name HIV. Jonathan Mann founded the WHO Global Programme on AIDS in 1987, framing the epidemic as a global development crisis rather than a clinical curiosity of wealthy countries.

The 1995 Ho-Shaw viral-dynamics papers ^[HoShaw1995], formalized by Perelson's mathematical modeling in 1996 ^{[Perelson1996]}, overturned the static picture of clinical latency and showed that the virus replicated continuously and rapidly throughout the asymptomatic years. This single insight motivated combination therapy: if the virus produced $10^{10}$ virions per day and explored the entire single-mutation landscape daily, only triple therapy could suppress resistance evolution. The Palella 1998 HOPS cohort ^{[Palella1998]} confirmed that this calculation was correct, with mortality falling about 70 percent within two years. Cohen's HPTN 052 trial in 2011 ^[Cohen2011] closed the loop between within-host dynamics and population-level transmission, proving that treatment is prevention.

The CCR5-Δ32 stem-cell transplantation cure of Timothy Brown, reported by Hütter in 2009 ^[Hutter2009], remains the most consequential single case report in modern HIV research. It proved that HIV is, in principle, curable and reframed the field from lifetime ART maintenance toward cure research. It also demonstrated the load-bearing role of the CCR5 co-receptor: a patient whose new CD4 T cells lack functional CCR5 is refractory to R5-tropic HIV, which has driven the development of CCR5 gene-editing approaches (CRISPR-CCR5, zinc-finger nucleases) and of the long-acting injectable CCR5-blocking strategies now in clinical trials.

Bibliography Master

@article{CDCMMWR1981,
  author = {{Centers for Disease Control}},
  title  = {{Pneumocystis pneumonia -- Los Angeles}},
  journal = {Morbidity and Mortality Weekly Report},
  volume = {30},
  number = {21},
  pages  = {250--252},
  year   = {1981},
}

@article{Montagnier1983,
  author = {Barre-Sinoussi, F. and Chermann, J. C. and Rey, F. and
            Nugeyre, M. T. and Chamaret, S. and Gruest, J. and
            Dauguet, C. and Axler-Blin, C. and Vezinet-Brun, F. and
            Rouzioux, C. and Rozenbaum, W. and Montagnier, L.},
  title  = {Isolation of a {T}-lymphotropic retrovirus from a patient
            at risk for acquired immune deficiency syndrome ({AIDS})},
  journal = {Science},
  volume = {220},
  number = {4599},
  pages  = {868--871},
  year   = {1983},
  doi    = {10.1126/science.6189183},
}

@article{Gallo1984,
  author = {Gallo, R. C. and Salahuddin, S. Z. and Popovic, M. and
            Shearer, G. M. and Kaplan, M. and Haynes, B. F. and
            Palker, T. J. and Redfield, R. and Oleske, J. and Safai, B. and
            White, G. and Foster, P. and Markham, P. D.},
  title  = {Frequent detection and isolation of cytopathic retroviruses
            ({HTLV-III}) from patients with {AIDS} and at risk for {AIDS}},
  journal = {Science},
  volume = {224},
  number = {4648},
  pages  = {500--503},
  year   = {1984},
  doi    = {10.1126/science.6200936},
}

@article{HoShaw1995,
  author = {Ho, D. D. and Neumann, A. U. and Perelson, A. S. and
            Chen, W. and Leonard, J. M. and Markowitz, M.},
  title  = {Rapid turnover of plasma virions and {CD4} lymphocytes
            in {HIV-1} infection},
  journal = {Nature},
  volume = {373},
  number = {6510},
  pages  = {123--126},
  year   = {1995},
  doi    = {10.1038/373123a0},
}

@article{Perelson1996,
  author = {Perelson, A. S. and Neumann, A. U. and Markowitz, M. and
            Leonard, J. M. and Ho, D. D.},
  title  = {{HIV-1} dynamics in vivo: virion clearance rate, infected
            cell life-span, and viral generation time},
  journal = {Science},
  volume = {271},
  number = {5255},
  pages  = {1582--1586},
  year   = {1996},
  doi    = {10.1126/science.271.5255.1582},
}

@article{Palella1998,
  author = {Palella, F. J. and Delaney, K. M. and Moorman, A. C. and
            Loveless, M. O. and Fuhrer, J. and Satten, G. A. and
            Aschman, D. J. and Holmberg, S. D.},
  title  = {Declining morbidity and mortality among patients with
            advanced human immunodeficiency virus infection},
  journal = {New England Journal of Medicine},
  volume = {338},
  number = {13},
  pages  = {853--860},
  year   = {1998},
  doi    = {10.1056/NEJM199803263381301},
}

@article{Cohen2011,
  author = {Cohen, M. S. and Chen, Y. Q. and McCauley, M. and
            Gamble, T. and Hosseinipour, M. C. and Kumarasamy, N. and
            Hakim, J. G. and Kumwenda, J. and Grinsztejn, B. and
            Pilotto, J. H. and Godbole, S. V. and Mehendale, S. and
            Chariyalertsak, S. and Santos, B. R. and Mayer, K. H. and
            Hoffman, I. F. and Eshleman, S. H. and Piwowar-Manning, E. and
            Wang, L. and Makhema, J. and Mills, L. A. and de Bruyn, G. and
            Sanne, I. and Eron, J. and Gallant, J. and Havlir, D. and
            Swindells, S. and Ribaudo, H. and Elharrar, V. and
            Burns, D. and Taha, T. E. and Nielsen-Saines, K. and
            Celentano, D. and Essex, M. and Fleming, T. R.},
  title  = {Prevention of {HIV-1} infection with early antiretroviral therapy},
  journal = {New England Journal of Medicine},
  volume = {365},
  number = {6},
  pages  = {493--505},
  year   = {2011},
  doi    = {10.1056/NEJMoa1105243},
}

@article{Hutter2009,
  author = {H{\"u}tter, G. and Nowak, D. and Mossner, M. and
            Ganepola, S. and M{\"u}ssig, A. and Allers, K. and
            Schneider, T. and Hofmann, J. and K{\"u}cherer, C. and
            Blau, O. and Blau, I. W. and Hofmann, W. K. and Thiel, E.},
  title  = {Long-term control of {HIV} by {CCR5} {Delta32}/{Delta32}
            stem-cell transplantation},
  journal = {New England Journal of Medicine},
  volume = {360},
  number = {7},
  pages  = {692--698},
  year   = {2009},
  doi    = {10.1056/NEJMoa0802905},
}

@article{DeeksLewin2013,
  author = {Deeks, S. G. and Lewin, S. R. and Havlir, D. V.},
  title  = {The end of {AIDS}: {HIV} infection as a chronic disease},
  journal = {Lancet},
  volume = {382},
  number = {9903},
  pages  = {1525--1533},
  year   = {2013},
  doi    = {10.1016/S0140-6736(13)61809-7},
}