Bad news for stop–start therapy?

An HIV-infected patient who was being treated with anti-retroviral drugs in a 'stop–start' protocol has become infected with a second HIV strain, raising questions about both the treatment strategy and vaccine development.

“This is terrible news,” was the response of one senior HIV immunologist at the Barcelona AIDS conference this July, on hearing the results now published on page 434 of this issue by Altfeld and colleagues1. These authors had studied a patient who was treated with anti-retroviral drugs soon after becoming infected with HIV-1; this therapy achieved good control of his HIV-1 levels. Later, according to the 'supervised treatment interruption' (STI) protocols developed by Walker and colleagues2, treatment was deliberately stopped, and then restarted when virus levels rose. This stop–start procedure was repeated twice more, for a total of three treatment-free periods. Altfeld et al. found that, during the first two of these periods, the patient's immune system controlled replication of the virus well, if only for a few months. This stands in marked contrast to the situation seen in HIV sufferers who receive anti-retroviral drugs only late in infection; here, within two weeks of stopping therapy, virus levels normally rebound to those seen before treatment.

Viral control during STI is thought to occur through enhancement of the immune response to HIV-1, particularly the response of T lymphocytes that kill HIV-infected cells (these lymphocytes are characterized by expression of the surface protein CD8), and possibly that of 'helper' T lymphocytes (which express the alternative surface protein CD4)2. Indeed, the CD8+ T cells from the patient studied by Altfeld et al. showed an excellent response, with some 6% recognizing and proliferating in response to HIV-1 proteins. His CD4+ T cells also proliferated markedly when exposed to HIV proteins in vitro — unusual in HIV-infected people.

So what's the bad news? Things started to go wrong during the third treatment interruption. Virus levels rebounded more rapidly than before, and the response of CD4+ T cells deteriorated. On further investigation, Altfeld et al. found that a new virus variant had emerged towards the end of the second treatment-free period. Although from the same HIV-1 'family' as the virus causing the original infection — namely the B clade, which predominates in the Western world — this new variant was quite different. The authors carried out a detailed study of 16 regions (epitopes) of HIV-1 that are recognized by CD8+ T cells, and found that 7 differed by at least one amino acid between the original and the new virus; the T cells could not detect these new epitopes. The other 9 epitopes were the same, and the responses of CD8+ T cells to them were maintained.

Further analysis of the sequence of the second virus showed that this variant was a distinct strain, probably a new infection (a superinfection) — indeed, the patient reported a recent episode of sexual exposure that was followed by a fever-associated illness. If this is the case, then superinfection took place despite the patient's strong HIV-specific immune response. There is, however, a small chance that the second virus was there from the beginning, possibly hiding in lymph nodes. Although in men the usual pattern is of initial infection with a single virus strain, it is not uncommon for women to be infected with multiple strains from the start — an intriguing example of gender-specific biology in HIV-1 infection3.

But why would a superinfection be such terrible news? It is widely held that CD8+ T cells play the major role in controlling HIV replication during a long-lasting infection. This notion is firmly based on evidence such as the rise in viral count that occurs after CD8-blocking antibodies are infused into macaques infected with SIV, the simian version of the virus4, and viral mutation to escape CD8+ T cells in HIV and SIV infections5. These data, together with findings that sex workers who are frequently exposed to HIV seem to be resistant to HIV infection and generate anti-HIV responses through CD8+ T cells6, have spurred efforts to develop preventive and therapeutic vaccines that stimulate this type of immune response. For instance, vaccinating macaques to induce SIV-specific responses of CD8+ T cells enables the animals to control infection more effectively after challenge with an aggressive SIV7. So the findings of Altfeld et al.1 might be worrying news for vaccine developers — don't they show that HIV infection can occur in the face of substantial activity of CD8+ T cells?

The answer is, not necessarily. First, we do not know the denominator. Recent cases of superinfection have attracted attention8, but there may be many more HIV-infected people who can repel an attempted superinfection. Moreover, studies of primates suggest that superinfection is rare: for instance, macaques infected with an attenuated strain of SIV that induces a strong cellular immune response are resistant to infection with more virulent strains9.

Even if the case studied by Altfeld et al. turns out to be representative, there are reasons to think that the immune activity generated by a healthy person in response to an HIV vaccine will be qualitatively distinct from that of an HIV-infected person — especially one whose immune system is already damaged by more than three years of infection. For instance, immune impairment can be seen from the very earliest stages of HIV-1 infection, particularly among CD4+ T cells; and increases in virus levels after treatment interruption are probably associated with preferential infection of HIV-specific CD4+ T cells10. So the response of CD4+ T cells was probably already weakened, before superinfection, in the patient studied by Altfeld et al. — indeed, the cells' proliferation when exposed to viral proteins in vitro did decline at about the time that the new virus appeared.

CD4+ helper T cells may have direct anti-HIV effects, and might also influence the effectiveness of CD8+ T cells. Although this patient's CD8+ T cells still recognized roughly half of the epitopes in the new virus1, including one that stimulated the strongest immune response, it seems that the cells failed to protect against superinfection. HIV-specific CD8+ T cells in infected people differentiate in an unusual way: they produce little of the membrane-puncturing molecule perforin, and their ability to specifically kill infected cells is consequently reduced11. And the dominant response is not necessarily the most protective, as illustrated by mice infected with lymphocytic choriomeningitis virus12. Finally, the specificity of the responses of CD8+ T cells to SIV and HIV molecules is different in vaccinated macaques13 and highly exposed but apparently HIV-resistant sex workers14 compared with infected animals and people15. So, the superinfection reported by Altfeld et al. might have occurred in the context of a very different immune response from that produced by vaccination of an uninfected person, in terms of T-cell specificity, the competence of CD8+ T cells and the strength of CD4+ helper T cells.

The new findings1 may in fact be worse news for the STI strategy than for vaccine development. The usefulness of this strategy very early in infection remains controversial10. A few patients have achieved remarkable control of HIV-1 and have discontinued therapy entirely, but this seems rare2, and as yet there has been no formal double-blind controlled clinical trial — the bedrock of evidence-based medicine. In one study of STI in a long-lasting HIV-1 infection, viral control was not improved despite the enhancement of HIV-specific immune responses16. The results of Altfeld et al. suggest that superinfection with a second HIV strain during a period off therapy could significantly undermine viral control, so patient commitment to safe sex practices will be an important adjunct to STI. It is not clear whether superinfection is only a risk during treatment interruption: more studies on this are needed.

Altfeld et al.'s work1 is a beautiful illustration of the power of modern techniques to explore the minute details of a virus-specific immune response. But the effort required will preclude studies of large numbers of patients. Certainly, this single-patient analysis raises many questions, but whether the news is bad, neutral or even good remains to be seen. Although causing a brief pause for thought, nothing here should slow or divert efforts to develop an HIV vaccine.


  1. 1

    Altfeld, M. et al. Nature 420, 434–439 (2002).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Rosenberg, E. S. et al. Nature 407, 523–526 (2000).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Long, E. M. et al. Nature Med. 6, 71–75 (2000).

    CAS  Article  Google Scholar 

  4. 4

    Schmitz, J. E. et al. Science 283, 857–860 (1999).

    ADS  CAS  Article  Google Scholar 

  5. 5

    Klenerman, P., Wu, Y. & Phillips, R. Curr. Opin. Microbiol. 5, 408–413 (2002).

    CAS  Article  Google Scholar 

  6. 6

    Rowland-Jones, S. L. et al. J. Clin. Invest. 102, 1758–1765 (1998).

    CAS  Article  Google Scholar 

  7. 7

    Amara, R. R. et al. Science 292, 69–74 (2001).

    ADS  CAS  Article  Google Scholar 

  8. 8

    Jost, S. et al. New Engl. J. Med. 347, 731–736 (2002).

    Article  Google Scholar 

  9. 9

    Daniel, M. D. et al. Science 258, 1938–1941 (1992).

    ADS  CAS  Article  Google Scholar 

  10. 10

    Douek, D. C. et al. Nature 417, 95–98 (2002).

    ADS  CAS  Article  Google Scholar 

  11. 11

    Appay, V. et al. J. Exp. Med. 192, 63–75 (2000).

    CAS  Article  Google Scholar 

  12. 12

    Gallimore, A. et al. J. Exp. Med. 187, 1647–1657 (1998).

    CAS  Article  Google Scholar 

  13. 13

    Ferrari, G. et al. Immunol. Lett. 79, 37–45 (2001).

    CAS  Article  Google Scholar 

  14. 14

    Kaul, R. et al. J. Clin. Invest. 107, 1303–1310 (2001).

    CAS  Article  Google Scholar 

  15. 15

    Vogel, T. U. et al. J. Immunol. 169, 4511–4521 (2002).

    CAS  Article  Google Scholar 

  16. 16

    Oxenius, A. et al. Proc. Natl Acad. Sci. USA 99, 13747–13752 (2002).

    ADS  CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

McMichael, A., Rowland-Jones, S. Bad news for stop–start therapy?. Nature 420, 371–373 (2002).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Sign up for the Nature Briefing newsletter for a daily update on COVID-19 science.
Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing