HIV treatment as prevention

Nice opening session discussing two contrasting strategies for HIV transmission prevention-- Pre-exposure prophylaxis, and immediate HAART.

PrEP provides substantial but incomplete protection against transmission. Rates are estimated from 39-62%. An alternate calculation looks at risk reduction. In these studies, PrEP reduces the three-year transmission risk from 0.04 to 0.01. Oral dosing (tenofovir) is effective, but not nearly as effective as a topical gel. The gel delivers much higher drug concentrations to the area concerned, with fewer chances for side effects. There is no evidence that PrEP selects for drug resistant strains. The gel also stops transmission of herpes. In all cases, adherence is the key factor in the real world. An interesting factor is that PrEP is associated with an increase in pregnancy and also with other STDs. The suggestion is that people feel protected, so do not take other measures.

HAART provides an alternative route. A functional therapy suppresses the virus to non-detectable levels. This means the person no longer transmits (real world check-- as long as the therapy is still working, and the patient is adhering ...). A problem is the common belief that the epidemic is driven by people in the primary phase of infection, many of whom (80%!!) are not aware that they are infected. At particular risk in EU are young homosexuals.

A suggested solution was mandatory (or at least strongly encouraged) universal HIV testing for all 15 year olds, with yearly followup. Anyone with positive results is immmediately put on HAART. Some epidemic models suggest that this would effectively stop the epidemic from spreading, and lead to almost no new infections by 2050. While this would initially cost more, after 10 years it would be substantially cheaper since we would have much fewer cases.

In one Kenya study, 84% of patients with primary infection did not know they were infected.

Finally, Brooks Nichols from Erasmus presented a mean field approximation of the epidemic in a rural African community of approx 150K. She used the model to compare the effectiveness of targeting PrEP to a small number of highly active individuals vs a large number of randomly selected individuals. Not surprisingly, targeting the hubs proved better control for significantly less cost. We are talking about simulating this using my zombie code.

rates of HIV evolution

Prabhav suggested comparing rates of viral evolution within patients to those within a transmission cluster.

We could measure/compare rates of adoptation to treatment to rates of adoptation to a new host.

The confound is that we identify the transmission clusters via genetic distance. It would be easier if we had a known transmission chain.

Alternately, EuResist does have ~6000 cases with two samples taken under one treatment. We could compare rates here with rates after a treatment change.

Prabhakaran (2011) HIV haplotype inference using a constraint-based DPM

This paper extends SHORAH to better infer global haplotypes.

The work is a collaboration between Niko Beerenwinkel's group and Volker Roth's group. Roth is at the the U Basel computer science department. Other authors are Melanie Rey, Osvaldo Zagordi, Huldrych Günthard, and Karin Metzner.

the paper really isn't clear on how they move from local to global reconstruction.

They introduce the following soft constraint. If two reads are assigned to different local haplotypes (i.e. differ in their local window) then they must be assinged to different haplotypes at the global level. Such do-not-link constraints are incorporated into the prior distribution of the subequent class assignments