Sunday, January 29, 2006

Quick vaccine cycle follow-up

As promised, I have now had a chance to read the paper by Rao et al. that has just appeared online in the Journal of Virology (Feb. 2006, pp. 1959 - 1964; thanks to Selise for .pdf). It describes the new quick cycle bird flu vaccine from the University of Pittsburgh group reported on the wireservices. After some back and forth on my part given the vagueness of the descriptions, it turns out my original impression was correct. This is what is usually called a DNA vaccine using a common cold virus as a vector.

Here are some of the details and explanations. The usual seasonal flu as well as the current experimental NIH H5N1 vaccines are inactivated flu viruses that carry the surface proteins of the H5N1 virus but the inner proteins of a more benign virus. They are grown in eggs, inactivated and then injected in two doses into arm muscles. The immune system "sees" the viral surface protein and makes protective antibodies against it. In the trials so far it took a lot of viral protein to produce a response (twelve times the usual amount found in the seasonal flu vaccine) and attempts to boost this with additives have been relatively disappointing. Given these numbers, there wouldn't be enough production capacity to satisfy demand globally for a long time, if ever. Egg supply would also be a problem. And to top it off, it takes 6 to 8 months to make the vaccine once a candidate pandemic strain has shown itself. We could go through several pandemic waves before a vaccine was ready.

So quicker, more efficient ways to produce a vaccine have been under intensive investigation. One appraoch would be to grow the vaccine in cell cultures rather than chicken eggs. This could produce capacity and reduce production cycles. And another has been to use DNA vaccines, which is what the Pittsburgh work is about.

Many of you know that the genetic material in the flu virus is RNA, not the DNA of human cells. But either one codes for proteins, so the tactic is to get the DNA of the viral surface protein (or some other component of the virus) into a human cell and let the cell make just that protein (not the whole virus), thus eliciting (it is hoped) the kind of immunity one gets from the inactivated virus. DNA can be made much faster and cheaper than growing viruses. But there are a couple of technical obstacles. The first is the matter of getting the DNA into the cell in the first place. Not such an easy task it turns out.

You might try to physically shoot it into the cell, and this is the basis of one approach being employed by PowerMed, Ltd. in the UK. They coat tiny particles with the DNA genes and shoot it into the skin with a superfast injector device (no needles). There are many immune cells in the skin and PowerMed claims good results. But they don't have FDA approval as yet and it sounds like there is still development work to do. You could also try to inject "naked" DNA (in a plasmid loop) into the skin, and this approach is being used for some other infectious diseases.

But the most efficient carrier of genetic material into a cell remains a virus, which needs to get at the cell's genetic machinery so it can make new copies of itself. So there has been a lot of work using viruses as vectors to get DNA into human cells, not just for vaccines but for gene therapy (correcting genetic defects). A number of different viruses have been used for this purpose (baculovirus, vaccinia virus, Adeno Associated Virus) but adenovirus type 5 has been one of the more successful, and it is this that the Pittsburgh group used as a vector (or vehicle) for DNA segments that code for the HA surface protein (the H5 part of H5N1).

The AD5 virus was modified so that it didn't cause disease and in fact couldn't replicate or be integrated into human genes. It also has a finite lifetime, as the body makes antibodies also to AD5, neutralizing it after a relatively short time. It is during this period that the body's immune response must go into action. It was not at all obvious this would work, although a human trial with AD5 incorporating an H1N1 gene was shown to be safe and to elicit a good immune response (see van Kampen et al.)

The Pittsburgh group followed the same approach, splicing into AD5 segments of the HA surface protein from two different H5N1 viruses, one a Vietnam isolate from 2004, the other a Hong Kong isolate from 1997. They obtained genetic sequences from CDC for either the whole HA proteins or two subsections of it, the HA1 segment (the far end of the protein which elicits an antibody response), and the HA2 (closer in) segment, important for entry and then uncoating of the virus within the cell. Thus they put in an adenovirus vector DNA sequences from either of two H5N1 HA proteins, each either full length or in pieces (HA1 or HA2), six combinations in all. They report that from the day the sequence information arrived to the day the vector AD5s were made was just 36 days, thus demonstrating this could be accomplished on a very short cycle.

They then injected the AD5 modified viruses into mice on day 0 and again on day 14 (a booster shot). Antibody response to the full length HAs was quite rapid and the authors believe it might be possible to get by without the booster for full length HA. In addition, there was also some cross-strain immunity, i.e., in the full length versions there was the strongest response to the strain from which the HA sequence was taken but also a measurable response to the other strain as well (i.e., the Vietnamese versus the Hong Kong strains). Response after boosting was also good for the HA1 versions, but not the HA2s, perhaps not surprising because the immune system recognizes HA through parts of the protein that are at the HA1 end.

Also of interest was the fact that another arm of the immune system, cell mediated immunity, was also stimulated. The exact function of cell-mediated immunity in prtoecting against influenza is still being worked out, but this response usually precedes the antibody response by several days and is initially relatively non-specific. Interestingly, HA2 also participated in cell mediated immunity and there was evidence of strain cross-reactivity as well.

Finally the Pittsburgh group demonstrated that the animals were truly protected against challenges with amounts of flu virus rapidly 100% lethal to unvaccinated animals. While some of the vaccinated animals did fall ill, all recovered.

All in all this is a promising proof of principle for a new approach to an H5N1 vaccine. It builds on other work and is parallel to several other efforts to do the same thing. Perhaps the most important signal here is that there is a great deal of interesting work being done on new vaccine technology. We note this work has been going on for some time and didn't need Bill Frist to hand Big Pharma the Christmas gift of liability immunity. But that's another subject.

Still, there is a long way to go. It has to be shown that such a vaccine is both safe and effective in humans. Protecting mice is relatively easy. Protecting humans is another matter. There may be a lot of natural immunity to adenovirus 5 that could interfere with the efficacy of the vaccine, which might require a move to another serotype or vector. Not an insurmountable problem but one that takes time. The safety issue isn't trivial when hundreds of millions or even billions might be vaccinated in a pandemic setting. A risk of death of only one in a million, while very favorable in the risk-risk trade-off calculus, might be a significant psychological barrier if in the first weeks of a mass immunization campaign the public sees several dozen deaths related to vaccination. Finally there is the problem of scaling up to produce adequate amounts of vaccine. This will be easier for this kind of vaccine than an inactivated whole virus one, but won't be trivial.

Let's hope we are closer and have enough time to get to where we need to go.