Re: Study suggests bird flu must be treated within 48 hours
de Jong's "hit it hard and early" paper
Category: Bird flu
Posted on: September 12, 2006 7:13 AM, by revere
http://www.scienceblogs.com/effectmeasure/
Over the weekend Menno de Jong and his many collaborators published a detailed clinical study of 18 H5N1 patients diagnosed and treated in Vietnam in 2004 - 2005 (Nature Medicine, subscription only). Thirteen of the cases died and five survived. Like many papers that have received media attention, it has important and interesting findings but has also been over interpreted, or at least its findings have been generalized too much. In this case, it is the fault of the authors.
de Jong et al. did virological and immunological studies in 18 bird flu cases and 8 other cases with the more usual human influenza subtype infections (5 were H3N2 and 3 were H1N1 cases). The two groups differed both in severity of disease (none of the non-H5N1 flu cases died) and in the timing of their presentation to the hospital. The authors speculate the non-H5N1 cases presented later because they were from more outlying areas. This is plausible as a less severe flu case is more likely to be hospitalized closer to home than a very severe one. But the difference is more than just average time of presentation. Examination of Supplementary Figure 1 in the online supporting material shows that all of the non-H5N1 cases presented before all of the H5N1 cases (with the exception of the two earliest H5N1 cases who presented on day four of illness, that same day as the single latest presenting non-H5N1 case).
One interesting finding was a non-finding: genetic differences did not show any signature changes characteristic of increased virulence in either the human cases of avian influenza or in the fatal cases, in particular. Mutations associated with PB2 (specificaly E627K) did occur, but in only half of the viruses isolated, and equally in the fatal and non-fatal cases. Some other mutations in the polymerase complex associated with mammalian adaption were seen, reinforcing the probable importance of viral genetic replication elements in determining host response. The same general finding was reported recently in an analysis by Chen et al.
Previously there has been much discussion of mutations in the HA protein, which has sites on it that bind to cellular docking sites ("receptors") thought to be specific to birds versus humans. The last set of papers to receive media attention even stated that the reason the virus was not easily transmitted from birds to humans and between humans was that the cells with avian virus receptors in humans were so deep in the lung the virus had a hard time getting there and getting out again. We have previously pointed out that there is good evidence that receptors for bird-associated HA (α2,3SA) exist in the upper respiratory tract of humans and perhaps other tissues and the de Jong paper seems to bear this out, with high viral loads in the nose, throat and trachea, all parts of the upper tract.
In the non-H5N1 infections, viral loads in the nose and throat were about equal, but in the H5N1 cases viral loads in both nose and throat were higher than in non-H5N1 cases and higher in the throat than the nose of H5N1 cases. Remember, however, that the H5N1 cases were all later in the course of their illness than the non-H5N1 cases. The point here is that with these data it isn't possible to know if the higher viral loads are due to increased efficiency of replication or because a host factor like decreased immunity allowed the infection to go on longer, or (less likely), that if the non-H5N1 viral loads had been measured at the same duration of illness, they too would have been higher. The difference in clinical course argues against this last, possibility, in our view.
Viral genetic material was also found in rectal swabs and in the blood of H5N1 cases. There were no rectal swabs available for the non-H5N1 cases, nor was viral isolation attempted for blood in these instances. No viral genetic material was found in the blood of non-H5N1 cases. These data are one more indication that influenza virus may infect tissues outside the respiratory tract and that H5N1 may be more likely to do so. We know very little about the tissues that H5N1 can or does infect and the question of infection of the intestinal tract, particularly, remains open and of vital consequence.
This paper also presents more data on the levels of cytokines in H5N1 and non-H5N1 cases. Cytokines are local chemical signals the immune system uses to synchronize the delicate choreography of its many moving parts. Some signals call for inflammatory cells to come to an area of infection to fight microbial invaders while other signals tell the system to stop sending reinforcements. If the signalling system gets out of kilter, there is danger of a "cytokine storm," a dysregulated signalling system that can have fatal consequences. Scientists believe cytokine dysregulation is one of the possible consequences of H5N1 infection and may contribute to a rapidly fatal outcome in some (but not all) cases. The de Jong paper found significantly higher levels of cytokines in the blood (likely reflective of increased local levels in the lung), especially those cytokines that call for more inflammatory cells. The signal to come help was there in abundance but the signal to stop sending inflammatory cells didn't happen. At the same time the balance of infection fighting white blood cells in the blood was altered, with a marked decrease in white blood cells in the peripheral blood. (This same effect was seen recently in a drug-induced cytokine storm syndrome we posted on here,) Cytokines were increased in both H5N1 and non-H5N1 cases, but more so in H5N1 and still more in the fatal cases versus the non-fatal H5N1 cases. The question remains what causes the cytokine dysregulation, in particular if it is a specific effect of the virus, and if so, how it works. The authors of this study state that viral load is the trigger for increased cytokines and they present data that show good correlations between cytokines and viral load in the throat. However both cytokines and viral load could be caused by a third factor. If the virus is able to disable the cytokine system and with it the cell mediated immune reaction, then this prior effect would produce both higher cytokine levels and higher viral loads.
This affects the last paragraph of the paper, which forms the punchline of much of the media coverage:
Our observations point to a central role for high viral burden in the pathogenesis of human H5N1 disease and suggest that timely suppression of viral replication soul remain the mainstay for treatment of influenza H5N1. This is supported by our previous data demonstrating that successful control of viral replication by antivirals was associated with a good clinical outcome.
The previous data come from his New England Journal article in 2005 on oseltamivir resistance, cited here by de Jong as support for hitting the disease hard and early with antivirals. It does not demonstrate that oseltamivir made a difference, only that in oseltamivir treated cases those whose viral load decreased to below detection survived and those whose viral load didn't, succumbed. It is neither a showing that oseltamivir was responsible for the decrease in viral load nor that the sustained viral load was the cause rather than a sign of a failing patient. Neither paper demonstrates that vigorous treatment with antivirals affects cytokine dysregulation, whose cause remains unknown.
In the last analysis de Jong may be right about all of this, and with some luck and hard work we will be able to find out. However we believe this study, like the earlier ones about deep lung involvement being the explanation for lack of transmissibility, has been over interpreted, not just by the press but by the authors.
