Saturday, October 15, 2005

Tamiflu resistance revisited

[Warning: High geek factor]

The story of oseltamivir (Tamiflu) resistance, like the virus itself, continues to circulate in one form or another. Yesterday there was a Brief Communication in the scientific journal, Nature (20 October 2005) which was widely noticed. Thanks to the excellent reporting of Canadian Press's Helen Branswell, we had already learned that earlier reports of a developing H5N1 Tamiflu resistance may have been a result of a misunderstanding, promulgated by Hong Kong University pharmacology professor William Chui who admitted he was quoting old data of partial resistance in a Vietnamese patient. The Nature article may be the Vietnamese case that was the basis of Chui's claim.

We noted in the previous post there are some biological reasons to believe influenza/A would not show the kind of resistance to Tamiflu characteristic of the other main influenza/A anti-viral, amantadine because the genetic mutations required to make the virus resistant also made it less fit. Yesterday's paper shows data consistent with that idea. It also suggests the degree of resistance in infected animals may not be as great as the test tube data suggest.

The patient was a 14 year old Vietnamese girl who cared for her 21 year old brother with documented H5N1 infection. The sister had no history of poultry contact and the timing of symptom onset suggested person-to-person transmission, according to the authors. The sister had been given a prophylactic dose (one 75 mgm tablet per day) for three days, increased to a therapeutic dose (2 tablets per day) for 7 days. She recovered and on discharge no longer had isolatable virus. Viral clones isolated from both brother and sister during their illnesses revealed that some of the sister's viral load was highly resistant to Tamiflu (three log increase in IC50). A specific amino acid change (H274Y) in the neuraminidase protein was found in the highly resistant clones, with a one to two log decrease in Tamiflu induced neuraminidase inhibition. The sub-therapeutic initial dose might have contributed to the development of resistance, a significant risk if Tamiflu were to be widely used for prophylaxis at that dosage.

The resistant and sensitive viruses both infected ferrets in the laboratory, but it is interesting to note the viral titers are about 10-fold lower (one log) for the resistant virus (paper, figure 1b, bottom). The data on Tamiflu response in the animal studies is curious. The figures (1b, top and bottom) show the ferrets infected with both sensitive and resistant viruses responded to Tamiflu, although the text says the drug did not reduce viral titers in the resistant virus animals. The reason for this (mis)interpretation appears to be that in sensitive-virus infected animals the reduction was "statistically significant" (p=.048, which when rounded gives p=.05, borderline significance), while for the resistant virus the p-value was 0.23, above the conventional cut-off of .05. However the resistant viral titer started lower (three logs versus four logs) and was reduced to the same level as the sensitive viral titer (one log), suggesting the authors may have been guilty of a (common) error in interpreting statistical results. There is clearly a titer reduction for both viruses after Tamiflu treatment, as shown in the figure. The statistical test shows the reduction in the resistant clone might have been a result either of chance alone or the drug (or both), while the reduction in titer of the sensitive virus was unlikely to be due to chance alone. Had the initial titer of the resistant virus been larger or more animals used, it would likely have achieved the same level of statistical significance as the treated animals infected with sensitive virus. Both viruses were sensitive to zanamivir (Relenza), showing that the Tamiflu-resistance conferring mutation did not affect the action of that drug, welcome news.

One other result in this short paper is of interest. The authors compared viral binding to both α-2,6 and α-2,3 sialyl glycopolymers by the H5N1 virus from the patient, an H5N1 bird flu virus isolated from a Mongolian duck in 2001, and a human flu virus. The α-2,6 sialyl glycopolymer is characteristic of viral receptor on human cells, while the α-2,3 sialyl glycopolymer is characteristic of bird viral receptors (receptors are chemicals on a cell's surface where the virus docks prior to entry and infection). The H5N1 viruses both bound to the avian receptor, but the Vietnamese virus also bound to some extent to the α-2,6 human receptor, while the H5N1 bird virus from 2001 did not. The human flu virus (A/Kawasaki/1/2001) bound strongly to α-2,6 (characteristic of human receptors) but only weakly to α-2,3 (bird receptor). This suggests that the Vietnamese H5N1 had evolved from its 2001 form to one better adapted to a human host.

Background information on the biology of influenza and the naming of influenza viruses can be found in our Basic Science entries on the Flu Wiki.