Intital descriptive and analytical data on an outbreak of norovirus infection at marine corps recruit depot Parris Island, South Carolina.

current understanding of norovirus-host interaction. In addition, several points should perhaps be reconsidered with respect to data analysis and interpretation of the results. The ability of noroviral virus-like particles to recognize human HBGAs is highly strain-specific [2, 3], and this fact is the basis of current exploration of noro-virus host range. Different HBGA binding patterns have been found in both GI and GII noroviruses [4]. Recent data even suggest variations in HBGA recognition within the GII-4 genotype [5]. Thus, it is critical for an outbreak study to have a single norovirus strain involved in each outbreak. Unfortunately, this does not appear to be the case for the study described in the Halperin et al. article [1]. The 2003 outbreak was caused by at least 2 strains that phylogenetically clustered with 2002B and 2003A of GII-4 [1], respectively. This could result in a misleading outcome if the 2 strains have different HBGA binding patterns. Similarly, data for 2 outbreaks that occurred in 2 different years and were caused by genetically distinct strains should not be pooled for analysis. Other concerns about this study include the following. First, the low reverse transcriptase polymerase chain reaction detection rates of norovirus in the 2 outbreaks (33% and 28%, respectively) [1] raises the possibility that additional causes, such as other noroviruses and/or non-norovirus pathogens, were involved, which could further dilute the results. Second, it is known that in addition to the ABO family, the Lewis and secretor families are also involved in norovirus-host interaction. The Halperin et al. study [1] focused only on the ABO family, and another reason for the negative results could be that the important information of the secretor family was missed, as shown by several previous studies [6 –9]. Third, GII is composed of at least 17 genotypes and many of them have not yet been studied in terms of virus-host interaction. Therefore , the conclusion that there is no association between HBGA and susceptibility to clinical infection with GII norovirus— based on only 2 GII-4 outbreaks—is overstated. Fourth, it would be informative if the HBGA binding patterns of the caus-ative strains involved in the outbreaks could be determined, which would address the key question about the strain-specific host range. Our group performed a similar study recently on 2 norovirus-associated gas-troenteritis outbreaks caused by a GII-3 and GII-4 virus, respectively [9]. A significant association was observed between symptomatic infection and host …