Rotavirus G9 Severity Data Revisited

TO THE EDITOR Linhares et al. [1] and the accompanying editorial commentary by Kang [2] describe an association between rotavirus serotype G9 (hereafter referred to as G9) infection and severity of diarrhea among infants from Brazil, Mexico, and Venezuela in 2001–2002. Such a finding is important when >100 million rotavirus infections and 440,000 deaths occur yearly among children <5 years of age [3]. G9 has emerged as the fifth most common rotavirus serotype; however, specific coverage of G9 strains is not included in 2 well-studied, currently available vaccines [4,5–6]. Several issues warrant additional comment. Linhares et al. [1] used data collected from 454 infants in the placebo group of a rotavirus vaccine study [7] who met criteria for the primary efficacy analysis; these were “subjects who received 2 doses of study vaccine or placebo [and] who had no rotavirus other than vaccine strain in stool samples collected from the first dose until 2 months after the second dose” (p. 809). Because the second placebo dose was administered at a mean of 18.1 weeks, the primary efficacy analysis data only include subjects between 5–12 months of age. Sixteen additional placebo recipients developed rotavirus infection (all due to the G1 serotype) between receipt of the first and second placebo doses; one-half of the infections were severe [7]. The primary efficacy analysis data excluded these recipients, thereby artificially elevating G9 prevalence to 29% (15 of 51; from 22% [15 of 67]). Eight of the excluded G1 episodes were severe, potentially impacting calculations and comparisons with G9. Two recipients of placebo each experienced 2 separate episodes of rotavirus gastroenteritis: 1 patient experienced 2 G1 infections, and the other experienced a G9 infection followed by a G1 infection [7]; however, it appears that these recipients were included in the primary efficacy analysis. Prior infection with rotavirus reduces subsequent infection severity [8], which is also important in vaccine protection against severe disease [5,6–7]. Inclusion of these 2 subjects who experienced second G1 infections in the severity analysis would, therefore, decrease G1 infection severity calculations. Data are provided regarding the age at hospitalization and not at infection [1]; however, data suggest that G1 disease might occur earlier. If all episodes of rotavirus are considered, it is likely that G1 infections occurred earlier than G9 infections—a fact that is difficult to attribute to maternal antibodies. Another possibility, as alluded to by Kang [2], is that a G9 epidemic occurred, as serotype data in Venezuela, Mexico, and Brazil are sparse and vary regionally [9,10–11]. The natural incidence of severe rotavirus gastroenteritis peaks among patients aged 9–11 months [12]; a G9 infection epidemic in 2002 would have encountered many infants in this age range. Although introduction of a new strain of rotavirus into a new population should result in disease across all age groups, as Kang argues [2], the failure to document cases of adult disease does not provide evidence against an epidemic. Adults are rarely tested for rotavirus, and serotype data are lacking, with few exceptions [11, 13, 14]. Thus, failure to appreciate an epidemic among adults does not exclude a late G9 epidemic, which could account for excess G9 infection severity. These important issues complicate the G9 infection severity calculations that have global implications for the development of several rotavirus vaccines (for which G9 efficacy data are limited) [5,6–7]. Voluntary ingestion of G1 versus G9 in a randomized, blinded trial [13], although unpalatable, is necessary to completely clarify the relationship between serotype and severity of diarrhea. I thank Ben Z. Katz, MD, for his careful review and suggestions. Potential conflicts of interest. E.J.A. has served as a consultant for, receives grant support from, and is on the speaker's bureau of Merck, regarding the subject of adult rotavirus.