LETTER

Top Letter References

Surveillance of rotavirus strains was first established in the United States in 1987, through the analysis of G and P types of isolates obtained in seven sentinel locations, mostly during multiple consecutive rotavirus seasons extending from 1987 to 1993 (4, 14). Subsequently, this initial surveillance was expanded to 37 centers participating in a system of national reporting on year-round rotavirus activity in order to assess geographic and temporal trends, but without monitoring strain types (8, 15).

The composition and distribution of rotavirus types may vary in distinct geographic and socioeconomic regions of the world, as demonstrated in a study that compared data obtained in the United States and in Brazil as models for the epidemiology of rotavirus in temperate, developed and tropical, developing countries, respectively (6). In this context, it was of interest to verify whether the recent data reported by Ramachandran et al. (12) were still in accordance with those obtained by Gouvea et al. (4) and Santos et al. (14) for the United States in the previous decade. Indeed, the major epidemiological characteristics were confirmed: an overwhelming prevalence of conventional strains (83% in the recent season versus 95% in previous seasons) with a predominance of P[8]G1 strains (66 versus 71%) and very low proportions of mixed infections (2.3 versus 3%) were found in the continental United States. Differences in the sensitivities of the assays employed in the two studies might have accounted for the small difference found in the efficiencies of typing rotavirus strains (94.5% in the recent study versus 100% in the old study). Nevertheless, contrary to Ramanchandran et al.'s statement, a few strains bearing the supposedly attenuated P[6] specificity had already been identified in association with infant diarrhea in the United States (14). Some of those isolates had been further sequenced and characterized, including the oldest known and cell culture-adapted American P[6] strain, SC2 (13).

Of interest in Ramachandran et al.'s study was the finding of a small percentage (7.2%) of strains of the G9 type in several Midwestern states. Remarkably, however, G9 strains were not detected in the more populated region of Philadelphia, where it had circulated a decade ago (1, 12). Following the first identification of rotavirus type G9 in the developed world (prototype strain WI61 in Philadelphia, Pa., and strain F45 in Japan), it was also recognized as a cause of infant diarrhea in the developing world (strain Mc323 in Chiang Mai, Thailand, and an unnamed strain in Belém, Brazil) (1, 10, 11, 16). Molecular analysis revealed very high genomic identity between strains WI61 and F45, suggesting that they might represent the same emerging virus that had been successfully introduced into the human population (5, 7). Most recently, genetically distinct G9 strains were identified as common human rotaviruses in India and frequent swine pathogens in Southern Brazil (2, 9), thus supporting the suggestion that G9 strains might have emerged in the human population from animal reservoirs by natural reassortment (3, 16).