A statistical approach was used to establish a new classification system of Marek's disease virus (MDV) on the basis of neurologic responses. To develop the system, neurologic response data from 15×7 chickens inoculated with 30 strains of serotype 1 MDV were statistically analyzed by a cluster analysis. The goal was to identify a statistical system that would verify if three neurovirulence groups correlated with the three pathotypes previously described. The system was also validated in two additional strains of specific-pathogen-free (SPF) chickens, SPAFAS and line SC (Hy-Vac).

The proposed system is based on analysis of three variables: 1) frequency of birds showing transient paralysis between 9 and 11 days postinoculation (dpi), (2) mortality before 15 dpi, and (3) frequency of birds showing persistent neurologic disease between 21 and 23 dpi. By use of this system, a MDV may be classified in one of three groups, designated neuropathotypes A, B, and C, which roughly correspond to the virulent, very virulent, and very virulent plus pathotypes, respectively. However, correlation between neuropathotype and pathotype was not absolute, and neuropathotyping is more a complement to the current pathotyping system than a replacement for it. Our results showed that neuropathotyping studies can be conducted in two types of commercial SPF chickens by the use of the same variables, although the system would first have to be standardized by the use of prototype viruses. Neuropathotypes can also be estimated without statistical analysis with reasonable accuracy. By use of this analysis, we established that MDV strains within the very virulent pathotype may be subdivided into neuropathotypes B and C, thus establishing a previously unrecognized pathotypic classification. This finding illustrates how neuropathotyping may extend important information not identified by conventional pathotyping.