VIRUS PARTICLES AND THEIR FUNCTIONAL ACTIVITY 331 



A. Assay Based on All-or-None Response 



The end-poiiit dilution method of assay based upon an all-or-none or 

 quantal response is used most conmionly for the titration of animal viruses. 

 Usually a standard volume of serial tenfold or fivefold dilutions of a virus 

 preparation is inoculated into test groups of five or six susceptible animal or 

 tissue culture hosts, and the number of positive and negative responses are 

 scored for each dilution. A positive response may be death as the result of 

 infection or some gross manifestation of infection, such as paralysis. In the 

 case of infected tissue culture, where each culture is analogous to an individual 

 animal host, readily observable cytopathic effects, such as complete celMar 

 degeneration or giant cell formation, may represent a positive score. A plot 

 of the per cent incidence of positive scores as a function of the log dilution of 

 virus suspension will yield an S-shaped dose-response curve. By interpolation, 

 one can estimate the dilution at which 50 % positive and 50 % negative 

 responses occur. This is called the 50 % infectivity end-point dilution (ID50) 

 and represents that point on the curve where the slope is steepest, i.e., where 

 the smallest change in virus concentration will produce the greatest difference 

 in response. The ID50 is therefore the most precisely measured point on the 

 curve. Relative virus concentrations are calculated from the inoculum volume 

 and dilution and expressed as the number of ID50 per milliliter of original 

 virus suspension. 



Too often a dose response curve is quite irregular because of variations in 

 host susceptibility at any given dilution of virus inoculated. The ID5Q is most 

 frequently calculated, therefore, not from actual incidence values but from 

 their accumulated sums, a procedure devised by Reed and Muench (1938). 

 Here the assumption is made that a host which gave a positive response at a 

 certain dilution of virus would also be positive at the next lower dilution, and 

 vice versa for negative responses. This method is only valid if the variation in 

 response reflects variation in host susceptibility. If, on the contrary, the 

 chance absence of an infective particle in the inoculum is responsible for the 

 negative response, the use of the cumulative procedure is not justifiable 

 (Luria, 1953). This procedure, nevertheless, is almost universally used in 

 estimating virus titers by the end-point dilution method. 



An ID50 is a statistic which does not lend itself readily to an estimate of its 

 variabihty. A distribution of individual end points can be determined experi- 

 mentally by repeated titrations of the same virus sample and a standard 

 deviation calculated therefrom. Such estimates have been made, for example, 

 on the variability of infectivity titers of influenza virus in mice (Lauffer and 

 Miller, 1944) and chick embryos (Knight, 1944); from these data levels of 

 probabilities for significance in the differences between end points were 

 determined. Five to ten replicate titrations are impractical, of course, for the 

 usual experimental work where some estimate of the reliability of an end 



