Calculation of the Clearance Factor
The downstream processing of biotechnically manufactured proteins derived from mammalian cell culture and other pharmaceutical products of biological, i.e. animal or human origin provides a clearance factor for the removal and inactivation of potential viruses which needs to be verified in appropriate validation studies. The measures which are applied in order to demonstrate the freedom from infective virus are conceptionally comparable to the proof of sterility. Absolute sterility of a product solution can only be demonstrated for the case that the entire production lot undergoes sterility testing, and that the test method is capable to detect all potential contaminants - but no product would be left. Therefore, it is accepted practice to test fractions of the production lot based on a statistical approach, where the result will not be absolute but will provide a reasonable probability of freedom from a contaminant. According to the well established methodology for sterility testing, the demonstration of an absolute exclusion of any potential virus is not pursued (and, of course, is technically even not feasible), but viral clearance provides for an acceptable limitation of the probability for a potential virus contamination of a single product dose. Hence, the manufacturing of pharmaceutical proteins is accepted, even though the occurrence of a virus contamination cannot be excluded absolutely for an unlimited time, much more it addresses the judgement on the acceptability for a level of drug safety and on a statistically evaluated probability for potential but extremely rare events.
Determination of the virus titer
Effectivity of virus clearance is determined by spiking experiments for the respective unit operations. Virus distribution is monitored and balanced for the individual intermediates of such an operation: the virus titer of the load is measured and compared to the (residual) virus titer of the product containing fraction after processing, e.g. the flowthrough or eluate of a chromatographic process or the permeate of a filtration process.
Prior to the titration of process samples the potential effect of the applied buffer solutions has to be investigated regarding interference with the detector cells or reduction of infectivity of the model virus. Typical detector cells for the titration of viruses are SC-1 cells (Retrovirus), CV-1 cells (SV 40), L 929 cells (Reovirus) and Vero cells (PI3). For an extended detection for retroviruses the XC plaque assay can be applied, where SC-1 cells are inoculated with the respective sample; after a defined period of cultivation the cell layer is UV-irradiated and overlaid with XC cells. After plaque formation the cell layer is fixed, stained and plaques are counted. With reference to the morphological shape of a retrovirus infected cell monolayer the titer is expressed in plaque or focus forming units (pfu/FFU):
pfu ml-1 = n x d
n = number of plaques or foci
Viruses not leading to the formation of plaques or foci are measured by their CPE (cytopathic effect) on the detector cells, and the titer is expressed as TCID50 (tissue culture infectious dose for 50 % of the entire cell number):
TCID50 = X0 - (dlog/2) + (dlog/n) x xi
X0 = positive exponent of the highest dilution showing
The balance of virus distribution throughout the complete process step, i.e. including corresponding fractions such as washing and regeneration steps of chromatography or the retentate of filtration, is unlikely for most processes due to denaturation of virus by caustic solutions typically used for regeneration or due to capture of virus particles within the membrane matrix of a filter.
The demonstration of virus clearance for a single validation experiment is limited mainly by two factors: the maximum available titer for the accepted viruses is in the range of 107 to 109 ml-1; it is further reduced by 1 log by the required spike of 1:10-20.
Another limitation is the technical difficulty to titrate the entire process fluids: the volume of a sample is depending on the detector cell. Typically the titration is performed using a sample volume of 0.1 - 1.0 ml.
A statistical approach has to be taken in order to calculate the probability that the investigated samples do not contain infective virus:
V = entire volume of the process fluid
Hence, samples taken from process fluids which contain a potential virus burden below 1 infective particle per ml may not contain any virus. Thus an additional statistical evaluation might become necessary for the case that no virus can be detected in any of the samples. The assessment of the Poisson Distribution at its 95 % upper confidence limits is widely used as an appropriate methodology. The virus titer is determined for V > v, the probability p is calculated as
p = e-cv
c = concentration of infective virus.
The reduction factor of a designated unit operation is calculated upon the volumes of the process fluids and the virus titers measured for the load and the product containing fraction after processing. The reduction factor is typically expressed in log 10 units, the "individual reduction factor" Ri for each unit operations is calculated as:
Ri = individual reduction factor
The overall reduction factor for a virus within the entire purification process is the cumulation of the individual reduction factors. However, the cumulation of virus clearance can only be claimed for process steps which represent different physico-chemical measures. Based on a cell assay variability which is measured in a logarithmic scale, a logarithmic reduction factor in the order of 1, i.e. a 90 % reduction in titer, is considered to be not significant for virus clearance.
In order to set a numerical figure for the virus burden of a cell
culture fluid at the time of harvest, electron microscopy is applied for
counting viral particles in a distinctive volume. However, this approach
raises additional questions: how representative can a few ml sample be for
a few hundred liter to several thousand liter scale of cell culture
fermentation? Furthermore, the cells in culture are grown to densities
between 106 to 107 ml-1; the number of
cells investigated by EM is reduced by several logs and is about
103 ml-1. The identification of virus particles and
their differentiation from particulate matter due to the preparation
procedure requires extensive experience, and a confirmation of the viral
nature is impossible.
Examples are artifacts which are originated from sample preparation, e.g. high speed centrifugation of cell culture supernatant; the pellets derived from centrifugation typically harbour complex aggregates which often conceal those details necessary to identify virus structures.
©1999 Charm Bioengineering, Inc.