In the 1960s, researchers discovered that organisms retained on the upstream side of a microporous membrane could penetrate
it after a period of time and appear on the downside. Investigations showed this occurred both when the liquid flow was ongoing,
as in dynamic aqueous systems, and during static periods, when there was no flow, but the filter was still in place in a wet
condition. Not all organisms exhibited this action, nor was the time interval the same for those that did.
Although grow-through was still a present concern, its imminence was assuaged by the US Food and Drug Administration. In 1976,
in its proposed good manufacturing practices (GMPs) for large-volume parenterals (LVPs), FDA addressed the problem by requiring
that the mixing and filtration of a batch be completed within the eight-hour period of a single shift, an interval too brief
for grow-through to occur (1). Although this proposal was never finalized, in the actual event it became the rule. (Indeed,
its application discouraged the long-term use of membrane filters at points-of-use in water systems before the validation
of sanitizing practices enabled its management.)
Wallhäusser described a companion phenomenon, "blow-through," as resulting from the imposition of pressure upon a wet membrane
within whose pores organisms had already partly grown or penetrated (2). For example, instead of a "normal" release of 1 or
2 bacteria per liter through a membrane, 29 penetrated after an overnight shutdown and restart under pressure the next day.
Following a flush of some liters, the bacterial passage declined to a "normal" number.
The grow-through phenomenon was rationalized as being a result of a binary fission whereby an organism cell divides into two
during its reproduction. Presumably, in this reduced cell size, the organism can penetrate an otherwise impermeable membrane
Another explanation recognized the existence of pore-size distributions in membranes. Researchers postulated that the continuing
growth of organisms on the wet membrane eventually yields numbers sufficient to encounter even the occasional large pore,
thereby leading to penetration. (Interestingly, this explanation was strongly advanced by a filter manufacturer who had the
good fortune to recognize the potential shortcoming in competitive filters that he asserted was absent from his.)
The grow-through experience proved difficult to investigate because not all organisms showed the effect, and those that did
differed in their response times. The total contact time between the organism and the drug is seldom known, as also whether
the same organism types have been involved in each of the exposures, and whether under roughly the same conditions such as
temperature. Indeed, some investigators dispute the "grow-through" and "blow-through" episodes. Carter and Levy discuss that
view, stating, "Both phenomena are hypotheses and have never been rigorously demonstrated to be real events in pharmaceutical
processes" (5). There is the belief that such events, culminating in organism passage through 0.2/0.22 μm-rated membranes,
are overstated as they pertain to pharmaceutical processing contexts.
Grow-through versus reverse osmosis membranes
Curiously, organisms may, on occasion, be found on the downside of reverse osmosis (RO) membranes, although these have smaller
"pores" than 0.1 μm-rated membranes and would seem far less likely to permit grow-through. Unlike the case of microporous
membranes, the integrity testing of RO membranes does not involve correlations with organism retentions. It is, therefore,
possible that "imperfect" RO membranes rather than grow-through is the cause of such occurrences. Carter hypothesized the
presence of pinholes in RO filters (6).
Manufacturing techniques have undoubtedly improved since 1976. Nonetheless, the intersegmental spaces among the convoluted
polymer molecules that constitute the pores of the RO membranes are exceedingly small. Their castings are, therefore, much
more liable to be compromised by the incorporation of miniscule particle-inclusions that are not significant in microporous
membranes. Tellingly, perhaps, RO membranes are not relied upon to produce sterile effluent. Indeed, FDA is generally regarded
as being wary even of the use of two-pass, product-staged RO, at least for the production of water for injection.