Attached is the response to T. Cundell's post dated 12/13/99.
Let me begin by noting we now appear to agree that penetration of sterilizing
grade filters occurs, and that size is only one of several factors governing
bacterial retention by filters. The issues raised in your follow-up post
(12/13/99) mostly pertain to:
1) the "suspected" process penetration event that prompted this study
(which was
the impetus for, but not the subject of, this paper),
2) the relevance of the study to "typical" pharmaceutical operations (in terms
of challenge levels and duration), and finally,
3) discussion of potential penetration mechanisms.
While these issues are addressed in detail below, the short answers to these
questions are that:
1) Pall Corporation cannot provide additional information on the suspected
process penetration event due to confidentiality constraints. These must be
obtained from the pharmaceutical manufacturer as noted among the co-authors of
the subject paper.
2) Readers should judge for themselves how the results and putative penetration
mechanism(s) presented in this paper may or may not apply to their own products
and processes, keeping in mind that there are other industry reports of
penetration that are not limited to high challenge levels or long
durations; and
3) We agree that mechanism(s) of penetration are probably complex and well
worth
additional investigation.
Question 1a:
" R. pickettii was isolated downstream of a 0.2 micron rated filter
during a production (validation) run with this drug product. "
What was the pre-fitration bioburden in the product when the failure
occurred? Was it checked at the end of the filtration?
Response:
The suspected process penetration event pre-dates by a few years the tests
described in the paper, so I am not privy to all the details, but it is my
understanding that bacterial contamination was detected in some filled vials
post-filtration. This microbial contaminant was subsequently identified as R.
pickettii. Given that there was at least one published case of penetration
of a
0.2 micron (microfibrous) sterilizing grade filter by R. pickettii (reference
29), a bacterial challenge test investigation was deemed to be appropriate.
All
other details are confidential.
Typically, the industry sees less than 10 cfu/100 mL in their pre-sterile
bulk solutions which is greater than 6 logs lower than the bacterial
challenges used in your publication. Also, can members of the website
comment if 2200 liter bulk solutions with operating parameters of 0.3 liters
per min up to 120 hours is typical? For a product that is not
bacteriostatic, a 120 hour filtration process seems awfully long. A bulk
solution of 2200 liters with a 1 mL container size would have a lot size of
2.2 million containers.
Response:
First, a bioburden level in counts per mL is meaningless without the culture
medium, incubation temperature, and incubation time also being specified. A
count of less than 10cfu/100mL on TSA at 35C for 48 hours does not preclude
higher counts of "starved" or "stressed" organisms that either cannot grow on
rich medium, or at the higher (than ambient) temperature, or are "slow"
growers.
Second, it should also be noted that we have established that R. pickettii
grows
in the drug solution to a (maximum) concentration of about 10^6 per mL in about
48 hours at ambient temperature. Granted, the inoculum in our growth curves
were higher than 10cfu/100mL, but given the 120 hour process time, a lower
starting concentration may still be sufficient. For example, if one were to
assume that the bacterial concentration reached this level (10^6 per mL), say
only just for the last 3 hours of the 120-hour process, this would still
correspond to a challenge level of about 10^7 cfu/cm^2 (and almost 10^8 cfu/cm2
if we assume this level for just the last 24 hours). From this perspective, we
do not think that the challenge levels employed in the study were
unrealistically high. Third, in terms of penetration being a function of time,
we have stated in the paper that the cells were "grown in TSA or SLB, both of
which are richer in nutrients (relative to the drug solution) before being
exposed to the drug solution. Under process conditions, the bioburden is
presumably already in a diminutive form due to the lack of nutrients, and thus,
penetration may strictly be a function of challenge level. For example, a
recent study with a naturally occurring diminutive organism, one that
maintained
its diminutive state (non-revertant) under the growth conditions used in the
study, showed that penetration of 0.2 and 0.22 mm rated filters occurred in
very
short challenge times (20-60 min), with penetration being a function of
challenge level (97)."
We also believe that this issue of growth conditions is also relevant to the
discussion about challenge levels since R. pickettii in its natural state may
well penetrate 0.2/0.22 micron rated filters at much lower challenge levels
than
something grown first on TSA/SLB and then resuspended in product (which may not
be as "penetrative" as native cells). In other words, "we grow gluttonous
bacteria using laboratory media and subject them to starvation conditions" (R.
Y. Morita, in Starvation in Bacteria, Ed., S. Kjellberg, Plenum Press, pp.17,
1993), and whether this is the most appropriate way to "mimic" the natural
state
of the bacteria (which in most cases is in an oligotropic environment) is
debatable. Clearly, the objective is to use a bacterial suspension that most
closely resembles what is likely to be found in the actual process.
It is well acknowledged in the "starvation" field, that "within limits, the
longer the starvation period, the smaller the cells become", and that it is
difficult to reproduce the "natural state" of bacteria by standard laboratory
conditions. For example, it has been shown that penetration of 0.2/0.22 micron
rated filters by R. pickettii strains (and other species) increased with
increasing periods of exposure (starvation) (M. Shahamat et. al., Abstracts of
the 99th General Meeting of American Society for Microbiology, pp. 140, ASM
Press; Washington, DC). In short, there is considerable experimental
literature
to indicate that a higher level of challenge with a "cultured
microorganism" may
not necessarily be a "worst-case" than a lower level challenge with the same
organism in its natural state.
Question 2:
Did you investigate the filter integrity properties, e.g., bubble point,
forward flow, pressure hold at the operating pressures with the filter
exposed to the drug product in contrast to water and IPA-water mixtures?
Question 5:
"If there had been some effect of the drug product on the filter, we would
expect both B. diminuta and R.pickettii to show penetration, which is not
the case."
The cause of the R. pickettii-drug product-filter failures are probably
complex. The number of challenge organisms on the recovery filter at failure
was approx. 10 cfu for a cumulative challenge of 10^8 cfu/sq. cm,. i.e., 1
cfu passing through the filter per 10^7 cfu retained per sq. cm. Small
changes in the size, shape and surface properties of the R. pickettii and/or
the filter properties or performance could result in a few bacteria not
being retained.
Response:
As stated before, all filter discs in question were integrity tested (in a
standard fluid) by a manual "bubble-point" type test, both pre- and
post-challenge. All tests were performed using 47-mm discs, and at this scale,
the Forward Flow (or Pressure Hold) is too low to measure accurately (for
example, the Forward Flow specification for the 47-mm nylon 6,6 filter disc, if
scaled by area, would be 20 ml/min). In all cases, the filter disc met the
corresponding "bubble point" specifications. In the post-challenge test, each
filter disc had been challenged with R. pickettii and exposed to the drug
solution in question for the entire duration of the challenge (ranging from
120-196 hours) before being tested. In every case, the filters passed this
manual "bubble point" test post-challenge.
It seems to me that you are suggesting that a filter integrity test in the drug
product is more informative than a post-challenge test performed on the same
filter in a standard fluid after flushing out the drug product. A passing
value
in ANY wetting fluid under appropriate test conditions for that fluid is
indicative of adequate wetting and membrane integrity. Nonetheless, both this
issue of a filter "defect" detectable somehow only during a integrity test in
the drug product, as well as the issue of "the effect of the unidentified
product on the filter" (which I interpret to mean a chemical incompatibility
with the filter) are addressed by the B. diminuta challenges. It is well
acknowledged that a bacterial (B. diminuta) challenge test is the "ultimate"
compatibility and integrity test for 0.2/0.22 micron rated filters. I would
also point out that up until this study, retention of B. diminuta under
product-
and process- specific conditions was automatically considered evidence of a
"sterilizing grade" filter as well (some still believe this despite this and
other referenced studies). Thus, retention of B. diminuta by these filters
under the same conditions that resulted in penetration by R. pickettii can be
used to rule out any compatibility and/or filter integrity issues. In other
words, since whatever "effect" the product has on the filter (in terms of
chemical incompatibility or compromising its integrity) is independent of the
test organism, one would expect that such an "effect" would result in
penetration of both the standard microorganism (B. diminuta) as well as R.
pickettii. If you are postulating that somehow this "effect" was
operational in
the R. pickettii challenges resulting in some retention failures, but failed to
show up in any of the B. diminuta challenge (in all 11 filters challenged),
this
is theoretically possible but the probability of non-detection in all of the B.
diminuta tests is quite small. In addition, we have since done mixed
challenges of 0.2/0.22 micron filters with B. diminuta and other "diminutive"
microorganisms since this study, and demonstrated penetration of these filters
by these diminutive organisms even when B. diminuta is fully retained (ref 97,
presentation at this year's PDA meeting posted at the PDA at
www.pda.org/presentations.html).
If size was important, you should obtain photomicrographs of the bacteria on
the surface after 192 hours and the bacteria that pass through the filter.
The latter could be obtained using the Scan RDI since it would be difficult
to find them using SEM techniques.
Response:
We have done SEMs on a nylon 6,6 filter disc post-challenge (looking at
both the
surface, as you suggest, and the cross-section for any evidence of
"growthrough") as well as cells recovered downstream (unpublished data which I
will be glad to discuss privately). The Scan RDI is not amenable to accurate
sizing, because its resolution is limited to that of a good fluoresence
microscope at best. Accurate sizing of small bacteria requires a minimum
magnification of 10,000X with high resolution.
Question 7:
I agree it is difficult to determine if cells are actively dividing from
SEMs. Dividing cells are important as you propose one of the mechanism of
smaller cell size is reductive cell divsion.
What contribution was made to the cell size by a combination of product
exposure and lower temperature?
Response:
Seeing "dividing cells" is not important since the numbers of cells shown in
these SEMs are a very small percentage of the population (20-30 cells out
of the
10^8 or so incident on the filter). These SEMs also represent single time
points. The literature on effects of starvation indicates that "dwarfing" or
"reductive division" occurs in the initial phase of starvation (over the first
few hours typically), and the few "dividing" cells we see in Fig 3B are
probably
indicative of the end of the "reductive division" process. Even if there
weren't any "dividing cells" in Fig 3B, it might just mean that process was
completed before the 24 hours of product exposure at ambient temperature (the
time point corresponding to the SEMs in 3B), or that the number of cells still
undegoing reductive division were at a level not detectable by any microscopic
technique.
The data certainly suggest that the lower temperature favors reductive
division,
but I am not sure whether the initial non-change at 32C that you refer to
is not
due to "carry-over" of nutrients from the SLB culture; note that we added
0.1 mL
of the SLB culture to about 700 mL of drug solution as inoculum. It might also
be the fact that the second step at 25C followed what I call the "4-fold
dilution" rule established by some workers to be important in starving cells.
Question 8:
"it can be safely argued that the changes in R. pickettii size and
morphology in the drug solution were, at a minimum, a necessary condition
for bacterial penetration. This does not, however, rule out the
contributions of other changes that may have occurred in R. pickettii as a
result of exposure to the drug solution, such as changes in cell-surface
structures, charge, or hydrophobicity, or changes in cell-wall permeability
and deformability."
With respect to R. pickettiii size, the publication presented the dimensions
of the bacterial cells after 48 hours in a solution of the drug product at
20 degree C. No data is presented on the R. pikettii size after 48 to 192
hours exposure to drug product at 20 degree C either in solution or on a
filter surface. It is well known that the metabolism and morphology of
bacteria change after attachment to a surface.
Response:
We stated that we have only looked at a maximum of 48 hours, and that we cannot
rule out additional changes upon prolonged exposure and that "tests to
determine
if such a prolonged exposure could lead to further size changes, beyond that
seen within the first 24-48 hours of exposure to the drug solution, are in
progress." This publication was primarily intended to be the "phenemenon"
paper, which provided a basic description of the phenomenon (in the context of
the historical literature both of penetration of sterilizing grade filters and
specifically the cases of nosocomial infections attributed to contamination of
"sterile" products by R. pickettii), as well as some preliminary information on
putative mechanism(s). More definitive statements on "mechanisms" cannot be
made at this time. Our initial efforts have been to evaluate this effect in
other non-proprietary "nutrient-deprived" or "nutrient-limited" fluids, and we
are starting to focus more on mechanisms. This type of a test program is
complex and is a slow process. In addition, our major efforts are focused on
the more critical concern of penetration in non-proprietary fluids (ref:
Abstracts of the 99th General Meeting of ASM, pp. 560; my presentation at the
recent PDA Annual Meeting posted on the PDA Website at
www.pda.org/presentations.html).
Question 8 (continued)
It is possible that the progressive loss of R. pickettii retention on 0.2
micron rated filters is due to 1) the cumulative challange of 0.8 X 0.4
micron bacterial cells especially at the higher pressures to maintain the
flow rate as the cells acumulate,
Response:
As stated in our paper (and my previous response), there was NO change in
differential pressure across the filter discs (<1 psi) through the entire
challenge, for the R. pickettii in drug product challenges.
Question 8 (continued)
2) grow through of the R. pikettii during a filtration process that takes up to
192 hours,
Response:
"Growthrough" is a term which is used to denote different hyopthetical and
largely unproven mechanisms. In one case, growthrough is used to describe the
permeation of the membrane due to bacterial division within the matrix of the
filter leading to bacteria spanning the entire thickness of the membrane. We
see no evidence of this in our SEM examination of the test discs (unpublished
data). Others have proposed that growthrough is the result of proliferation of
organisms to the point where the population is large enough to find the larger
pores of a filter membrane. Still others use growthrough to mean the
permeation
of the membrane by organisms that have assumed smaller sizes due to binary
division. Both these "mechanisms" are inconsistent with the absence of
penetration by R. pickettii when grown and challenged in SLB. Clearly, growth
is SLB is more vigorous than in the drug solution (and the challenge levels are
definitely higher), and Figures 2A and 2B show plenty of evidence of binary
division...yet, penetration is not observed. In other words, R. pickettii in
SLB did not "growthrough" the filters even during long-term (up to 196 hour)
challenges.
This confusion in the literature regarding "growthrough" has resulted in its
erroneous use to explain any penetration event that shows a
time-dependency. It
has been pointed out that in filtration, time correlates to volume filtered and
since longer filtrations correspond to higher challenge levels, penetration may
simply be the result of the filter's retention capability being exceeded.
Question 8 (continued)
3) the changes to the filter when exposed to the drug product,
Response:
To reiterate, we see no evidence for this (see response under question 3 and 5
above), nor have we seen it in the several hundred product validations that
Pall
has performed.
Question 8 (continued) and/or 4) progressive changes to the surface properties
of R. picketti.
Response:
We agree.
If size is important but not the full story, I would recommend that you
investigate the relative surface properties of bacteria of a similar size
range that are retained or pass through 0.22 and 0.45 micron rated filter at
cumulative challenges greater than 10^7 cfu/sq. cm
Response:
The citations that you have faxed to me on quantitating bacterial surface
properties are going to be very valuable for this, and I thank you for
forwarding them to me.
Sri Sundaram, Ph.D.
Senior Staff Scientist
Scientific and Laboratory Services
Pall Corporation
E-mail: Sri_Sundaram@pall.com
|