Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
OPS H. pseudoflava Filter paper 11/3/2009
1
1
Filters with a 0.2 µm rating: Hydrogenophaga pseudoflava penetration 2
can be used to quantitatively assess impact of filtration parameters.* 3
4
5
Lee, A. 1; J. McVey2 P. Faustino; S. Lute, 1; N. Sweeney3; V. Pawar2; 6
M. Khan4; K. Brorson1; D. Hussong2 7
8
Office of Biotechnology Products1; Office of Pharmaceutical Science 9
New Drug Microbiology Staff2, Office Generic Drugs3, Office of Testing 10
and Research4; 10903 New Hampshire Ave.; Silver Spring MD 20903; 11
7520 Standish Place, Rockville, Maryland 20855. 12
13
14
15
16
17
18
19
* This scientific contribution is intended to support regulatory 20
policy development. The views presented in this article have 21
not been adopted as regulatory policies of the Food and Drug 22
Administration at this time. Inclusion or exclusion of filter 23
brands or types in this study does not constitute an 24
endorsement or recommendation by the FDA or US 25
Government. 26
Copyright © 2009, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved.Appl. Environ. Microbiol. doi:10.1128/AEM.01825-09 AEM Accepts, published online ahead of print on 4 December 2009
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
2
Abstract 27
Filters rated 0.2 µm are used in laboratory and manufacturing settings 28
for diverse bacterial and particle removal applications from process 29
fluids, analytical test articles and gasses. Using Hydrogenophaga 30
pseudoflava, a diminutive bacteria with an unusual geometry (i.e. very 31
thin), we evaluated passage through 0.2 µm rated filters and the 32
impact of filtration process parameters and bacterial challenge density. 33
We show that consistent H. pseudoflava passage occurs through 0.2 34
µm rated filters. This is in contrast to an absence of significant 35
passage of nutritionally challenged bacteria that are of similar size 36
(i.e., hydrodynamic diameter), but dissimilar geometry. 37
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
3
Introduction 38
39
The 0.2 µm filter class includes a large and diverse set of products 40
(22). They include air filters, particle reduction filters, filters used for 41
bioburden reduction, lab grade filters, and “sterilizing grade” filters 42
used in sterile dosage form manufacture. ASTM F 838-05, the 43
Brevundimonas diminuta challenge test, is a standard for the 44
“sterilizing grade” filters (4), a subset of the 0.2 µm rated filters. The 45
“0.2 µm” designation is applied to the larger and more diverse set of 46
products. This designation is based on physical measurements (e.g., 47
bubble point, the force necessary to extrude air through the capillary 48
network of wet filter) and mathematical extrapolations (5, 14, 29). 49
50
The current filter validation approach for parenteral pharmaceuticals 51
involves a demonstration of removal of 7 log10 cfu/cm2 of nutritionally 52
starved B. diminuta from bulk drug product liquids (4, 8, 11, 29). B. 53
diminuta can penetrate 0.2 µm rated filters, but only sporadically and 54
at low levels (12, 21). Larger bacteria (Listeria monocytogenes) have 55
been demonstrated to be able to penetrate 0.2 µm filters after long-56
term exposure (27). Recently, a species of small water-borne 57
bacteria, H. pseudoflava, have been shown to penetrate 0.2 µm rated 58
filters (31-36) to a greater extent than the above described bacteria. 59
None of these bacteria are actually physically smaller than 0.2 µm, 60
even H. pseudoflava (25, 37, 38). 61
62
Because H. pseudoflava penetrates 0.2 µm rated filters in a potentially 63
quantifiable manner, it can be used to study filtration efficiency. In 64
this report, we evaluate the impact of filtration process parameters 65
and bacterial challenge density on passage. We benchmark H. 66
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
4
pseudoflava passage against nutritionally challenged bacteria which 67
are of similar size (i.e., hydrodynamic diameter), but dissimilar 68
geometry. 69
70
71
Materials and Methods 72
73
Bacteria. Hydrogenophaga (38) (formerly Pseudomonas) pseudoflava 74
(A.T.C.C. 700892); Brevundimonas (37) (also formerly Pseudomonas) 75
diminuta (A.T.C.C. 19146); Serratia marcescens (10) (A.T.C.C. 13880 76
& A.T.C.C. 8100) and Ralstonia pickettii (39) (A.T.C.C. 700590 & ATCC 77
49129) were purchased from American Type Culture Collection 78
(Manassas VA) or Fisher (Waltham MA). 79
80
Growth conditions. H. pseudoflava stock was prepared by seeding 81
with a 1:100 overnight culture and growth for 2 days in R2A media at 82
28oC with mixing at 100 RPM. B. diminuta was cultured 2 days in 83
Saline-Lactose Broth (SLB) at 30oC with mixing at 100 RPM after 84
seeding with a 1:100 TSB overnight culture . S. marcescens and R. 85
pickettii were cultured at 30oC in high purity water for 2 days with a 86
1:100 seed from a NB overnight culture. 87
88
Filters. Filters discs for bioprocessing & cell culturing from several 89
manufacturers were studied (Millipore Corp., Bedford, MA; Pall Corp., 90
Ft. Washington, NY; Whatman, Maidstone, UK; Sartorius, Goettingen, 91
Germany; Nalgene, Rochester, NY). These filter discs (0.2 µm rated) 92
were selected and purchased based on availability from lab supply 93
sources. Small disc filters (25-35 mm) were chosen to represent 94
multiple membrane chemistries (Table 1). 95
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
5
96
Bubble point procedure. Filters were pre-wetted by flowing 150 mL 97
deionized water from a pressurized reservoir at 1 Bar, followed by a 5 98
minute equilibrium pause. The remaining water in the reservoir was 99
sent through the filter at 5-10 psi until the water flow stopped at the 100
filter surface. Pressure was increased gradually (~ 2 psi/min) until air 101
bubbles were seen passing from the filter outlets. A minimum of six 102
filters were tested to establish the water wet bubble point for each lot. 103
104
Growth Media and Chemicals. Glycerol, sucrose and NaCl were 105
purchased from Fisher Scientific (Waltham MA). Bacterial growth 106
media were purchased from Fisher. R2A was produced using ATCC 107
recipe 2258 with components purchased from Sigma (St Louis MO) 108
and Fisher. 109
110
Viscosity, pH, conductivity & osmolality Instruments. Challenge 111
solution viscosity was measured with Brookfield LVDV-III Ultra 112
Programmable Rheometer (Middleboro MA) with Enhanced UL Adapter 113
(1.0-2000 cP). The pH was measured with an Oakton pH6 Acorn 114
series pH meter (Singapore). The conductivity was measured with 115
Oakton Con 100 Series Conductivity Meter (Singapore). Osmolality 116
was measured with Wescor Vapro 5520 Vapor Pressure Osmometer 117
(Logan UT). 118
119
Model challenge fluids. H. pseudoflava was grown for 2 days and 120
then was mixed 1:1 with Minimal Media Davis (MMD), or with 60% 121
sucrose in MMD, or with 100% glycerol or with 1M NaCl in MMD. These 122
challenge solutions were mixed and allowed to equilibrate to room 123
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
6
temperature for approximately 60 minutes prior to filtration 124
experiments. 125
126
Filter studies. Filter discs were challenged with bacterial loads 127
(ranging from 1 x 108 – 3 x 1010 cfu/cm2) from pressurized reservoirs 128
at a constant pressure of 2 bar. Challenge solutions were filtered into 129
sterile beakers inside a tissue culture laminar air flow hood. 130
Throughput was measured using precision balances. The instantaneous 131
liquid flow rate was calculated from the throughput readings and 132
filtering time measured by precision timers. Filtration was allowed to 133
proceed until the entire reservoir (200 mL) was filtered or the flux 134
decayed to as low as 0.1% of its initial water flux. Bacterial titers in 135
effluents and test solutions were measured by colony counting media 136
plates that had been overlaid with bacterial trapping membranes from 137
Millipore MicrofilTM filter cups (membrane/agar devices). Colony 138
morphology after 2-7 days incubation verified the correct species of 139
bacteria. A minimum of three filter discs were tested for each 140
challenge solution/ filter combination. 141
142
Log10 Reduction Value (LRV). LRV is calculated as the log10 (pre-143
filter bacterial titer ÷ post-filter bacterial titer). 144
145
Dynamic Light Scattering (DLS) Instrument. Light scattering was 146
performed with a Malvern Zetasizer nano S (Worcestershire, UK) with 147
an He-Ne laser (633nm) powered at 4mW. The instrument takes 12 148
or more consecutive measurements per sample, checking them for 149
reproducibility. The software (Dispersion Technology Software DTS 150
v5.02) includes an “expert advice” error checking function that allowed 151
elimination of readings that failed machine criteria. Bacterial test 152
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
7
articles were 5 µm filtered prior to analysis. Each sample was read in 153
triplicate at one or more dilutions (technical replicates); results 154
passing machine criteria were averaged. Biological replicates were 155
read on different days and then averaged. 156
157
Scanning Electron Microscopy (SEM). SEM of bacteria was 158
performed by JFE Enterprises (Brookeville MD). Briefly, after the 159
specimen was fixed in 2.5% glutaraldehyde, it was then rinsed in 160
phosphate buffer. The free floating samples were positioned on a filter 161
before continuing. Biological samples were post-fixed in Osmium 162
Tetroxide for 1 hour and then washed in distilled H2O. They were then 163
dehydrated through a graded series of ethanol washes, critical point 164
dried, mounted on SEM stubs and coated with Au/Pd. A Hitachi 165
S-4700 FESEM instrument was used at the settings of an accelerating 166
voltage of 5.0KV, and a working distance of approximately 12mm. 167
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
8
Results 168
Recently, H. pseudoflava has been shown to pass through 0.2 µm 169
rated filters with a variable LRV between 4-7 log10 (31-36). Because 170
H. pseudoflava passage was observed to be quantitative, we wanted to 171
see if it can be used to evaluate filtration efficiency under various 172
conditions. We first confirmed the observation that H. pseudoflava 173
passage occurred consistently across several 0.2 µm rated filter types 174
(Table 1). Under similar filtration conditions passage of three other 175
benchmarking bacteria (B. diminuta, S. marcesens, R. picketti) did not 176
occur. 177
178
Test fluid composition. 179
Extremes of process fluid composition (e.g. osmolarity, ionic strength, 180
and viscosity) can, in theory, impact bacterial passage. Previous 181
studies have shown that osmotic shock (after adjustment with sucrose 182
or NaCl) can result in either shrinkage or swelling of bacterial cells, 183
which presumably could impact filter passage (3, 6, 20). Impacts on 184
the membrane itself, such as charge shielding, are also in theory 185
possible. Discussions of the diversity of potential impact of these 186
parameters have been published (23, 24). However, it was noted 187
that retention of B. diminuta was robust even under extremes of these 188
parameters and the impact of parameters was discussed on a 189
theoretical level, in the absence of quantitative data actually assessing 190
impact. 191
192
To test these hypothesis with H. pseudoflava, model process fluids 193
were prepared to test extremes of osmolarity (30% sucrose, 1285 194
mmol/kg vs. MMD diluted culture, 99 mmol/kg); ionic strength (500 195
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
9
mM NaCl, 54.4 mS/cm vs. MMD diluted culture, 9.1 mS/cm), and 196
viscosity (50% glycerol, 6.7 cP vs. MMD diluted culture, 1.1 cP). 197
198
Challenge studies of filters with various membrane chemistries by 199
high-density stock cultures of H. pseudoflava were performed with all 200
three challenge solutions as well as an MMD control (Fig. 1). For some 201
membranes, the passage did not appear to be impacted by extreme 202
process fluid physicochemical attributes; these membranes were 203
Aluminum Oxide (AlO3), Cellulose Acetate derivatives (CA’s), 204
Polyethersulfones (PES). For Nitrocellulose derivative (NC) 205
membranes, high levels of viscosity, osmolality and conductivity 206
increased H. pseudoflava retention. For Polysulfone (PS) membranes 207
50% glycerol increased bacterial retention. 208
209
Note the error bars (standard error of the mean) in our small scale 210
experiments, suggesting there can be variability in LRV between filter 211
runs. This may be due to (1) variation between small disc devices, (2) 212
inherent variability of H. pseudoflava passage or (3) effects from the 213
high load challenge conditions in our studies. Despite the variation, 214
the differences seen between process fluids for PS and NC membranes 215
are statistically significant (p < 0.05 in a t-test). 216
217
Bacterial challenge density. To evaluate the impact of challenge 218
density on filter performance, PES membranes from three 219
manufacturers were challenged with three levels of H. pseudoflava 220
(Table 2). High, medium and low challenges were prepared to be 221
about 1 log10 apart (undiluted bacterial culture, 1/10, 1/100 dilution). 222
For all three filter types, the lower challenge density leads to lower 223
LRV. There was some variability between vendors at the highest 224
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
10
challenge density (range 4.5 log10 to 8.3 log10). SEM micrographs 225
taken of filters challenged at these high bacterial densities showed 226
some caking of bacteria on the membrane surface (data not shown). 227
In contrast, the retention levels at the lower challenge densities were 228
tighter between vendors (range 1.8 log10 to 3.6 log10), where the 229
lowest degree of retention (the highest passage of bacteria) occurred 230
in the filter type with the highest initial water flux. 231
232
These observations argue that filter type differences have less of an 233
impact on LRV than challenge density, possibly due to variable effects 234
of “caking”. This observation is unlikely to be an experimental artifact 235
given that (1) the same trend was observed for all 3 vendors, and (2) 236
the low LRV was not due to small experimental window (i.e., 237
H.pseudoflava was present in the filtrate even at lowest challenge 238
density). 239
240
Size and geometry of H. pseudoflava vs. three benchmarking bacteria. 241
To determine if shape differences (i.e., aspect ratio) can explain H. 242
pseudoflava passage vs. the other benchmarking bacteria (B. 243
diminuta, S. marcescens and R. pickettii), SEM micrographs were 244
taken from all four bacteria grown under low nutrient conditions (SLB, 245
R2A, H20; Fig. 2, Table 3). H. pseudoflava (Fig 2A) appears thinner 246
and longer than the other three bacteria, perhaps it is thin and flexible 247
enough to pass through the 0.2 µm filter pore network, unlike the 248
other three thicker bacteria (Fig, 2B, C and D). 249
250
We also measured the bacteria by Dynamic Light Scattering (DLS). 251
DLS measures Brownian motion of particles typically in the submicron 252
region, like bacteria, and relates this to the size, i.e., the larger the 253
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
11
particle, the slower the Brownian motion (7). The size of a particle is 254
calculated from the translational diffusion coefficient by using the 255
Stokes- Einstein equation. 256
257
As can be seen, H. pseudoflava has approximately the same 258
hydrodynamic diameter as that of B. diminuta, R. pickettii, and S. 259
marcescens within the error of the assay (Table 3). The diameter that 260
is measured in DLS is a value that refers to how a particle diffuses 261
within a fluid (i.e., hydrodynamic diameter). This value will depend 262
not only on the size of the particle core, but also on the aspect ratio, 263
any surface structure, as well as the concentration and type of ions in 264
the medium. It should be noted that club-shaped bacteria are not 265
spheres, and can clump. Thus, sizes from DLS are approximations of 266
overall volume, not an exact diameter measurement. In this case, 267
long, thin bacteria can have a similar hydrodynamic diameter as short, 268
thick bacteria. 269
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
12
Discussion 270
271
In this study we confirmed that H. pseudoflava passage occurred 272
consistently across 0.2 µm rated filters of several membrane types 273
from different vendors and with different matrix chemistries. 274
H. pseudoflava challenge testing results in quantitative result (LRV) vs. 275
qualitative result (+ or -) typical of a B. diminuta validation study (8, 276
29). 277
278
For a subset of the 0.2 µm rated membrane types studied, H. 279
pseudoflava passage was impacted by extremes in physicochemical 280
attributes (i.e., viscosity, osmolality and conductivity) of the model 281
process fluids. In theory, filtration or process fluid parameters that 282
modify the physicochemical attributes of the filter membrane or 283
bacteria can impact passage. For example, fluid osmolality is 284
predicted to shrink bacteria, increasing passage (23, 24). In other 285
publications, high viscosity has either been suggested to either 286
decrease retention by hampering Brownian motion-mediated 287
adsorptive effects (26) or by increasing processing time (23, 24). 288
Ionic strength has been suggested to decrease adsorptive effects by 289
shielding charge groups (24, 26), again potentially increasing passage. 290
Here, however, we demonstrate that the opposite predicted by theory 291
happened. While this observation cannot be ascribed to any specific 292
mechanism, one can speculate that our challenge solutions either 293
aggregate bacteria or have an impact on the filter membrane that 294
inhibits passage. The fact that we observed an impact of process fluid 295
on LRV for some but not other membrane types argues that it is most 296
likely that we were observing an effect on the filter, not the bacteria. 297
298
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
13
Lower H. pseudoflava challenge densities in the model fluids correlate 299
with lower LRVs. This observation provides supportive evidence for 300
the theory that a caking effect, seen at high challenge densities, can 301
act as a “prefilter” to trap bacteria and inhibit passage. 302
303
We examined size, shape and 0.2 µm rated filter passage of three 304
other bacteria, B. diminuta, S. marcescens and R. pickettii as 305
benchmarked against H. pseudoflava. Only H. pseudoflava 306
experienced consistently passed though 0.2 µm rated filters. When 307
measured by DLS, all four bacteria were within 800-1200 nm in 308
hydrodynamic diameter, although SEM analysis revealed that they 309
possessed different shapes. The SEM results suggest that H. 310
pseudoflava is probably particularly prone to 0.2 µm rated filter 311
passage due to its geometry rather than its net volume; it is long and 312
thin relative to the other bacteria. Filters possess complex and 313
irregular matrix structures, not series of symmetrical, cylinder pores of 314
uniform sizes. Because of the “pore” shape, torturosity and 315
connectivity, and because of H. pseudoflava’s geometry, it may be 316
particularly prone to penetrate the passages of 0.2 µm rated filters. 317
318
In addition, the quantitative nature of H. pseudoflava passage is 319
particularly useful for process design and Quality by Design (QbD) 320
studies (13). Parameters such as processing time, transmembrane 321
pressure or fluid characteristics such as osmolality or conductivity can 322
be varied for assessment of their quantitative impact on H. 323
pseudoflava retention. 324
325
Arguments have been made that pharmaceutical facility flora are the 326
only relevant bacterial species to consider when designing filter 327
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
14
validation studies (1, 2, 16). Also, there is no assumption by industry 328
or regulators that 0.2 µm rated membranes alone provide sterility 329
assurance in an absolute sense for injectable pharmaceuticals, rather 330
filtration is considered to be one part of a larger, multi-layer sterility 331
risk-mitigation strategy (8, 29). Thus, it is not warranted to conclude 332
that the observation of H. pseudoflava passage through 0.2 µm rated 333
filters should prompt a re-evaluation of filtration approaches for 334
injectable medicines (i.e., switching to 0.1 uM rated filters; (1, 15-335
19)). 336
337
The “0.2 µm” designation relies on an inverse relationship between the 338
largest pores in the membrane and the pressure needed to overcome 339
the capillary pressure in the pore network (5, 29). Based on the 340
bubble point pressure (P), a few assumptions described below, 341
parameters dependent on the filter and process fluid physico-chemical 342
attributes (e.g. liquid surface tension, γ; advancing contact angle of 343
liquid with respect to the pore wall, cos Θ), and a correction factor for 344
pore shape (k), the bubble point equation 345
346
P = (4kγ cos Θ) ÷ d 347
348
can be used to calculate d, the largest pores in the membrane (5, 14, 349
29). Pore sizes calculated from physical measurements depend on a 350
variety of assumptions, most notable of which is that the pores in a 351
typical filter membrane are simple circular capillaries extending from 352
one side of the membrane to the other. 353
354
However, almost all filters possess complex irregular matrix structures, 355
not a series of symmetrical, cylindrical pores of uniform size (14, 40) 356
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
15
and capture particles by a variety of mechanisms (9). Because of this, 357
pore shape, torturosity and connectivity factors and aspect ratios 358
impact pore geometry, making the pore-size concept difficult to 359
interpret. Pore structure geometry varies widely between filter types 360
and brands, making an estimate of the k parameter difficult. In fact, 361
average pore sizes actually measured by scanning electron microscopy 362
(SEM) tend to be larger than pore size ratings (40). Studies with latex 363
beads have shown that 0.2 µm rated filters can allow passage of 0.5 364
µm beads under certain conditions (28). Further, the physico-365
chemical attributes that are parameters for the above equation (i.e., γ 366
and cos Θ) will vary with matrix chemistry, test fluid attributes and 367
filter type (14, 29, 30). 368
369
In light of the above described limitations of the 0.2 µm rating (i.e., 370
reliance on assumptions about pore geometry, influence of difficult to 371
measure physical factors like “wetting angle”) and the quantitative 372
nature of H. pseudoflava passage, a functional rating based on a set 373
level of H. pseudoflava retention may be a meaningful supplement to 374
the 0.2 µm designation. 375
376
It should be noted that these studies are preliminary and do not 377
constitute regulatory recommendations or policy. Current 378
recommendations and procedures for filter validation, including 379
challenge testing with B. diminuta should be followed as stipulated by 380
guidance. Filtration is one part of comprehensive sterility assurance 381
programs in place at pharmaceutical facilities. Further, these 382
observations do not imply that H. pseudoflava passage constitutes a 383
significant pharmaceutical safety issue, that H. pseudoflava is a 384
routine facility isolate or that the current filter validation methods and 385
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
16
sterility assurance procedures are inadequate. It should also be noted 386
that in our studies model challenge solutions were chosen based on 387
simplicity, and not on actual process fluids. Further, the filters used in 388
this study were tested “as provided”; lab-grade disc devices rated as 389
0.2 µm, not pharmaceutical grade capsules. The membranes were 390
presumed to be representative of membranes in sterilizing grade 391
cartridges, despite potential differences in composite structure or QC 392
testing. However, none of the above listed limitations detract from the 393
central observation that H. pseudoflava challenges result in a 394
quantitative result (LRV). 395
396
Conclusions 397
In conclusion, we find H. pseudoflava passage occurred consistently 398
across filters of several 0.2 µm membrane types from different 399
vendors and with different matrix chemistries. Factors that appear to 400
impact filter passage included: filter and process fluid physicochemical 401
attributes, bacterial load level and bacterial geometry. 402
403
Acknowledgments 404
This project was supported in part by (1) a Regulatory Science and 405
Review Enhancement program grant from CDER/FDA and (2) the 406
Research Participation Program at the Center for Drug Evaluation and 407
Research administered by the Oak Ridge Institute for Science and 408
Education through an interagency agreement between the U.S. 409
Department of Energy and the U.S. Food and Drug Administration. 410
411
We acknowledge Agnes Nguyen (CDER/FDA) for instruction in 412
operating the viscometer and osmometer. We thank Hans G. 413
Schroeder for stimulating discussions concerning 0.2 µm filtration. We 414
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
17
also acknowledge the following individuals for careful peer of this 415
manuscript Keith Webber, Dennis Guilfoyle. 416
417
418
Figure Legends 419
420
Figure 1. Results of high H. pseudoflava challenge tests with three 421
model process fluids. Shown are polyethersulfone (n = 3), cellulose 422
acetate derivatives from two manufacturers (n =3), aluminum oxide (n 423
= 3) nitrocellulose derivative (n =3), polysulfone (n =3). Replicate 424
challenge tests were generally performed with independent cultures on 425
different days. Bars represent standard error of the mean. 426
427
Figure 4. SEM (20,000 x magnification) of bacteria grown under low 428
nutrient conditions. A) H. pseudoflava cultured 2 days in R2a. B) B. 429
diminuta cultured 2 days in SLB. C) R. pickettii cultured 2 days in H2O. 430
D) S. marcescens cultured 2 days in H2O. 431
432
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
18
References 433
434
1. Agalloco, J. P. 1998. "It just doesn't matter, it just doesn't matter, it just doesn't 435
matter,..." PDA J Pharm Sci Technol 52:149-50. 436
2. Akers, J. E., and J. P. Agalloco. 2002. Recent inspectional trends: are regulatory 437
requirements for sterile products becoming scientifically undoable or unpractical? 438
PDA J Pharm Sci Technol 56:179-82. 439
3. Alemohammad, M. M., and C. J. Knowles. 1974. Osmotically induced volume 440
and turbidity changes of Escherichia coli due to salts, sucrose and glycerol, with 441
particular reference to the rapid permeation of glycerol into the cell. J Gen 442
Microbiol 82:125-42. 443
4. ASTM. 2005. Standard Test Method for Determining Bacterial Retention of 444
Membrane Filters Utilized for Liquid Filtration. F838-83. American Society for 445
Testing and Materials, West Conshohoken, PA. 446
5. ASTM. 2003. Standard test method for pore size characteristics of membrane 447
filters by bubble point and mean flow pore test F316-03. American Society for 448
Testing and Materials, West Conshohoken, PA. 449
6. Baldwin, W. W., M. J.-T. Sheu, P. W. Bankston, and C. L. Woldringh. 1988. 450
Changes in buoyant density and cell size of Escherichia coli in response to 451
osmotic shocks. J Bacteriol 170:452-5. 452
7. Chu, B. 1991. Laser light scattering, 2nd ed. Academic Press, Inc., New York. 453
8. FDA. 2004. Sterile Drug Products Produced by Aseptic Processing — Current 454
Good Manufacturing Practice US Food and Drug Administration, Department of 455
Health and Human Services, Silver Spring, MD. 456
9. Grant, D. C., B. Y. H. Liu, W. G. Fisher, and R. A. Bowling. 1989. Particle capture 457
mechanisms in gases and liquids - an analysis of operative mechanisms in 458
membrane fibrous filters. Journal of Environmental Sciences 32:43-51. 459
10. Grimont, F., and P. Grimont. 2009. Serratia. In G. Garrity (ed.), Bergey's Manual 460
of Systematic Bacteriology. Springer, New York. 461
11. HIMA. 1982. Microbiological Evaluation of Filters for Sterilizing Liquids, vol. 4. 462
Health Industry Manufacturing Association, Task Force on Filtration Sterilization, 463
Washington, DC. 464
12. Howard, G. J., and R. Duberstein. 1980. A case of penetration of 0.2-micrometer 465
rated membrane filters by bacteria. J Parenter Drug Assoc 34:95-102. 466
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
19
13. ICH. 2008. Q8(R): Pharmaceutical Development. The International Conference 467
on Harmonisation of Technical Requirements for the Registration of 468
Pharmaceuticals for Human Use Geneva Switzerland. 469
14. Jornitz, M., J. Agalloco, J. E. Akers, R. E. Madsen, and T. H. Meltzer. 2001. Filter 470
integrity testing, revisited part I. Pharm Tech 26:34-50. 471
15. Jornitz, M. W., J. P. Agalloco, J. E. Akers, R. E. Madsen, and T. H. Meltzer. 472
2002. Considerations in sterile filtration--Part I: The changed role of filter integrity 473
testing. PDA J Pharm Sci Technol 56:4-10. 474
16. Jornitz, M. W., J. E. Akers, J. P. Agalloco, R. E. Madsen, and T. H. Meltzer. 475
2003. Considerations in sterile filtration. Part II: the sterilizing filter and its 476
organism challenge: a critique of regulatory standards. PDA J Pharm Sci Technol 477
57:88-96. 478
17. Jornitz, M. W., and T. H. Meltzer. 2000. Letter to the editor. PDA J Pharm Sci 479
Tech 55:58. 480
18. Jornitz, M. W., and T. H. Meltzer. 2002. The mechanics of removing organisms 481
by filtration: The importance of theory. Med Device Technol 13:14-8. 482
19. Jornitz, M. W., and T. H. Meltzer. 2006. The Sterilizing Filter, p. 3-21, 483
Pharmaceutical Filtration: The Management of Organism Removal Parenteral 484
Drug Association (PDA/DHI 2006), Bethesda, MD. 485
20. Koch, A. L. 1984. Shrinkage of growing Escherichia coli cells by osmotic 486
challenge. J Bacteriol 159:919-24. 487
21. Lee, S. H., S. S. Lee, and C. W. Kim. 2002. Changes in the cell size of 488
Brevundimonas diminuta using different growth agitation rates. PDA J Pharm Sci 489
Technol 56:99-108. 490
22. Levy, R. V., and M. W. Jornitz. 2006. Types of filtration. Adv Biochem 491
Engin/Biotechnol 98:1-26. 492
23. Levy, R. V., K. S. Souza, D. Hyde, S. Gambale, and C. B. Neville. 1991. The 493
matrix approach: Microbial retention testing of sterilizing-grade filters with final 494
parenteral products. Part 2. Pharm Technol 15:58-68. 495
24. Levy, R. V., K. S. Souza, and C. B. Neville. 1990. The matrix approach: Microbial 496
retention testing of sterilizing-grade filters with final parenteral products. Part 1. 497
Pharm Technol 14:160-173. 498
25. McLauchlin, J., and C. Rees. 2009. Listeria. In G. Garrity (ed.), Bergey's Manual 499
of Systematic Bacteriology, vol. Three. Springer, New York. 500
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
20
26. Mittelman, M. W., M. W. Jornitz, and T. H. Meltzer. 1998. Bacterial cell size and 501
surface charge characteristics relevant to filter validation studies. PDA J Pharm 502
Sci Technol 52:37-42. 503
27. Nakazawa, K., H. Hasegawa, Y. Nakagawa, M. Terao, and T. Matsuyama. 2005. 504
Factors influencing the ability of Listeria monocytogenes to pass through a 505
membrane filter by active infiltration. Appl Environ Microbiol 71:7571-4. 506
28. Pall, D. B., E. A. Kirnbauer, and B. T. Allen. 1980. Particulate retention by 507
bacteria retentive membrane filters. Colloids and Surfaces 1:235-256. 508
29. PDA. 2008. Technical report 26 revised 2008: Sterilizing filtration of liquids, vol. 509
62. Parenteral Drug Association, Bethesda MD. 510
30. SAE. 2001. Bubble-Point Test Method: Aerospace Recommended Practice, 511
ARP901. Engineering Society for Advancing Mobility Land Sea Air and Space, 512
Warrendale PA. 513
31. Sundaram, S., J. Eisenhuth, G. Howard, Jr., and H. Brandwein. 2001. Method for 514
qualifying microbial removal performance of 0.1 micron rated filters. Part I: 515
Characterization of water isolates for potential use as standard challenge 516
organisms to qualify 0.1 micron rated filters. PDA J Pharm Sci Technol 55:346-517
72. 518
32. Sundaram, S., J. Eisenhuth, G. Howard, Jr., and H. Brandwein. 2001. Retention 519
of water-borne bacteria by membrane filters. Part III: Bacterial challenge tests on 520
0.1 micron rated filters. PDA J Pharm Sci Technol 55:114-26. 521
33. Sundaram, S., J. Eisenhuth, M. Lewis, G. Howard, Jr., and H. Brandwein. 2001. 522
Method for qualifying microbial removal performance of 0.1 micron rated filters. 523
Part III: bacterial challenge tests on 0.2/0.22 and 0.1 micron rated filter cartridges 524
with Hydrogenophaga (formerly Pseudomonas) pseudoflava. PDA J Pharm Sci 525
Technol 55:393-416. 526
34. Sundaram, S., J. Eisenhuth, M. Steves, G. Howard, Jr., and H. Brandwein. 2001. 527
Method for qualifying microbial removal performance of 0.1 micron rated filters. 528
Part II: preliminary characterization of Hydrogenophaga (formerly Pseudomonas) 529
pseudoflava for use as a standard challenge organism to qualify 0.1 micron rated 530
filters. PDA J Pharm Sci Technol 55:373-92. 531
35. Sundaram, S., M. Lewis, J. Eisenhuth, G. Howard, Jr., and B. Larson. 2002. 532
Method for qualifying microbial removal performance of 0.1 micron rated filters. 533
Part IV: Retention of hydrogenophaga pseudoflava (ATCC 700892) and 534
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
OPS H. pseudoflava Filter paper 11/3/2009
21
Ralstonia pickettii (ATCC 700591) by 0.2 and 0.22 micron rated filters. PDA J 535
Pharm Sci Technol 56:150-71. 536
36. Sundaram, S., S. Mallick, J. Eisenhuth, G. Howard, Jr., and H. Brandwein. 2001. 537
Retention of water-borne bacteria by membrane filters. Part II: Scanning electron 538
microscopy (SEM) and fatty acid methyl ester (FAME) characterization of 539
bacterial species recovered downstream of 0.2/0.22 micron rated filters. PDA J 540
Pharm Sci Technol 55:87-113. 541
37. Vancanneyt, M., P. Segers, W. Abraham, and P. De Vos. 2009. Brevundimonas. 542
In G. Garrity (ed.), Bergey's Manual of Systematic Bacteriology. Springer, New 543
York. 544
38. Willems, A., and M. Gillis. 2009. Hydrogenophaga. In G. Garrity (ed.), Bergey's 545
Manual of Systematic Bacteriology. Springer, New York. 546
39. Yabuuchi, E., Y. Kawamura, and T. Ezaki. 2009. Ralstonia. In G. Garrity (ed.), 547
Bergey's Manual of Systematic Bacteriology, vol. Two, part C. Springer, New 548
York. 549
40. Zeman, L. 1992. Characterization of microfiltration membranes by image analysis 550
of electron micrographs. Part II. Functional and morphological parameters. J. 551
Membrane Sci. 71:233-246. 552
553
554
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
Table 1. Log10 reduction value for 0.2 µm filtration of H. pseudoflava and three
benchmarking bacteria.
LRV (log10) Chemical Derivativea Vendor
H. pseudoflava Benchmarking bacteriab
Polyethersulfone (PES) A 4.6 ± 0.4 >9
B 6.7 ± 0.4 >9
C 4.1 ± 0.4 >9
Cellulose Acetate (CA) A 3.9 ± 0.9 >9
E 4.6 ± 0.4 >9
Polyvinylidene fluoride
(PVDF)
B 6.8 ± 2.3 >9
Cc 7.5 ± 0.3 >9
Nitrocellulose (NC) B 5.5 ± 1.2 >9
Nylon C 5.6 ± 0.3 >9
Cc 6.5 ± 1.1 >9
Polysulfone (PS) C 6.9 ± 0.9 >9
Aluminum Oxide (AlO3) D 5.2 ± 0.5 >9
a. All filters were lab grade intended for tissue culture (TC): “lab-grade
filtration (sterilizing for tissue culture)”, ”Small volume liquid
sterilization, Sterile laboratory filtration devices”, “lab, sterilizing”,
“Sterilizing filter unit, also for TC” “cold sterilization, sterile filtration”,
“Small volume liquid sterilization, Sterile laboratory filtration devices”,
etc.
b. Data represents an average LRV from 3-8 filtration runs.
c. Benchmarking bacteria were B. diminuta, S. marcescens, and R.
picketti grown two days under low nutrient conditions (B. diminuta,
saline lactose broth; S. marcescens and R. picketti, distilled water).
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
Generally, complete clearance was observed (LRV >9 log10). In a
minority of cases, sporadic, low level breakthrough (<1 cfu/10 ml)
occurred in minority of replicates.
d. Stated by manufacturer to be a positively charged membrane.
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
Table 2. H. pseudoflava passage through PES 0.2 µm membranes at three challenge levels.
PES
Type
Mean
Bubble
Point (PSI)
Mean Water
Fluxa
(mL/min/cm2)
Test Undilutedb 1:10 Dilution 1:100 Dilution
Challenge log10(cfu)/cm2 10.0 to 10.4
9.3 to 9.7
8.3 to 8.7
A 46.1 ± 0.8 75.4 ± 3.4 LRV
c (log10) 4.5 ± 0.3 2.3 ± 0.1 1.8 ± 0.3
Challenge log10(cfu)/cm2 9.7 to 9.8
9.1 to 9.2 8.0 to 8.2
B 73.4 ± 0.2 38.1 ± 1.4 LRV
c (log10) 8.3 ± 1.5 5.3 ± 0.1 3.5 ± 0.2
Challenge log10(cfu)/cm2 9.2 to 9.8 8.9 to 9.6 8.1 to 8.8
C 49.4 ± 0.7 49.4 ± 1.4 LRV
c (log10) 5.3 ± 0.8 3.8 ± 0.8 3.6 ± 0.2
a. Pre-filtration water flux at 15 psi.
b. When H. pseudoflava was filtered undiluted (~109 cfu/mL) the volumetric throughput was limited by filter blockage.
c. Instantaneous LRV after extended throughput.
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from
Table 3. Summary of bacterial size information.
Bergey’s Manuala (µm) SEM Measurementc DLS Bacteria
Width Length Width Length Hydrodynamic diameterd
H. pseudoflava 0.3-0.6 0.6-5.5b 0.24 ± 0.01 2.48 ± 1.04 994 ± 193
B. diminuta 0.4-0.5 1-2 0.41 ± 0.08 1.03 ± 0.58 780 ± 89
S. marcescens 0.5-0.8 0.9-2.0 0.51 ± 0.05 1.03 ± 0.30 976 ± 67
R. pickettii 0.5-0.6 1.5-3.0 0.39 ± 0.03 0.89 ± 0.23 827 ± 85
a. Sections in Bergey’s Manual are (10, 37-39)
b. H. pseudoflava is 0.25 ± 0.03 x 1.65 ± 0.35 µm and B. diminuta is 0.31 ± 0.03 x 0.88 ± 0.19 µm
according to Sundaram (31-36).
c. Averages and standard deviations of 45 to 105 different bacteria measured in the SEM micrographs
in Fig. 2.
d. Measured on 2-7 independent samples on different days. Each measurement represents an average
of 3 or more independent measurements of each sample.
on June 5, 2020 by guesthttp://aem
.asm.org/
Dow
nloaded from