The Guideline on Sterile Drug Products Produced by Aseptic Processing (FDA, 1987) refers to media fills as an "acceptable method of validating the aseptic assembly process." By 1994, the Guideline to Industry for the Submission Documentation for Sterilization Process Validation in Applications for Human and Veterinary Drug Products (FDA, 1994) said that specifications for media fills should be among the information submitted in support of sterility assurance for products manufactured by aseptic processing.
In the U.K., the 1983 "Orange" Guide (Department of Health and Social Security, 1983) gave media fills as an example (albeit the only example provided) of how the "efficacy of aseptic procedures should be validated." This has been succeeded by the 1992, 1997 and 2003 editions of the Commission of the European Communities' Good Manufacturing Practice for Medicinal Products (CEC, 1992, 1997, 2003) which state that "validation of aseptic processing should include simulating the process using a nutrient medium." Should is a strongly directive verb in the language of these requirements.
In the last ten years media fills have, in the eyes of the regulatory bodies, developed from a reasonably good way of validating aseptic processes, through to the preferred way of validating aseptic processes, to an essential requirement of a properly validated aseptic process. It is now highly unlikely that any regulatory submission for a new aseptically filled sterile pharmaceutical product would be acceptable without supportive media fill data. It is also unlikely that a manufacturer of an existing aseptically filled sterile product would escape severe regulatory criticism if media fill data were unavailable.
It is now well accepted in the pharmaceutical manufacturing industry that validation is an exercise intended to confirm that a process is capable of operating consistently. As far as asepsis is concerned, the consistency of the contamination control "engineering" of a process is qualified by three successive replicate media fills done on separate days. Completion of the media fills is usually the factor that dictates the time of handover of the process for routine usage.
New aseptic processes require validation by media fill. Any process (irrespective of the equipment being old or new) beginning in a new clean room requires media fills as part of validation. A new filling machine in an established clean room requires validation media fills.
The trickier decisions arise over container sizes. It is quite probable that a range of container sizes may be filled on the same filling line. The question then arises over the necessity to perform media fills on all sizes, and in validation in particular, whether it is necessary to replicate each size through three media fills. The glib answer is that media fills should only be necessary for the container size that takes longest to fill and has the widest neck diameter. This combination presents the greatest potential for contamination and therefore addresses the contamination potential for all smaller sizes. However, this is not necessarily true. Wide-neck containers may be more stable than narrow-neck containers. Therefore the wide-neck filling process may be arguably less susceptible to contamination because there are fewer personnel intrusions necessary for rectifying fallen containers. Glass moulders often use a common neck or flange mold for different capacity vials; it would be usual for vials with capacities from 10 ml to 100 ml to have identical necks and flanges. There is probably no sensible way of rationalizing media fills to fewer than two container sizes on a multicontainer filling line. The decision over what and how many sizes to include in a media fill validation protocol is judgmental. For regulatory purposes the reasons for taking particular decisions must be justified and documented. If the rationale for performing media fills on more than one container size is based on the risks of contamination arising from a different source, or different balance of sources, rather than from a scale-up of risks from the same source, then it is logical that the three replicate media fills thought necessary to verify consistency of control must be performed on each container size.
The initial significant formality of validating media fills is the protocol, based on three principles.
1. The first principle of the protocol is that the process that is to be validated has to have been already defined and documented. In other words, draft operating SOPs have been prepared and personnel have been trained in them.
2. The second principle is that the test method, in this case the media fill, has been defined and documented. Importantly the protocol must define the number of units that are to be filled.
The minimum number of units expected is 3000. The origins of this figure are worth justifying. In principle, it is an expression of the minimum number of units for which a contamination rate of no more than one contaminated unit in 1000 units (0.1%) can be demonstrated with 95% confidence. But, why a contamination rate of no more than one contaminated unit in 1000 units (0.1%)? And why with 95% confidence?
In 1971, Tallentire and co-authors wrote that "sterility testing has several serious defects, not least amongst them being the high frequency of spurious results, sometimes called "false positives," due to contamination during testing. When measured using a population of items known to be sterile under best known test conditions, this frequency is approximately 1 in 103" (Tallentire et al., 1971).
The view that processes involving aseptic manipulation are limited by test-related contamination at or around a frequency of 1 in 1000 originate in this 1971 paper. The one in 1000 level also ties in with the regulatory expectation of sterility test failures within any particular laboratory as no greater than 0.5% of all tests conducted.
This assertion is based on the typical sterility test involving aseptic transfer from 20 product units; therefore 0.5% test failure represents aseptic transfer from 1000 units.
The contention was based on the technology of the 1970s. Asepsis has moved on considerably since then but the one in 1000 limit has become attached to media fills, probably because it is a practical benchmark for the number of units that can be filled or incubated, etc.
The PDA (PDA, 1981) supported a limit of no more than 0.1% contamination for media fills in its 1981 monograph (an unofficial voluntary standard). It added that this should be demonstrated with 95% confidence and that at least 3000 filled units are required to achieve this. No reason for choosing 95% confidence rather than 99% confidence or 90% confidence is given.
The idea of 3000 units and 95% confidence reappeared in the FDA 1987 Guide and has become part of the regulatory industry and expectation of media fills.
The association of 3000 units with 95% confidence of assuring a contamination rate of no more than 0.1% has been elaborated by Halls (1994), supported from two different mathematical positions.
• The PDA (1981) references the following equation of an "operating characteristic" curve to describe the probability of detecting one or more contaminated items in a sample size N taken from a population with a contamination rate of 0.1%:
When P(x>0) is made equal to 95%, this equation describes how large a sample size, N, needs to be taken from a universe in which there is 0.1% of contaminated units to find at least one contaminated unit on at least 95% of occasions when samples are taken. In practice, 95% confidence cannot be achieved with a sample size of less than 2996.
Alternatively, the measured contamination rate in a media fill may be regarded as an estimate of the true contamination rate (P) in the underlying population that may be higher or lower than the measured rate (Pest). The reliability with which Pest can be claimed to be a true reflection of P can be calculated from the confidence limits of Pest.
The 95% confidence limits around Pes, may be calculated from the expression,
Pest - hPestQes/N < P < Pest + PestQJN, where h is the number of standard deviations appropriate to particular confidence limits (1.96 for 95% confidence).
If 0.01% is regarded as the upper 95% confidence limit of the lowest measurable number of contaminants obtainable in a media fill (one contaminated unit) the lowest value of N can be calculated to be close to 3000 units.
The number of units (if any) in excess of the 3000 required to be filled is a further important decision, and there are several views on how it should correctly be made.
• One view holds that the number of units filled should be related to the product batch size. This is difficult to reconcile with the fact that the media fill is a process test and should not logically be related to product batching. Different products filled into the same containers on the same filling machine could easily have different batch sizes, perhaps dictated by some complexity of compounding. To which of these batch sizes should the number of media fill units be related? If this approach is taken, the pragmatic answer is usually the largest of the batch sizes. Guidance on media fill dimensions in relation to product batch size given in ISO/IS 13408 Aseptic Processing of Health Care Products (ISO, 1997) is summarized as Table 3.3. In practical terms this guidance applies only to small batch sizes; for normal production batch sizes ISO supports only a minimum media fill size of 3000 units.
Table 3.3 Minimum Numbers of Media Fill Units Related to Production Batch Size from ISO/IS
13408 Aseptic Processing of Health Care Products (ISO, 1997)
Table 3.3 Minimum Numbers of Media Fill Units Related to Production Batch Size from ISO/IS
13408 Aseptic Processing of Health Care Products (ISO, 1997)
Number of units in |
Minimum number of units |
Minimum number of units |
production batch |
for validation media fills |
for periodic media fills |
< 500 |
5000 in ten or more runs |
Maximum batch size per run |
> 500-2999 |
5000 in three or more runs |
Maximum batch size per run |
> 3000 |
9000 in three runs |
3000 per run |
• A second view is that the media fill should be run over the same time as an operating shift. In many cases this amount of elapsed time may be necessary to simulate all of the potential contaminating events arising in a process. In other cases, e.g., with high-speed ampoule filling lines, it could result in vast numbers of units being filled. As long as there are no contaminated units present, this approach to filling ampoules gives good assurance of asepsis. Its logic breaks down when contaminated units are identified; perhaps three or four contaminated units would be insignificant in comparison to the overall large numbers filled. Regardless, they may be significant to contaminating events that occurred during filling but the effect of the large dimensions of the media fill is to dilute their impact.
• The third view, and the one supported by the author, is that the dimension of the media fill should be dictated by the time necessary to allow simulation of all of the potential contaminating events. The identification of all potential contaminating events is in the long run a matter of opinion. Nonetheless, there are techniques such as failures modes and effects analysis that can be used to create a documented structure around the development of these opinions. This type of approach adds to the knowledge of the process if done properly. If done in a cursory manner it is open to abuse.
3. The third principle of the protocol is that acceptance criteria must be predetermined. In the case of media fills a maximum number of contaminated units must be specified for the media fill, and indeed the underlying aseptic process, to be acceptable. If the acceptance number is exceeded in any one of the three validation media fills, appropriate action must be taken and the media fill(s) repeated until three successive successful media fills are obtained. Ideally the appropriate action is preventive, i.e., action appropriate to preventing a further recurrence should be taken, probably involving some change in working practice and to the operating SOP.
In the real world the action is most often corrective — something like a re-disinfection of the filling room or retraining of personnel. This is because it is not usually easy to accurately diagnose the source of contamination in a media fill, and this difficulty is greatest for a new process (and one would expect validation to be done for a new process in its broadest sense).
The question arising out of the predetermination of acceptance criteria is exactly how many contaminated units are tolerable? This is not an easy question to answer.
As a starting point, if we take the statistic of no more than one contaminated unit in 1000 as the acceptance limit, and 3000 units as the minimum number of units in a media fill, then we might reasonably expect that zero, one, two or three contaminated units in 3000 would be acceptable, and four or more contaminated units unacceptable. Up to four contaminated units would be acceptable in 4000, up to five in 5000, etc.
This approach was overtaken by the PDA recommendation (1981) that the limit of no more than one contaminated unit in 1000 should be met with 95% confidence. In relation to 3000 units filled, compliance with this modification to the one-in-1000 limit would only be acceptable with zero or one contaminated units.
Slightly different (but in practical terms insignificant) mathematical treatments result in recommendations of "pass zero, fail one or more contaminated units in 3000" or "pass one or fewer contaminated units, fail two or more contaminated units in 3000."
When media fills require numbers of units larger than 3000 it might be considered reasonable to increase the number of contaminated units permissible beyond zero or one. The guidance in ISO/IS 13408 allows maximum numbers of contaminated units ranging from one in a 3000 unit media fill to 11 in a 17,000-unit (approximately) media fill (ISO, 1997).
Bernuzzi et al. (1997) examined these recommended limits and concluded that they became weaker as the total number of filled units increased. (One positive in a 5000-unit media fill does not have the same meaning as 10 units positive when 16,970 units are filled.)
These authors attempted to develop an alternative set of limits for media fills, but in all cases found the same statistical frailty as numbers of units filled increased. The limits from ISO/IS 13408 (ISO, 1997) and from the most rigorous plan of Bernuzzi et al. (1997) are summarised in Table 3.4.
Table 3.4 Maximum Permissible Numbers of Contaminated Units in Media Fill According to ISO/IS 13408 Aseptic Processing of Health Care Products (ISO, 1997) and to the More Rigorous Scheme of Bernuzzi et al. (1997)
Number of units filled Maximum permissible number of contaminated units
ISO/IS 13408 (1997) Bernuzzi et al. (1997)
3000 |
1 |
0 |
4750 |
2 |
- |
6300 |
3 |
- |
7200 |
- |
1 |
7760 |
4 |
- |
9160 |
5 |
- |
10520 |
6 |
- |
11500 |
- |
2 |
11850 |
7 |
- |
13150 |
8 |
- |
14440 |
9 |
- |
15710 |
10 |
- |
15800 |
- |
3 |
16970 |
11 |
- |
20200 |
- |
4 |
It is, however, all very well in principle and in statistics to present limits such as these. In practice it is unrealistic that, for example, a manufacturer of aseptically filled ampoules would repeatedly tolerate (or be allowed by the regulatory agencies to tolerate) six contaminated units in media fills of 10,000 units as these recommendations appear to suggest. It is also unrealistic that a manufacturer of blow-fill-sealed ampoules would repeatedly tolerate even four contaminated units in media fills of 20,000 units.
In practice any number of contaminated units in excess of zero or one would have to be investigated seriously by any conscientious pharmaceutical manufacturer, irrespective of the overall dimensions of the media fill. This is particularly true in validation. It is outside the experience and belief of the author that any ethical pharmaceutical manufacturer would approve validation of an aseptic process in which three or more contaminated units appeared in validation media fills.
Contaminated units are the stimulus for process improvement. The practical limit in all media fills is that there should be no more than one, possibly (in very unusual circumstances) two contaminated units. Larger numbers of contaminated units must elicit preventive action and improved control.
In summary, it makes the best sense that validation media fills should be composed of a number of units in excess of 3000 sufficient to allow for enough elapsed time to simulate all predicted potential contaminating events, and no more than one contaminated unit should be allowed in any single run no matter how many units are filled in total.
Was this article helpful?
One of the main home remedies that you need to follow to prevent gallstones is a healthy lifestyle. You need to maintain a healthy body weight to prevent gallstones. The following are the best home remedies that will help you to treat and prevent gallstones.