Microbial Recovery Studies – 50% or 70%?

Scott Sutton, Ph.D.
Vectech Pharmaceutical Consultants
This article first appeared in the PMF Newsletter of August, 2007 and is protected by copyright to PMF.  It appears here with permission.

Introduction

The PMFList is a source of great ideas for review and for further thought.  One that keeps coming up on the list is the question of 70% recovery (as described in USP chapter <1227> Validation of Microbial Recovery from Pharmacopeial Articles) and 50% recovery as described in the harmonized chapter <61> Microbiological Examination Of Nonsterile Products: Microbial Enumeration Tests.

The questions and discussion seem to fall into two distinct groups – the first a discussion about when to apply 70% and when to apply 50% as your recovery criteria (with frequent complaints about the inferred lack of consistency in USP) and the second a discussion of what types of tests we are talking about.    We will look at these issues separately. 

What are we talking about?

The first thing to do is to establish the scope of the discussion.  For starters, let’s begin by stating that the compendial chapters are, by definition, validated.  This refers to those chapters in the USP that number under 1000.   We therefore cannot really “validate” the test method, instead we are trying to demonstrate the suitability of the recovery method.  This has been referred to as “verification” (Porter 2007) and in the harmonized Microbial Limits chapters as “method suitability.” 

The point of a method suitability study in microbiology is not to validate the assay, but rather to demonstrate that our specific test method is suitable; that the recovery scheme allows recovery of viable microorganisms.  In other words, microorganisms are not prevented from growing in the experimental system by residual antimicrobial activity of the product

 This demonstration is critical in accurate determination of disinfecting efficacy, bioburden, sterility or any test that requires determination of surviving microorganisms in a product containing antimicrobial properties.  Failure to confirm adequate neutralization and recovery could result in under-reporting of surviving microorganisms.   This expectation of 70% recovery can also be applied to media growth promotion studies, where a new batch of media is compared to a previously qualified batch for its ability to support at least 70% of a standard inoculum.

A convenient method for this neutralization is through the use of recovery diluents designed to neutralize commonly used antimicrobials.  A number of reagents are used in this regard (reviewed by Russell 1981; Furr & Rogers 1987). However, some of these compounds may be toxic to the test organisms (Reybrouck 1978) and so it is also important to determine the potential toxicity of the neutralizing medium (recovery diluent). These two activities, neutralizer efficacy and growth promotion (or neutralizer toxicity), are equally important in this consideration.  A schematic of a design for this type of study is presented below, where a consistent inoculum is added to the product in the recovery diluent, peptone in the recovery diluent (use the same volume of peptone as that of the product), and into peptone.  These are then plated 5-6 times to provide a good estimate of the number of organisms present (Wilson and Kullman 1931).   The Neutralizer Efficacy is determined by comparing the recovery in the peptone suspension to that in the Product + Recovery Diluent suspension, Neutralizer Toxicity by comparing the Peptone suspension to the Peptone + Recovery Diluent (USP 2007a).

What is not part of this discussion?

It should be obvious from the previous discussion that the “method suitability” study is highly controlled.  A standard inoculum is added to three tubes, and then replicate aliquots are removed and immediately plated.  In a perfect world the numbers would be in agreement 100% of the time, but we work in microbiology.  Even in such a simple design the opportunity for variability is enormous, and there are workers in the field who are vehement that no better than 50% should be expected between replicates of this type.  One wonders if this is a limitation of the test system or of their laboratory training program.  In any event, the discussion of 50% to 70% between the populations applies only to this design (and those closely related to it).

 The recommendation in USP of 70% recovery was never meant to apply to studies of microbial recovery from solid surfaces.  These studies are extremely complicated, and are confounded by issues of recovery efficacy of swabs, contact plates, and other methods (Buggy, et al 1983, Rose et al 2004, Whyte 1989).  In addition, if vegetative cells are used for the study, there is the additional problem of die-off due to dessication (Potts 1994). 

 Recovery studies looking at bioburden of solid surfaces (facility, equipment, medical device or personnel) are not part of the 50% to 70% debate.  They have their own set of issues and will be discussed in a later newsletter.

Is there any support for these numbers?

 There are two studies which directly support the 70% recovery acceptance criterion.

Proud and Sutton (1992) describe the development of a “universal” diluting fluid for membrane filtration sterility testing using a modification of the design described above.  The product was placed in a filtration apparatus containing 100 mL of the diluting fluid, and then passed through the membrane, followed by two additional 100 mL rinses.  The membrane was then removed and placed on the surface of a nutrient agar plate for incubation and enumeration.  Each treatment was performed at least three times.  CFU were converted to their log₁₀ values, and ANOVA analysis performed on the replicates.  When all was said and done, a recovery of 75% of the inoculum count (raw CFU – untransformed) passed the ANOVA analysis. 

 Sutton, et al.  (2002) conducted a large study on methods to recover microorganisms in the presence of surface disinfectants.   “Neutralizer efficacy (NE) ratios were determined [in this study] by comparing the recovery of identical inocula from the neutralizing solution in the presence, or the absence, of a 1:10 dilution of the biocide. Neutralizer toxicity (NT) ratios were determined between recovery of viable  microorganisms incubated for a short period in peptone, and in the neutralizing medium without the biocide. An effective and non-toxic neutralizer was initially identified by NE and NT ratios of ≥ 0.75. Statistical evaluation of the data was performed by ANOVA, with Dunnett’s test for multiple comparisons used to confirm failures. By this analysis, 239/244 identified failures were confirmed by ANOVA of 588 NT and NE comparisons (5 presumptive failures were not confirmed by statistical analysis). We therefore conclude that recovery of 75% is a suitable criterion (2% false negative rate) for neutralizer evaluations.”

A side issue to this discussion is the occasional use of 70-130% recovery as the acceptance criteria.  I have trouble with this one – would you really disqualify a method because it improves your recovery over expectations?  In my opinion the acceptance criteria should be that the test treatment should recover at least 70%, with no consideration of recovery by the test in excess of the comparator treatment.

Which should you use?

 I am of the opinion that 70% is easily attainable if the technicians are proficient and the recovery method works.  This may require 5-6 replicates, rather than the usual duplicate plates per sample.  However, this is a “verification” study or a “method suitability” study (or whatever we wish to call it) and so may be worth a bit more work. 

 So, how did they get different criteria in the USP?   Chapter <1227> was developed to address a specific concern – that of providing information on microbial recovery studies (not limited to neutralizer efficacy) for use in the pharmaceutical industry.  This work was well in progress by 1996 (USP 1996).  The harmonization program discussed this point much later, and after negotiation agreed to the 50% so that agreement could occur.  No data was presented to support the assertion that 50% was appropriate (by my records), it was, however, the number that could be accepted.

 The harmonized USP chapter <61> (USP 2006b) cites a 50% recovery frequency and so this is the official acceptance criteria for this test.  If you wish to use 50% for the acceptance criteria for all method suitability studies (non-compendial bioburden tests, method suitability studies for disinfectancy tests, Antimicrobial Efficacy tests, media growth promotion, etc) I would strongly urge a solid rationale for failing to observe the recommendation of chapter <1227>.  In addition, I would be prepared to answer questions of technician proficiency as the suspicion may be that your lab is not confident of reproducibility to 70% even between identical samples. 

References

Buggy, B. et al/i>.  1983.  Comparison of Methods for Recovery of Clostridium Difficile From an Environmental Surface.  J Clin Microbiol.  18(2):348-352.