Establishing Scientifically JustifiedAcceptance Criteria for CleaningValidation of Finished Drug Products

II>

~- .~~~L m

This article presents an approach to calculatinglimits for cleaning validation based on calculatinglimits first in the subsequently manufacturedproduct, then on the cleaned equipment surfaces,and finally in the sample actually analyzed by theappropriate analytical technique. The article dis-cusses issues in detennining each of these limitsand covers related issues including recoveryfactors, visual cleanness, nonunifonn contamin-ation, and microbiological contamination.

,.". A. L..,.., is vice-pnsident of scientific technical sup-port at STERIS Corporation, PO Box 147. St. Louis. MO63166. tel. (314) 535-3392. fax (314) 535-2998, e-mail( destin_leblanc@steris.com).

ne key to effective cleaning validation is determin-ing: How clean is clean enough? This is usually de-tennined by establishing specific analytical limitsfor target residues. The limits chosen when setting

acceptance criteria must be scientifically justified. Arbitrarysetting of limits is just that, arbitrary, and may raise concernduring an FDA audit. This situation is somewhat complicatedby the fact that the term limit sometimes is used loosely to re-fer either to the acceptance limit in the next product, to theacceptance limit of surface contamination, or to the acceptancelimit of the analyzed sample. Although all of these limits areinterrelated, they are not necessarily measured in the sameunits nor are they of the same magnitude. For example, con-tamination of the next product (the product subsequentlymanufactured in the cleaned equipment) typically is given inparts per million or micrograms per gram. surface contamina-tion usually is measured in micrograms per square centime-ter, and the analyzed sample typically is given in microgramsor micrograms per gram.

It should be clear that limits per surface area are differentunits and cannot be compared directly without other pieces ofinformation such as batch size and equipment surf~ area. Inaddition, limits in the subsequent product may be given in thesame units as limits in the analyzed sample, but they also arenot comparable without other information such as areaswabbed and swab recovery factor. This means that an accep-tance limit of 3.2 ppm in the subsequent product is not neces-sarily the same as 3.2 ppm in the analyzed sample preparedby a swabbing procedure. One must make these proper dis-tinctions when discussing residue limits so that, for example,analytical methods are properly validated near the appropriatelimits for the residue in the analyzed sample. This article dis-cusses the calculation of residue limits for the different parts ofa validated system and explores the limitations and applicabil-ity of such calculations as applied to finished drug products.

One of the main objectives of the cleaning process in drugmanufacture is to remove residue of the just-manufacturedproduct so that those residues are not transferred to the sub-

. '"

sequently manufactured product. A key complicating feature ofcleaning is that it involves not only the product being cleanedbut also the product subsequently manufactured in the cleanedequipment. The starting point for any determination of accep-tance residue limits is the amount of residue from the cleaningprocess that could be present in the subsequently manufacturedproduct without posing an unreasonable risk. Clearly, one wouldprefer that no residue be present; however, it is impossible tomeasure "no" residue. Even the criterion of being below the lim-its of detection of the analytical procedure is not by itself a verygood method for selecting residue limits. As methods are im-proved and have even lower limits of detection, a cleaningprocess that was previously viewed as acceptable can becomeunacceptable. Therefore, it is preferable to express that limit as,for example, ~8 ppm, rather than specifying "undetected:'

FDA? s guidance for detennining residue limits dictates that lim-its be logical (i.e., based on an understanding of the process),practical, achievable, and verifiable (1). Fortunately for those in-volved in cleaning validation, the same FDA document mentionslimits proposed by industry representatives "such as 10 ppm, bio-logical activity levels such as 1/1000 of the normal therapeuticdose, and organoleptic levels such as no visible residue:' This isclearly a reference to work:. done by Fourman and Mullen at EliLilly (2), whose paper is listed in the reference section of the FDAguidance document. Although it is not officially endorsed byFDA, the Lilly method of establishing residue limits (or somevariation of it based on the same principles) is widely used withinthe pbannaceutical industry for detennining acceptable levels ofchemical residues (3-5).

The published Lilly criteria are that. the equipment be visually clean. any active agent be present in a subsequently produced prod-

uct at maximum levels of 10 ppm. any active agent be present in a subsequently produced product

at maximum levels of 1/1 (XX) of d1e minimum daily dose of theactive agent in a maximum daily <k>se of d1e subsequent productAlthough Fourman and Mullen directly calculate the surface

area contamination based on these latter two criteria, the analy-sis in this article uses the same assumptions to arrive separatelyat subsequent product residue limits, surface area residue limits,and analytical sample residue limits. Such a discrete analysismay be more conducive to understanding contributions to resi-due limits from various factors.

UMIT .1 SUBSEQUENT PRODUCT

To calculate the limit of the active agent in any subsequently

manufactured product (LSP), the only infonnation needed is the

minimum daily dose of the active being cleaned and the maxi-

mum daily dose of the subsequently manufactured drug product.

For purposes of illustration, product A will refer to the product

being cleaned, and product B will refer to the subsequently pro-

duced product. The LSP (L]) can be expressed as

L _ (0001 min dailydoseofactiveioproductA-. )X [1.11 max. dailydoseofproductB

For an example using orally dosed liquids, assume that prod-

oct A has an active at a level of 2000 ~g/mL and is dosed at 5 mLfrom dtree to five times daily. The minimum daily dose would be

(2<XX> ~g/mL) X (5 mUdose) X (3 doses/day) =30,000 ~g/day

If one also assumes that product B is dosed at 5 mL from twoto four times a day, then the maximum daily dose of product Bwould be

5 mUdose X 4 doses/day = 20 mUday

Note that this calculation for the subsequent product is inde-pendent of what the active is or at what level that active is pre-sent. The residue limit in the subsequent product can be calcu-lated by substituting these values into Equation I:

0.001 x 30,000 JA.g/day = 1.5L I = 20 mL/day ~g/mL

This value of 1.5 jJ.g/mL (or -1.5 ppm, assuming a specificgravity of 1.0) is independent of both batch size and equipmentsurface area. Therefore this calculation can be done as soon asinformation is available about the composition and relevant dos-ing of the two products. The calculated value of 1.5 ppm shouldbe compared to the 10 ppm default value from Lilly's criterion,and the lower of the two values should be used in subsequentcalculations. For this example, 1.5 ppm would be chosen forsubsequent calculations.

One could select safety factors other than 0.00 1. For example,safety factors of 0.001 for orally dosed products and 0.0001 forparenterals have been suggested (6). Although more stringentsafety factors may easily be justified, use of a safety factor lessstringent than 0.001 would probably require significantjustifica-tion. However, different safety factors may be appropriate if theyare applied to dosing factors other than the minimum dailypharmacological dose. In addition to different safety factors,companies may also base limits on parameters other than thera-peutic doses. For example, rather than using the minimum dailydose of the active, firms may select other measures, such as theno observable effect level (NOEL) or the minimum pharmaco-logical effect level. Because the use of these factors will resultin residue limits more stringent than limits based on the mini-mum daily dose, there is no scientific reason not to select thesecriteria. It should be noted, however, that arbitrarily selectingmore stringent criteria may result in unreasonable cleaningoverkill and may also stretch the detection limits of availableanalytical methods.

UMIT PER SURFACE AREA

Once the residue limit in the subsequent product is determined(using the lower values in Lilly's second and third criteria), thenext step is to determine the residue limit in terms of active in-gredient contamination level per surface area of equipment. Thislimit (L2, in ~g/cm2) depends on LSP (determined as the lowerof L) and 10 ppm), the batch size of the subsequent product B(BSSP, in kg), and the shared equipment surface area (SESA. incm2). This can be expressed mathematically as

L = LSP X BSSP X 1<XX>2 SESA [2]

where 1(0) is a conversion factor to account for ppm (the limit

of the subsequent batch) and to convert kg to ILg (for BSSp). Forthis example, if LSP is 1.5 ppm and assuming BSSP is 200 kgand SESA is 60,000 cm2, then the ~ limit is

~g/cm2L2 = 1.5 ppmx200q X 1(XM) =S.O

(i).(XX) cm2

When determining the surface area, one should consider allshared product contact surfaces, including piping, baffles, etc.Estimates of surface area should be on the high side becauselarger surface areas result in lower ~ residue limits.

Note that this calculation for L2 assumes that the residue willbe evenly distributed over all surfaces. This is generally not thecase. However, this assumption still represents the worst casein the sense that if one is doing swab sampling on the most dif-ficult to clean locations, then the assumption of an even distrib-ution will make it more difficult to meet the residue limits forthose most difficult to clean locations. In addition, it would bedifficult to scientifically justify one area of the equipment hav-ing a residue limit of lO l1.g/cm2 and other areas having limits of2.5 l1.g/cm2. If a true sampling rinse is used, then issues of dis-tribution do not arise because one is sampling the entire equip-ment surface (7).

If more than one product (e.g., products B, C, D, and E) couldpossibly be manufactured following product A, then surface-area limits (Lv for cleaning following product A should be cal-culated for each subsequent product. For purposes of cleaningvalidation, the residue limit should be set at the lowest of theseL2 surface-area limits. This gives the manufacturing depart-ment greater flexibility to make products in any order. It shouldbe recognized, however, that there may be circumstances inwhich residue limits result in restrictions on which productscan follow product A on the manufacturing schedule.

UMIT II THE ANALYZED SAMPLE

Although one establishes residue limits for the active in the sub-sequent product (LJ) and for the active per surface area followingcleaning (L2), these limits typically are not measured directly bythe analytical procedure. The analytical procedure typically mea-sures the active agent in solution as a result of either swabbingand desorbing that swab into a suitable solvent or by doing rinsesampling and measuring the active in the rinse solvent (8). Forpurposes of expediency, this article focuses on swab sampling.The reader is referred to other sources for examples related torinse sampling (7).

For swab sampling, one assumes that a fixed surface area ofthe equipment is sampled and the swab is then desorbed into afixed amount of solvent. To determine the residue limit (Lj. inmicrograms per gram or parts per million) for the analyticalsample (the solvent the swab is desorbed into), one must knowthe surface-area residue limit (L2)' the surface area swabbed (insquare centimeters), and the amount of solvent the swab is des-orbed into (in grams). The limit in the analyzed sample is

L = L 2 x swabbed surface areaj amount of desorption solvent

Continuing with the example used so far. if ~ is 5.0 JA.g/cm2. the

surface area swabbed is 25 cm2, and the amount of solvent usedfor desorption is 20 g, then the limit LJ is

- 5.0 ~p2 x 25 cm2 = 6.3 'tJ.g/g or6.3 ppmLJ - 2Og

RECOVERY FACTORS

The limit LJ is the acceptance limit in the analyzed sample. Itshould be adjusted by a swab recovery factor, which can be donein two ways. One method is to include the swab recovery fac-tor in the actual analytical calculation by dividing the analyti-cal result by the recovery factor. For example, if the swab re-covery factor is 0.80 (80%) and the analytical procedure givesa value of 1.3 ppm, then the limit should be adjusted by dividingthe analytical result by the recovery factor to arrive at a deter-mination of 1.3 ppm -':- 0.80 = 1.6 ppm. The alternative is to

include the recovery factor in Equation 3. In this case the re-covery factor of 0.80 should be included in the numerator. Al-though the numbers used will be different, the net effect of com-paring the analytical result to the calculated limit will belogically the same. One should standardize how this will be per-formed in order to avoid situations in which the recovery factoris used in the calculations of both the LJ limit and the determi-nation of the analytical result on the desorbed solvent sample.

EFFECTS 01 AlALYTICAL METIIOD VALIDAnOIIt should be noted that, at least in this example, LSP (L), 1.5ppm) was significantly different from the limit in the analyzedsample (y, 6.3 ppm). This is because L) reflects the residue ofthe active being evenly distributed in a batch of the subsequentproduct, whereas L 3 reflects the residue of the active concen-trated, in most cases, in a smaller volume of matrix material (thedesorbing solvent). One can see that this effect can be leveragedby either sampling a larger surface area or by desorbing the swabinto a smaller amount of solvent. When sampling a larger area,one must consider whether sampling this larger area might alsolower the swab recovery factor, thus adding more uncertainty tothe determination. This is significant for analytical method pur-poses because the analytical method chosen for determining theresidue of the active agent should be validated (at least in theexample used) not based on the LJ limit of 1.5 ppm but ratheron the L3limit of 6.3 ppm adjusted appropriately by the recoveryfactor. In this case, one might validate dte analytical method notin the range of 0.5 to 1.5 ppm but rather in the range of 2.1 to6.3 ppm. This fourfold factor in limit of quantitation may besignificant for analytical method selection and validation.

VISUAL CLEAMESS

The issue of visual cleanness is also significant If a surface isvisually dirty, either the cleaning procedure is not acceptable oran acceptable procedure is now out of control. The standard ofvisually clean can be used for purposes of both validation andmonitoring. The dividing line between visually clean and visu-ally dirty is regarded as being in the range of 4 ~g/cm2 (2). If theL2 surface contamination acceptance limit is calculated and isfound to be significantly higher than 4 ~g/cm2. and providingthat critical surfaces are readily visible. it may be possible to de-

Pl

fault to visually clean as the only acceptance criteria. For po-tent drugs for which the L2 acceptance criteria would typicallybe well below I ~g/cm2, a determination of visual cleannesswould have no significance regarding the adequacy of clean-ing. In such a case, a visually dirty surface would still be an in-dication of cleaning failure, but a visually clean surface couldnot clearly indicate whether the residue was at an acceptablelevel.

For cases in which a determination of visually clean is crit-ical, it may be appropriate to actually determine the highestlevel at which the specific residue is not visible. This can bedone by spiking model surfaces (e.g., stainless steel coupons)with different levels of the residue (e.g, 0.5. 1.0,2.0,4.0, and8.0 ~g/cm2) and asking a trained panel to observe the couponsin a "blinded" manner to determine whether the coupons arevisually clean. This should be done under viewing conditions(of lighting, angle, and distance) that simulate the viewing ofactual equipment. The highest residue level at which all panelmembers consider the coupons to be visually clean estab-lishes the acceptance level for visual cleanness for that par-ticular residue.

NONUNIFINIM COITAMllAno.The cleaning validation acceptance criteria used by Lilly and re-ferred to in the FDA guidance document on cleaning validation doprovide one logical construct for detennining residue limits forchemical residues of active ingredients in finished drug manufac-ture. One issue that may arise involves possible nonuniform con-tamination of the subsequent product (9). It should be noted thatthis is different from (although it may be related to) nonunifonncontamination of the cleaned equipment Nonunifonn contamina-tion of dIe subsequent product is more likely to arise during con-tinuous processes rather than batch processes. For example, in abatch process involving liquid drug blending, dIe contaminatingresidue (provided dIere is adequate mixing) is likely to be evenlydistributed throughout the subsequently manufactured productIn a continuous process, such as flow through a piping system andthrough a filling nozzle during a packaging operation. it is morelikely that the contaminating residue would appear in the first vialsfilled. This, of course, depends on the solubility of the residue inthe subsequently manufactUred product and on product flow char-acteristics. A special case of a continuous process is a tablet press,because any residues on dIe surfaces of dIe press are not likely tobe distributed evenly over all tablets manufactUred in a batch.

In dealing with nonuniform contamination at least two alter-natives are possible. If it is reasonable that the contaminatingresidue would appear in the first portion of a product batch (offilled vials, for example), then calculations can be made to deter-mine the maximum number of vials filled that could theoreti-cally be at the maximum allowable contamination level. Thosevials, plus as a safety measure a reasonable number of subse-quently produced vials, should be discarded or destroyed. Forexample, suppose the acceptance limit for the target residue is5 ppm in the subsequently manufactUred product filled as 5-mLvials. If the calculated residue levels in the filling equipment andassociated piping were 0.3 IJ.glcm2 (determined from a swab-

bing procedure, for example), and if the surface area of the fill-

ing equipment and associateAi piping is 5(XX) cm2, then if all of the

contamination in the equipment were concentrated in the first vial

produced, the concentration of residue in that one vial would be

0.3 fJ.8/Cm2 X S<XX}Cm2 = 300 fA.g/mL or -300ppm

SmL

This would clearly be above the L I acceptance limit of 5 ppm. Ifall the contaminating residue in the equipment were evenly di-vided in the first 60 vials, then each vial would contain 5 ppm ofresidue; therefore those 60 vials and a reasonable number ofsubsequently produced vials should be discarded.

These determinations were based on theoretical considera-tions of what could happen, and thus they represent a worstcase. If the number of vials to be discarded is unacceptablylarge, a study could be done to deal with the nonuniform con-tamination (of the filled vial, in this example). This studywould involve cleaning the filling equipment and then filling aplacebo product. Vials number 1, 10,20, and so forth wouldthen be analyzed for the target residue. If vial number 10 wereat 6 ppm and vial number 20 were at 2 ppm, then those first20 vials and a reasonable number of subsequently producedvials should be discarded.

MICROBIOLOOICAL CONTAMIIATiOIA second issue in setting limits is what to do about microbio-logical contamination, for which setting acceptable limits ismore difficult. The main issue is that merely one organism inthe equipment could possibly result in a significantly highercontamination level in the product manufactured in that equip-ment. No clear guidelines exist for microbiological contami-nation of process equipment. This may be the reason why theFDA guidance document explicitly states that the document ap-plies only to chemical residues (1). One cannot expect theequipment to be free of all microorganisms, especially if any fi-nal rinse involves non sterile water, unless a final sanitizing orsterilization step is used. As a minimum, one should employthe criteria used for critical cleanroom surfaces (10). A secondconcern is with the species of microorganism present. Obvi-ously, the presence of enteric organisms such as E. coli or En-terococcus should ordinarily be unacceptable.

Other concerns beyond the scope of this article include set-ting limits for residues from chemical sources other than activeingredients (e.g., from excipients or cleaning agents), settinglimits for residues in active phannaceutical ingredients, and ac-counting for residues from multiple process steps. However,consideration of residue limits in these cases should be based onthe same principles discussed here. The first consideration is thepossible effects of any residue when it is present in any subse-quently manufactured finished drug product.-It is then possibleto work backwards to determine limits for equipment surfacesand/or bulk actives. Any residue limit determination should bebased on considerations and principles similar to those usedfor finished drug products, modified as they apply to those situ-ations. The key, as in any validation activity, should be the useof sound scientific and logical reasoning.

fault to visually clean as the only acceptance criteria. For p0-tent drugs for which the L2 acceptance criteria would typicallybe well below 1 ~g/cm2, a determination of visual cleannesswould have no significance regarding the adequacy of clean-ing. In such a case, a visually dirty surface would still be an in-dication of cleaning failure, but a visually clean surface couldnot clearly indicate whether the residue was at an acceptable

level.For cases in which a determination of visually clean is crit-

ical, it may be appropriate to actually determine the highestlevel at which the specific residue is not visible. This can bedone by spiking model surfaces (e.g., stainless steel coupons)with different levels of the residue (e.g, 0.5,1.0,2.0,4.0, and8.0 ~g/cm2) and asking a trained panel to observe the couponsin a "blinded" manner to determine whether the coupons arevisually clean. This should be done under viewing conditions(of lighting, angle, and distance) that simulate the viewing ofactual equipment. The highest residue level at which all panelmembers consider the coupons to be visually clean estab-lishes the acceptance level for visual cleanness for that par-

ticular residue.

NONUNIFORM COITAMllAnO.The cleaning validation acceptance criteria used by Lilly and re-

fen-ed to in the FDA guidance document on cleaning validation do

provide one logical construct for determining residue limits for

chemical residues of active ingredients in finished drug manufac-

ture. One issue that may arise involves possible nonuniform con-

tamination of the subsequent product (9). It should be noted that

dlis is different from (although it may be related to) nonuniform

contamination of the cleaned equipment. Nonuniform contamina-

tion of the subsequent product is more likely to arise during con-

tinuous processes rather than batch processes. For example, in a

batch process involving liquid drug blending, the contaminating

residue (provided there is adequate mixing) is likely to be evenly

distributed throughout the subsequently manufactured product.

In a continuous process, such as flow through a piping system and

through a filling nozzle during a packaging operation, it is more

likely that the contaminating residue would appear in the first vials

filled. This, of course, depends on the solubility of the residue in

the subsequently manufactmed product and on product flow char-

acteristics. A special case of a continuous process is a tablet press,

because any residues on the swfaces of the press are not likely to

be distributed evenly over all tablets manufactured in a batch.

In dealing with nonuniform contamination at least two alter-

natives are possible. If it is reasonable that the contaminating

residue would appear in the first portion of a product batch (of

filled vials, for example), then calculations can be made to deter-

mine the maximum number of vials filled that could theoreti-

cally be at the maximum allowable contamination level. Those

vials, plus as a safety measure a reasonable number of subse-

quently produced vials, should be discarded or destroyed. For

example, suppose the acceptance limit for the target residue is

5 ppm in the subsequently manufactured product filled as 5-mL

vials. H the calculated residue levels in the filling equipment and

associated piping were 0.3 ~g/cm2 (determined from a swab-

bing procedure, for example). and if the surface area of the fill-

ing equipment and associated piping is 5<XX> cm2, then if all of the

contamination in the equipment were concentrated in die first vial

produced, the concentration of residue in that one vial would be

0.3 ~g/cm2 X 5<XX> cm2300 g/mL 300= ~ or- ppm

5mL

This would clearly be above the L 1 acceptance limit of 5 ppm. Ifall the contaminating residue in the equipment were evenly di-vided in the first 60 vials, then each vial would contain 5 ppm ofresidue; therefore those 60 vials and a reasonable number ofsubsequently produced vials should be discarded.

These determinations were based on theoretical considera-tions of what could happen, and thus they represent a worstcase. If the number of vials to be discarded is unacceptablylarge, a study could be done to deal with the nonuniform con-tamination (of the filled vial, in this example). This studywould involve cleaning the filling equipment and then filling aplacebo product. Vials number 1, 10,20, and so forth wouldthen be analyzed for the target residue. If vial number 10 wereat 6 ppm and vial number 20 were at 2 ppm, then those first20 vials and a reasonable number of subsequently producedvials should be discarded.

MICROBIOLtMIICAl CONTAMIUnoNA second issue in setting limits is what to do about microbio-logical contamination, for which setting acceptable limits ismore difficult. The main issue is that merely one organism inthe equipment could possibly result in a significantly highercontamination level in the product manufactured in that equip-ment. No clear guidelines exist for microbiological contami-nation of process equipment. This may be the reason why theFDA guidance document explicitly states that the document ap-plies only to chemical residues (1). One cannot expect theequipment to be free of all microorganisms, especially if any fi-nal rinse involves nonsterile water, unless a final sanitizing orsterilization step is used. As a minimum, one should employthe criteria used for critical cleanroom surfaces (10). A secondconcern is with the species of microorganism present. Obvi-ously, the presence of enteric organisms such as E. coli or En-terococcus should ordinarily be unacceptable.

Other concerns beyond the scope of this article include set-ting limits for residues from chemical sources other than activeingredients (e.g., from excipients or cleaning agents), settinglimits for residues in active phannaceutical ingredients, and ac-counting for residues from multiple process steps. However,consideration of residue limits in these cases should be based onthe same principles discussed here. The first consideration is thepossible effects of any residue when it is present in any subse-quently manufactured finished drug product. It is then possibleto work backwards to determine limits for equipment surfacesand/or bulk actives. Any residue limit determination should bebased on considerations and principles similar to those usedfor finished drug products, modified as they apply to those situ-ations. The key, as in any validation activity, should be the useof sound scientific and logical reasoning.

REF£REICES1. FDA, "Guide to Inspections of Validation of Cleaning Processes,"

Division of Investigations, Office of Regional operations, Office ofRegulatory Affairs, July 1993.

2. G.L. Fourman and M. V. Mullen, "Determining Cleaning ValidationAcceptance Umits for Pharmaceutical Manufacturing Operations,"Phann. Techno/. 17 (4), 54-60 (1993).

3. K.M. Vitale, "Cleaning Validation Acceptance Criteria," presentedat the 15th Annual Pharm Tech Conference '95, East Brunswick,New Jersey, 18-21 September 1995.

4. R. Brewer. "Regulatory Aspects of Cleaning Validation," presentedat lSPE seminar, Rockville. Maryland, 6-8 March 1996.

5. PDA Biotechnology Cleaning Validation Subcommittee, Cleaningand Cleaning Validation: A Biotechnology Perspective (PDA,Bethesda, MD, 1996).

STERIg.Corporation

6. W.A. Hall, "Cleaning for Bulk Pbamlaceutical Chemicals (BPCs),"in Validation of Bulk Pharmaceutical Chemicals, D. Harpaz andI.R. Barry, Eds. (Intet"pharm Press, Buffalo Grove, u.., 1997), pp.335-370.

7. D.A. leBlanc, "Rinse Sampling for Cleaning Validation Studies,"Pharm. TechnoL 12 (5), 66-74 (1998).

8. R.B. Kirsch, "Validation of Analytical Methods Used in Pharma-ceutical Cleaning Assessment and Validation," in 1998 AnalyticalValidation in the Pharmaceutical Industry, supplement to Pharm.TechnoL 40-46 (1998).

9. K.M. Jenkins and A.J. Vanderweilen, "Cleaning Validation: AnOverall Perspective," Pharm. TechnoL 18 (4),60-73 (1994).

10. R. Dabbah, "Microbial Evaluation of Cleanrooms and Other Con-trolled Environments - In-Process Revision," Pharm. Forllm 23

(1),3494-3520 (Jan.-Feb. 1997).D

S TEl S@

. 5960 Road. , OH . 44060-1834 . USA. 440-354-2600Heisley

.10-800-0015 (10/98)