t is very important to biopharma-ceutical manufacturing that processingequipment be cleaned using validatedprocedures to avoid cross

contamination from material remaining froma previously manufactured product (13).Potential residues remaining on the surfacesof equipment following processing steps(fermentation, for example) include, inaddition to the product itself, a multitude oforganic cellular components such as nucleicacids, proteins, and carbohydrates.

Together with appropriate surfacesampling procedures, analytical techniquesthat can determine trace levels of

contaminants are required to support thevalidation of equipment cleaningprocedures. Currently, the preferred methodof sampling is to directly swab theequipment surface (4). In this approach, theswab head is wetted with an effective agentto facilitate transfer of contaminants fromthe surface to the swab, a defined area of theequipment surface is swabbed, andcontaminants acquired by the swab areextracted into solution before analysis.Analytical methodology is then needed todetermine trace levels of contaminants in thepresence of both a solubilizing medium anda swab.

Total organic carbon analysis (TOC) is amethod that has been used successfully formonitoring water quality (57), includingthe quality of water for injection inpharmaceuticals (8). TOC analysis offers anumber of distinct practical advantages overother commonly used residual testingmethods such as enzyme-linkedimmunosorbent assays (ELISA) and proteinassays (Lowry, Bradford, and bicinchoninicacid, for example) because of its highsample throughput, lack of interferingsubstances, and inherent sensitivity. Onestudy demonstrating the potentialapplication of TOC for the analysis of rinsewater samples for cleaning validation

determined trace levels of proteins,nucleic acids, amino acids, sugars, anddetergents (9).

In this article we present a procedure forthe TOC analysis of swab samples for thecleaning validation of fermentationequipment used in the production ofbiopharmaceuticals. This procedure is basedin part on a method developed previouslyfor the colorimetric determination of totalprotein in swab samples (10).

ExperimentalMaterials. Potassium biphthalate and sodiumhydroxide were purchased from Sigma (St.Louis, MO) and Mallinckrodt (Paris, KY),respectively. Texwipe Alpha Swab polyesterswabs were acquired from Baxter ScientificProducts (McGaw Park, IL), and E. colicells were obtained from Eli Lilly andCompany (Indianapolis).

It is important to note that the heads ofthe polyester swabs used in this study werethermally bonded to the handles, eliminatingthe need for adhesives that can causecontamination during extraction. The swabswere also laundered by the manufacturer tominimize inherent nonvolatile residues orparticulates that might otherwise havesignificantly decreased the sensitivity of theanalysis (11).

Equipment. The TOC analyzer used in thisstudy was a TOC-5000 (Shimadzu,Columbia, MD) equipped with a 74-position

Mark A. Strege, Terry L.Stinger, Brett T. Farrell, andAvinash L. Lagu

Total Organic Carbon Analysis of SwabSamples for the Cleaning Validation ofBioprocess Fermentation Equipment

Corresponding author Mark A. Strege is anassociate senior analytical chemist, Terry L.Stinger is a technician, Brett T. Farrell is anassociate senior analytical chemist, and Avinash L.Lagu is a senior research scientist at Lilly ResearchLaboratories, Lilly Corporate Center, Indianapolis,IN 46285, (317) 276-9116, fax (317) 276-5499.

I

Validated cleaning procedures areneeded to ensure the absence ofcontaminants from bioprocessingequipment, and these proceduresmust be supported by appropriateanalytical methodology. Thisarticle describes the developmentof a quantitative total organiccarbon (TOC) assay for residualcarbon-containing materials onstainless steel surfaces usingE. coli cells as a model substance.

Figure 1. Plot of TOC-measured carbonconcentration compared with preparedcarbon concentration (in aqueous sucrosesolutions) that demonstrates methodlinearity from 0 to 30 g/mL carbon.

Accuracy. The accuracy of an analyticalmethod is the closeness of test resultsobtained by that method to the true value.Accuracy may often be expressed aspercent recovery by the assay of known,added amounts of analyte. Accuracy is ameasure of the exactness of the analyticalmethod.

Limit of detection. The limit of detectionis the lowest concentration of analyte in asample that can be detected, but notnecessarily quantitated, under the statedexperimental conditions. Investigatorsmay measure the magnitude of analyticalbackground response by analyzing anumber of blank samples and calculatingthe standard deviation of this response.The standard deviation multiplied by afactor, usually 3, provides an estimate ofthe limit of detection.

Limit of quantitation. Limit ofquantitation is a parameter of quantitativeassays for low levels of compounds. It isthe lowest concentration of analyte in asample that can be determined withacceptable precision and accuracy underthe stated experimental conditions.Investigators may measure the magnitudeof analytical background response byanalyzing a number of blank samples and

calculating the standard deviation of thisresponse. The standard deviationmultiplied by a factor of 10 provides anestimate of the limit of quantitation.

Linearity. The linearity of an analyticalmethod is its ability to elicit test resultsthat are directly, or by a well-definedmathematical transformation, proportionalto the concentration of analyte in sampleswithin a given range. Linearity is usuallyexpressed in terms of the variance aroundthe slope of the regression line calculatedaccording to an established mathematicalrelationship from test results obtained bythe analysis of samples with varyingconcentrations of analyte.

Precision. The precision of an analyticalmethod is the degree of agreementamong individual test results when theprocedure is applied repeatedly tomultiple samplings of a homogenoussample. The precision of an analyticalmethod is usually expressed as thestandard deviation or relative standarddeviation (coefficient of variation).

Source: United States Pharmacopeia XXII,National Formulary XVII, pp. 17101712(1990).

ANALYTICAL PERFORMANCE PARAMETERSautosampler. The TOC-5000 employsacidification of the sample withhydrochloric acid followed by sparging withpurified air to remove inorganic carbon fromsolution as carbon dioxide gas. The organiccompounds remaining in solution are thenoxidized to carbon dioxide in a combustiontube packed with a catalyst and incubated at680 C. The carbon dioxide produced in thismanner passes through a dehumidifier andhalogen scrubber before it reaches a cellwhere it is quantified by a nondispersiveinfrared detector tuned to the highly specificcarbonyl absorption frequency in the1,700-cm1 range. The signal response wasstored in a microprocessor and comparedwith a calibration curve generated byanalysis of solutions of 0, 26.6, and 53.2ppm potassium biphthalate corresponding tolevels of 0, 12.5, and 25 ppm organiccarbon, respectively. Between three and fivereplicate measurements were obtained foreach sample so that in all cases, coefficientsof variation of less than 2% were obtainedfor the series of replicate injections.

Methods. Potassium biphthalate standardswere made up and diluted in purified waterin glass volumetric flasks that had beenpreviously rinsed with 50% nitric acid andcopious amounts of purified water to removeany trace of organic contaminants. Purifiedwater was obtained through the use of aMilli-Q water system (Millipore, Bedford,MA). All standards, blanks, spikes, andcontrols were prepared at least in triplicate.

Stainless steel pans were coated withE. coli cells in the following manner. Wetcell paste (1.43 mL), obtained aftercentrifugation of fermentation broth in aTJ-6 centrifuge (Beckman Instruments,Fullerton, CA) for 20 minutes at 1,000 g,was suspended to volume with purifiedwater in a 100 mL volumetric flask to makea 70-diluted cell suspension. A 3.825 mLaliquot of well-mixed 70-diluted cellsuspension was then diluted to 4.0 L finalvolume with purified water, making a73,200-diluted cell suspension. A 1.0 Lvolume of well-mixed 73,200-diluted cellsuspension was transferred to a25.2 50.6-cm stainless steel pan and wasthen lyophilized to dryness, depositingE. coli cells corresponding to approximately267 g of wet cell mass per 25 cm2 area.

Ten 25 cm2 sections of the stainless steelpan surface were sampled with swabs using

the following protocol. The fabric head of aswab was wetted in 1 N sodium hydroxide,and excess liquid was removed by gentlypressing the swab head against the walls ofthe reagent reservoir. 5 cm 5 cm sectionsof the pan were swabbed in their entirety,after which the swab stem was cutapproximately 1 cm above the swab head,dropping each swab head into a glass 5 mLTOC autosampler vial (Shimadzu) alsocontaining a small stir bar. Before analysis,1.0 mL of 1 N sodium hydroxide, 3.0 mL ofMilli-Q water, and a small stir bar wereadded to tubes containing swab samples.The samples were then stirred on a stir plateat room temperature for one hour, afterwhich time 1.0 mL of 2 N hydrochloric acidwas added before sparging and injection bythe instrument.

Swab blanks were prepared in a mannersimilar to that described above for thepreparation of stainless steel pan samples,

with the exception that clean swab heads notexposed to the pan surface were used. Forthe determination of a limit of detection anda limit of quantitation, a set of 10 swabblanks were prepared and analyzed.Linearity was evaluated by analyzing fourlevels of sucrose standards in purified watercorresponding to concentrations of 0, 10, 20,and 25 g/mL carbon.

Assay precision was determined bypreparing a set of 25 swabs by spiking100 L of 233 diluted cell suspensiononto the swab heads and then analyzing fivespiked swabs per day for five days. Tovalidate recovery, two sets of three levels ofdiluted cell paste (100 L volumes of 70-,140-, and 280-diluted cell paste werealiquoted four times each to sample tubes)were prepared, one set in the presence ofswab heads (spikes) and one set in theabsence of swab heads (controls).

Results and DiscussionWe determined the TOC signals generatedby the sucrose standards prepared in theabsence of swab heads to increase linearlywith carbon concentration, with a simplecurve fit correlation coefficient of 0.999(Figure 1). Because the preparation of swabsamples required extraction and dilution to a5.0 mL volume before analysis, the linearitydetermined for the concentrations of 025g carbon/mL from the sucrose solutionscorresponded to levels of 0125 g/swab. Incomparison with a standard curve, the 10swab blanks generated signals equivalent toa mean background of 8.0 g/swab. Allsample determinations discussed in theremainder of this section were corrected forthe swab blank background. A limit ofdetection and limit of quantitation wereestablished at 9.2 g/swab and 12.1g/swab, respectively, for this analysis.

A determination of analysis precisionindicated relative standard deviations (RSD)to range between 2.05% and 7.18% within aset of five replicate analyses of swabs spikedwith diluted cell paste, as assayed daily forfive days (Table 1). Day-to-day precisionwas determined to equal 7.15% RSD for the

pooled five-day data set.The results of the evaluation of accuracy

using diluted cell paste are presented inTable 2. At the three levels of dilution (70,140, and 280 in purified water), theTOC signals from E. coli cell paste weredetermined to range from approximately 20to 100 g/swab, corresponding to74.9103% recovery in the presence of swabheads.

Table 3 shows the recoveries from eachof the 10 swab samples taken from thesurface of the stainless steel pan, which hadbeen coated with a layer of E. coli cells. Asdescribed above, we calculated thatapproximately 267 g of wet cell mass wasdeposited per 25 cm2 area on the pansurface. A direct TOC determination of the73,200 diluted cell suspension revealedthat 267 g of wet cells corresponded to64.3 g of organic carbon, against whichnumber the percent recoveries of the assayed

PrecisionDay No. (% RSD)

1 6.872 2.053 2.744 7.185 5.83

Pooled five-day precision = 7.15%RSD.

Table 1. Precision of the analysis of fivereplicates of swabs spiked with dilutedE. coli cell paste analyzed on each offive days.

Dilution of Mean Control Mean Spike Mean % RecoveryCell Paste Signal (g) Signal (g) (5 Replicates)

70 97.8 100.6 102.9140 50.4 44.7 88.7280 25.9 19.4 74.9

Table 2. Method accuracy determined by the analysis of diluted E. coli cell paste both inthe presence (spikes) and absence (controls) of swab heads expressed as percentrecovery of spike signal compared with control signal.

TOC Recovered %Replicate (g/swab) Recovery

1 48.1 74.82 60.0 93.33 57.4 89.34 55.0 85.55 59.1 91.96 46.1 71.77 47.9 74.58 44.1 68.69 38.2 59.4

10 71.3 110.9

Table 3. Stainless steel pan swab samplerecovery determined by the analysis ofE. coli cell paste deposited on a stainlesssteel pan. The pan swab signal wascompared with that generated by anaqueous suspension of the cell paste andis expressed as percent recovery.

swab samples were then calculated. As displayed in Table 2, recovery ranged from 59.4% to 110.9% for the set of 10 replicates,with a mean recovery of 81.9%. A lower TOC recovery (67%) was achieved by other researchers who used a swab wetted with dilute phosphoric acid to sample a surface upon which whole cells had been applied (12). A similar degree of high variability of recovery was also encountered in a previous study (10), in which it was believed to refl ect limitations associated with precise sampling as well as the limitations in the establishment of a consistent layer of cells on the surface during drying. General Applications The method described in this article was developed as a quantitative assay for trace amounts of carbon-containing components that may be present on the stainless steelsurfaces

of bioprocess equipment. E. coli cells, which are present at levels of approximately 10131014 cells/L, together with a multitude of soluble carbon-containing components during a typical fermentation, were chosen to represent a most challenging case model for this study because of their insoluble particulate nature.In comparison with the fermentor cell concentration mentioned above, the TOC method accuracy was validated by the recovery of levels of carbon corresponding to approximately 109 cells/25 cm2 surface area. It is important to note that the method will determine the presence of cellular material in swab samples irrelevant

TOC is less laborintensive and offersgreater generalapplication than otherassays that havebeen used to supportcleaning validation.

References(1) The Pink Sheet, F-D-C Reports, 55 (22), T&G-

13 (1993).(2) H.L. Avallone, Drug Substance Manufacture

and Control, Pharm. Eng. 9 (2), 3740, 57(1989).

(3) S.W. Harder, The Validation of CleaningProcedures, Pharm. Tech. 8 (5), 2934(1984).

(4) FDA Mid-Atlantic Region Inspection Guide:Cleaning Validation (Food and DrugAdministration, Washington, DC, 1993), 3.

(5) Y. Egozy and J. P. Denoncourt, Trace Level

of the condition of the cells (intact compared with lysed, for example).

TOC is less labor intensive and offers greater general application than other assays that have been used to support cleaning validation. For example, TOC can be used to determine the presence of compounds such as amino acids, nucleic acids, and sugars that are incapable of generating a response in assay methods such as a total protein determination. At Eli Lilly and Co., TOC has been routinely used for the analyses of swabs taken from process equipment exposed to a variety of water-soluble drug substances (such as antibiotics), excipients (such as lactose), and formulated cleaning agents (such as surfactants), and the methods used for these assays have undergone general

validation, with a specifi c focus on recovery, in a manner similar to that described in this article (13).

Analysis of High-Purity Water, II. TotalOrganic Carbon, Ultrapure Water 3 (6)(1986).

(6) A. Henriksen, D.F. Brakke, and S.A. Norton,Total Organic Carbon Concentrations inAcidic Lakes in Southern Norway, Environ.Sci. Technol. 22 (1988), 11031105.

(7) C.D. Reach Jr. and J.T. OConnor, Total

Organic Carbon and Total Organic Halide asBroad-Spectrum Parameters for DrinkingWater Control, Proc. AWWA Water Qual.Technol. Conf. 13 (1986), 339.

(8) R. Matsuda et al., Total Organic Carbon as anIndex for Specification of Water for Injection,J. Assoc. Off. Anal. Chem. 70 (1987), 681686.

(9) R. Baffi et al., A Total Organic CarbonAnalysis Method for Validating CleaningBetween Products in BiopharmaceuticalManufacturing, J. Parent. Sci. Tech. 45(1991), 1319.

(10) M.A. Strege et al., Total Protein Analysis ofSwab Samples for the Cleaning Validation ofBioprocess Equipment, BioPharm 7 (9),4042 (1994).

(11) CleanTips Polyester Swabs productdescription, Texwipe Inc., 1995.

(12) S. Lombado et al., Development of SurfaceSwabbing Procedures for a CleaningValidation Program in a BiopharmaceuticalManufacturing Facility, Biotechnol. Bioeng.48 (1995), 513519.

(13) B.T. Farrell, unpublished results, Eli Lilly andCo. (1995). BP

Reprinted from BIOPHARM INTERNATIONAL, April 1996 Printed in U.S.A.

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/CreateJDFFile false /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice