MicrobiologyTopics.

]
ScottSutton
The MostProbable Number
Methodand Its Uses in
Enumeration, Qualification,
and Validation
ScottSutton
"Microbiology Topics" discusses various topics in
microbiology ofpractical use in validation and com-
pliance. We intend this column to be a useful
resourcefordailyworkapplications.
Readercomments, questions, and suggestions are
needed to help us fulfill our objective for this col-
umn. Please send your comments and suggestions
tocolumn coordinatorScottSutton atscott.sutton@
microbiol.org or journal coordinating editor Susan
Haigneyatshaigney@advanstar.com.
KEYPOINTS
Thefollowingkeypointsarediscussed:
Themostprobablenumber(MPN)methodisuseful
forestimating quantitativebioburdenifplatingfor
colonyformingunitsisnotadvised.
ThismethodisdescribedinUnited States Pharma-
copeia chapter<61>andbytheUS FoodandDrug
AdministrationintheBacteriological Analytical Man-
ual. Detailsof themethodarediscussed.
TheMPNmethodhasdirectapplicationinquali-
ficationstudiesformediaandforalternate(rapid)
microbiologicalmethods. Ithasalsobeensuggested
asaconsiderationforanalternatemethodtotrend
environmentalmonitoringstudies.
INTRODUCTION
The"mostprobablenumber"(MPN)methodisauseful,
if underutilized, tool forthe microbiologist. It ispart of
theharmonizedcompendialchapteronbacterialenu-
meration(1) andhasbeenpartof the"MicrobialLimits
Test"chapterintheUnited States Pharmacopeia sincethe
chapterinceptioninUSP XVIII (2). Thetestisamethod
toestimatetheconcentrationofviablemicroorganisms
ina sampleby means of replicate liquidbroth growthin
ten-folddilutionsandisparticularlyusefulwithsamples
thatcontainparticulaternaterialthatinterfereswith
platecountenumerationmethods.
ThebasicconcepttotheMPNmethodissimilarto
thefractionnegativemethodofD-valuedetermination.
Nutrientbrothwillsupportgrowthoforganismsand
turnturbid. Thebasicpatternof growthvs. no-growth
canprovideinformationasitisareflectionof sampling
error.For example,ifone replicate tube of media receives
a dilutionofthe sample that contains a bacertial cell,the
tubewillturnturbid.Itsneighbor,an"identical"replicate,
maynotreceiveanybacteriainitssampleduetopipetting
orsamplingandwould notturnturbid.Ibisinformation
is particularly useful at lownumbersof organisms. How-
ever,thisaccuracycanbegreatlyincreasedbydiluting the
inoculumandthencomparingtherecoveriesof alltubes
inthedilutionseries. ThisisthebasisoftheMPNmethod
FormoreAuthor
information.
goto
gxpandjvl.com/bios
ABOUTTHEAUTHOR
ScottSutton,Ph.D., is ownerand operatorofThe MicrobiologyNetwork(www.microbiol.org),
which providesservicestomicrobiology-related user'sgroups. DrSutton maybereached bye-mailat
scott.sutton@microbiol.org.
JOURNAL OF VALIDATION TECHNOLOGY [SUMMER 2010l 35 gXjl3mJjvLcom
r
MicrobiologyTopics.
Table: MPNtableforathree-replicatedesignfromFDA's Bacterial Analytical Manual.
Positive Tubes Positive TUbes
0.1 0.01 0.001 MPN 95% Confidence Range
0 0 0 <3.0 0-9.5
0 0 1 3 0.15-9.6
0 1 0 3 0.15-11
0 1 1 6.1 1.2-18
0 2 0 6.2 1.2-18
0 3 0 9.4 3.6-38
1 0 0 3.6 0.17-18
1 0 1 7.2 1.3-18
1 0 2 11 3.6-38
1 0 '7.4 1.3-20
1 1 1 11 3.6-38
1 2 0 11 3.6-42
1 2 1 15 4.5-42
1 3 0 16 4.5-42
2 0 0 9.2 1.4-38
2 0 1
2 0 2 20 4.5-42
2 1 0 15 3.7-42
2 1 1 20 4.5-42
2 1 2 27 8.7-94
0.1 0.01 0.001 MPN 95% ,vvRange
2 0 21 4.5-42
8.7-94
8.7-94
8.7-94
8.7-94
4.6-94
8.7-110
17-180
9-180
17-200
37-420
40-420
18-420
37-420
40-430
90-1000
42-1000
90-2000
180-4100
420-4000
2 2 1 28
2 2 2 35
2 3 0 29
2 3 1 36
3 0 0 23
3 0 1 38
3 0 2 64
3 1 0 43
3
3 1 2 120
3 1 3 160
3 0 93
3 2 1 150
3 2 2 210
3 2 3 290
3 3 0 240
3 3 1 460
3 ._ 3 2 1100
3 3 3 >1100
(alsolmownandmultipletube,dilutiontube,ordilution
tubemethods). Themethodoffersrealopportunitiesas
anenumerationtool. Itcanalsobeemployedforsemi-
quantitativeestimationofgrowth-promotion capabilityof
liquid mediaandinestimationofprecision foralternate
microbiologicalmethodswithasimplemodification.
THEMETHOD
Inthecompendialversionof theMPNtest,thesample
tobetestedispreparedin 10-folddilutionseries, and
then1mLsamplesofeachdilutionareinoculatedinto
triplicatebrothculturetubesforincubation. Asthedilu-
tionsincrease,thepossibilitythatthebrothtubeswillfail
tobeinoculatedwithanymicroorganismincreases. At
somepoint therefore,very fewof thereplicatetubes will
beinoculatedwithviablemicroorganisms.
Followingincubation,alltubesareexaminedfortur-
bidityandthepatternofgrowthinthetubesis scored
against a tableofsuchvalues(3). The MPNtable fromthe
US FoodandDrugAdministration'sBacterial Analytical
Manual (BAM) (http://www.fda.gov)isprovidedabove. A
typicaldesignusesthreereplicateswithathree-log
lO
unit
dilutionseries (althoughvaryingnumbersofreplicates
anddifferentdilutionseriesmayalsobeused). Inthis
design,if alltubesshowedgrowth,thentheresultswill
be noted as 333. Ifonlyone tube in each replicate shows
growthitwouldbedenotedas 111. Thepatternofgrowth
is thenreadfromthetabletoprovidethemostprobable
numberand95%confidenceinterval. Bythis,theresult
of210wouldreflectanMPNof21, andaresultof322
wouldbeinterpretedasanMPNof21O.
Figures1and2showthisingraphicdepiction. Asthe
incubatedtubeswouldberead321,theMPNwouldbe
recordedas150.
TheMPNtablenormallyonlypresentsresultsforthree
dilutions in sequence (e.g., 10', 10',10
3
), but the dilution
seriestestedmighthavebeenfromthe10
2
to10-
4
tubes
(seetheFDABAMdiscussiononhowtoselectappro-
priatetubestoread). Theworkerwillneedtotakethe
dilutionfactorsinthetableandintheactualexperiment
intoaccounttoderivethemostprobablenumberfrom
thisstudy. Theresultsof thistestshouldbeexpressedas
"MPN"ratherthanCFU(colonyformingunit)toreflect
thecapabilitiesof themethod.
36 JOURNAL OF VALIDATION TECHNOLOGY [SUMMER 2010]
Figure1:Three-tubedesignfor
MPN (unincubated).
= = =
10-
1
= = =
10-
2
= = =
10-
3
Themethodassumesarandomdistributionof micro-
organismsinthesampleandanaccuratedilutionof the
samplethroughthedilutionseries. It alsoassumesthat
themicroorganismsareseparateanddonotaffecteach
other(i.e.,attractorrepel). Inaddition,itmustbeassumed
thateverytube(orplate,etc.)whoseinoculumhasasingle
viableorganismwillresultinvisiblegrowth.
Althoughthecompendialversionutilizesthreereplicates
andaten-folddilutionseries,thereisnotheoreticalreason
fortheseparameters. Infact,itiswellknownthattheaccu-
racy ofthemethodincreasesdramaticallywhenincreasing
thenumberofreplicatesanddecreasing theintervalofthe
dilutionseries(five-foldortwo-fold) (4,5). cTheFDABAM
websitereferencedprovidesanExcelspreadsheettoassist
increatingdifferentMPNtablesasneeded.
APPLICATIONSOFMPNIN
ENVIRONMENTALMONITORINGDATA
Environmentalmonitoringdataisaproblemformicro-
biology. Weareurgedto"qualify"ourcontrollevelsand
oursamplesites. However,weareusingatechnology
(platecount)thatisexceedinglyimpreciseatnumbersof
CFU below25. The aseptic coreof a modern facility will
commonlyyieldcountsofzero,withconcernexpressedif
ScottSutton, Coordinator.
Figure2:Three-tubedesignfor
MPN (incubated).
10'
10
2
= =
10
3
thecountisapproachingthree. Thesecontrollevelsarein
truthoflittlevaluedespitetheirpopularityinregulatory
circles (6-8). Oneapproachsuggestedtodealwiththis
mismatchbetweenregulatoryexpectationsandplate
countcapabilitieshasbeentoexplorethepossibilityof
lookingatafrequencydistributionmodelstoestablish
controllevelsintheseareas (9) orincidentmodels(10).
ArecentpublicationbySunetal. (11) pointedoutthe
possibilities in usingMPNmethods for evaluation of clean
roommonitoringdata. Thebasicideaistousethefun-
damental statisticsasif only asingle dilution were being
considered. Inthisapproach, theapplicationisnotdis-
similar to fractionnegative studiesofbiological indicators
forsterilizationstudies. Preliminarystudiespresentedby
thisgrouplookpromising,andthisis anapproachthat
couldbepursuedwithexistentdataforevaluation.
MPNINQUALIFICATIONOFBROTHMEDIA
ThecompendialchaptersoftheUSP onmicrobiallimits
testsandsterilitytests(revisedin2009)bothplacegreat
emphasisonmediagrowthpromotionstudiesasarequi-
sitequalitycontrolactivity. Thishasbeenreinforcedby
theUSP chapter<1117>(revisedin2010) (12) discussion
oftheimportanceofmediacontrolinthelab. While
JOURNAL OF VAl.IDATION TECHNOLOGY [SUMMER 2010] 37
MicrobiologyTopics.
themethodsathandtocomparebacterialgrowthon
solidmediaarequantitativeinnature(recoverywithin
50%orwithin70%byCFU),thetoolsdescribedinthe
compendiaforbrothgrowthpromotionarequalitativeat
best."Liquid media are suitableif clearlyvisible growth
of themicroorganismscomparabletothatpreviously
obtainedwithapreviouslytestedandapprovedbatch
of mediumoccurs" (12).
Weenk(13)reviewedtheMPNmethodinhisextensive
reviewofmethodsusedtoqualifymediaandis recom-
mendedforitsdetaileddescriptionof howtoimprove
theprecisionofthemethod. Thebasicapproachrec-
ommendedis toapproachtheMPNmethodfromthe
oppositedirectionthanthatof thebioburdenMPN. In
thebioburdentest,wehaveasamplewithanunknown
bioburdenandwearetryingtodeducethemostprobable
number ofcells. In the recommended growth promotion
testforliquidmedia,wehavetwobatchesdispensedin
tubes. The"sample"is aknowninoculuminaknown
dilutionseries. Theinoculumdilutionsareseededinto
thetwomedia,andthenincubated. If themediaexhibit
identicalgrowthpromotingproperties,thenthe95%con-
fidenceintervalsofthe two MPN determinations should
overlap. Inthismanner,a (semi)-quantitativegrowth
promotionstudymaybeperformedforliquidmedia.
MPNINQUALIFICATION OFALTERNATE
(RAPID) MICROBIOLOGICALMETHODS
It shouldbeobvioustothereaderthatthepreviousdiscus-
sion on the useofMPN in growth promotion studies has
immediateapplicationfordeterminationof therelative
limitof detectionfortwomicrobiologicalmethods,for
examplea"traditional"methodandan"alternate"method.
ThisisinfactreferencedinUSPchapter<1223>(14).
TheUSP chapterrecommendsthisapproachevenfor
quantitativemethods. Althoughthismightatfirstseem
counter-intuitive,theMPNmethod(whenusedwitha
dilutionseries)canactuallybemoreaccuratethanplate
countsatlownumbers. Theonlymodificationthatneeds
tobemadeistoignorethecountsandtreateveryplateor
membraneasaseparate"tube"-theMPNmethodfits
rightintotheexperimentaldesign.
SUMMARY
ThebasicconceptfortheMPNmethodistodilutethe
sampletosuchadegreethatinoculainthetubeswill
sometimes(butnotalways) containviableorganisms.
Byreplicates,anddilutionseries,thiswillresultinafairly
accurateestimateofthemost probablenumberofcellsin
thesample. Whilethismethodismostcommonlyused
inthepharmaceuticalindustryforwatertestingorthe
38 JOURNAL OF VALIDATION TECHNOLOGY [SUMMER 2010]
compendialbioburdentest,ithassignificantpotential
tothequalitycontrolmicrobiologylab. Thesepossible
applicationsofMPNincludeenvironmentalmonitor-
ing,mediagrowthpromotionstudiesandaspectsofthe
validationof rapidmicrobiologicalmethods.
REFERENCES
1. USP, Chapter <61> "Microbiological Examinationof Non-
sterileProducts: MicrobialEnumerationTests," USP 32.
vol. 1 pp71-75, 2009.
2.USP, "MicrobialLimitTests," USPXVIII, p. 846,1970.
3. FDA, Bacterial Analytical Manual, Appendix2availableon-
lineathttp://www.fda.gov/Food/ScienceResearch/Labo-
ratoryMethodsjBacteriologicalAnalyticalManualBAMj
ucm109656.htm.
4.Eisenhart,C.andW. Perry,"StatisticalMethodsandControl
InBacteriology,"Bacterial Rev. 7:57-137, 1943.
5. Cochran, W., "Estimation ofBacterial Densities by Meansof
theMostProbableNumber,"Biometrics. 6:105-116, 1950.
6. Hussong,D. andRE Madsen, "AnalysisofEnvironmental
MicrobiologyDataFromCleanroomSamples," Phann Tech-
nol AsepticProc:1O-15, 2004.
7. Farrington, JK., "Environmental Monitoringin Pharmaceuti-
calManufacturing-AProductRiskIssue,"Amer Pharm Rev
8(4):26-30,2005.
8.Agalloco,J. andAkers,J., "The'Myth'CalledSterility,"Phann
Tech, BioprocessSterileManufact:s44-s45,2010.
9. Caputo,RAandA. Huffman,"EnvironmentalMonitoring:
DataTrendingUsingaFrequencyModel,"PDAJPharm Sci
Tech. 58(5):254-260,2004.
10.USP, Chapter<1116> "MicrobiologicalControlandMonitor-
ingEnvironmentsUsedfortheManufactureofHealthcare
Products,"Pharm Forum 33(3),2007.
11.Sun,Xetal., "TheExpandedApplicationof MostProbable
NumbertotheQuantitativeEvaluationofExtremelyLow
MicrobialCount,"PDA J Pharm Sci Tech 60(2):124-134,
2006.
12.USP, Chapter<1117> "MicrobiologicalBestLaboratoryPrac-
tices," USP 33,Reissue,SecondSupplement,pp.R-ll00-R-
1105,2010.
13. Weenk, C.H., "MicrobiologicalAssessmentofCulture
Media:Comparisonand Statistical EvaluationofMethods,"
IntlJ Food Microbiol., 17:159-181,1992.
14. USP, Chapter<1223>"ValidationofAlternativeMicrobio-
logicalMethods,"USP 32,vol. 1 pp730-3,2009.JVT
ARTICLEACRONYM LISTING
BAM Bacterial Analytical Manual
CFU Colony Forming Unit
MPN Most Probable Number
USP United States Pharmacopeia