Revised Criteria for Reducing Observer Error in the Histological Assessment of Linear Microcracks

Full text

!
Victoria M. Dominguez1, Amanda M. Agnew1,2
The Ohio State University: 1Skeletal Biology Research Laboratory; 2Department of Anthropology
"#$%&'()*+),)%-.
/)0)/)%#).1#2-)*
CONCLUSIONS
MATERIALS AND METHODS
INTRODUCTION
•Though often overlooked, clear and validated criteria for histomorphological
analyses are essential.Linear microcracks are aprimary aspect of bone
microdamage important for improved understanding of bone quality and
health, with no existing validated criteria.
•Previous works present only brief and unvalidated descriptions for counting
and measuring microcracks, never assessing observer error in microcrack
quantification.
•The aim here is to provide revised criteria with an accompanying
photographic atlas as ameans to reduce observer error. A preliminary
assessment of observer error was conducted using the proposed criteria.
&3456756 56686 49:;<
•10 slides prepared with basic fuchsin for microcrack analysis
•Two observers for intra-and inter-observer analysis
•Observer 1: Novice in microcracks, read all slides twice
•Observer 2: Experienced in microcracks, read all slides once
•Two variables considered, crack number (Cr.N) and crack length (Cr.Le)
•When the coefficient of variation was less than 10%, error rate was deemed
acceptable
aThe classification criteria here presented are for the assessment of linear microcracks.These criteria are based in part on the
definitions laid out by Burr and Stafford (1990)and revised in Lee et al. (1998), but also include additional considerations for
reliable counting.Figures are numbered clockwise beginning from the top.
-=3>51?@1#>=44ABAC=9A8D1
C6A956A==
AQlinearQmicrocrackQisQaQsharplyQdefinedQlineQwithQaQhaloQofQbasicQfuchsin.QQLinearQmicrocracksQareQmoreQrobustQinQ
appearanceQthanQcanaliculi,QwhichQoftenQresembleQaQ
spiderweb criss-crossing theQbone,QbutQsmallerQthanQvascularQ
channelsQlikeQ
Volkman’s canals,QwhichQareQwider.QQ(Figs.Q1—4)
OnlyQcracksQthatQvisiblyQopenQonQtheQsurfaceQofQtheQsectionQbeingQanalyzedQshouldQbeQcounted.QQWhenQviewedQ
underQtheQmicroscope,QtheQfeatureQinQquestionQshouldQbeQcountedQifQitQappearsQtoQopenQwhenQtheQdepthQofQfocusQisQ
changed,QifQnot,QitQshouldQbeQexcluded.QQThisQisQbestQseenQinQbrightQfield.
MicrocracksQthatQareQnotQ
artifactual willQbeQstainedQthroughQtheQdepthQofQtheQsection,QappearingQdarkerQthanQtheQ
surroundingQarea.Q(Fig.Q5)
SingleQcracksQthatQareQdistinguishableQfromQtipQtoQtipQshouldQbeQcountedQasQone.QQLengthQmeasurementsQshouldQ
followQtheQpathQofQtheQcrackQ(Fig.Q8).QQ
InQsomeQcases,QcracksQappearQtoQcoalesce,QresultingQinQwhatQlooksQlikeQtreeQbranching.QQInQtheseQinstances,QtheQ
longestQaxisQshouldQbeQcountedQasQoneQandQmeasuredQasQtheQprimaryQcrack.QQBranchesQshouldQbeQcountedQasQ
independentQcracks,QwithQeachQbranchQmeasuredQfromQtheQpointQofQintersectionQwithQtheQprimaryQcrackQtoQtheirQ
freeQendQ(Fig.Q9).QQ
MicrocracksQareQsometimesQseenQtoQbeQpartiallyQstained.QQThisQisQattributedQtoQoneQofQtwoQcauses:Q(1)QtheQstainQdidQ
notQfullyQimpregnateQtheQexistingQmicrocrackQduringQtheQstainingQprocess,QorQ(2)QtheQexistingQmicrocrackQcontinuedQ
toQpropagateQduringQtheQprocessQofQslideQpreparationQafterQtheQstainingQprocessQwasQcompleted,QresultingQinQaQ
portionQofQtheQmicrocrackQbeingQunstained.QQ
(1)QIfQtheQcrackQisQnotQstainedQthroughQtheQdepthQofQtheQsection,QdoQnotQcountQit.Q(Figs.Q5QandQ13)
(2)QIfQaQpartQofQtheQcrackQisQstainedQthroughQtheQdepthQofQtheQsection,QtheQmicrocrackQshouldQbeQ
counted,QbutQonlyQtheQstainedQportionQshouldQbeQmeasured.QQ(Figs.Q6QandQ12)
Debonding
ofQlamellarQlayersQpresentsQwithQaQsimilarQappearanceQtoQlinearQmicrocracks.QQThisQisQgenerallyQseenQ
betweenQlayersQofQconcentricQlamellaeQ(withinQtheQsecondaryQosteons)QorQlongitudinalQlayersQofQlamellarQappositionQ
(atQtheQperiostealQbordersQofQtheQbone).QQ
Debonding shouldQnotQbeQcountedQasQmicrocracks,QwithQtwoQexceptions.Q
(Fig.Q11)
(1)QIfQtheQplaneQofQseparationQisQgreatQenoughQthatQitQappearsQopenQonQtheQsurfaceQofQtheQsection,QitQ
shouldQbeQcountedQandQmeasured.Q(Fig.Q10)
(2)QIfQtheQmicrocracksQrunQintoQandQpropagateQthroughQsuchQplanesQofQseparation,QitQshouldQbeQ
counted.QQThisQhappensQmostQoftenQwhenQmicrocracksQbreakQthroughQorQdeflectQaroundQanQ
osteon’sQreversalQline.QQ(Fig.Q6)
InstancesQwhereQaQseriesQofQfineQlinesQthatQareQdifficultQtoQdistinguishQareQseenQtoQrunQthroughQeachQotherQareQ
knownQasQcross
-hatching.QQDamageQthatQbearsQaQcross-hatchedQappearanceQisQnotQopenQtoQtheQsectionQsurfaceQandQ
cannotQbeQreliablyQtraced,QandQthereforeQshouldQnotQbeQcountedQinQlinearQmicrocrackQassessment.Q(Fig.Q7)
RESULTS
2D96=83456756 56686
•Differences between the first and second read were significantly different in
four instances for Cr.N and two instances for Cr.Le
2D95683456756 56686
•Differences between Observer 1and 2were significantly different in nine
instances for Cr.N and two instances for Cr.Le based on Observer 1’s initial
reads, and significantly different in seven instances for Cr.N and three
instances for Cr.Le based on Observer 1’s second read
•Results indicate that observer differences are aserious concern for linear
microcrack studies, even with carefully defined criteria, and observer
experience is amajor factor in reducing error.The criteria and atlas are
presented here to encourage discussion for further improvement.
Figure'1. A)'A'single'microcrack in'the'cortex'near'the'endosteal
border,'indicated'by'the'black'arrow..'B)'Multiple'linear'
microcracks radiating'near'the'periosteal'border.'The'two'
microcracks on'the'left'are'stained'through'the'depth'of'the'
section,'while'the'microcrack on'the'right'is'incompletely'
stained.'C)'A'small'cluster'of'microcracks that'are'still'
individually'distinguishable.
Figure'2.'Linear'microcrack illuminated'in'bright'field'(left)'and'
fluorescence'(right).'Unstained'portions'appear'green'in'FITC'
fluorescence,'while'the'cracks,'osteocytic lacunae,'Haversian'
canals,'and'canaliculi'are'red'due'to'their'uptake'of'basic'fuchsin.
Figure'3.'A)'Close-up'of'canaliculi'connecting'osteocytic lacunae.'B)'Less'magnified'
example'of'a'linear'microcrack running'through'an'osteocytic lancuna.'
Figure'4.'Comparison'of'a'
Volkman’s canal'(A)'and'two'
linear'microcracks (B).'The'
Volkman’s canal'is'wider,'and'
when'the'image'is'zoomed'out'
(top),'it'can'clearly'be'seen'to'
work'as'a'transverse'connection'
between'the'Haversian'canals'of'
the'two'secondary'osteons'that'
surround'it.
Figure'5.'A)'Bright'field'and'B)'fluorescent'
images'showing'an'inQvivoQmicrocrack (blue'
arrow)'versus'an'artifactual microcrack from'
sample'preparation'(white'arrow).'
Figure'8.'An'example'of'how'to'measure'linear'microcracks
Figure'13.'Bright'field'(left)'and'fluorescent'(right)'images'of'a'linear'microcrack not'
stained'through'the'depth'of'the'section.''Basic'fuchsin has'not'permeated'the'depth'of'
the'section'and'any'evidence'of'a'halo'is'minimal'and'constrained'to'the'ends'of'the'crack.'
Figure'12.'Bright'field'(left)'and'fluorescent'(right)'images'of'a'crack'that'is'partially'
stained.'In'this'linear'microcrack,'the'upper'portion'is'infused'with'basic'fuchsin,'while'the'
lower'portion'is'not.'The'point'at'which'the'stain'ceases'is'indicated'by'the'black'arrow.''
Above'the'arrow'is'a'crack'acquired'inQvivo,'as'indicated'by'the'presence'of'basic'fucshin,'
while'below'the'arrow'is'artifactual damage.'It'is'likely'that'the'inQvivo crack'continued'to'
propagate'during'slide'preparation.'
Figure'11.'A)'Debonding between'the'appositional'layers'of'lamellae'at'the'periosteal'
border,'highlighted'by'the'four'white'arrows.'B)'Debonding between'the'layers'of'
circumferential'lamellae'within'a'secondary'osteon'that'does'not'open'onto'the'surface,'
highlighted'by'the'three'white'arrows.'Linear'structures'that'do'not'open'onto'the'
surface'of'a'section'likely'reflect'an'open'defect'on'another'plane'of'the'section.
Figure'10.'Similar'to'Figure'12B,'this'image'
demonstrates'debonding between'the'
lamellar'layers'of'a'secondary'osteon.'
However,'as'seen'in'the'bright'field'image,'
the'sharp'borders'of'an'open'crack'are'
visible'and'should'be'counted'during'
analysis.'
Figure'6.'Example'of'a'linear'microcrack
hitting'the'reversal'line'of'secondary'osteon'
and'being'deflected.''Stain'uptake'indicates'
that'the'crack'likely'propagated'to'the'right'
of'the'arrow'inQvivo,'but'that'the'opening'of'
the'crack'to'the'left'of'the'arrow'may'have'
resulted'after'the'staining'process'during'
the'course'of'sample'preparation.''
Figure'7.'Example'of'cross-hatching,'which'
resembles'linear'microcracks,'but'does'not'
open'onto'the'surface'and'has'indistinct'
borders.'
"
Figure'9.'A)'Complex'clustering'of'linear'microcracks that'appear'to'run'together'and'branch'off'each'other.'B)'The'
same'image'now'showing'how'measurements'would'be'taken'according'to'the'criteria'laid'out'here.'The'primary'
crack'will'be'defined'along'its'longest'axis'and'any'additional'cracks'will'be'measured'as'branches'from'this'line.
BurrQDB,QStaffordQTQ(1990)QValidityQofQtheQbulk-stainingQtechniqueQtoQseparateQartifactual fromQ
in'vivo boneQmicrodamage.QClin Orthop Relat Res,Q260,Q305-308; LeeQTC,QMyersQER,QHayesQWCQ
(1999)QFluorescence-aidedQdetectionQofQmicrodamage inQcompactQbone.QJQAnat,Q193,179-184.
TheQauthorsQwouldQlikeQtoQacknowledgeQtheQgenerousQgiftsQofQtheQ
donors.QQThanksQasQwellQtoQtheQmembersQofQtheQSkeletalQBiologyQ
ResearchQLaboratoryQforQhelpQwithQsampleQpreparation.
#8D9=C9E
*8FADG:5H@IJK84:@5;: