revbrashematolhemoter.2017;39(3):288–289
w w w . r b h h . o r g
Revista
Brasileira
de
Hematologia
e
Hemoterapia
Brazilian
Journal
of
Hematology
and
Hemotherapy
Letter
to
the
Editor
Assessment
of
the
Matos
&
Carvalho
index
for
distinguishing
thalassemia
from
iron
deficiency
anemia
DearEditor,
Inarecentissueofthisjournal,Matosetal.proposedanew indexfordiscriminatingbetweenirondeficiencyanemia(IDA) and thethalassemia trait.1 Weread theirpaperwith great
interest,since we haveperformed similar researchin this field.2Recentlywecompletedastudyinwhichwecompared
25discriminantformulasincorporatingonlybasicredblood cell(RBC)parametersusingalargedatabaseofpatientswith mildmicrocyticanemia,whoeitherhadadiagnosisofIDAor werethalassemiacarriers.3 Thisdatabasenowgivesusthe
opportunitytoinvestigatethediagnosticperformanceofthe Matos&Carvalho(M&C)indexinamuchlargerpatientcohort thantheauthorsusedfortheirdevelopmentandvalidation,1
andwepresentourfindingsinthiscommunication.
Ourdatabasecomprises2664patientswithmicrocytic ane-mia,definedashemoglobin<13g/dLandmeancorpuscular volume(MCV)<80fL,irrespectiveofgender.Itwasbuiltovera periodofnineyears,duringwhichtimetheBeckmanCoulter LH780wasthemainhematologyanalyzerinuse;other hema-tologyanalyzerswereoperatedforshorterperiods.Standard biochemicalandhematologicalmethodswereusedtomake thediagnosis;thediagnosisof␣-thalassemiawasbasedon moleculartechniques.Ofourpatients,1196had“pure” tha-lassemia(319␣-thalassemiaand877-thalassemia),1259had simpleIDA,117 hadthalassemia withconcomitantIDA,33 hadthalassemiawithconcurrentanemiaofchronicdisease, 23hadcomplexthalassemiaand36hadotherdiseases asso-ciatedwithmicrocyticanemia.
WhenapplyingtheM&Cformulatoourdatabaseinthe receiveroperatingcharacteristic(ROC)analysis,wefoundan areaunderthecurve(AUC)of0.892(95%confidenceinterval: 0.879–0.904);asFigure1demonstratesthisAUCissignificantly (p-value<0.0001) lower than the threebest performing for-mulasinour studythat all hadAUCsaround 0.95.3 Atthe
optimalcut-offof23.6,theM&Cindexresultedin0.854 sen-sitivity,0.795specificityand aYoudenindexof0.649.If we hadincludedtheM&Cindexinourstudy,thenitwouldhave
DOIoforiginalarticle:http://dx.doi.org/10.1016/j.bjhh.2016.05.011.
0 20 40 60 80 100
100 80 60 40 20 0
100-Specificity
S
en
si
tivi
ty
Figure1–Receiveroperatingcharacteristic(ROC)analysis
oftheMatos&Carvalho(M&C)index(black)incomparison
withtheJanel(green),Green&King(red)andJayabose
(blue)formulasfordifferentiatingirondeficiencyanemia
fromthethalassemiatrait.
ranked18outofthe26formulasinvestigated.Thisisinclear contrastwiththefindingsofMatosetal.,whoreported excel-lentdiagnosticaccuracyoftheirnewindex.However,itisnot atallunusualforauthorsofnewdiscriminantindicesto over-estimatetheperformanceoftheirinvention,asevidencedby independentvalidationstudies.3
revbrashematolhemoter.2017;39(3):288–289
289
geneticbackgrounds,sothisnumberofthalassemiapatients intheirstudymayhavebeentoolowtoproperlycharacterize thevariouscarriersubtypes;alternatively,thedisagreement maybeduetolocalcharacteristicsofthepopulationinthe area investigated. Secondly, Matoset al.excluded patients withconcomitantIDAandthethalassemiatrait;ithasbeen repeatedlydemonstratedthatthisconditioncausesmany dis-criminantformulastogiveanincorrectclassificationandby excluding this group, they may have introduced a bias in favoroftheirnewindex. If weexcludesuchpatientsfrom ouranalysis, the performanceofall discriminant formulas wouldincrease,butthiswouldnolongerrealisticallyreflect clinicalpractice,asapproximately5%ofpatientswith micro-cyticanemiasufferfromIDAassociatedwiththethalassemia trait.Thirdly, theM&C indexincludesthe mean corpuscu-larhemoglobinconcentration(MCHC),acalculatedparameter that isknown to haverelatively weak correlation between differenthematologyanalyzers.Althoughthetypeof hema-tologyanalyzerwasnotfoundtohavemuchimpactonthe performanceofdiscriminantfunctions,2apossiblerole
can-notbeexcluded.Inthiscontext,itisofinterestthattheMCHC alonehasaverypoorperformancefordistinguishingIDAand thalassemia.2
In conclusion, we are unable to confirm the finding by Matosetal.thattheirnewM&Cindexhasexcellent perfor-mancefor discriminating iron deficiency anemia from the thalassemiatrait.Asamatteroffact,wedonotrecommend laboratoriesuseanynewlypublishedindexwithoutadditional validationintheirownpatientpopulation.
Conflicts
of
interest
JohannesHoffmannisascientificemployeeofAbbott Diag-nostics; Eloísa Urrechaga has no conflicts of interest to declare.
r
e
f
e
r
e
n
c
e
s
1.MatosJF,DusseLM,BorgesKB,deCastroRL,Coura-VitalW,
CarvalhoMdG.Anewindextodiscriminatebetweeniron
deficiencyanemiaandthalassemiatrait.RevBrasHematol
Hemoter.2016;38(3):214–9.
2.HoffmannJJ,UrrechagaE,AguirreU.Discriminantindicesfor
distinguishingthalassemiaandirondeficiencyinpatients
withmicrocyticanemia:ameta-analysis.ClinChemLabMed.
2015;53(12):1883–94.
3.UrrechagaE,HoffmannJJ.Criticalappraisalofdiscriminant
formulasfordistinguishingthalassemiafromirondeficiency
inpatientswithmicrocyticanemia.ClinChemLabMed.
2017;55[aheadofprint].Availablefrom:
https://www.degruyter.com/view/j/cclm.ahead-of-print/cclm-2016-0856/cclm-2016-0856.xml?format=INT.
JohannesJ.M.L.Hoffmanna, EloísaUrrechagab,∗ aGlobalMedicalandScientificAffairs,AbbottHematology,
Hoofddorp,TheNetherlands
bHematologyLaboratory,HospitalGaldakao,Galdakao,Spain
∗Corresponding author at: Hematology Laboratory, Hospital
Galdakao-Usansolo,48960Galdakao,Vizcaya,Spain. E-mailaddress:eloisa.urrechagaigartua@osakidetza.net
(E.Urrechaga).
Received28March2017 Accepted24April2017 1516-8484/
©2017Associac¸ ˜aoBrasileiradeHematologia,Hemoterapiae TerapiaCelular.PublishedbyElsevierEditoraLtda.Thisisan openaccessarticleundertheCCBY-NC-NDlicense(http://
creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.bjhh.2017.04.003