<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.1 20151215//EN" "https://jats.nlm.nih.gov/publishing/1.1/JATS-journalpublishing1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink" specific-use="sps-1.9" dtd-version="1.1" xml:lang="pt" article-type="research-article">
  <front>
    <journal-meta>
      <journal-id journal-id-type="publisher-id">abc</journal-id>
      <journal-title-group>
        <journal-title>Arquivos Brasileiros de Cardiologia</journal-title>
        <abbrev-journal-title abbrev-type="publisher">Arq. Bras. Cardiol.</abbrev-journal-title>
      </journal-title-group>
      <issn pub-type="ppub">0066-782X</issn>
      <issn pub-type="epub">1678-4170</issn>
      <publisher>
        <publisher-name>Sociedade Brasileira de Cardiologia - SBC</publisher-name>
        <publisher-loc>S&#227;o Paulo, SP, Brazil</publisher-loc>
      </publisher>
    </journal-meta>
    <article-meta>
      <article-id pub-id-type="publisher-id" specific-use="scielo-v3">PCMfXQQLg4mhknqV5LJFXHz</article-id>
      <article-id pub-id-type="other">00006</article-id>
      <article-id pub-id-type="publisher-id" specific-use="scielo-v2">S0066-782X2004001800006</article-id>
      <article-id pub-id-type="publisher-id" specific-use="scielo-v1">S0066-782X(04)08300606</article-id>
      <article-id pub-id-type="doi">10.1590/S0066-782X2004001800006</article-id>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>Artigos Originais</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Avalia&#231;&#227;o da dosagem direta do colesterol-LDL em amostras de sangue de 10.664 pacientes em compara&#231;&#227;o com o uso da f&#243;rmula de Friedewald</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <name>
            <surname>Cordova</surname>
            <given-names>Caio Maur&#237;cio Mendes de</given-names>
          </name>
          <xref ref-type="aff" rid="aff01"/>
          <xref ref-type="aff" rid="aff02"/>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Schneider</surname>
            <given-names>Carlos Rudi</given-names>
          </name>
          <xref ref-type="aff" rid="aff01"/>
          <xref ref-type="aff" rid="aff02"/>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Juttel</surname>
            <given-names>Iara Deise</given-names>
          </name>
          <xref ref-type="aff" rid="aff01"/>
          <xref ref-type="aff" rid="aff02"/>
        </contrib>
        <contrib contrib-type="author">
          <name>
            <surname>Cordova</surname>
            <given-names>Maur&#237;cio Mendes de</given-names>
          </name>
          <xref ref-type="aff" rid="aff01"/>
          <xref ref-type="aff" rid="aff02"/>
        </contrib>
      </contrib-group>
      <aff id="aff01">
        <addr-line>
          <named-content content-type="state"/>
        </addr-line>
        <institution content-type="orgname">Universidade Regional de Blumenau</institution>
        <institution content-type="orgdiv1">Departamento de Ci&#234;ncias Farmac&#234;uticas </institution>
        <country country="BR">Brazil</country>
      </aff>
      <aff id="aff02">
        <addr-line>
          <named-content content-type="city">Blumenau</named-content>
          <named-content content-type="state">SC</named-content>
        </addr-line>
        <institution content-type="orgname">Laborat&#243;rio Santa Isabel de An&#225;lises Cl&#237;nicas</institution>
      </aff>
      <pub-date date-type="collection" publication-format="electronic">
        <month>12</month>
        <year>2004</year>
      </pub-date>
      <volume>83</volume>
      <issue>6</issue>
      <fpage>476</fpage>
      <lpage>481</lpage>
      <history>
        <date date-type="accepted">
          <day>18</day>
          <month>02</month>
          <year>2004</year>
        </date>
        <date date-type="received">
          <day>10</day>
          <month>09</month>
          <year>2003</year>
        </date>
      </history>
      <permissions>
        <license license-type="open-access" xlink:href="http://creativecommons.org/licenses/by-nc/4.0/" xml:lang="en">
          <license-p>This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.</license-p>
        </license>
      </permissions>
      <abstract>
        <p>OBJETIVO: Avaliar a dosagem direta do colesterol LDL (LDL-C) por um m&#233;todo homog&#234;neo, em compara&#231;&#227;o com a estimativa pela f&#243;rmula de Friedewald, em uma grande popula&#231;&#227;o heterog&#234;nea. M&#201;TODOS: As dosagens do colesterol total (ColT) e dos triglicer&#237;deos (Trig) foram realizadas por m&#233;todos enzim&#225;ticos tradicionais. As dosagens do HDL-C e do LDL-C foram realizadas por m&#233;todos diretos, sem precipita&#231;&#227;o, e a estimativa da fra&#231;&#227;o LDL-C calculada pela f&#243;rmula de Friedewald. RESULTADOS: Por an&#225;lise de regress&#227;o linear, os dois m&#233;todos apresentaram coeficientes de correla&#231;&#227;o extremamente significativos (p&lt;0,001). Entretanto, a f&#243;rmula de Friedewald apresentou um bias positivo em rela&#231;&#227;o ao m&#233;todo direto, mais pronunciado com n&#237;veis de ColT &gt; 201 mg/dL. Este bias positivo tamb&#233;m ocorreu com rela&#231;&#227;o a n&#237;veis de Trig &lt; 150 mg/dL. Com n&#237;veis de Trig entre 151-200 mg/dL e entre 201-300 mg/dL, n&#227;o foi observado bias entre os dois m&#233;todos. Por outro lado, com n&#237;veis de Trig entre 301-400 mg/dL, este bias da f&#243;rmula de Friedewald tornou-se negativo. CONCLUS&#195;O: Foi poss&#237;vel demonstrar que a f&#243;rmula de Friedewald n&#227;o apresenta um desempenho homog&#234;neo para a estimativa do LDL-C em amostras com diferentes n&#237;veis de Trig, em compara&#231;&#227;o com o m&#233;todo direto, podendo causar d&#250;vidas na classifica&#231;&#227;o quanto ao risco de desenvolver doen&#231;a arterial coronariana.</p>
      </abstract>
      <kwd-group xml:lang="pt" kwd-group-type="author-generated">
        <kwd>colesterol</kwd>
        <kwd>LDL-C</kwd>
        <kwd>LDL direto</kwd>
        <kwd>f&#243;rmula de Friedewald</kwd>
      </kwd-group>
      <pub-date publication-format="electronic" date-type="pub">
        <day>07</day>
        <month>01</month>
        <year>2005</year>
      </pub-date>
    </article-meta>
  </front>
  <body>
    <p>
      <bold>ARTIGO ORIGINAL</bold>
    </p>
    <p>Avalia&#231;&#227;o da dosagem direta do colesterol-LDL em amostras de sangue de 10.664 pacientes em compara&#231;&#227;o com o uso da f&#243;rmula de Friedewald</p>
    <p>
      <bold>Caio Maur&#237;cio Mendes de Cordova; Carlos Rudi Schneider; Iara Deise Juttel; Maur&#237;cio Mendes de Cordova</bold>
    </p>
    <p> Blumenau, SC </p>
    <p>Departamento de Ci&#234;ncias Farmac&#234;uticas/FURB e Laborat&#243;rio Santa Isabel de An&#225;lises Cl&#237;nicas, Blumenau, SC </p>
    <p>
      <xref ref-type="corresp" rid="end">Endere&#231;o para correspond&#234;ncia</xref>
    </p>
    <p>
      <bold>RESUMO</bold>
    </p>
    <p><bold>OBJETIVO: </bold>Avaliar a dosagem direta do colesterol LDL (LDL-C) por um m&#233;todo homog&#234;neo, em compara&#231;&#227;o com a estimativa pela f&#243;rmula de Friedewald, em uma grande popula&#231;&#227;o heterog&#234;nea. </p>
    <p><bold>M&#201;TODOS: </bold>As dosagens do colesterol total (ColT) e dos triglicer&#237;deos (Trig) foram realizadas por m&#233;todos enzim&#225;ticos tradicionais. As dosagens do HDL-C e do LDL-C foram realizadas por m&#233;todos diretos, sem precipita&#231;&#227;o, e a estimativa da fra&#231;&#227;o LDL-C calculada pela f&#243;rmula de Friedewald. </p>
    <p><bold>RESULTADOS: </bold>Por an&#225;lise de regress&#227;o linear, os dois m&#233;todos apresentaram coeficientes de correla&#231;&#227;o extremamente significativos (p&lt;0,001). Entretanto, a f&#243;rmula de Friedewald apresentou um bias positivo em rela&#231;&#227;o ao m&#233;todo direto, mais pronunciado com n&#237;veis de ColT &gt; 201 mg/dL. Este bias positivo tamb&#233;m ocorreu com rela&#231;&#227;o a n&#237;veis de Trig <underline>&lt;</underline> 150 mg/dL. Com n&#237;veis de Trig entre 151-200 mg/dL e entre 201-300 mg/dL, n&#227;o foi observado bias entre os dois m&#233;todos. Por outro lado, com n&#237;veis de Trig entre 301-400 mg/dL, este bias da f&#243;rmula de Friedewald tornou-se negativo. </p>
    <p><bold>CONCLUS&#195;O: </bold>Foi poss&#237;vel demonstrar que a f&#243;rmula de Friedewald n&#227;o apresenta um desempenho homog&#234;neo para a estimativa do LDL-C em amostras com diferentes n&#237;veis de Trig, em compara&#231;&#227;o com o m&#233;todo direto, podendo causar d&#250;vidas na classifica&#231;&#227;o quanto ao risco de desenvolver doen&#231;a arterial coronariana. </p>
    <p><bold>Palavras-chave: </bold>colesterol, LDL-C, LDL direto, f&#243;rmula de Friedewald</p>
    <p>A doen&#231;a arterial coronariana &#233; respons&#225;vel pelo maior n&#250;mero de mortes de indiv&#237;duos adultos no mundo<sup>1</sup>. V&#225;rios estudos t&#234;m demonstrado a rela&#231;&#227;o de n&#237;veis aumentados do colesterol presente nas lipoprote&#237;nas de baixa densidade (LDL-C) com o risco de desenvolvimento da doen&#231;a<sup>2,3</sup>. O III Consenso Brasileiro sobre Dislipidemias estratifica faixas de valores de LDL-C para avalia&#231;&#227;o do risco para o desenvolvimento da doen&#231;a arterial coronariana: desej&#225;vel abaixo de 130 mg/dL, lim&#237;trofe entre 130-159 mg/dL, e alto a partir de 160 mg/dL<sup>4</sup>. Estas faixas de valores s&#227;o bastante estreitas, de modo que o <italic>National Colesterol Education Program &#150; NCEP </italic>estabeleceu que os laborat&#243;rios cl&#237;nicos devem utilizar metodologias para a dosagem do LDL-C com um erro anal&#237;tico total que n&#227;o exceda 12%, com imprecis&#227;o &lt;4% e inexatid&#227;o &lt;4%<sup>1</sup>. O m&#233;todo considerado de refer&#234;ncia para a determina&#231;&#227;o do LDL-C &#233; a b-quantifica&#231;&#227;o<sup>5</sup>, que requer ultracentrifuga&#231;&#227;o das amostras, impratic&#225;vel para os laborat&#243;rios de rotina. Por isso, a maioria dos laborat&#243;rios utiliza a estimativa do LDL-C pela f&#243;rmula de Friedewald, a partir das concentra&#231;&#245;es do colesterol total, do colesterol presente nas lipoprote&#237;nas de alta densidade (HDL-C), e dos triglicer&#237;deos<sup>6</sup>. Entretanto, o erro da determina&#231;&#227;o do LDL-C por esta estimativa compreende o erro anal&#237;tico somado dos tr&#234;s par&#226;metros utilizados no c&#225;lculo, e normalmente n&#227;o atende os crit&#233;rios de erro total do <italic>NCEP</italic>. Al&#233;m disso, o uso da f&#243;rmula tem severas limita&#231;&#245;es, n&#227;o podendo ser aplicada em amostras com triglicer&#237;deos (Trig) &gt;400 mg/dL, em amostras com quilom&#237;crons, nem em amostras de pacientes com disbetalipoproteinemia (Fredrickson Tipo III)<sup>7</sup>. Alguns autores demonstraram inclusive que a f&#243;rmula n&#227;o deve ser usada em certos grupos de pacientes, como diab&#233;ticos e pacientes com hepatopatias ou nefropatias, mesmo com n&#237;veis de triglicer&#237;deos &lt;400 mg/dL<sup>8</sup>. </p>
    <p>Recentemente, v&#225;rios m&#233;todos homog&#234;neos t&#234;m sido desenvolvidos por diferentes fabricantes para a dosagem direta do LDL-C, na expectativa de que sejam atendidos os crit&#233;rios do <italic>NCEP</italic> e as necessidades da comunidade m&#233;dica na preven&#231;&#227;o da doen&#231;a arterial coronariana e do infarto do mioc&#225;rdio. Estes m&#233;todos parecem ser melhores que os m&#233;todos anteriores que utilizam precipita&#231;&#227;o qu&#237;mica seletiva ou imunoprecipita&#231;&#227;o, que s&#227;o trabalhosos e apresentam um bias significativo em compara&#231;&#227;o com o m&#233;todo de refer&#234;ncia<sup>7, 9</sup>. No entanto, principalmente devido ao seu custo, a utiliza&#231;&#227;o destes reagentes nos laborat&#243;rios cl&#237;nicos n&#227;o vem sendo amplamente disseminada, o que acarreta a escassez de dados sobre o desempenho e a valida&#231;&#227;o dos m&#233;todos. No presente estudo propusemo-nos a avaliar o desempenho de um m&#233;todo direto homog&#234;neo para a dosagem do LDL-C em compara&#231;&#227;o com a estimativa pela f&#243;rmula de Friedewald, analisando uma grande amostragem, obtida em dois anos de experi&#234;ncia com estes reagentes. </p>
    <p>
      <bold>M&#233;todos</bold>
    </p>
    <p>Foram analisadas amostras de sangue de pacientes que procuraram nosso laborat&#243;rio, para realizar exames de colesterol total, LDL-C, HDL-C e triglicer&#237;deos, de janeiro 2000 a dezembro 2002. Foram analisadas amostras de 10.664 pacientes, sendo 5.846 (54,82%) do sexo feminino, e 4.818 (45,18%) do sexo masculino, com idades entre 14 e 93 anos. As amostras de sangue foram coletadas ap&#243;s jejum de 12 a 14h, incubadas em banho-maria por 15 minutos para coagula&#231;&#227;o, e centrifugadas a 2.000 x <italic>g </italic>por 5 minutos. O soro foi separado e os ensaios foram realizados no mesmo dia da coleta. </p>
    <p>As dosagens dos triglicer&#237;deos e do colesterol total foram realizadas com os reagentes <italic>Triglycerides FS (DiaSys Diagnostic Systems GmbH &amp; Co. KG, Holzheim, Germany), e Cholesterol (BioSystems S.A., Barcelona, Spain)</italic>, respectivamente, de acordo com as especifica&#231;&#245;es dos fabricantes, em um equipamento Spectrum CCX II <italic>(Abbott Diagnostics, Abbott Park, IL, USA)</italic>. Os testes foram calibrados com o <italic>CCX Multicalibrator Set</italic> (Abbott), com curvas de tr&#234;s pontos. </p>
    <p>A dosagem do LDL-C homog&#234;neo foi feita com o reagente LDL-C Select FS (DiaSys), de acordo com as especifica&#231;&#245;es do fabricante. O m&#233;todo &#233; baseado na prote&#231;&#227;o seletiva do LDL-C com a adi&#231;&#227;o do reagente 1 (Tamp&#227;o Good's pH 6,8 22 mmol/L, colesterol esterase <underline>&gt;</underline> 2 kU/L, colesterol oxidase <underline>&gt;</underline> 2 kU/L, N-(2-hidroxi-3-sulfopropil)-3,5-dimetoxianilina (H-DAOS) 0,43 mmol/L, catalase <underline>&gt;</underline> 400 kU/L, concentra&#231;&#245;es finais), enquanto o colesterol das outras lipoprote&#237;nas &#233; processado pela colesterol oxidase, e o per&#243;xido de hidrog&#234;nio formado &#233; degradado pela catalase. Ap&#243;s 5 minutos, com a adi&#231;&#227;o do reagente 2 (Tamp&#227;o Good's pH 7,0 22 mmol/L, 4-aminoantipirina 0,68 mmol/L, peroxidase <underline>&gt;</underline> 3 kU/L, concentra&#231;&#245;es finais), o LDL-C &#233; liberado para o processamento enzim&#225;tico e revela&#231;&#227;o pela rea&#231;&#227;o de Trinder. Todos os reagentes s&#227;o l&#237;quidos est&#225;veis. N&#227;o h&#225; interfer&#234;ncia com n&#237;veis de triglicer&#237;deos at&#233; 1.000 mg/dL, bilirrubina at&#233; 50 mg/dL, hemoglobina at&#233; 500 mg/dL ou &#225;cido asc&#243;rbico at&#233; 50 mg/dL, segundo o fabricante. Os testes foram realizados em um equipamentos Spectrum CCX II (Abbott) atrav&#233;s de programa&#231;&#227;o especial, e calibrados com TruCal (DiaSys). </p>
    <p>O LDL-C foi estimado pela f&#243;rmula de Friedewald: LDL-C = ColT&#150; HDL-C &#150; (Trig/5), para as amostras que tiveram resultado de triglicer&#237;deos &lt; 400 mg/dL <sup>6</sup>. </p>
    <p>A dosagem do HDL-C foi feita por um m&#233;todo homog&#234;neo sem precipita&#231;&#227;o, utilizando o reagente HDL-C Immuno FS (DiaSys). O m&#233;todo &#233; baseado na forma&#231;&#227;o de imunocomplexos das lipoprote&#237;nas LDL, VLDL e quilom&#237;crons com anticorpos anti-b lipoprote&#237;nas humanas ap&#243;s a adi&#231;&#227;o do reagente 1 (Tamp&#227;o Good's pH 7,0 26 mmol/L, 4-aminoantipirina 0,60 mmol/L, peroxidase 1.600 U/L, ascorbato oxidase 1.800 U/L, concentra&#231;&#245;es finais, e anticorpos de carneiro anti-b lipoprote&#237;nas humanas), seguida do processamento enzim&#225;tico do HDL-C com a adi&#231;&#227;o, ap&#243;s 5 minutos, do reagente 2 (Tamp&#227;o Good's pH 7,0 26 mmol/L, colesterol esterase 800 U/L, colesterol oxidase 4.000 U/L, N-etil-N-(2-hidroxi-3-sulfopropil)-3,5-dimetoxi-4-fluoranilina s&#243;dica 0,16 mmol/L). Todos os reagentes s&#227;o l&#237;quidos est&#225;veis. N&#227;o h&#225; interfer&#234;ncia com n&#237;veis de triglicer&#237;deos at&#233; 1.200 mg/dL, bilirrubina at&#233; 40 mg/dL, hemoglobina at&#233; 500 mg/dL ou &#225;cido asc&#243;rbico at&#233; 50 mg/dL, segundo o fabricante. Os testes foram realizados em um equipamentos Spectrum CCX II (Abbott) atrav&#233;s de programa&#231;&#227;o especial, e calibrados com TruCal (DiaSys). </p>
    <p>A determina&#231;&#227;o do coeficiente de varia&#231;&#227;o dos testes realizados foi feita atrav&#233;s da an&#225;lise dos resultados obtidos durante 20 dias consecutivos, utilizando al&#237;quotas de soro controle Accumark &#153; (<italic>Sigma Diagnostics, St. Louis, MO, USA</italic>), lote 111K6403. </p>
    <p>A compara&#231;&#227;o entre os m&#233;todos de dosagem do LDL-C homog&#234;neo e estimativa do LDL-C pela f&#243;rmula de Friedewald foi feita segundo m&#233;todo de Passing e Bablok<sup>10</sup>, atrav&#233;s de an&#225;lise de correla&#231;&#227;o, expressa pela equa&#231;&#227;o <italic>y =bx + a</italic>, onde <italic>b </italic>&#233; a inclina&#231;&#227;o da reta, representando o erro proporcional, e <italic>a </italic>&#233; a intersec&#231;&#227;o no eixo <italic>y</italic>, representando o erro constante. Para melhorar a compara&#231;&#227;o entre os m&#233;todos, as amostras foram estratificadas de acordo com os n&#237;veis de colesterol total: 70-150 mg/dL, 151-200 mg/dL, 201-250 mg/dL, 251-550 mg/dL, e com os n&#237;veis de triglicer&#237;deos: <underline>&lt;</underline> 150 mg/dL, 151-200 mg/dL, 201-300 mg/dL, e 301-400 mg/dL. </p>
    <p>A an&#225;lise estat&#237;stica dos resultados obtidos foi feita com o aux&#237;lio dos <italic>softwares GraphPad InStat &#153;</italic> e <italic>GraphPad Prism &#153;</italic> (San Diego, CA, USA). Foram considerados significativos valores de <italic>p </italic>&lt; 0,05. M&#233;dia, desvio padr&#227;o e coeficiente de varia&#231;&#227;o foram calculados com o aux&#237;lio do <italic>Software Excel</italic> (Microsoft). </p>
    <p>
      <bold>Resultados</bold>
    </p>
    <p>O coeficiente de varia&#231;&#227;o da dosagem do LDL-C pelo m&#233;todo homog&#234;neo foi de 4%. Para as dosagens do colesterol total, dos triglicer&#237;deos e do HDL-C, os coeficientes de varia&#231;&#227;o foram de 3%, 4% e 3%, respectivamente. As recomenda&#231;&#245;es do <italic>National Cholesterol Education Program &#150; NCEP </italic>s&#227;o dosagens do LDL-C com imprecis&#227;o <underline>&lt;</underline> 4%<sup>1</sup> e as dosagens realizadas em nosso laborat&#243;rio atenderam o crit&#233;rio. </p>
    <p>As concentra&#231;&#245;es de colesterol total e lipoprote&#237;nas obtidas neste estudo em rela&#231;&#227;o aos diferentes n&#237;veis de triglicer&#237;deos encontram-se na <xref ref-type="table" rid="tab01">tabela I</xref>. As concentra&#231;&#245;es de triglicer&#237;deos e lipoprote&#237;nas em rela&#231;&#227;o aos diferentes n&#237;veis de colesterol total na <xref ref-type="table" rid="tab02">tabela II</xref>. </p>
    <table-wrap id="tab01">
      <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gt01.jpg"/></table-wrap>
    <table-wrap id="tab02">
      <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gt02.jpg"/></table-wrap>
    <p>A compara&#231;&#227;o dos m&#233;todos [LDL-C homog&#234;neo (<italic>x</italic>) <italic>versus</italic> LDL-C calculado (<italic>y</italic>)] resultou em equa&#231;&#245;es de regress&#227;o de <italic>y</italic> = 0,6905 <italic>x </italic>+ 27,9 para n&#237;veis de colesterol total entre 70 e 150 mg/dL, de <italic>y</italic> = 0,6387 <italic>x </italic>+ 44,8 para n&#237;veis de colesterol entre 151 e 200 mg/dL, de <italic>y</italic> = 0,6039 <italic>x </italic>+ 63,7 para n&#237;veis de colesterol entre 201 e 250 mg/dL, e de <italic>y</italic> = 0,7256 <italic>x </italic>+ 60,7 para n&#237;veis de colesterol &gt; 250 mg/dL. Os coeficientes de correla&#231;&#227;o entre os m&#233;todos, de acordo com os n&#237;veis de colesterol total, de 0,6105, 0,6160, 0,6735 e 0,7822, respectivamente, foram extremamente significantes (&lt; 0,001). Entretanto, a estimativa do LDL-C pela f&#243;rmula de Friedewald tende a produzir resultados um pouco mais elevados em compara&#231;&#227;o com a dosagem do LDL-C pelo m&#233;todo homog&#234;neo. Este bias praticamente n&#227;o ocorre com n&#237;veis de colesterol at&#233; 150 mg/dL, onde foi observada uma diferen&#231;a m&#233;dia de 7 &#177; 12,1 mg/dL (11,5% &#177; 16,1%), com um erro proporcional de &#150; 30,9 % [(inclina&#231;&#227;o da reta - 1) x 100), e um erro constante de + 27,9 mg/dL. Mas, com n&#237;veis de colesterol entre 151 e 200 mg/dL, observamos um aumento do erro constante, que passa para + 44,8 mg/dL, com um erro proporcional de - 36 %, resultando num desvio m&#233;dio de 8 &#177; 14,5 mg/dL (9,5% &#177; 13,3%). Da mesma forma, com n&#237;veis de colesterol entre 201-250 mg/dL e &gt; 250 mg/dL, observamos um aumento do erro constante para + 63,7 mg/dL e + 60,7 mg/dL, com erros proporcionais de &#150; 40% e &#150; 27%, resultando em desvios m&#233;dios de 10 &#177; 15,3 mg/dL (8,3% &#177; 10,5%) e de 10 &#177; 20,4 mg/dL (6,5% &#177; 10,5%), respectivamente. Os resultados destas compara&#231;&#245;es est&#227;o resumidos na (<xref ref-type="table" rid="tab03">tab. III</xref> e <xref ref-type="fig" rid="fig1a">fig. 1</xref>). </p>
    <table-wrap id="tab03">
      <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gt03.jpg"/></table-wrap>
    <fig id="fig1a"/>
    <p>
      <graphic xlink:href="1678-4170-abc-83-06-0476-00006-e01.jpg"/></p>
    <p>
      <fig id="fig1b">
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gf01.jpg"/></fig>
    </p>
    <p>
      <fig id="fig1c">
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gf02.jpg"/></fig>
    </p>
    <p>A compara&#231;&#227;o dos m&#233;todos [LDL-C homog&#234;neo (<italic>x</italic>) <italic>versus</italic> LDL-C calculado (<italic>y</italic>)] resultou em equa&#231;&#245;es de regress&#227;o de <italic>y</italic> = 0,9746 <italic>x </italic>+ 17,6 para n&#237;veis de triglicer&#237;deos <underline>&lt;</underline> 150 mg/dL, de <italic>y</italic> = 0,9593 <italic>x </italic>+ 12,6 para n&#237;veis de triglicer&#237;deos entre 151-200 mg/dL, de <italic>y</italic> = 0,9459 <italic>x </italic>+ 6,7 para n&#237;veis de triglicer&#237;deos entre 201-300 mg/dL, e de <italic>y</italic> = 0,8999 <italic>x </italic>- 0,1 para n&#237;veis de triglicer&#237;deos entre 301-400 mg/dL. Os coeficientes de correla&#231;&#227;o entre os m&#233;todos, de acordo com os n&#237;veis de triglicer&#237;deos, de 0,9426, 0,9332, 0,9345 e 0,9072, respectivamente, foram extremamente significantes (&lt; 0,001). Entretanto, a estimativa do LDL-C pela f&#243;rmula de Friedewald tamb&#233;m tende a produzir resultados mais elevados em compara&#231;&#227;o com as dosagens do LDL-C pelo m&#233;todo homog&#234;neo para n&#237;veis de triglicer&#237;deos at&#233; 150 mg/dL, onde encontramos uma diferen&#231;a m&#233;dia de 14 &#177; 13 mg/dL (12,8% &#177; 9,2%), com um erro constante de + 17,6 mg/dL e um erro proporcional de &#150; 3%. Com n&#237;veis de triglicer&#237;deos entre 151-200 mg/dL este desvio tende a diminuir, apresentando uma diferen&#231;a m&#233;dia de 7 &#177; 14,5 mg/dL (5,3% &#177; 9,5%), onde observamos um erro proporcional de - 4 % com + 12,6 mg/dL de erro constante. Com n&#237;veis de triglicer&#237;deos entre 201 e 300 mg/dL este desvio praticamente n&#227;o existe: 2 &#177; 15,3 mg/dL (0,8% &#177; 10%), com um erro proporcional de - 5 % e um erro constante de + 6,7 mg/dL. Por outro lado, com n&#237;veis de triglicer&#237;deos entre 301-400 mg/dL, a estimativa do LDL-C pela f&#243;rmula de Friedewald tende a produzir resultados mais baixos, com uma diferen&#231;a m&#233;dia de &#150; 16 &#177; 19,4 mg/dL (- 10,3% &#177; 12,4%), em compara&#231;&#227;o com a dosagem do LDL-C pelo m&#233;todo homog&#234;neo, onde encontramos um erro proporcional de &#150; 10 %, com &#150; 0,1 mg/dL de erro constante (<xref ref-type="table" rid="tab03">tab. III</xref> e <xref ref-type="fig" rid="fig2a">fig.2</xref>). </p>
    <fig id="fig2a"/>
    <p>
      <graphic xlink:href="1678-4170-abc-83-06-0476-00006-e02.jpg"/></p>
    <p>
      <fig id="fig2d">
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gf03.jpg"/></fig>
    </p>
    <p>
      <bold>Discuss&#227;o</bold>
    </p>
    <p>Neste trabalho, propusemo-nos a avaliar o desempenho de um m&#233;todo homog&#234;neo para a dosagem direta do LDL-C, em compara&#231;&#227;o com a estimativa do LDL-C pela f&#243;rmula de Friedewald. Apesar das inova&#231;&#245;es tecnol&#243;gicas, a f&#243;rmula de Friedewald continua sendo usada por muitos laborat&#243;rios, e sua aplica&#231;&#227;o foi inclusive recomendada pelo III Consenso Brasileiro sobre Dislipidemias, para amostras com triglicer&#237;deos at&#233; 400 mg/dL. Entretanto, como outros autores<sup>9,11-17</sup> e por n&#243;s demonstrado, m&#233;todos homog&#234;neos e a f&#243;rmula de Friedewald n&#227;o s&#227;o capazes de fornecer resultados id&#234;nticos. Esta conclus&#227;o tamb&#233;m fica evidente analisando os resultados das amostras do presente estudo, de acordo com seus respectivos n&#237;veis de colesterol total ou de triglicer&#237;deos. </p>
    <p>Avaliando os resultados das amostras de acordo com diferentes n&#237;veis de colesterol total, pudemos observar que a estimativa do LDL-C pela F&#243;rmula de Friedewald apresenta uma correla&#231;&#227;o extremamente significativa (<italic>P </italic>&lt; 0,001) em compara&#231;&#227;o com o m&#233;todo direto (<xref ref-type="table" rid="tab02">tab. II</xref>). Por&#233;m, os coeficientes de correla&#231;&#227;o entre os dois m&#233;todos n&#227;o s&#227;o extremamente pr&#243;ximos (0,6105 a 0,7822) (<xref ref-type="table" rid="tab03">tab. III</xref>). De fato, a f&#243;rmula de Friedewald apresenta um desvio, ou bias, positivo em rela&#231;&#227;o ao m&#233;todo direto. Esse bias n&#227;o &#233; muito pronunciado com n&#237;veis de colesterol total entre 70-150 mg/dL (desvio m&#233;dio de 7 &#177; 21 mg/dL) nem com n&#237;veis de colesterol total entre 151-200 mg/dL (8 &#177; 14,5 mg/dL). Mas, a partir de n&#237;veis de colesterol total entre 201-250 mg/dL, este desvio tende a aumentar, com a m&#233;dia de 10 &#177; 15,3 mg/dL. Da mesma forma, com n&#237;veis de colesterol total &gt; 250 mg/dL, temos um bias positivo de 10 &#177; 20,4 mg/dL. Assim, um paciente que teria, por exemplo, um resultado de LDL-C de 125 mg/dL pelo m&#233;todo direto, poderia ter um resultado de 139 mg/dL pela estimativa da f&#243;rmula de Friedewald (<xref ref-type="fig" rid="fig1b">fig. 1B</xref>), considerando a equa&#231;&#227;o de regress&#227;o linear obtida para n&#237;veis de colesterol total entre 201-250 mg/dL (<italic>y</italic> = 0,6039 <italic>x</italic> + 63,7). Deste modo, apesar de n&#227;o haver signific&#226;ncia estat&#237;stica, teoricamente uma parcela da popula&#231;&#227;o poderia sair do limite desej&#225;vel para valores de LDL-C (&lt; 130 mg/dL) para um valor lim&#237;trofe (130-159 mg/dL), sujeita a controles da dieta ou at&#233; mesmo tratamento com estatinas. Da mesma forma, pacientes classificados dentro de valores lim&#237;trofes pelo m&#233;todo direto, poderiam cair na faixa de valores altos (160-189 mg/dL) com a estimativa pela f&#243;rmula de Friedewald (<xref ref-type="fig" rid="fig1c">fig. 1C</xref>). </p>
    <p>Por outro lado, avaliando os resultados destas amostras de acordo com diferentes n&#237;veis de triglicer&#237;deos, observamos um padr&#227;o de invers&#227;o deste bias (<xref ref-type="table" rid="tab01">tab. I</xref>), apesar de os dois m&#233;todos apresentarem excelentes coeficientes de correla&#231;&#227;o (0,9072 a 0,9426) (<xref ref-type="table" rid="tab03">tab. III</xref>). Com n&#237;veis de triglicer&#237;deos at&#233; 150 mg/dL, observamos uma bias positivo m&#233;dio de 14 &#177; 13 mg/dL da f&#243;rmula de Friedewald. Assim, mesmo sem haver signific&#226;ncia estat&#237;stica, teoricamente um paciente com resultado de LDL-C de 125 mg/dL pelo m&#233;todo direto e triglicer&#237;deos &lt; 150 mg/dL, poderia ter um valor de LDL-C estimado pela f&#243;rmula de Friedewald de 139 mg/dL (<xref ref-type="fig" rid="fig2a">fig. 2A</xref>), considerando a equa&#231;&#227;o de regress&#227;o linear para esta faixa de triglicer&#237;deos (<italic>y</italic> = 0,9746 <italic>x</italic> + 17,6). Esse paciente passaria de um valor desej&#225;vel para um valor lim&#237;trofe. O mesmo aconteceria com um paciente com 150 mg/dL de LDL-C pelo m&#233;todo direto, que passaria da faixa de valores lim&#237;trofes para a faixa de valores altos (164 mg/dL), pela estimativa da f&#243;rmula. </p>
    <p>V&#225;rios autores encontraram resultados mais baixos para m&#233;todos diretos de dosagem do LDL-C em compara&#231;&#227;o com a  b-quantifica&#231;&#227;o ap&#243;s ultracentrifuga&#231;&#227;o<sup>11, 12</sup>, enquanto outros observaram este bias com um determinado reagente, e uma correla&#231;&#227;o perfeita com outro reagente, apesar de ambos utilizarem detergentes, mas com princ&#237;pios diferentes<sup>13</sup>. Os reagentes utilizados no presente estudo seguem um princ&#237;pio de prote&#231;&#227;o com detergentes semelhante ao m&#233;todo utilizado no estudo citado, que n&#227;o apresentou bias em compara&#231;&#227;o com o m&#233;todo de refer&#234;ncia. Alguns autores tamb&#233;m n&#227;o encontraram esta varia&#231;&#227;o do bias em rela&#231;&#227;o a diferentes n&#237;veis de triglicer&#237;deos com um m&#233;todo que tamb&#233;m utiliza tensoativos espec&#237;ficos<sup>12</sup>, ao contr&#225;rio da maioria dos m&#233;todos diretos<sup>9, 14-18</sup>, o que pode ser devido a uma diferen&#231;a de componentes dos reagentes. </p>
    <p>Para n&#237;veis de triglicer&#237;deos entre 151-200 mg/dL, com nossos resultados este desvio diminui para 7 &#177; 14,5 mg/dL, e para n&#237;veis de triglicer&#237;deos entre 201-300 m/dL, este bias praticamente inexiste (2 &#177; 15,3 mg/dL). </p>
    <p>Por outro lado, nas amostras com n&#237;veis de triglicer&#237;deos entre 301-400 mg/dL, este bias da f&#243;rmula de Friedewald torna-se negativo, com um desvio m&#233;dio de &#150; 16 &#177; 19,4 mg/dL. Desta maneira, como pode ser demonstrado com nossos resultados, um paciente com triglicer&#237;deos entre 301-400 mg/dL e um resultado de LDL-C dosado pelo m&#233;todo direto de 140 mg/dL, teoricamente poderia ter um valor de LDL-C estimado pela F&#243;rmula de Friedewald de 126 mg/dL (<xref ref-type="fig" rid="fig2d">fig. 2D</xref>), considerando a equa&#231;&#227;o de regress&#227;o linear para esta faixa de triglicer&#237;deos (<italic>y</italic> = 0,8999<italic>x</italic> &#150; 0,1), apesar de n&#227;o haver signific&#226;ncia estat&#237;stica. O paciente passaria de um valor lim&#237;trofe para um valor desej&#225;vel de LDL-C. O mesmo aconteceria com um paciente com 170 mg/dL de LDL-C pelo m&#233;todo direto, que passaria da faixa de valores altos para a faixa de valores lim&#237;trofes (153 mg/dL). </p>
    <p>Uma das explica&#231;&#245;es para estes resultados mais altos de LDL-C pelo m&#233;todo direto em compara&#231;&#227;o com a f&#243;rmula de Friedewald, segundo alguns autores, seria a diferen&#231;a na rela&#231;&#227;o triglicer&#237;deos/colesterol nas part&#237;culas de VLDL, em pacientes com dislipidemias do tipo IIb, III e IV de Fredrickson. Part&#237;culas de VLDLs ricas em triglicer&#237;deos induziriam os m&#233;todos diretos a um bias negativo<sup>13.</sup> VLDLs ricas em colesterol, seriam respons&#225;veis por um bias positivo<sup>19</sup>. Entretanto, isto n&#227;o explicaria o bias positivo do m&#233;todo direto em pacientes com altos n&#237;veis de triglicer&#237;deos, encontrado em nossos resultados. </p>
    <p>Outra explica&#231;&#227;o seria a possibilidade da dosagem do colesterol presente nas part&#237;culas de lipoprote&#237;nas de densidade intermedi&#225;ria (IDL) como LDL-C, nos casos de pacientes dislipidemia do tipo III de Fredrickson, por alguns m&#233;todos diretos<sup>13,</sup> o que poderia estar acontecendo com o m&#233;todo em quest&#227;o, mas esta hip&#243;tese n&#227;o pode explicar esta tend&#234;ncia em todos os pacientes com triglicer&#237;deos entre 301-400 mg/dL, demonstrada por nossos resultados. De fato, como argumentam esses autores, esses pacientes com dislipidemia do tipo III tenderiam a ser erroneamente classificados no tipo IIb utilizando o m&#233;todo direto para a dosagem do LDL-C, superestimando a fra&#231;&#227;o LDL-C. Por&#233;m, a estimativa pela f&#243;rmula de Friedewald n&#227;o auxilia a classifica&#231;&#227;o correta, superestimando a fra&#231;&#227;o VLDL (triglicer&#237;deos/5), pois tamb&#233;m &#233; incapaz de indicar a presen&#231;a das IDLs, que poderia ser suspeitada apenas pelos altos n&#237;veis de triglicer&#237;deos. No caso espec&#237;fico dos pacientes com dislipidemia do tipo III, uma boa comunica&#231;&#227;o do cl&#237;nico com o laborat&#243;rio, que teria condi&#231;&#245;es de informar se o m&#233;todo que utiliza pode superestimar a fra&#231;&#227;o LDL-C, no caso do m&#233;todo direto, ou a fra&#231;&#227;o VLDL, no caso da estimativa pela f&#243;rmula de Friedewald, seria uma excelente contribui&#231;&#227;o para o diagn&#243;stico. </p>
    <p>Para a estimativa do LDL-C, a f&#243;rmula de Friedewald padroniza o valor da fra&#231;&#227;o VLDL como o valor dos triglicer&#237;deos dividido por cinco. Entretanto, as part&#237;culas encontradas nos pacientes com hipertrigliceridemia (tipos IIb, III, IV e V) s&#227;o normalmente uma mistura heterog&#234;nea de remanescentes de quilom&#237;crons, VLDL e remanescentes de VLDL (IDLs). Como &#233; sabido, a rela&#231;&#227;o triglicer&#237;deos/colesterol &#233; bastante vari&#225;vel entre esta gama de part&#237;culas. De fato, segundo nossos resultados, a estimativa do LDL-C pela f&#243;rmula de Friedewald apresenta um bom desempenho em amostras com n&#237;veis de triglicer&#237;deos entre 151-300 mg/dL, em compara&#231;&#227;o com o m&#233;todo direto. Entretanto, nos extremos desta faixa, nas amostras com &lt; 150 mg/dL e entre 301-400 mg/dL de triglicer&#237;deos, a f&#243;rmula n&#227;o apresenta um bom desempenho. De fato, a f&#243;rmula de Friedewald pode classificar erroneamente, segundo alguns autores, at&#233; 25% dos pacientes com valores de triglicer&#237;deos entre 301-400 mg/dL<sup>13</sup>. Esta percentagem pode ser ainda maior dependendo do m&#233;todo utilizado para a dosagem do HDL-C. </p>
    <p>Assim, pacientes com triglicer&#237;deos &lt;150 mg/dL com resultados desej&#225;veis de LDL-C pelo m&#233;todo direto, poderiam estar recebendo tratamento com base em resultados de LDL-C obtidos com a estimativa pela f&#243;rmula de Friedewald. Por outro lado, pacientes com triglicer&#237;deos entre 301-400 mg/dL e valores de LDL-C em n&#237;veis lim&#237;trofes ou altos pelo m&#233;todo direto, n&#227;o estariam recebendo o tratamento adequado com uso da f&#243;rmula de Friedewald. Obviamente os limites de valores de LDL-C para a classifica&#231;&#227;o de risco para doen&#231;a arterial coronariana s&#227;o bastante estreitos, mas como s&#227;o os valores recomendados pelo III Consenso Brasileiro sobre Dislipidemias<sup>4</sup>, o laborat&#243;rio cl&#237;nico deve esfor&#231;ar-se ao m&#225;ximo para realizar a dosagem desta fra&#231;&#227;o com o melhor desempenho diagn&#243;stico poss&#237;vel, isto &#233;, com o menor coeficiente de varia&#231;&#227;o, e com a probabilidade de melhor classificar corretamente os pacientes para a avalia&#231;&#227;o do risco de doen&#231;a arterial coronariana. </p>
    <p>Com este trabalho, analisando amostragem de mais de 10.000 pacientes, foi poss&#237;vel demonstrar que o m&#233;todo direto utilizado para dosagem do LDL-C tem um desempenho muito bom, com boa reprodutibilidade e coeficiente de varia&#231;&#227;o dentro das exig&#234;ncias do <italic>NCEP</italic>, dificilmente obtido com a f&#243;rmula de Friedewald. Entretanto, apesar de um dos objetivos ter sido discutir que os dois m&#233;todos n&#227;o apresentam resultados id&#234;nticos, &#233; importante salientar que ainda permanece o consenso de que a f&#243;rmula de Friedewald pode ser utilizada nos pacientes com n&#237;veis de triglicer&#237;deos at&#233; 400 mg/dL, que n&#227;o apresentem quilom&#237;crons e que n&#227;o tenham IDLs (classifica&#231;&#227;o de Fredrickson tipo III)<sup>4, 20</sup>. De fato, muitos laborat&#243;rios ainda continuam utilizando a estimativa do LDL-C pela f&#243;rmula, devido ao custo dos reagentes para a dosagem do LDL-C pelos m&#233;todos diretos existentes. Com a diminui&#231;&#227;o destes custos e com uma melhor avalia&#231;&#227;o do desempenho destes reagentes, os m&#233;todos diretos tenderiam a ser mais amplamente utilizados pelos laborat&#243;rios, possibilitando uma melhor classifica&#231;&#227;o dos pacientes, como resultados mais confi&#225;veis de LDL-C, dentro dos crit&#233;rios do <italic>NCEP</italic>. Certas popula&#231;&#245;es seriam extremamente beneficiadas com a utiliza&#231;&#227;o dos m&#233;todos diretos, como por exemplo os diab&#233;ticos, que s&#227;o naturalmente propensos a desenvolver doen&#231;a arterial coronariana, e cujos valores de LDL-C n&#227;o s&#227;o corretamente estimados pela f&#243;rmula<sup>8</sup>. </p>
    <p>No futuro, novos m&#233;todos podem estar sendo implementados para a determina&#231;&#227;o do LDL-C, como um m&#233;todo descrito recentemente, sem reagentes, baseado na absor&#231;&#227;o espectrofotom&#233;trica das lipoprote&#237;nas no infravermelho<sup>21</sup>. Este m&#233;todo apresentou excelente desempenho, e sua avalia&#231;&#227;o pode lan&#231;ar novas perspectivas no debate sobre a introdu&#231;&#227;o ou n&#227;o de novos m&#233;todos para a determina&#231;&#227;o do LDL-C. </p>
    <p>Recentemente, alguns autores demonstraram que dois m&#233;todos diretos para a dosagem do LDL-C n&#227;o apresentam uma boa recupera&#231;&#227;o do subtipo de LDL pequena e densa (sdLDL), previamente separada por ultracentrifuga&#231;&#227;o<sup>22</sup>. A determina&#231;&#227;o dos n&#237;veis deste subtipo de LDL tem sido implicada como um fator na avalia&#231;&#227;o do risco de desenvolvimento de doen&#231;a arterial coronariana mais sens&#237;vel do que a determina&#231;&#227;o do LDL-C total. M&#233;todos para a determina&#231;&#227;o rotineira de sdLDL ainda est&#227;o em desenvolvimento, mas possivelmente, num futuro pr&#243;ximo, os laborat&#243;rios estar&#227;o incluindo mais este par&#226;metro no perfil de lip&#237;deos e lipoprote&#237;nas &#224; disposi&#231;&#227;o dos cl&#237;nicos<sup>23</sup>. </p>
    <p>Da mesma forma, o desempenho do m&#233;todo direto para a dosagem do HDL-C apresentou um desempenho excelente, com um coeficiente de varia&#231;&#227;o muito baixo. Em nossa experi&#234;ncia, este m&#233;todo apresenta um &#243;timo desempenho na rotina do laborat&#243;rio, &#233; facilmente automatiz&#225;vel, e n&#227;o apresenta os problemas de interfer&#234;ncias, como os m&#233;todos de precipita&#231;&#227;o<sup>1</sup>, que acarretam um erro adicional na estimativa do LDL-C pela f&#243;rmula de Friedewald. Outros autores puderam demonstrar que o m&#233;todo direto n&#227;o apresenta resultados mais baixos de HDL-C como os m&#233;todos de precipita&#231;&#227;o, em compara&#231;&#227;o com o m&#233;todo de refer&#234;ncia<sup>24</sup>. </p>
    <p>
      <bold>Agradecimentos</bold>
    </p>
    <p>&#192; Empresa DiaSys Diagnostic Systems pelo suporte parcial na realiza&#231;&#227;o deste trabalho.</p>
    <p>
      <fn id="end" fn-type="corresp">
        <label>
          <bold>Endere&#231;o para correspond&#234;ncia</bold>
        </label>
        <p> Caio M. M. de Cordova </p>
        <p> Departamento de Ci&#234;ncias Farmac&#234;uticas/FURB Campus III </p>
        <p> R. Ant&#244;nio da Veiga, 140 </p>
        <p> CP 1507 - Cep 89010-971 - Blumenau, SC </p>
        <p> E-mail: </p>
        <p>
          <email xlink:href="cmcordova@furb.br">cmcordova@furb.br</email>
        </p>
      </fn>
    </p>
    <p>Recebido para Publica&#231;&#227;o em: 10/09/2003 </p>
    <p> Aceito em: 18/02/2004</p>
  </body>
  <back>
    <ref-list>
      <ref id="B1">
        <mixed-citation>1. NCEP. National Cholesterol Education Program.    Second report of the expert panel on detection, evaluation and treatment of    high blood cholesterol in adults (Adult Treatment Panel II). Circulation 1994;    89:1329-1345. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Second report of the expert panel on detection, evaluation and treatment of high blood cholesterol in adults (Adult Treatment Panel II)</article-title>
          <source>Circulation</source>
          <date>
            <year>1994</year>
          </date>
          <fpage>1329</fpage>
          <lpage>1345</lpage>
          <volume>89</volume>
        </element-citation>
      </ref>
      <ref id="B2">
        <mixed-citation>2. Grundy SM. Role of low-density lipoproteins    in atherogenesis and development of coronary heart disease. Clin. Chem. 1995;    41:139-146. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Role of low-density lipoproteins in atherogenesis and development of coronary heart disease</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>1995</year>
          </date>
          <fpage>139</fpage>
          <lpage>146</lpage>
          <volume>41</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Grundy</surname>
              <given-names>SM</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B3">
        <mixed-citation>3. Gordon T, Kannel WB, Castelli WP et al. Lipoproteins,    cardiovascular disease and death. The Framingham Study. Arch. Intern. Med. 1981;    141:1128-1130. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Lipoproteins, cardiovascular disease and death: The Framingham Study</article-title>
          <source>Arch. Intern. Med.</source>
          <date>
            <year>1981</year>
          </date>
          <fpage>1128</fpage>
          <lpage>1130</lpage>
          <volume>141</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Gordon</surname>
              <given-names>T</given-names>
            </name>
            <name>
              <surname>Kannel</surname>
              <given-names>WB</given-names>
            </name>
            <name>
              <surname>Castelli</surname>
              <given-names>WP</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B4">
        <mixed-citation>4. SBC. Sociedade Brasileira de Cardiologia.    Resumo das III Diretrizes Brasileiras sobre Dislipidemias e Diretriz de Preven&#231;&#227;o    da Aterosclerose da Sociedade Brasileira de Cardiologia. Arq. Bras. Cardiol.    2001; 77:1-48. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Resumo das III Diretrizes Brasileiras sobre Dislipidemias e Diretriz de Preven&#231;&#227;o da Aterosclerose da Sociedade Brasileira de Cardiologia</article-title>
          <source>Arq. Bras. Cardiol.</source>
          <date>
            <year>2001</year>
          </date>
          <fpage>1</fpage>
          <lpage>48</lpage>
          <volume>77</volume>
        </element-citation>
      </ref>
      <ref id="B5">
        <mixed-citation>5. Bachorik PS. Measurement of low density lipoprotein    cholesterol. In: Rifai N, Warnick GR, Dominiczak MH, eds. Handbook of Lipoprotein    Testing. Washington DC: AACC Press, 1997:145-160. </mixed-citation>
        <element-citation publication-type="book">
          <source>Handbook of Lipoprotein Testing</source>
          <date>
            <year>1997</year>
          </date>
          <fpage>145</fpage>
          <lpage>160</lpage>
          <person-group person-group-type="author">
            <name>
              <surname>Bachorik</surname>
              <given-names>PS</given-names>
            </name>
            <name>
              <surname>Rifai</surname>
              <given-names>N</given-names>
            </name>
            <name>
              <surname>Warnick</surname>
              <given-names>GR</given-names>
            </name>
            <name>
              <surname>Dominiczak</surname>
              <given-names>MH</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B6">
        <mixed-citation>6. Friedewald WT, Levi RI, Fredrickson DS. Estimation    of the concentration of low density lipoproteins cholesterol in plasma without    use of the ultracentrifuge. Clin. Chem. 1972; 18:499-502. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Estimation of the concentration of low density lipoproteins cholesterol in plasma without use of the ultracentrifuge</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>1972</year>
          </date>
          <fpage>499</fpage>
          <lpage>502</lpage>
          <volume>18</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Friedewald</surname>
              <given-names>WT</given-names>
            </name>
            <name>
              <surname>Levi</surname>
              <given-names>RI</given-names>
            </name>
            <name>
              <surname>Fredrickson</surname>
              <given-names>DS</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B7">
        <mixed-citation>7. McNamara JR, Conh JS, Wilson PWF et al. Calculated    values of low-density lipoprotein in the assessment of lipid abnormalities and    coronary disease risk. Clin. Chem. 1990; 36:36-42. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Calculated values of low-density lipoprotein in the assessment of lipid abnormalities and coronary disease risk</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>1990</year>
          </date>
          <fpage>36</fpage>
          <lpage>42</lpage>
          <volume>36</volume>
          <person-group person-group-type="author">
            <name>
              <surname>McNamara</surname>
              <given-names>JR</given-names>
            </name>
            <name>
              <surname>Conh</surname>
              <given-names>JS</given-names>
            </name>
            <name>
              <surname>Wilson</surname>
              <given-names>PWF</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B8">
        <mixed-citation>8. Rubies-Prat J, Rever&#233; RJ, Senti M et    al. Calculated low-density lipoprotein cholesterol should not be used form management    of lipoprotein abnormalities in patients with diabetes mellitus. Diabetes Care    1993; 16:1081-1086. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Calculated low-density lipoprotein cholesterol should not be used form management of lipoprotein abnormalities in patients with diabetes mellitus</article-title>
          <source>Diabetes Care</source>
          <date>
            <year>1993</year>
          </date>
          <fpage>1081</fpage>
          <lpage>1086</lpage>
          <volume>16</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Rubies-Prat</surname>
              <given-names>J</given-names>
            </name>
            <name>
              <surname>Rever&#233;</surname>
              <given-names>RJ</given-names>
            </name>
            <name>
              <surname>Senti</surname>
              <given-names>M</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B9">
        <mixed-citation>9. Harris N, Neufeld E, Newburgwe JW et al. Analytical    performance and clinical utility of a direct LDL-cholesterol assay in a hyperlipidemic    pediatric population. Clin. Chem. 1996; 42:1182-1188. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Analytical performance and clinical utility of a direct LDL-cholesterol assay in a hyperlipidemic pediatric population</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>1996</year>
          </date>
          <fpage>1182</fpage>
          <lpage>1188</lpage>
          <volume>42</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Harris</surname>
              <given-names>N</given-names>
            </name>
            <name>
              <surname>Neufeld</surname>
              <given-names>E</given-names>
            </name>
            <name>
              <surname>Newburgwe</surname>
              <given-names>JW</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B10">
        <mixed-citation>10. Passing H, Bablock W. Comparison of several    regression procedures for method comparison studies and determination of sample    size. J. Clin. Chem. Biochem. 1984; 22:431-445. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Comparison of several regression procedures for method comparison studies and determination of sample size</article-title>
          <source>J. Clin. Chem. Biochem.</source>
          <date>
            <year>1984</year>
          </date>
          <fpage>431</fpage>
          <lpage>445</lpage>
          <volume>22</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Passing</surname>
              <given-names>H</given-names>
            </name>
            <name>
              <surname>Bablock</surname>
              <given-names>W</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B11">
        <mixed-citation>11. Rifai N, Iannoti E, DeAngelis K et al. Analytical    and clinical performance of a homogeneous enzymatic LDL-cholesterol assay compared    with ultracentrifugation-dextran sulfate-Mg2+ method. Clin. Chem. 1998; 44:1242-1250.    </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Analytical and clinical performance of a homogeneous enzymatic LDL-cholesterol assay compared with ultracentrifugation-dextran sulfate-Mg2+ method</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>1998</year>
          </date>
          <fpage>1242</fpage>
          <lpage>1250</lpage>
          <volume>44</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Rifai</surname>
              <given-names>N</given-names>
            </name>
            <name>
              <surname>Iannoti</surname>
              <given-names>E</given-names>
            </name>
            <name>
              <surname>DeAngelis</surname>
              <given-names>K</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B12">
        <mixed-citation>12. Nauck M, Graziani MS, Bruton D et al. Analytical    and clinical performance of a detergent-based homogeneous LDL-cholesterol assay:    a multicenter evaluation. Clin. Chem. 2000; 46:506-514. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Analytical and clinical performance of a detergent-based homogeneous LDL-cholesterol assay: a multicenter evaluation</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>2000</year>
          </date>
          <fpage>506</fpage>
          <lpage>514</lpage>
          <volume>46</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Nauck</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Graziani</surname>
              <given-names>MS</given-names>
            </name>
            <name>
              <surname>Bruton</surname>
              <given-names>D</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B13">
        <mixed-citation>13. Esteban-Sal&#225;n M, Guim&#243;n-Berdesi    A, de la Viuda-Unzueta JM et al. Analytical and clinical evaluation of two homogeneous    assays for LDL-cholesterol in hyperlipidemic patients. Clin. Chem. 2000; 46:1121-1131.    </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Analytical and clinical evaluation of two homogeneous assays for LDL-cholesterol in hyperlipidemic patients</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>2000</year>
          </date>
          <fpage>1121</fpage>
          <lpage>1131</lpage>
          <volume>46</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Esteban-Sal&#225;n</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Guim&#243;n-Berdesi</surname>
              <given-names>A</given-names>
            </name>
            <name>
              <surname>de la Viuda-Unzueta</surname>
              <given-names>JM</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B14">
        <mixed-citation>14. Yu HH, Ginsbrug GS, Harris N et al. Evaluation    and clinical application of a direct low density lipoprotein cholesterol assay    in normolipemic and hyperlipidemic adults. Am. J. Cardiol 1997; 80:1295-1299.    </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Evaluation and clinical application of a direct low density lipoprotein cholesterol assay in normolipemic and hyperlipidemic adults</article-title>
          <source>Am. J. Cardiol</source>
          <date>
            <year>1997</year>
          </date>
          <fpage>1295</fpage>
          <lpage>1299</lpage>
          <volume>80</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Yu</surname>
              <given-names>HH</given-names>
            </name>
            <name>
              <surname>Ginsbrug</surname>
              <given-names>GS</given-names>
            </name>
            <name>
              <surname>Harris</surname>
              <given-names>N</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B15">
        <mixed-citation>15. Sheikh M, Miller NE. Evaluation of commercial    reagent for precipitating human serum low density lipoprotein. Clin. Chem. Acta    1985; 152:213-217. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Evaluation of commercial reagent for precipitating human serum low density lipoprotein</article-title>
          <source>Clin. Chem. Acta</source>
          <date>
            <year>1985</year>
          </date>
          <fpage>213</fpage>
          <lpage>217</lpage>
          <volume>152</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Sheikh</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Miller</surname>
              <given-names>NE</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B16">
        <mixed-citation>16. Demacker PN, Hijmans AG, Brenninkmeijer BJ    et al. Five methods for determining low density lipoprotein cholesterol compared.    Clin. Chem. 1984; 30:1797-1800. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Five methods for determining low density lipoprotein cholesterol compared</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>1984</year>
          </date>
          <fpage>1797</fpage>
          <lpage>1800</lpage>
          <volume>30</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Demacker</surname>
              <given-names>PN</given-names>
            </name>
            <name>
              <surname>Hijmans</surname>
              <given-names>AG</given-names>
            </name>
            <name>
              <surname>Brenninkmeijer</surname>
              <given-names>BJ</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B17">
        <mixed-citation>17. Assmann G, Jabs HU, Nolte W et al. Precipitation    of LDL with sulphopolyanions: a comparison of two methods for LDL cholesterol    determination. J. Clin. Chem. Clin. Biochem. 1984; 22:781-785. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Precipitation of LDL with sulphopolyanions: a comparison of two methods for LDL cholesterol determination</article-title>
          <source>J. Clin. Chem. Clin. Biochem.</source>
          <date>
            <year>1984</year>
          </date>
          <fpage>781</fpage>
          <lpage>785</lpage>
          <volume>22</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Assmann</surname>
              <given-names>G</given-names>
            </name>
            <name>
              <surname>Jabs</surname>
              <given-names>HU</given-names>
            </name>
            <name>
              <surname>Nolte</surname>
              <given-names>W</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B18">
        <mixed-citation>18. Mulder K, van Leeuwen C, Schouten JA et al.    An evaluation of three commercial methods for the determination of LDL cholesterol.    Clin. Chem. Acta 1984; 143:29-35. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>An evaluation of three commercial methods for the determination of LDL cholesterol</article-title>
          <source>Clin. Chem. Acta</source>
          <date>
            <year>1984</year>
          </date>
          <fpage>29</fpage>
          <lpage>35</lpage>
          <volume>143</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Mulder</surname>
              <given-names>K</given-names>
            </name>
            <name>
              <surname>van Leeuwen</surname>
              <given-names>C</given-names>
            </name>
            <name>
              <surname>Schouten</surname>
              <given-names>JA</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B19">
        <mixed-citation>19. Okada M, Matsui H, Ito Y et al. Low-density    lipoprotein cholesterol can be chemically measured: a new superior method. J.    Lab. Clin. Med. 1998; 132. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Low-density lipoprotein cholesterol can be chemically measured: a new superior method</article-title>
          <source>J. Lab. Clin. Med.</source>
          <date>
            <year>1998</year>
          </date>
          <volume>132</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Okada</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Matsui</surname>
              <given-names>H</given-names>
            </name>
            <name>
              <surname>Ito</surname>
              <given-names>Y</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B20">
        <mixed-citation>20. Nauck M, Warnick GR, Rifai N. Methods for    measurement of LDL-cholesterol: a critical assessment of direct measurement    by homogeneous assays versus calculation. Clin. Chem. 2002; 48:236-54. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Methods for measurement of LDL-cholesterol: a critical assessment of direct measurement by homogeneous assays versus calculation</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>2002</year>
          </date>
          <fpage>236</fpage>
          <lpage>54</lpage>
          <volume>48</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Nauck</surname>
              <given-names>M</given-names>
            </name>
            <name>
              <surname>Warnick</surname>
              <given-names>GR</given-names>
            </name>
            <name>
              <surname>Rifai</surname>
              <given-names>N</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B21">
        <mixed-citation>21. Liu KZ, Shaw RA, Man A et al. Reagent-free,    simultaneous determination of serum cholesterol in HDL and LDL by infrared spectroscopy.    Clin. Chem. 2002; 48:499-506. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Reagent-free, simultaneous determination of serum cholesterol in HDL and LDL by infrared spectroscopy</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>2002</year>
          </date>
          <fpage>499</fpage>
          <lpage>506</lpage>
          <volume>48</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Liu</surname>
              <given-names>KZ</given-names>
            </name>
            <name>
              <surname>Shaw</surname>
              <given-names>RA</given-names>
            </name>
            <name>
              <surname>Man</surname>
              <given-names>A</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B22">
        <mixed-citation>22. Usui S, Kakuuchi H, Okamoto M et al. Differential    reactivity of two homogeneous LDL-cholesterol methods to LDL and VLDL subfractions,    as demonstrated by ultracentrifugation and HPLC. Clin. Chem. 2002; 48:1946-54.    </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Differential reactivity of two homogeneous LDL-cholesterol methods to LDL and VLDL subfractions, as demonstrated by ultracentrifugation and HPLC</article-title>
          <source>Clin. Chem.</source>
          <date>
            <year>2002</year>
          </date>
          <fpage>1946</fpage>
          <lpage>54</lpage>
          <volume>48</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Usui</surname>
              <given-names>S</given-names>
            </name>
            <name>
              <surname>Kakuuchi</surname>
              <given-names>H</given-names>
            </name>
            <name>
              <surname>Okamoto</surname>
              <given-names>M</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B23">
        <mixed-citation>23. Le NA. Small, dense low-density lipoprotein:    Risk or Mith. Current Atherosclerosis Reviews 2003; 5:22-28. </mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Small, dense low-density lipoprotein: Risk or Mith</article-title>
          <source>Current Atherosclerosis Reviews</source>
          <date>
            <year>2003</year>
          </date>
          <fpage>22</fpage>
          <lpage>28</lpage>
          <volume>5</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Le</surname>
              <given-names>NA</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
      <ref id="B24">
        <mixed-citation>24. Jensen T, Truong Q, Frandsen M et al. Comparison    of a homogeneous assay with a precipitation method for the measurement of HDL    cholesterol in diabetic patients. Diabetes Care 2002; 25:1914-8.</mixed-citation>
        <element-citation publication-type="journal">
          <article-title>Comparison of a homogeneous assay with a precipitation method for the measurement of HDL cholesterol in diabetic patients</article-title>
          <source>Diabetes Care</source>
          <date>
            <year>2002</year>
          </date>
          <fpage>1914</fpage>
          <lpage>8</lpage>
          <volume>25</volume>
          <person-group person-group-type="author">
            <name>
              <surname>Jensen</surname>
              <given-names>T</given-names>
            </name>
            <name>
              <surname>Truong</surname>
              <given-names>Q</given-names>
            </name>
            <name>
              <surname>Frandsen</surname>
              <given-names>M</given-names>
            </name>
          </person-group>
        </element-citation>
      </ref>
    </ref-list>
  </back>
  <sub-article article-type="translation" id="TRen" xml:lang="en">
    <front-stub>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>Original Articles</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title xml:lang="en">Comparison of LDL-cholesterol direct measurement with the estimate using the Friedewald formula in a sample of 10,664 patients</article-title>
      </title-group>
      <abstract xml:lang="en">
        <p>OBJECTIVE: To compare direct measurement of LDL-cholesterol (LDL-C) determined by a homogeneous method with LDL-cholesterol estimation determined by the Friedewald formula in a large heterogeneous population. METHODS: The measurements of total cholesterol (TC) and triglycerides (TG) were performed using traditional enzymatic methods. The measurements of HDL-C and LDL-C were performed using direct methods with no precipitation, and the estimation of the LDL-C fraction was calculated using the Friedewald formula. RESULTS: On linear regression analysis, the 2 methods had extremely significant correlation coefficients (P &lt; 0.001). However, the Friedewald formula had a positive bias in regard to the direct method, more pronounced with TC levels &gt; 201 mg/dL. This positive bias also occurred in regard to TG levels &lt; 150 mg/dL. No bias was observed between the methods for TG levels ranging from 151 to 200 mg/dL and from 201 to 300 mg/dL. On the other hand, for TG levels ranging from 301 to 400 mg/dL, this bias of the Friedewald formula became negative. CONCLUSION: The Friedewald formula did not have a homogeneous performance for estimating LDL-C levels in samples with different TG levels as compared with that of the direct method, what could launch doubts on patients classification on the risk of developing coronary artery disease.</p>
      </abstract>
      <kwd-group xml:lang="en">
        <kwd>cholesterol</kwd>
        <kwd>LDL-C</kwd>
        <kwd>direct LDL</kwd>
        <kwd>Friedewald formula</kwd>
      </kwd-group>
    </front-stub>
    <body>
      <p>
        <bold>ORIGINAL ARTICLE</bold>
      </p>
      <p>Comparison of LDL-cholesterol direct measurement with the estimate using the Friedewald formula in a sample of 10,664 patients</p>
      <p>
        <bold>Caio Maur&#237;cio Mendes de Cordova; Carlos Rudi Schneider; Iara Deise Juttel; Maur&#237;cio Mendes de Cordova</bold>
      </p>
      <p> Blumenau, SC - Brazil </p>
      <p>Departamento de Ci&#234;ncias Farmac&#234;uticas/FURB and Laborat&#243;rio Santa Isabel de An&#225;lises Cl&#237;nicas, Blumenau, SC </p>
      <p>
        <xref ref-type="corresp" rid="end-body2">Correspondence</xref>
      </p>
      <p>
        <bold>ABSTRACT</bold>
      </p>
      <p><bold>OBJECTIVE:</bold> To compare direct measurement of LDL-cholesterol (LDL-C) determined by a homogeneous method with LDL-cholesterol estimation determined by the Friedewald formula in a large heterogeneous population.</p>
      <p><bold>METHODS: </bold>The measurements of total cholesterol (TC) and triglycerides (TG) were performed using traditional enzymatic methods. The measurements of HDL-C and LDL-C were performed using direct methods with no precipitation, and the estimation of the LDL-C fraction was calculated using the Friedewald formula. </p>
      <p><bold>RESULTS: </bold>On linear regression analysis, the 2 methods had extremely significant correlation coefficients (P &lt; 0.001). However, the Friedewald formula had a positive bias in regard to the direct method, more pronounced with TC levels &gt; 201 mg/dL. This positive bias also occurred in regard to TG levels <underline>&lt;</underline> 150 mg/dL. No bias was observed between the methods for TG levels ranging from 151 to 200 mg/dL and from 201 to 300 mg/dL. On the other hand, for TG levels ranging from 301 to 400 mg/dL, this bias of the Friedewald formula became negative. </p>
      <p><bold>CONCLUSION:</bold> The Friedewald formula did not have a homogeneous performance for estimating LDL-C levels in samples with different TG levels as compared with that of the direct method, what could launch doubts on patients classification on the risk of developing coronary artery disease. </p>
      <p><bold>Key words: </bold>cholesterol, LDL-C, direct LDL, Friedewald formula</p>
      <p>Coronary artery disease accounts for the greatest number of deaths of adult individuals worldwide <sup>1</sup>. Several studies have shown the correlation existing between the increased levels of cholesterol in low-density lipoproteins (LDL-C) and the risk of developing that disease <sup>2,3</sup>. The III Brazilian Consensus on Dyslipidemias stratifies the following ranges of LDL-C levels for assessing the risk of developing coronary artery disease: desirable, below 130 mg/dL; borderline, between 130 and 159 mg/dL; and high, above 160 mg/dL <sup>4</sup>. These ranges are very narrow, and the National Cholesterol Education Program (NCEP) established that clinical laboratories should use methodologies for measuring LDL-C levels with a total analytical error &lt; 12%, imprecision &lt; 4%, and inaccuracy &lt; 4% <sup>1</sup>. The reference method for determining LDL-C is b-quantitation <sup>5</sup>, which requires ultracentrifugation of the samples, which is not feasible in laboratory routine. Therefore, most laboratories estimate LDL-C levels with the Friedewald formula, based on the concentrations of total cholesterol, the cholesterol present in high-density lipoproteins (HDL-C), and triglycerides <sup>6</sup>. However, the error of determining LDL-C through that estimation comprises the addition of the analytical errors of the 3 parameters used in the calculus, which usually does not meet the NCEP criteria for total error. In addition, the use of that formula has severe limitations and cannot be applied to samples containing triglyceride (TG) levels &gt; 400 mg/dL, to samples with chylomicrons, and to samples of patients with dysbetalipoproteinemia (Fredrickson Type III) <sup>7</sup>. Some authors have demonstrated that that formula should not be used in certain groups of patients, such as patients with diabetes, hepatopathies, or nephropathies, even with triglyceride levels &lt; 400 mg/dL <sup>8</sup>. </p>
      <p>Recently, several homogeneous methods have been developed by different manufacturers for the direct measurement of LDL-C levels, expecting that the NCEP criteria are met, as well as that the medical community's need to prevent coronary artery disease and myocardial infarction are fulfilled. These methods seem to be better than the previous ones that use selective chemical precipitation or immunoprecipitation, which are laborious and have a significant bias as compared with the reference method <sup>7, 9</sup>. However, mainly due to the costs of the reagents, their use in clinical laboratories has not been largely disseminated, resulting in scarcity of data about the performance and validation of those methods. This study aimed at assessing the performance of a direct homogeneous method for measuring LDL-C and at comparing it with the estimation of LDL-C levels using the Friedewald formula, analyzing a large sample obtained over 2 years of experience with those reagents.</p>
      <p>
        <bold>Methods</bold>
      </p>
      <p>This study assessed the blood samples of 10,664 patients who sought our laboratory to undergo total cholesterol, LDL-C, HDL-C, and triglyceride measurements from January 2000 to December 2002. Their ages ranged from 14 to 93 years; 5,846 (54.82%) were females, and 4,818 (45.18%) were males. Blood samples were collected after a 12- to 14-hour fast, incubated in a hot-water bath for 15 minutes for coagulation, and centrifuged at 2,000 x g for 5 minutes. The serum was separated and the assays were performed on the same day of sample collection. </p>
      <p>The measurements of the triglycerides and total cholesterol were performed with the reagents Triglycerides FS (DiaSys Diagnostic Systems GmbH &amp; Co. KG, Holzheim, Germany), and Cholesterol (BioSystems S.A., Barcelona, Spain), respectively, according to the specifications of the manufacturers, in a Spectrum CCX II device (Abbott Diagnostics, Abbott Park, IL, USA). The tests were calibrated with the CCX Multicalibrator Set (Abbott), with curves of 3 points. </p>
      <p>The LDL-C measurement with the homogeneous method was performed with the reagent LDL-C Select FS (DiaSys), according to the specifications of the manufacturer. The method is based on the selective protection of LDL-C with the addition of reagent 1 [Good's buffer, pH 6.8, 22 mmol/L, cholesterol esterase <underline>&gt;</underline> 2 kU/L, cholesterol oxidase <underline>&gt;</underline> 2 kU/L, N-(2-hydroxy-3-sulfopropyl)-3,5-dimethoxyaniline (H-DAOS) 0.43 mmol/L, catalase <underline>&gt;</underline> 400 kU/L, final concentrations]. The cholesterol of the other lipoproteins is processed by cholesterol oxidase, and the hydrogen peroxide formed is broken down by catalase. After 5 minutes, with the addition of reagent 2 (Good's buffer, pH 7.0, 22 mmol/L, 4-aminoantipyrine 0.68 mmol/L, peroxidase <underline>&gt;</underline> 3 kU/L, final concentrations), LDL-C is released for enzymatic processing and development by the Trinder reaction. All reagents are stable fluids. According to the manufacturer, no interference occurs with triglyceride levels up to 1,000 mg/dL, bilirubin up to 50 mg/dL, hemoglobin up to 500 mg/dL, or ascorbic acid up to 50 mg/dL. The tests were performed using special programming in a Spectrum CCX II device (Abbott) calibrated with TruCal (DiaSys). </p>
      <p>For samples with triglyceride levels &lt; 400 mg/dL, the LDL-C level was estimated using the Friedewald formula: LDL-C = TC &#150; HDL-C &#150; (TG/5) <sup>6</sup>. </p>
      <p>The HDL-C measurement was performed using a homogeneous method without precipitation with the HDL-C Immuno FS reagent (DiaSys). The method is based on the formation of immunocomplexes of LDL and VLDL lipoproteins and chylomicrons with human anti-b lipoprotein antibodies after the addition of reagent 1 (Good's buffer, pH 7.0, 26 mmol/L, 4-aminoantipyrine 0.60 mmol/L, peroxidase 1,600 U/L, ascorbate oxidase 1,800 U/L, final concentrations, and sheep antibodies anti human-b lipoproteins). Then, enzymatic processing of HDL-C occurred with the addition, after 5 minutes, of reagent 2 [Good's buffer, pH 7.0, 26 mmol/L, cholesterol esterase 800 U/L, cholesterol oxidase 4,000 U/L, sodium N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3,5-dimethoxy-4-fluoroaniline 0.16 mmol/L]. All reagents are stable fluids. According to the manufacturer, no interference occurs with triglyceride levels up to 1,200 mg/dL, bilirubin up to 40 mg/dL, hemoglobin up to 500 mg/dL, or ascorbic acid up to 50 mg/dL. The tests were performed using special programming in a Spectrum CCX II device (Abbott) calibrated with TruCal (DiaSys). </p>
      <p>The coefficient of variation of the tests performed was determined by analysis of the results obtained during 20 consecutive days using Accumark control serum aliquots (Sigma Diagnostics, St. Louis, MO, USA), 111K6403 lot. </p>
      <p>The comparison between the LDL-C measurement methods, homogeneous and estimation through the Friedewald formula, was performed according to the Passing and Bablok method <sup>10</sup>, using analysis of correlation expressed by the equation y = bx + a, where <italic>b </italic>is the gradient of the line and represents the proportional error, and <italic>a </italic>is the intersection in the <italic>y</italic> axis and represents the constant error. To improve the comparison between the methods, the samples were stratified according to total cholesterol levels (70-150 mg/dL, 151-200 mg/dL, 201-250 mg/dL, and 251-550 mg/dL) and to triglyceride levels (<underline>&lt;</underline> 150 mg/dL, 151-200 mg/dL, 201-300 mg/dL, and 301-400 mg/dL). </p>
      <p>The statistical analysis of the results was performed with the aid of GraphPad InStat and GraphPad Prism software (San Diego, CA, USA). The significance level adopted was P &lt; 0.05. Mean, standard deviation, and coefficient of variation were calculated with the aid of Excel software (Microsoft). </p>
      <p>
        <bold>Results</bold>
      </p>
      <p>The coefficient of variation of the LDL-C measurement using the homogeneous method was 4%. The coefficients of variation of the total cholesterol, triglyceride, and HDL-C measurements were 3%, 4%, and 3%, respectively. In regard to LDL-C measurements, the National Cholesterol Education Program (NCEP) recommends imprecision <underline>&lt;</underline> 4% <sup>1</sup>. The measurements performed in our laboratory met that criterion. </p>
      <p>The concentrations of total cholesterol and lipoproteins obtained in this study in regard to the different triglyceride levels are shown in <xref ref-type="table" rid="tab01-body2">table I</xref>. The concentrations of triglycerides and lipoproteins in regard to the different levels of total cholesterol are shown in <xref ref-type="table" rid="tab02-body2">table II</xref>. </p>
      <table-wrap id="tab01-body2">
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gt01-en.jpg"/></table-wrap>
      <table-wrap id="tab02-body2">
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gt02-en.jpg"/></table-wrap>
      <p>The comparison of the methods [homogeneous LDL-C (<italic>x</italic>) versus calculated LDL-C (<italic>y</italic>)] resulted in the following regression equations: <italic>y</italic> = 0.6905 <italic>x </italic>+ 27.9, for total cholesterol levels between 70 and 150 mg/dL; <italic>y</italic> = 0.6387 <italic>x </italic>+ 44.8, for total cholesterol levels between 151 and 200 mg/dL; <italic>y</italic> = 0.6039 <italic>x </italic>+ 63.7, for total cholesterol levels between 201 and 250 mg/dL; and <italic>y</italic> = 0.7256 <italic>x </italic>+ 60.7, for total cholesterol levels &gt; 250 mg/dL. According to the total cholesterol levels, the correlation coefficients between the methods of 0.6105, 0.6160, 0.6735, and 0.7822, respectively, were extremely significant (&lt; 0.001). However, the LDL-C estimate determined with the Friedewald formula tends towards producing slightly greater results as compared with those from LDL-C measurements using the homogeneous method. This bias practically does not occur with cholesterol levels up to 150 mg/dL, in which a mean difference of 7 &#177; 12.1 mg/dL (11.5% &#177; 16.1%) was observed, with a proportional error of &#150;30.9% [(gradient of the line - 1) x 100), and a constant error of +27.9 mg/dL. However, with cholesterol levels between 151 and 200 mg/dL, the constant error increased to +44.8 mg/dL, and the proportional error was -36%, which resulted in a mean deviation of 8 &#177; 14.5 mg/dL (9.5% &#177; 13.3%). In the same way, with cholesterol levels between 201 and 250 mg/dL and &gt; 250 mg/dL, the constant error increased to +63.7 mg/dL and +60.7 mg/dL, respectively, with proportional errors of &#150;40% and &#150;27%, respectively. This resulted in mean deviations of 10 &#177; 15.3 mg/dL (8.3% &#177; 10.5%), and 10 &#177; 20.4 mg/dL (6.5% &#177; 10.5%), respectively. The results of these comparisons are summarized in <xref ref-type="table" rid="tab03-body2">table III</xref> and <xref ref-type="fig" rid="fig1a-body2">figure 1</xref>. </p>
      <table-wrap id="tab03-body2">
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gt03-en.jpg"/></table-wrap>
      <fig id="fig1a-body2"/>
      <p>
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-e01-en.jpg"/></p>
      <p>
        <fig id="fig1b-body2">
          <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gf01-en.jpg"/></fig>
      </p>
      <p>
        <fig id="fig1c-body2">
          <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gf02-en.jpg"/></fig>
      </p>
      <p>The comparison of the methods [homogeneous LDL-C (<italic>x</italic>) versus calculated LDL-C (<italic>y</italic>)] resulted in the following regression equations: <italic>y</italic> = 0.9746 <italic>x </italic>+ 17.6, for triglyceride levels <underline>&lt;</underline> 150 mg/dL; <italic>y</italic> = 0.9593 <italic>x </italic>+ 12.6, for triglyceride levels between 151 and 200 mg/dL; <italic>y</italic> = 0.9459 <italic>x </italic>+ 6.7, for triglyceride levels between 201 and 300 mg/dL; and <italic>y</italic> = 0.8999 <italic>x </italic>- 0.1, for triglyceride levels between 301 and 400 mg/dL. In those different triglyceride levels, the correlation coefficients between the methods of 0.9426, 0.9332, 0.9345, and 0.9072, respectively, were extremely significant (&lt; 0.001). However, the LDL-C estimate using the Friedewald formula also tends towards producing greater results as compared with those obtained in LDL-C measurements through the homogeneous method for triglyceride levels up to 150 mg/dL, in which a mean difference of 14 &#177; 13 mg/dL (12.8% &#177; 9.2%) was found, with a constant error of +17.6 mg/dL, and a proportional error of &#150;3%. With triglyceride levels between 151 and 200 mg/dL, this deviation tends towards decreasing, with a mean difference of 7 &#177; 14.5 mg/dL (5.3% &#177; 9.5%), a proportional error of -4%, and a constant error of +12.6 mg/dL. With triglyceride levels between 201 and 300 mg/dL, this deviation almost does not exist: 2 &#177; 15.3 mg/dL (0.8% &#177; 10%), with a proportional error of -5% and a constant error of +6.7 mg/dL. On the other hand, with triglyceride levels between 301 and 400 mg/dL, the LDL-C estimate with the Friedewald formula tends towards producing lower values, with a mean difference of -16 &#177; 19.4 mg/dL (-10.3% &#177; 12.4%), as compared with the LDL-C measurement using the homogeneous method, in which a proportional error of -10% and a constant error of -0.1 mg/dL are observed (<xref ref-type="table" rid="tab03-body2">tab. III</xref> and <xref ref-type="fig" rid="fig2a-body2">fig.2</xref>). </p>
      <fig id="fig2a-body2"/>
      <p>
        <graphic xlink:href="1678-4170-abc-83-06-0476-00006-e02-en.jpg"/></p>
      <p>
        <fig id="fig2d-body2">
          <graphic xlink:href="1678-4170-abc-83-06-0476-00006-gf03-en.jpg"/></fig>
      </p>
      <p>
        <bold>Discussion</bold>
      </p>
      <p>This study aimed at assessing the performance of a homogeneous method for direct LDL-C measurement, as compared with the LDL-C estimate by using the Friedewald formula. Despite the technological innovations, the Friedewald formula continues to be used in many laboratories, and its application has been recommended by the III Brazilian Consensus on Dyslipidemias for samples with triglyceride levels up to 400 mg/dL. However, as already reported by other authors <sup>9,11-17</sup> and ourselves, the homogeneous methods and the Friedewald formula are not capable of providing identical results. This conclusion is also evident when the results of the samples of this study are analyzed according to their respective levels of total cholesterol and triglycerides. </p>
      <p>Assessing the results according to the different levels of total cholesterol, the LDL-C estimate through the Friedewald formula was observed to have an extremely significant correlation (P<italic> &lt;</italic> 0.001), as compared with the direct method (<xref ref-type="table" rid="tab02-body2">tab. II</xref>). However, the correlation coefficients between the 2 methods are not extremely close (0.6105 to 0.7822) (<xref ref-type="table" rid="tab03-body2">tab. III</xref>). In fact, the Friedewald formula has a positive deviation or bias in regard to the direct method. This bias is neither very pronounced with total cholesterol levels between 70 and 150 mg/dL (mean deviation of 7 &#177; 21 mg/dL), nor with total cholesterol levels between 151 and 200 mg/dL (8 &#177; 14.5 mg/dL). But, from total cholesterol levels between 201 and 250 mg/dL onwards, this deviation tends towards increasing, with a mean of 10 &#177; 15.3 mg/dL. Similarly, with total cholesterol levels &gt; 250 mg/dL, a positive bias of 10 &#177; 20.4 mg/dL occurs. Therefore, if a patient had an LDL-C level of 125 mg/dL on the direct method, the result by using the estimate of the Friedewald formula could be 139 mg/dL (<xref ref-type="fig" rid="fig1b-body2">fig. 1B</xref>), considering the linear regression equation obtained for total cholesterol levels between 201 and 250 mg/dL (<italic>y</italic> = 0.6039 <italic>x</italic> + 63.7). Therefore, although with no statistical significance, theoretically, LDL-C levels of a part of the population could pass from desirable values (&lt; 130 mg/dL) to borderline values (130-159 mg/dL), which are subject to dietary control and even treatment with statins. Similarly, patients classified as within borderline values through the direct method could pass into the high value range (160-189 mg/dL) through estimate calculation by using the Friedewald formula (<xref ref-type="fig" rid="fig1c-body2">fig. 1C</xref>). </p>
      <p>On the other hand, when assessing the results according to the different levels of triglycerides, a pattern of inversion of that bias is observed (<xref ref-type="table" rid="tab01-body2">tab. I</xref>), although the 2 methods have excellent correlation coefficients (0.9072 to 0.9426) (<xref ref-type="table" rid="tab03-body2">tab. III</xref>). With triglyceride levels up to 150 mg/dL, a mean positive bias of 14 &#177; 13 mg/dL is observed for the Friedewald formula. Therefore, theoretically, even without statistical significance, a patient with triglycerides &lt; 150 mg/dL and an LDL-C level of 125 mg/dL, by using the direct method, could have an LDL-C level estimated through the Friedewald formula of 139 mg/dL (<xref ref-type="fig" rid="fig2a-body2">fig. 2A</xref>), considering the linear regression equation for that triglyceride range (<italic>y</italic> = 0.9746<italic>x</italic> + 17.6). That hypothetical patient would pass from a desirable LDL-C value to a borderline value. The same would happen to a patient with an LDL-C level of 150 mg/dL measured by the direct method, who would pass from the range of borderline values to that of high values (164 mg/dL) according to the estimate through the formula. </p>
      <p> Several authors reported lower results obtained through the direct methods of LDL-C measurement as compared with those obtained through b-quantitation after ultracentrifugation <sup>11,12</sup>. Others, however, reported that bias with a certain reagent, and a perfect correlation with another reagent, although both used detergents, but with different principles <sup>13</sup>. The reagents used in this study follow a principle of detergent protection similar to the method used in the study cited, which reported no bias, as compared with the reference method. Some authors have also not found that variation in bias in regard to different levels of triglycerides with a method that also uses specific tensoactive agents <sup>12</sup>, differently from most direct methods <sup>9, 14-18</sup>, which may be due to a difference in the components of the reagents. </p>
      <p> In our study, for triglyceride levels between 151 and 200 mg/dL, that bias decreased to 7 &#177; 14.5 mg/dL, and, for triglyceride levels between 201 and 300 mg/dL, that bias almost did not exist (2 &#177; 15.3 mg/dL). </p>
      <p> On the other hand, in the samples with triglyceride levels between 301 and 400 mg/dL, that bias of the Friedewald formula became negative, with a mean deviation of -16 &#177; 19.4 mg/dL. This way, as can be demonstrated with our results, a patient with triglyceride levels between 301 and 400 mg/dL and an LDL-C level measured through the direct method of 140 mg/dL could theoretically have an LDL-C level estimated through the Friedewald formula of 126 mg/dL (<xref ref-type="fig" rid="fig2d-body2">fig. 2D</xref>), considering the linear regression equation for that range of triglyceride levels (<italic>y</italic> = 0.8999<italic>x</italic> &#150; 0.1), although no statistical significance was observed. The patient would pass from a borderline LDL-C value to a desirable LDL-C value. The same would happen to a patient with an LDL-C level of 170 mg/dL measured through the direct method, who would pass from the range of high values to the range of borderline values (153 mg/dL). </p>
      <p>According to other authors <sup>13,19</sup>, one explanation for these higher LDL-C levels obtained through the direct method as compared with those obtained through the Friedewald formula could be the difference in the triglyceride/cholesterol ratio in the VLDL particles in patients with types IIb, III, and IV Fredrickson dyslipidemias. Triglyceride-rich VLDL particles could induce a negative bias in the direct methods <sup>13</sup>. Cholesterol-rich VLDL particles could induce a positive bias <sup>19</sup>. However, this would not explain the positive bias of the direct method in patients with high triglyceride levels in our study. </p>
      <p>Another explanation would be the possibility of measuring the cholesterol present in the particles of intermediate-density lipoproteins (IDL), such as LDL-C, in patients with type III Fredrickson dyslipidemia through some direct methods <sup>13</sup>. This could be happening with the method in question, but that hypothesis cannot explain the tendency in all patients with triglyceride levels between 301 and 400 mg/dL, shown in our study. In fact, as reported by those authors, those patients with type III dyslipidemia would tend to be erroneously classified as type IIb using the direct method for measuring LDL-C, overestimating the LDL-C fraction. However, the estimate through the Friedewald formula does not help in the correct classification, overestimating the VLDL fraction (triglycerides/5), because it cannot indicate the presence of IDLs, which could also be suspected only because of the high triglyceride levels. In the specific case of patients with type III dyslipidemia, good communication between the clinician and the laboratory is an excellent contribution to the diagnosis, because the latter would be able to determine whether the method used may overestimate the LDL-C fraction, in the case of the direct method, or the VLDL fraction, in the case of estimate through the Friedewald formula. </p>
      <p>For LDL-C estimate, the Friedewald formula standardizes the value of the VLDL fraction as the triglyceride level divided by 5. However, the particles found in patients with hypertriglyceridemia (types IIb, III, IV, and V) are usually a heterogeneous mixture of remaining chylomicrons, VLDL, and remaining VLDL (IDLs). As is already known, the triglyceride/cholesterol ratio varies a lot within that range of particles. In fact, according to our results, the LDL-C estimate through the Friedewald formula has good performance in samples with triglyceride levels between 151 and 300 mg/dL, as compared with the direct method. However, in the extremes of that range, in the samples with triglyceride levels &lt; 150 mg/dL and between 301 and 400 mg/dL, the formula does not have good performance. In fact, according to some authors, the Friedewald formula may erroneously classify up to 25% of the patients with triglyceride levels between 301 and 400 mg/dL <sup>13</sup>. That percentage may be even higher, depending on the method used for measuring HDL-C levels. </p>
      <p>Thus, patients with triglyceride levels &lt; 150 mg/dL and desirable LDL-C levels measured through the direct method could be being treated based on an LDL-C level estimated through the Friedewald formula. On the other hand, patients with triglyceride levels between 301 and 400 mg/dL and borderline or high LDL-C levels measured through the direct method would not be receiving the appropriate treatment with the results obtained through the use of the Friedewald formula. Obviously, the limits of LDL-C levels for risk classification for coronary artery disease are very narrow, but as those are the values recommended by the III Brazilian Consensus on Dyslipidemias <sup>4</sup>, the clinical laboratory should strive to perform the measurement of that fraction with the best possible diagnostic performance, ie, with the lowest coefficient of variation, and with the best likelihood of correctly classifying the patients for assessing the risk for coronary artery disease. </p>
      <p>This study, analyzing a sample with more than 10,000 patients, demonstrated that the direct method used for measuring LDL-C has very good performance, with good reproducibility and a coefficient of variation within the requirements of the NCEP, which is hardly obtained with the Friedewald formula. It is worth noting, however, that although one of the objectives was to discuss that the 2 methods do not have identical results, it is still a consensus that the Friedewald formula may be used in patients with triglyceride levels up to 400 mg/dL, who have neither chylomicrons, nor IDLs (type III of the Fredrickson classification) <sup>4,20</sup>. In fact, many laboratories still continue to estimate LDL-C levels by using that formula, because of the costs of the reagents for the existing direct LDL-C measuring methods. With a decrease in those costs and a better assessment of the performance of the reagents, the direct methods tend to be more widely used in laboratories, providing a better classification of the patients, with more reliable LDL-C level results, according to the NCEP criteria. Certain populations would benefit extremely by the use of the direct methods, such as diabetic patients, who are naturally prone to developing coronary artery disease, and whose LDL-C levels are not correctly estimated with the Friedewald formula <sup>8</sup>. </p>
      <p>In the future, new methods may be implemented for determining LDL-C levels, such as one recently described, which is reagent-free and based on infrared spectrophotometric absorption of lipoproteins <sup>21</sup>. That method had excellent performance, and its assessment may shed light on the debate about the introduction of new methods for LDL-C measurement. </p>
      <p>Recently, some authors reported that 2 direct methods for LDL-C measurement did not have good retrieval of the small, dense LDL subtype (sdLDL), previously separated by ultracentrifugation <sup>22</sup>. The determination of that LDL subtype levels has been implicated as a more sensitive factor for assessing the risk of developing coronary artery disease than the determination of total LDL-C has. Methods for the routine sdLDL measurement are still being developed, but, in the near future, the laboratories may also be including that parameter in the lipid and lipoprotein profile available for clinicians <sup>23</sup>. </p>
      <p>Similarly, the direct method for HDL-C measurement had excellent performance, with a very low coefficient of variation. In our experience, that method has excellent performance in laboratory routine, is easily automated, and does not suffer from interference, as occurs in precipitation methods <sup>1</sup>, which cause additional error in LDL-C estimated with the Friedewald formula. Other authors <sup>24</sup> reported that, in comparison with the reference method, the direct method does not provide lower HDL-C levels, as occurs in precipitation methods. </p>
      <p>
        <bold>Acknowledgements</bold>
      </p>
      <p>We thank the DiaSys Diagnostic Systems by their partial support.</p>
      <p>
        <bold>References</bold>
      </p>
      <p>
        <fn id="end-body2" fn-type="corresp">
          <label>
            <bold>Correspondence to</bold>
          </label>
          <p> Caio M. M. de Cordova </p>
          <p> Departamento de Ci&#234;ncias Farmac&#234;uticas/FURB, Campus III </p>
          <p> Rua Ant&#244;nio da Veiga, 140 - CP 1507 </p>
          <p> Cep 89010-971 - Blumenau, SC, Brazil </p>
          <p> E-mail: </p>
          <p>
            <email xlink:href="cmcordova@furb.br">cmcordova@furb.br</email>
          </p>
        </fn>
      </p>
      <p>Received for publication: 09/10/2003 </p>
      <p> Accepted for publication: 02/18/2004 </p>
      <p>English version by Stela Maris Costalonga</p>
    </body>
  </sub-article>
</article>
