{"id":7915,"date":"2011-04-23T00:51:22","date_gmt":"2011-04-23T04:51:22","guid":{"rendered":"http:\/\/1boringoldman.com\/?p=7915"},"modified":"2011-04-23T01:11:36","modified_gmt":"2011-04-23T05:11:36","slug":"personalized-medicine-the-shoals-of-fuzzy-math","status":"publish","type":"post","link":"https:\/\/1boringoldman.com\/index.php\/2011\/04\/23\/personalized-medicine-the-shoals-of-fuzzy-math\/","title":{"rendered":"personalized medicine: the shoals of fuzzy math&#8230;"},"content":{"rendered":"\n<div align=\"justify\">This business of &quot;biomarkers&quot; is a slippery slope. I think in the back of my mind, I&#8217;ve always retained an envy for my old medical specialty &#8211; rheumatology. Like Psychiatry, the disorders are heavily weighted towards &#8216;diseases of unknown etiology&#8217; diagnosed with &#8216;criteria.&#8217; But as a rheumatologist, we had biomarkers [LE prep, ANA, RA, sed rate, serum complement]. People still argue about what they mean, but they were there. They said &quot;disease&quot; in neon lights. &quot;Biomarkers&quot; in psychiatry have always been an elusive dream. They mean more than just as diagnostic tools. It&#8217;s as if they would &#8216;legitimize&#8217; us &#8211; give us something tangible. And it must be even stronger for the &quot;medical model&quot; people who built the DSMs. So this business of <strong><font color=\"#300030\">personalized medicine<\/font><\/strong> has a back story as well &#8211; an added dimension. When I read that article appended to the end of my last post last night, I was thinking that:<\/div>\n<ul>\n<div align=\"justify\"><sup>A tentative integrative model showed that a combination of N1 amplitude  at Pz and verbal memory performance accounted for the largest part of  the explained variance. These markers may serve as new biomarkers suitable for the prediction of antidepressant treatment outcome.<\/sup><\/div>\n<\/ul>\n<div align=\"justify\">&#8230;just wasn&#8217;t going to get it. And listening to the speakers at the <a target=\"_blank\" href=\"http:\/\/1boringoldman.com\/index.php\/7604-2\/%20\"><u><strong><font color=\"#400040\">Mayflower Action Group Initiative<\/font><\/strong><\/u><\/a> talk in the <em>definite<\/em> about things that fit better in the <em>maybe<\/em> or, at best, <em>tentative<\/em> gave me the feeling, &quot;Uh Oh. There they go again.&quot; I was pleased to see that more rational forces were at work in our universe when <a target=\"_blank\" href=\"http:\/\/1boringoldman.com\/index.php\/2011\/04\/21\/personalized-medicine-beyond-the-blockbuster-era\/#comment-199845\"><u><strong>SteveBMD<\/strong><\/u><\/a> sent us this:<\/div>\n<blockquote>\n<div align=\"center\"><u><a target=\"_blank\" href=\"http:\/\/www.nature.com\/mp\/journal\/v16\/n5\/full\/mp201038a.html\"><strong><font color=\"#200020\">Poor replication of candidate genes for major depressive disorder using genome-wide association data<\/font><\/strong><\/a><\/u><br \/>                             <sup>by F J&nbsp;Bosker, C A&nbsp;Hartman, I M&nbsp;Nolte, B P&nbsp;Prins, P&nbsp;Terpstra, D&nbsp;Posthuma, T&nbsp;van Veen, G&nbsp;Willemsen, R H&nbsp;DeRijk, E J&nbsp;de Geus, W J&nbsp;Hoogendijk, P F&nbsp;Sullivan, B W&nbsp;Penninx, D I&nbsp;Boomsma, H&nbsp;Snieder and W A&nbsp;Nolen<\/sup><br \/>                             <strong><font color=\"#200020\">Molecular Psychiatry<\/font><\/strong> [2011] 16, 516&ndash;532.<\/div>\n<p>                             <\/p>\n<div><em><strong><font color=\"#200020\">Abstract:<\/font><\/strong><\/em><\/div>\n<div align=\"justify\"><sup>Data  from the Genetic Association Information Network [GAIN] genome-wide  association study [GWAS] in major depressive disorder [MDD] were used to  explore previously reported candidate gene and single-nucleotide  polymorphism [SNP] associations in MDD. A systematic literature search  of candidate genes associated with MDD in case&ndash;control studies was  performed before the results of the GAIN MDD study became available.  Measured and imputed candidate SNPs and genes were tested in the GAIN  MDD study encompassing 1738 cases and 1802 controls. Imputation was used  to increase the number of SNPs from the GWAS and to improve coverage of  SNPs in the candidate genes selected. Tests were carried out for  individual SNPs and the entire gene using different statistical  approaches, with permutation analysis as the final arbiter. In all, 78  papers reporting on 57 genes were identified, from which 92 SNPs could  be mapped. In the GAIN MDD study, two SNPs were associated with MDD: <em>C5orf20<\/em> [rs12520799; <em>P<\/em>=0.038; odds ratio [OR] AT=1.10, 95% CI 0.95&ndash;1.29; OR TT=1.21, 95% confidence interval [CI] 1.01&ndash;1.47] and <em>NPY<\/em> [rs16139; <em>P<\/em>=0.034; OR C allele=0.73, 95% CI 0.55&ndash;0.97], constituting a direct replication of previously identified SNPs. At the gene level, <em>TNF<\/em> [rs76917; OR T=1.35, 95% CI 1.13&ndash;1.63; <em>P<\/em>=0.0034]  was identified as the only gene for which the association with MDD  remained significant after correction for multiple testing. For SLC6A2  [norepinephrine transporter [NET]] significantly more SNPs [19 out of  100; <em>P<\/em>=0.039] than expected were  associated while accounting for the linkage disequilibrium [LD]  structure. Thus, we found support for involvement in MDD for only four  genes. However, given the number of candidate SNPs and genes that were  tested, even these significant may well be false positives. The poor  replication may point to publication bias and false-positive findings in  previous candidate gene studies, and may also be related to  heterogeneity of the MDD phenotype as well as contextual genetic or  environmental factors.<\/sup><\/div>\n<\/blockquote>\n<div align=\"justify\">Looking at the original full text version, I was impressed that they did it right &#8211; and used the honest statistics to correct for the fact that they had a lot of independent variables. The new technologies, particularly the DNA\/SNP\/Gene technologies offer us a fascinating field to play on, but as I&#8217;ve been writing about, they open the door to an area where the sloppy science of recent years could have a field day in the search for a way to continue Pharma&#8217;s pressure to increase sales. The possibility of expensive tests to find your genotype\/phenotype in order to pick a <strong><font color=\"#300030\">personalized <\/font><\/strong>designer anti-whatever drugs seems to be drawing together a coalition of formidable forces. I&#8217;m glad to see that there are watchdogs who have &quot;<em>Data  from the Genetic Association Information Network [GAIN]  genome-wide  association study [GWAS] in major depressive disorder [MDD]<\/em>&quot; to keep people honest.<\/div>\n<p>                        <\/p>\n<div align=\"justify\">I wanted to say something about corrections of data, because it&#8217;s a key point in these studies that survey a lot of SNPs [<strong><a target=\"_blank\" href=\"http:\/\/en.wikipedia.org\/wiki\/Single-nucleotide_polymorphism\"><em>Single-nucleotide polymorphism<\/em><\/a><\/strong>] in the search for a genetic biomarker, or for that matter any biomarker. In applying statistics to data, the ubiquitous p value actually makes a statement that&#8217;s pretty simple to understand. <\/div>\n<ul>\n<div align=\"justify\"><strong><sup><em>p<\/em>&lt;0.05 says that the observed difference could only occur by chance one time in twenty.<\/sup><\/strong><\/div>\n<div><strong><sup><em>p<\/em>&lt;0.01 says that the observed difference could only occur by chance one time in a hundred.<\/sup><\/strong><\/div>\n<div><strong><sup><em>p<\/em>&lt;0.001  says that the observed difference could only occur by chance one time in a thousand.<\/sup><\/strong><\/div>\n<\/ul>\n<div align=\"justify\">But what if you test 50 different SNPs? Using a standard of p&lt;0.05 for significance, you&#8217;d expect two or three to be &quot;significant&quot; just by chance which is of course absurd. How do you define real significance in that situation? There is a correction called the <a target=\"_blank\" href=\"http:\/\/en.wikipedia.org\/wiki\/Bonferroni_correction\"><strong>Bonferroni Correction<\/strong><\/a> that&#8217;s often used. It&#8217;s logic is fairly straightforward:<\/div>\n<ul>\n<div align=\"justify\"><sup><strong>The Bonferroni correction is derived by observing Boole&#8217;s inequality. If you perform <em>n<\/em> tests, each of them significant with probability <em>&beta;<\/em>, (where <em>&beta;<\/em> is unknown) then the probability that at least one of them comes out significant is [by Boole&#8217;s inequality] &le; <em>n&sdot;&beta;<\/em>. Now we want this probability to equal <em>&alpha;<\/em>, the significance level for the entire series of tests. By solving for <em>&beta;<\/em>, we get <em>&beta; = &alpha;\/n<\/em>.<\/strong><\/sup><\/div>\n<\/ul>\n<div align=\"justify\">So when somebody goes on a fishing expedition and assays a bunch of SNPs looking for a biomarker, the p value should be divided by the number of SNPs [or other variables] tested [<em>n<\/em>]. So:<\/div>\n<table cellspacing=\"0\" cellpadding=\"3\" border=\"0\" align=\"center\">\n<tr>\n<td>&nbsp;<\/td>\n<td valign=\"top\" align=\"center\" colspan=\"5\"><strong><font color=\"#200020\">Bonferroni Correction<\/font><\/strong><\/td>\n<\/tr>\n<tr>\n<td width=\"20%\">&nbsp;<\/td>\n<td valign=\"middle\" align=\"center\"><em>n=1<\/em><\/td>\n<td valign=\"middle\" align=\"center\"><em>n=10<\/em><\/td>\n<td valign=\"middle\" align=\"center\"><em>n=20<\/em><\/td>\n<td valign=\"middle\" align=\"center\"><em>n=50<\/em><\/td>\n<td valign=\"middle\" align=\"center\"><em>n=100<\/em><\/td>\n<\/tr>\n<tr>\n<td>&nbsp;<\/td>\n<td colspan=\"5\">\n<hr size=\"1\" \/><\/td>\n<\/tr>\n<tr>\n<td valign=\"top\"><strong><font color=\"#200020\">p&lt;0.05<\/font><\/strong><\/td>\n<td valign=\"top\"><em>p&lt;0.05<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.005<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.0025<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.001<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.0005<\/em><\/td>\n<\/tr>\n<tr>\n<td valign=\"top\"><strong><font color=\"#200020\">p&lt;0.01<\/font><\/strong><\/td>\n<td valign=\"top\"><em>p&lt;0.01<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.001<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.0005<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.0002<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.0001<\/em><\/td>\n<\/tr>\n<tr>\n<td valign=\"top\"><strong><font color=\"#200020\">p&lt;0.001<\/font><\/strong><\/td>\n<td valign=\"top\"><em>p&lt;0.001<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.0001<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.00005<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.00002<\/em><\/td>\n<td valign=\"top\"><em>p&lt;0.00001<\/em><\/td>\n<\/tr>\n<\/table>\n<p>               <\/p>\n<div align=\"justify\">There are <u><a target=\"_blank\" href=\"http:\/\/en.wikipedia.org\/wiki\/Bonferroni_correction#Alternatives_to_Bonferroni_correction\"><strong>other ways<\/strong><\/a><\/u> of correcting things, but the results are similar. The point is not to understand funny-named statistics or even to understand the math. The point is that there&#8217;s a wide open door to fudge results when people are screening a lot of potential biomarkers looking for the elusive pot of gold. If the correction method isn&#8217;t mentioned, be very suspicious that trickery is afoot. In this abstract, this is how this important part reads:<\/div>\n<ul>\n<div align=\"justify\"><sup>At the gene level, <em>TNF<\/em> [rs76917; OR T=1.35, 95% CI 1.13&ndash;1.63; <em>P<\/em>=0.0034]   was identified as the only gene for which the association with MDD   remained significant <u>after correction for multiple testing<\/u>.<\/sup><\/div>\n<\/ul>\n<div>There&#8217;s a lot of fuzzy math going around these days&#8230;<\/div>\n","protected":false},"excerpt":{"rendered":"<p>This business of &quot;biomarkers&quot; is a slippery slope. I think in the back of my mind, I&#8217;ve always retained an envy for my old medical specialty &#8211; rheumatology. Like Psychiatry, the disorders are heavily weighted towards &#8216;diseases of unknown etiology&#8217; diagnosed with &#8216;criteria.&#8217; But as a rheumatologist, we had biomarkers [LE prep, ANA, RA, sed [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_bbp_topic_count":0,"_bbp_reply_count":0,"_bbp_total_topic_count":0,"_bbp_total_reply_count":0,"_bbp_voice_count":0,"_bbp_anonymous_reply_count":0,"_bbp_topic_count_hidden":0,"_bbp_reply_count_hidden":0,"_bbp_forum_subforum_count":0,"footnotes":""},"categories":[2],"tags":[],"class_list":["post-7915","post","type-post","status-publish","format-standard","hentry","category-politics"],"_links":{"self":[{"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/posts\/7915","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/comments?post=7915"}],"version-history":[{"count":43,"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/posts\/7915\/revisions"}],"predecessor-version":[{"id":7958,"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/posts\/7915\/revisions\/7958"}],"wp:attachment":[{"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/media?parent=7915"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/categories?post=7915"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/1boringoldman.com\/index.php\/wp-json\/wp\/v2\/tags?post=7915"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}