Ontology engineering

NEWS feature© 2010 Nature America, Inc. All rights reserved.tumors respond to Herceptin alone 2 . “It’s notjust a simple translation of gene overamplificationto susceptibility to the drug,” says LarryNorton, of Memorial Sloan-Kettering CancerCenter in New York.Others wonder if the test is even necessary.Studies from two clinical trials (NSABP B31and NCCTG N9831) presented at the AmericanSociety of Clinical Oncologists (ASCO) annualmeeting in 2007 (ref. 3) suggested that someindividuals with HER2-negative tumorscan benefit from Herceptin. In these trials,which comparedchemotherapyalone with chemotherapyplus CentromereaHerceptin, onlywomen who wereHER2 positivecould participate.Upon retesting,however, sometumor samplescame up negative.Nonetheless,some womenwith negativetest results benefitedfrom thedrug, which hasspurred a nowl ong-runningdebate. Althoughmany expertsbelieve this findingto be an artifactof variationin test accuracy,others think thismay be anotherimportant clue.“It’s easy todismiss a findingyou can’texplain, but thisbis forcing us to reexamine our notions of whatbeing HER2 positive or negative means,” saysNorton.The controversy around these particular trialfindings may be resolved soon. Samples fromthe NCCTG N9831 trial are being retested in around-robin fashion by three different groups.Results will then be sent to a central monitoringgroup to identify any discrepancies and totry to pinpoint their cause. Soonmyung Paikof the National Surgical Adjuvant Breast andBowel Project (a National Cancer Institute(NCI)-sponsored cooperative based at theUniversity of Pittsburgh) postulates that individualswith HER2-negative primary tumorsmay have circulating tumor cells that are HER2Her2 (ERBB2)~190 kbLSI HER2positive, but he admits that the discrepancycould just reflect problems with the tests. Hisgroup is doing further microarray analysis ofthe NSABP trial samples. If the earlier findingsare confirmed, a trial could be launched in thesummer of 2010 to test Herceptin in patientswith HER2-negative tumors. “We have an NCIapprovedprotocol,” Paik writes in an e-mail.Abundance of richesFrom the time Herceptin was launched, expertshave warned that existing tests have problems.The most commonlyused test,an immunohis-Telomere tochemistry-17q11.2–q12 regionbased assay,happens to bethe least dependable,especiallywhen performedin laboratoriesthat do onlyoccasional tests.The immunohistochemistrytestmeasures proteinlevels, whereasthe newer FISHbasedtests measuresgene copynumber andare believed tobe more reliable,especiallyin expert hands.Some thinkthe FISH assayshould be thestandard, butclearly, it makesa big differencewho is doing thetesting. In 2006,ASCO and theCollege of American Pathologists releasedstricter guidelines which, according to Ross,forced many laboratories that had low test volumesto send the samples to laboratories withhigher volumes and more experience. But someexperts were dismayed that the new guidelinesdid not recommend FISH over immunohistochemistry.Others, including Norton, are skeptical of allavailable tests. The FISH probe, he points out,is large (190 kilobases in the case of the Vysisprobe), spanning the gene and then some (Fig. 1).“When you see changes in HER2 at the genecopy number level, is that a reflection of HER2itself or of generalized genomic instability?” heasks. Genomic instability, he points out, is notFigure 1 FISHing for HER2. (a) Probe map shows therelative size of the Vysis LSI HER2 probe and the gene.(b) An example of a FISH test for HER2 amplificationshows multiple copies of the HER2 gene (red clusterssignals) compared to chromosome number (green signals).Source: Abbott.a random event and the region where HER2is found is a ‘hot spot’. Norton thinks that theprescribing of HER2-targeting drugs won’t beimproved until we understand how specificmutations influence a tumor’s susceptibilityand sequencing can be routinely done onbiopsies. Comparative genome hybridization(CGH) studies done in his laboratory suggestthat simple ‘amplified’ and ‘nonamplified’ readingsavailable from FISH do not adequatelyreflect the complex changes that can occur inthis region. For example, CGH studies revealedthat in some samples that were HER2 positiveon FISH, the amplified area was actually adjacentto HER2 and not within it 4 . It may endup that higher resolution methods like CGHare needed to get the right information abouta tumor’s status.Meanwhile, companies like Genomic Healthin Redwood City, California, and Labcorp’sMonogram Sciences of S. San Francisco,California, are jumping in with new approachesto testing breast cancer patients. GenomicHealth claims that an advantage of its quantitativePCR-based test, OncotypeDX, is its accuracy.“Our test is more than 95% concordantwith reference labs’ assessment by FISH,” saysSteve Shak, chief medical officer at GenomicHealth. HER2 is one of 21 genes included inOncotypeDX, which is used to quantify therisk of recurrence of early breast cancer and theresponse to particular types of chemotherapy.Starting in 2008, the company began includingestrogen receptor, progesterone receptor andHER2 status in every report it provides.Monogram’s HERMark test measures HER2total protein as well as functional homodimersin a dual-antibody format. The company claimsthe test has advantages over FISH because it is adirect measurement of the protein, and that itis seven to ten times more sensitive than immunohistochemistrytesting. Albany MedicalCollege’s Ross counters, “Monogram has noprospective randomized data to support thattheir test is better.”Neither company’s test is FDA approvedfor use with Herceptin. For now, the tests’use may be confined to confirming or clarifyingresults obtained using other tests. “Weneed not just technical accuracy but to knowif these tests are actually clinically relevant,”says Edith Perez, of the Mayo Clinic Floridain Jacksonville. But Genomic Health seemsoptimistic. “We have extremely positivefeedback on the value of being able to lookat that result, especially in those cases whenthe results for HER2 testing are uncertain,”says Shak. The company is doing additionalstudies of the test’s ability to predict whetherparticular individuals will benefit fromHER2-targeted therapy.118 volume 28 number 2 february 2010 nature biotechnology