jnci.oxfordjournals.org
JNCI
|
News
7
NEWS
repeat and repeat and repeat if you think you’ve found something that’s real, asopposed to patient samples where you getbasically one shot at doing something,”said Houghton.
Power To Predict
The drug industry, in the last few years,has recognized the potential utility of tumorgrafts, and drug candidate screeningusing tumorgraft models is now becomingroutine. “Some of the large pharmaceuticalcompanies have established very large pan-els of [tumorgrafted mice] in the particularcancer types that they’re interested in … tosee how their drugs are working,”Houghton said. For example, at the AACRmeeting, Frederic de Sauvage, Ph.D., of Genentech in South San Francisco, Calif.,described using such models to decipherhow experimental inhibitors of the hedge-hog signaling path- way target tumorconnective tissue.How strong isthe evidence thattumorgrafts canaccurately predictactual human response to drugs, giventhat standard xenografts have done sopoorly? Heinz-Herbert Fiebig, M.D.,Ph.D., a medical oncologist and cancerresearcher in Freiburg, Germany, hasbeen creating direct transfer xenograftmodels since the late 1970s and retro-spectively testing their predictive powerfor human treatments. Fiebig has reportedconsistently good results. For example, in2004 he reported in the
European Journal of Cancer
that his xenografts correctly predicted response in 90% of patients (19of 21 tumors) and resistance in 97% (57of 59). The failure of standard cell line–de-rived xenografts to predict, said Fiebig, iscaused by changes in culture. Tumor cellsin culture tend to undergo selective pres-sure to become less differentiated andmore homogeneous, so that the tumorsthey give rise to in mice no longer faith-fully reflect the original tumor. Mosttumorgrafts, on the other hand, closely mirror their parent human tumors micro-scopically, he said. And Houghton’s grouphas shown that they change minimally onthe molecular level, when taking intoaccount the absence of surrounding humanconnective tissue, or stroma. “They havemaintained the [gene] expression profilesand genomic profiles very, very accu-rately,” Houghton said. Tumorgrafts, though, have their ownproblems, including the high transplanta-tion failure rate and the labor required formultiple transfers, or “passages,” frommouse to mouse (needed to propagateeach human tumor in multiple mice). Theoverall “take rate” varies from 40% to60% for non–small-cell lung cancer, coloncancer, and melanomas down to 12%– 20% for breast cancer and just 3% forprostate cancer, said Fiebig. So tumor-grafts cost much more than conventional xenografts.Both Houghtonand Fiebig stress theimportance of limit-ing the number of mouse-to-mouse pas-sages to prevent thetumors in the mousehost from gradually mutating away fromthe parent tumors. Frozen tumors make upa permanent repository that allows unlim-ited replenishment. “As a measure of pre-caution, we go back to the frozen masterstock after 10–12 passages” to generate newmouse tumors, said Fiebig.
Mining the Model
The main knock against tumorgrafts is theneed for more prospective trials showingthat they predict drug activity in humans.Several such tumorgraft validation studiesare ongoing or planned. But for many researchers, the evidence is already con- vincing, and use of the models is growing.In 2005, Houghton’s group reported aclinical trial in pediatric neuroblastoma with the drug topotecan, guided by indi- vidual drug level adjustments and dosingschedules tested in a direct transfer mouse xenograft model. “The response rates inthe clinic were exactly as predicted fromthe models—a 60% response rate,”Houghton said.On the basis of those results, the NCI isfunding a multicenter drug developmentinitiative, the Pediatric Preclinical TestingProgram, which uses 60 mouse tumormodels, with most of the tumors takendirectly from patients rather than from cellculture. “Basically the drugs that work inhuman cancers, childhood cancers, work inthe mice,” said Houghton. “Now we’reusing those models to identify new drugs, which are being fast-tracked into pediatrictrials.”Similar work is under way at Fiebig’scompany, Oncotest GmbH in Freiburg,Germany, which now has a bank of morethan 200 tumors established directly frompatients. Oncotest has also derived morethan 40 cell lines from these tumors, whichare used for biochemical and molecularstudies. U.S. drug companies especially have shown “a hugeinterest” in these toolsfor drug development,said Fiebig, beginning4 years ago when geneexpression chips werefirst used to character-ize these tumors on amolecular level.Personalized ther-apy, based on such gene expression signa-tures, is the ultimate goal. Fiebig hasidentified signatures in his models thatpredict response to 14 cytotoxic drugs andtwo targeted agents, bevacizumab andcetuximab. Overall response rates for bothtargeted drugs and most of the cytotoxics were close to those of past clinical results.Fiebig is planning to validate these signa-tures further using samples from clinicaltrials that have been completed and for which the outcomes are known, and even-tually in prospective clinical trials.In 2006, Sidransky and Hidalgo alsocofounded a company, BioMerk, to com-mercialize the tumorgraft platform. Thefollowing year, Champions Biotechnology in Arlington, Va., acquired BioMerk,later naming Sidransky chairman andHidalgo chief scientific officer.Champions, in collaboration with South Texas Accelerated Research Therapeutics,a San Antonio company, is establishing
Peter Houghton, Ph.D.
“There’s this mantra:‘Xenografts don’t predict for human effects.’ ”
Leave a Comment