In the late 19th century, when thepioneering architect Daniel H.Burnham was planning some ofthe first modern skyscrapers, his associ-ates were skeptical about erecting build-ings that soared into the clouds. Burn-ham reportedly warned the skepticsagainst making “little plans,” having“no magic to stir men’s blood.” Heurged them to reach beyond traditionalarchitectural boundaries, to think onceinconceivable thoughts and to performpreviously unimagined deeds—the hall-marks of revolutions.Revolutionary changes have also oc-curred in medicine over the past fewcenturies. Witness the new understand-ings and practices that issued from theintroduction of microscopy, anesthesia,vaccination, antibiotics and transplanta-tion. Medicine is now preparing to un-dergo another epochal shift: to an era inwhich genes will be delivered routinelyto cure or alleviate an array of inheritedand acquired diseases.Preparing for a radical change, yes,but not yet in the midst of it. By empha-sizing hopes and downplaying uncer-tainties, some overzealous researchers,representatives of industry and mem-bers of the lay and scientific media haveimplied that gene therapy is already ad-vanced enough for widespread applica-tion. It is not.Arguably, the conceptual part of thegene therapy revolution has indeed oc-curred. Whenever a new gene is discov-ered, researchers and nonscientists im-mediately ask whether it can be used totreat some disorder, even when more tra-ditional approaches might be applied.But the technical part of the revolution—the ability to correct disease—is anotherstory. Investigators have accomplishedthe requisite first steps: they have shownthat transferred genes can be induced tofunction in the human body, at times forseveral years. So far, however, no ap-proach has definitively improved thehealth of a single one of the more than2,000 patients who have enrolled ingene therapy trials worldwide.This lack of a convincing therapeuticbenefit is sobering. Yet it would be amistake to doubt gene therapy’s power-ful future. Remember, the field is young;in the U.S., trials in patients have beencarried out for fewer than 10 years. Amore realistic interpretation of the un-spectacular clinical results thus far isthat they reflect researchers’ imperfectinitial gropings toward a difficult newtechnology and that the obstacles aremore formidable than many of us hadexpected.A central challenge, as a federallycommissioned critique of the gene-ther-apy research effort noted in 1995, is per-fecting methods for delivering therapeu-tic genes to cells. Often genes introducedinto patients do not reach enough ofthe appropriate cells or, for reasons thatare not always clear, function poorly orshut off after a time. Under those con-ditions, a gene that could potentially behelpful would have little chance of af-fecting a disease process. In this article I will outline some of themost pressing technological stumblingblocks to successful gene transfer andthe strategies being considered to copewith those difficulties. I will deal onlywith therapy affecting somatic cells, thekinds that are neither sperm nor egg.To date, research aimed at human genetherapy has avoided manipulations thatwould deliberately affect descendants ofthe treated individuals, perhaps in unin-tended ways. The need for enlightenedpublic debate over the merits and risksof germ-line therapy has, however, beenmade more urgent by the recent cloningof an adult sheep [see “What CloningMeans for Gene Therapy,” by SteveMirsky and John Rennie, on page 122].How Genes and Gene Therapy WorkAnyone who wants to understand theobstacles to gene therapy shouldfirst know a bit about what genes doand about how attempts at gene thera-py are currently carried out. An individ-ual gene in the human cell is a stretch ofDNA that, in most cases, acts as a blue-print for making a specific protein; itspells out the sequence of amino acidscomposing that protein. All cells in abody carry the same genes in the chro-mosomes of the nucleus. But neurons,say, behave unlike liver cells because dif-ferent cells use, or express, distinct sub-sets of genes and hence make separatesets of proteins (the main functionariesof cells). Put more precisely, each cellcopies only selected genes into individu-al molecules of messenger RNA, whichthen serve as the templates from whichproteins are constructed. If a particular gene is mutated, itsprotein product may not be made at allor may work poorly or even too aggres-sively. In any case, the flaw may disturbvital functions of cells and tissues thatuse the normal gene product and canthereby cause symptoms of disease.96 Scientific American June 1997 Overcoming the Obstacles to Gene TherapyOvercoming the ObstaclesTreating disease by providing neededgenes remains a compelling idea, but clinical and basic researchers still havemuch to do before gene therapy can live up to its promiseby Theodore FriedmannCopyright 1997 Scientific American, Inc.Historically, physicians have treateddisorders stemming from inherited ge-netic mutations not by altering genes butby intervening in the biological eventsresulting from a mutation. For example,dietary restriction has long been pre-scribed for phenylketonuria, in whichloss of a gene leads to the toxic buildupof the metabolic products of the aminoacid phenylalanine. Unfortunately, non-genetic manipulations are usually onlypartly effective against inherited ills.In the early 1970s this fact—combinedwith growing understanding of howgenes function and with discovery of thegenes underlying many inherited ills—led to the suggestion that better resultsmight be achieved by attacking inborndiseases at their source. Among the ge-netic diseases that have been studiedare cystic fibrosis (which mainly affectsthe lungs), muscular dystrophy, adeno-sine deaminase deficiency (which severe-ly impairs immunity), and familial hy-percholesterolemia (which leads to theearly onset of severe atherosclerosis).Surprisingly, as time went on, it be-came clear that even acquired maladiesoften have a genetic component that cantheoretically be a target of a genetic cor-rection strategy. Indeed, quite unexpect-edly, more than half of all clinical trialsfor gene therapy these days aim at can-cer, which in most cases is not inheritedbut results from genetic damage accu-mulated after birth [see “Gene Therapyfor Cancer,” by R. Michael Blaese, onpage 111]. A number of trials also fo-cus on AIDS, which is caused by thehuman