The past decades have seen a spectacular burgeoning of a new cognitive field—molecular genetics—beginning with Watson and Crick's^1,2 description of the structure of DNA. However, translating new biomedical knowledge into clinical applications has been difficult. Effective therapy through molecular medicine will require: (1) new biomedical knowledge to allow the description, classification and explanation of disease entities on a molecular level; (2) new tools to diagnose molecular findings in patients; and (3) new therapies to intervene on the same molecular level on which disease entities were defined and diagnosed.

Scientific theories must be translated into diagnostic and therapeutic applications in order to effectuate the transition from biomedical explanatory models to molecular medical practice.³ It is often assumed that new technologies are the hinges on which the doors of molecular medicine may finally begin to swing. At first glance, this view seems to be correct. In a remarkably short period of time, laboratory technologies of molecular biology designed to identify and locate genes were developed for in vitro use in the clinical laboratory. Since then, they have become the diagnostic backbone of predictive testing and genetic counseling. The problem has been in translating molecular and recombinant technologies into therapeutic procedures for in vivo application to patients. The difficulties associated with the development of therapeutic applications are not just technologic in nature. To understand the problems that have prevented a smooth transition to molecular practice, we must discuss the three levels of medical innovation that were previously mentioned.

First, models in molecular genetics are still inadequate for defining clear-cut molecular disease entities. In the case of cancer, despite tremendous research efforts in the scientific field, molecular genetics still has not shown a strong correlation between genotype and disease.⁴ Second, on the level of diagnosis, tests for susceptibility-conferring genotypes share this problem of limited causal correlation. Low positive predictive values of tests do not usually justify tailoring treatments to genotypes (except perhaps for monogenic disorders with unusually high penetration like Huntington's disease).⁵ Third, it is still uncertain if treatments can be reliably tailored to genetic conditions. On May 25, 2000, only 3,476 patients worldwide were registered in clinical studies of somatic gene therapy.⁶ Despite the latest reports on single successful therapeutic trials,⁷ controlled gene delivery, reliable gene expression and potentially lethal reactions to therapy—the last resulting in the highly publicized death of Jesse Gelsinger^8–11—remain major obstacles in the vast majority of experimental approaches to somatic gene therapy.¹²

Notwithstanding the fundamental problems that have plagued molecular medical practice for decades, enthusiasm for molecular medicine as such does not wane. Taking into account all we know thus far about the functions of genes, some practical hope for molecular medicine appears to be justified, especially with respect to diseases clearly rooted in molecular genetics. Nevertheless, we should be aware that the discourse on molecular medicine—even the apparently scientific discourse—is, in its current form, driven by politics.¹³ Thus, it is sometimes difficult to distinguish wish from reality.

We can identify two distinct approaches to innovation: utility- and evidence-based. Utility-based approaches are more characteristic of the public domain, while evidence-based modes occur within the context of science and technology. The utility-based mode refers simultaneously to an independent market and professional practice, and the legal regulations, ethical considerations and consumer forces used to determine utilization of new procedures (e.g., BRCA1 or 2 testing, or somatic gene therapy), given its uncertain practical significance. In contrast, the evidence-based mode refers simultaneously to scientific and biomedical knowledge, the definition and feasibility of particular pragmatic goals of applications, and the prioritization of these goals by relating them to the socially defined tasks of science, technology and medicine.¹⁴

At this point, the evidence-based approach is not ready to provide definite answers with regard to the molecular vision of health and disease. In other words, science is lagging behind molecular genetics' promise of public benefit. For now, advances in the implementation of molecular genetics must rely on the utility-based mode, with the main purpose being to establish consensus about possible benefits of innovative procedures. The main goal is to balance uncertainty and keep the momentum of research projects going.

A focus on the public interest is a legitimate purpose, a legitimate goal and a necessary part of medical innovation. At the same time, it is important to remember that utility-based and evidence-based arguments are not the same thing. In the current discussion about molecular medicine, we must distinguish between issues we address in our role as members of the community and issues we have to be concerned with as professional health care providers. Professional decision making requires evidence-based discourse.

Research for this paper has been generously supported by the Alexander von Humboldt Foundation, Bonn, Germany, the Stanford University Program in History and Philosophy of Science and the Stanford Program in Genomics, Ethics, and Society.