n engl j med 376; 11 nejm. org March 16, 2017 to so much medical progress, we may never see stem-cell therapy reach its full potential. The safety and efficacy of the use of stem cells derived from peripheral blood or bone marrow for hematopoietic reconstitution are well established. Increasingly, however, hematopoietic stem cells and stem cells derived from sources such as adipose tissue are being used to treat multiple orthopedic, neurologic, and other diseases. Often, these cells (whether derived from autologous or allogeneic sources) are being used in practice on the basis of minimal clinical evidence of safety or efficacy, sometimes with the claim that they constitute revolutionary treatments for various conditions. Despite the absence of compelling evidence from adequate, well-controlled clinical trials, some practitioners assert that stem cells have a unique capacity to restore health because they can sense their environment and differentiate in a manner that repairs any defect. A separate argument is that conducting controlled trials and meeting regulatory standards for such promising therapies is too complex for all except large industrial sponsors and that therefore broad use in clinical practice should be allowed and encouraged while evidence regarding efficacy is gathered. Proponents of both arguments generally assert that stem-cell therapies are quite safe, particularly when the cells are derived from an autologous source. Outside the setting of hematopoietic reconstitution and a few other well-established indications, the assertion that stem cells are intrinsically able to sense the environment into which they are introduced and address whatever functions require replacement or repair—whether injured knee cartilage or a neurologic deficit—is not based on scientific evidence. Published data derived primarily from small, uncontrolled trials plus a few well-controlled, randomized trials have not reliably demonstrated the effectiveness of stem-cell treatments even in some of the most systematically studied conditions, such as heart failure and graft-versus-host disease. 1, 2

This lack of evidence is worrisome. The literature is replete with instances of therapeutic interventions pursued on the basis of expert opinion and patient acceptance that ultimately proved ineffective or harmful when studied in well-controlled trials comparing them with the standard of care. One of the most unfortunate therapeutic misadventures in contemporary times was the widespread use of autologous stem-cell transplantation to treat metastatic breast cancer, a practice ultimately shown to be ineffective, costly, and risky.