COVID-19 Show message

Skip to content

Post by Former NIMH Director Thomas Insel: Lost in Translation

By on

Marfan syndrome is a rare genetic disease of connective tissue caused by a mutation in FBN1, the gene that encodes the protein fibrillin-1. People with Marfan syndrome tend to be unusually tall with long fingers and toes and often have curvature of the spine. But the most serious problems involve the heart and especially the aorta. Weakening of the connective tissue in the aorta leads to dilation or sometimes aneurysms, which can be fatal if they rupture. Mice with a Marfan mutation of FBN1 show changes in fibrillin-1 and develop the same changes in the aorta found in humans with the syndrome.

In fact, Marfan syndrome has been my favorite example of the cycle of translational research: find the genetic mutation in patients, induce the mutation in mice, identify the effects of the mutation in mice, find a treatment for the mice, then move that treatment back into patients. This translational cycle showed all the signs of success in Marfan syndrome. Not only did the mice develop a key aspect of the human syndrome, but losartan, a drug already available on the market, blocked the effects of misfolded fibrillin-1 and consistently prevented the aortic disease in mice. When initial studies of losartan in patients with Marfan syndrome looked promising, there was every reason to believe that the translational cycle would deliver the first treatment to target the fundamental defect in this genetic disease. So the report last month in the New England Journal of Medicine that losartan was not more effective than the usual treatment with a beta-blocker was deeply disappointing.1 Although the usefulness of losartan is not foreclosed, neither is it a magic bullet.2 Disappointing but perhaps not entirely surprising. The final step in this translational cycle appears to be the hardest. That has been especially true for brain disorders. Over the past decade we have cured nearly every brain disorder, from autism to Alzheimers, in mice, but these effects rarely translate to patients. In fact, the cynical joke among clinical researchers is that if you are going to develop a brain disorder, first become a mouse.

Why have mouse cures not become human cures? In many cases, such as in Rett syndrome, the mouse cure requires a genetic manipulation that is not yet possible for human patients. In some cases, as with Marfan syndrome, a drug is already available, but the drug effect observed in mice does not translate to humans. For some disorders, like schizophrenia, mice may not have the relevant brain areas to represent the effect needed in humans. There are a variety of reasons, but the result is now becoming more the rule than the exception: treatment effects in mice do not predict treatment effects in humans.

That is not to say that mice are not critical for progress in understanding human disease, especially brain diseases. Once a genetic mutation is discovered in patients, studying the effects of this gene in mice is essential to define the mechanisms of disease. The first parts of the translational cycle work predictably well. Studies of the genes involved in autism, epilepsy, schizophrenia, and many rare neurological disorders have mapped the process by which a mutation leads to altered proteins with subsequent changes in cell and circuit function. The problem seems to be that the complexity grows and the predictability drops as we move further away from the genetic mutation. Mutant mice sometimes show the behavioral changes associated with brain disorders, but often only select aspects of these disorders and rarely a precise copy of a syndrome. The idea that mice can provide a “model” of human brain disorders has been nearly always useful at the molecular level, occasionally helpful at the behavioral level, and too often misleading in the development of medications.

What can be done to improve this final, critical step in translation? One approach has been to move to monkeys, assuming that the non-human primate brain will be a better model for the human brain. The Brain/MINDS (Brain Mapping by Integrated Neurotechnologies for Disease Studies) project just announced in Japan will develop marmosets as a new model for studying genetic diseases that cause autism and dementia.3 Another approach is to work with human cells, using induced pluripotent cells from patients that can be differentiated into neurons in a dish. This “disease in a dish” approach just received a big boost with the demonstration that human neurons grown in vitro can be induced to form circuits and that these can be used for screening medications. 4 A third approach is to grow synthetic human organs, through projects like the Tissue Chip for Drug Screening effort at NIH’s National Center for Advancing Translational Sciences. This looks promising for the lung and liver, but we are a long way from growing human cortex.

Finding better ways to predict treatment responses has become an urgent need for treatment development. The Marfan syndrome story, which I have always used as the model for translational research, now serves as a cautionary tale. Mice will continue to be essential for the early steps in translation, but we need a better approach to treatment development.


1 Lacro RV et al. Atenolol versus losartan in children and young adults with Marfan's syndrome. N Engl J Med. 2014 Nov 27;371(22):2061-71. doi: 10.1056/NEJMoa1404731. Epub 2014 Nov

2 Bowen JM, Connolly HM. Of Marfan's syndrome, mice, and medications. N Engl J Med. 2014 Nov 27;371(22):2127-8. doi: 10.1056/NEJMe1412950. Epub 2014 Nov 18.

3 Cyranoski D. Marmosets are stars of Japan's ambitious brain project. Nature. 2014 Oct 9;514(7521):151-2. doi: 10.1038/514151a.

4 Choi SH et al. A three-dimensional human neural cell culture model of Alzheimer's disease. Nature. 2014 Nov 13;515(7526):274-8. doi: 10.1038/nature13800. Epub 2014 Oct 12.