Even though my background is in neuroscience, I rarely write about this topic. But wo papers on amyotrophic lateral sclerosis (ALS) from the latest issue of the Journal of Neuroscience struck me as interesting to talk about.

In the first one, Fiona Laird and her colleagues generated transgenic mice that express wild-type and mutant forms of the human protein dynactin p150-Glued. As mutant forms of this molecule had been linked to ALS, they decided to explore the mechanism whereby dynactin p150-Glued contributes to the pathology. They found that expression of dynactin p150-Glued carrying a mutation that had been linked to the disease in patients led to motor neuron disease in transgenic mice, something that was not seen in mice overexpressing the wild-type form of the human protein.

The paper is very nice in that it provides a very detailed account of the neuropathology the authors see in the mouse, including some intriguing evidence of autophagic cell death. The picture below, which comes from the paper, is a silver-stained section of the spinal cord from a mutant mouse, showing dark, presumably dying, motor neurons (arrowheads) that are not seen in control mice. Unfortunately, the authors didn't get to explore the hardcore molecular mechanisms that account for the motor neuron death. But they now have a useful system to ask more mechanistic questions to understand the role of dynactin p150-Glued in cell death and investigate its actual relationship to human ALS.

The second study deals with a question that has occupied the field for some time. We know that mutations in superoxide dismutase (SOD) are linked to familial forms of ALS, but where does SOD need to be expressed to cause disease: in neurons, in glia, in muscle? Dick Jaarsma and his colleagues tried to get at this question by generating transgenic mice that expressed mutant SOD only in neurons. The figure below, from the original paper, shows spinal cord sections from mice that expressed the mutant protein only in neurons (top left and bottom right) or ubiquitously (top middle).

This is not the first time that neuron-specific expression of SOD has been tried, but it is perhaps the first time in which it is found to effectively kill the motor neurons. In other words, these findings fly in the face of other studies reporting no motor neuron death in mice with neuron-specific expression of mutant SOD and of papers specifically identifying a contribution of extraneuronal SOD to ALS. Not unexpectedly, there is at present no definitive way to reconcile these disparate observations, other than invoking technical differences in the studies or stating that the cell-autonomous effect reported by Jaarsma et al. does not negate an additional contribution from glial SOD. What we can say for sure is that we don't yet understand the neuron/glia/muscle interplay in ALS, and that it will be quite hard to establish if the contributions of mutant SOD from each of these sources in transgenic mice are indeed relevant to the human condition.

Posted by Juan Carlos Lopez at 02:51 PM | Permalink | Comments (0)

Have a look at this report from Siri Carpenter that just came out in Science's "Science Careers". It's about how to publish translational research, focusing on what journals look for in the submissions they receive. It features interviews with (and pictures of) several editors, including yours truly.

Not a bad read to kick off the weekend, I reckon.

Posted by Juan Carlos Lopez at 10:43 AM | Permalink | Comments (0)

As I approach the age at which the word 'prostate' starts sounding like a funereal drum, I become more interested in studies such as those published this week in the NEJM and about three weeks ago in Nature Genetics.

The NEJM paper, by Lilly Zheng and colleagues, shows that single-nucleotide polymorphisms (SNPs) in five chromosomal regions, each of which had previously and independently been associated with prostate cancer, have a cumulative association with the disease when considered in combination. The authors estimate that the five SNPs and a family history of prostate cancer account for as many as 46% of the cases in the Swedish population they studied.

The three Nature Genetics papers, by Julius Gudmundsson et al., Gilles Thomas et al. and Rosalind Eeles et al., all of which are nicely summarized in the journal's March editorial, disclose multiple new susceptibility loci associated with prostate cancer that, together with other loci identified in 2006 and 2007, give us plenty of new avenues to explore in order to understand the disease.

The most immediate implication of findings of this sort is often diagnostic -- if you identify gene variants that are linked to a disease, you can ask questions about how good these variants are at predicting onset and/or progression of the pathology. Validating the diagnostic value of these genomic data often requires blinded samples analyzed in a prospective (preferably longitudinal) fashion.

The findings could also help us understand the biology of the disease, although this almost always takes more time and is not always pursued, as it is very challenging: you need to identify with precision the protein whose gene harbors the relevant SNP, then establish how the SNP affects protein function, and finally look at how this altered function modifies the physiology of the cell as it becomes tumorigenic in an in vivo setting.

This is what we at Nature Medicine look for when we evaluate submissions that report new associations of SNPs or mutations with disease, which is why we don't tend to publish too many of these kind of studies. That said, these ruminations do not take anything from the value of these four studies, which shine some more light on the black box that prostate cancer has turned out to be.

Posted by Juan Carlos Lopez at 04:58 PM | Permalink | Comments (0)

Today's paper in PLoS Medicine reporting on a meta-analysis of clinical data on SSRI inhibitors for the treatment of depression really made a splash.

The article, by Irving Kirsch and his colleagues, showed that, when you look at data from 35 randomized clinical trials testing the efficacy of four of these "new-generation antidepressants", the only differences between the drug and the placebo groups are seen in severely depressed patients.

Although this is not the first time that a meta-analysis has provided evidence against the eficacy of SSRIs, it is perhaps the largest study of its kind available so far, immediately fuelling existing concerns about their widespread use in medical practice.

Not unexpectedly, the companies that sell the drugs have already issued statements supporting the efficacy of their products, pointing out that the meta-analysis didn't look at all of the available data. It wasn't clear, however, if the companies were referring to postmarketing surveillance data, something that is entirely possible, as the paper was based strictly on data received by the US Food and Drug Admnistration (FDA) before approval of the drug.

In any case, this is not the first time that we experience this scenario. You can bet that we will now start hearing accusations of negligence against the FDA for approving medicines that don't work. The public image of pharma companies will be further tarnished by negative claims against their products. Psychiatrists will have to reassure their patients, trying to encourage them to remain compliant. Iin fact, some associations of psychiatrists have already issued statements urging patients not to stop taking their medications until they discuss the situation with their doctors.) And people with depression? Well, I guess they'll have every reason to be depressed.

Now, is it really the case that these drugs don't work but in a small subset of patients? These drugs are globally available, and several world regions have their own regulatory agencies, each of which carries its own analysis of efficacy before approving a drug. Are we supposed to accuse not only the FDA, but every other regulatory agency of not doing a good job making sure that a drug works? It's possible, but unlikely.

OK, let's say that the lack of efficacy could not be detected with the data submitted to the regulators, but only with the large number of patients that you look at when you do a meta-analysis. Should the regulatory agencies then ask for much larger (and therefore much more expensive) trials before approving an antidepressant? Maybe, but then it might be too harsh to use a post hoc analysis that had the opportunity to look at data from at least four different companies to then rebuke the FDA for their approval of the drug or each of the individual companies for their trial design.

To my mind, the key issue now is to see what the regulatory agencies are going to do after the publication of these findings. Will they look at postmarketing surveillance data to try to confirm or rebut the alleged lack of efficacy? Will they modify their approval of the drug so that it is only prescribed to the small subset of patients in whom an effect was observed? In several Nature Medicine Editorials, some of which are here (1, 2, 3, 4, 5), we have been critical of the FDA's leadership and commented on the need of the agency to regain the confidence of consumers. So, the publication of the PLoS Medicine paper should not necessarily be construed as a new embarrassment for the FDA, but as a fresh opportunity for the agency to show that it can respond to public concerns, and handle the situation efiiciently and wisely.

Posted by Juan Carlos Lopez at 03:39 PM | Permalink | Comments (2)

It's always gratifying when something you published in your journal is regarded by others as an important contribution. This report in yesterday's New York Times discusses extensively the trial we published last year showing that an agonist of metabotropic glutamate receptors was beneficial in people with schizophrenia.

It's a shame that the NYT didn't identify Nature Medicine as the place in which the original paper was published, but so be it. I'm delighted to see that the findings are receiving the attention they deserve.

Posted by Juan Carlos Lopez at 04:35 PM | Permalink | Comments (2)