. GM-CSF upregulated in rheumatoid arthritis reverses cognitive impairment and amyloidosis in Alzheimer mice. J Alzheimers Dis. 2010;21(2):507-18. PubMed.


Please login to recommend the paper.


  1. Boyd and coworkers present an interesting set of data demonstrating that both direct intracerebral administration and indirect subcutaneous injections of the cytokine GM-CSF significantly ameliorate β amyloid plaque burden in Tg2576 mice. Surprisingly, the authors actually noted the greatest reductions in Aβ deposits—57 percent—in their peripheral (subcutaneous) GM-CSF treatment paradigm. Coincident with reduced cerebral amyloid deposits, Boyd and colleagues noted near-complete amelioration of behavioral impairment and a modest but significant increase in Iba1+ microglia after daily subcutaneous GM-CSF injections. The authors deduce that GM-CSF promoted innate immune (likely microglial) clearance of Aβ plaques.

    This work is particularly interesting from a translational medicine standpoint, as the related molecule G-CSF was just approved to move forward into a Phase 2 trial for AD. The present authors make a compelling case that, because GM-CSF is more effective than G-CSF in AD model mice, GM-CSF/leukine may be the better candidate for treatment of human AD. The authors encountered potential safety issues (hyperplasia near the cerebral injection site that may indicate tumor formation) in AD model mice given intracerebral injections of another related molecule, M-CSF, which seem to rule out its use in the clinic.

    I’d like to offer a few comments on the basic science front. The authors make reference to rheumatoid arthritis (RA) as a “negative risk factor” for AD. While there seems to be a negative association between RA and AD, at least in my mind, it is still controversial whether RA is truly a negative risk factor for AD. While many of the early epidemiologic studies found an inverse association between RA and AD, it was shown two decades ago that RA was a surrogate for NSAID use, as most arthritics are prescribed NSAIDs. This shifted the collective focus to NSAIDs as the protective factor.

    Just recently, it has been suggested by Christine Szekely and Peter Zandi (Szekely and Zandi, 2010) that there may be some independent “protection” from AD in RA patients that are non-NSAID users, but we await confirmation of this result by other groups. Even if we accept that RA is inversely associated with AD independently of NSAIDs, there are myriad innate immune cytokines produced during the course of RA, including pro-inflammatory neurotoxic molecules such as TNFα and IL-1β, and it seems a tad over-reaching for Boyd and colleagues to conclude that a negative association between RA and AD is owed to one molecule—GM-CSF. The authors build on this idea by commenting that NSAID clinical trials have proven unsuccessful. I am assuming that the authors are partly referring to results from the only primary prevention NSAID RCT for AD—ADAPT. I would offer caution when interpreting the ADAPT results. At the ICAD meeting in 2007, John Breitner presented data showing that, with extended follow-up time (two years of treatment and ~two years of follow-up), naproxen produced a positive signal on mitigating conversion to AD. There are other factors at play here, as well. For example, it is not unfathomable that NSAIDs produce different outcomes depending on whether they are administered to cognitively normal individuals or to folks ”on the verge” of conversion to AD.

    Finally, the authors speculate on mechanism(s) of GM-CSF action in their experimental paradigm. In the direct brain injection experiment, it is possible that GM-CSF changed the phenotype of the microglia from non-phagocytic to Aβ-clearing. However, the authors invoke a different mechanism in their peripheral GM-CSF injection model. There, they raise the possibility that peripheral innate immune cells may have been coaxed into the brains of AD model mice. Though not discussed by the authors, we have found that blocking another cytokine—TGFβ—specifically on peripheral innate immune cells produces this effect in two different mouse models of AD (Town et al., 2008). I wonder whether the authors have carefully examined, either by confocal microscopy or flow cytometry, whether there is evidence of bona fide peripheral macrophage infiltration into brains of AD model mice given GM-CSF?


    . Non-steroidal anti-inflammatory drugs and Alzheimer's disease: the epidemiological evidence. CNS Neurol Disord Drug Targets. 2010 Apr;9(2):132-9. PubMed.

    . Blocking TGF-beta-Smad2/3 innate immune signaling mitigates Alzheimer-like pathology. Nat Med. 2008 Jun;14(6):681-7. PubMed.

Make a Comment

To make a comment you must login or register.

This paper appears in the following:


  1. GM-CSF Leukine


  1. At AAIC, Yet Another Phase 3 Flop While Phase 1 Trials Forge Ahead