The media is buzzing after recent publication of an article in the journal Science: Translational Medicine, which suggests that we can treat Alzheimer’s disease non-invasively with ultrasound. As usual though, they may be making more hype than is warranted given a thorough look at the data; or if they would have read the last paragraph, where the authors discuss (some of) the study’s weaknesses.

The study authors, from Queensland Brain Institute in Australia, used repeated scanning ultrasound in conjunction with intravenous injection of microbubbles (a gas core wrapped in a lipid shell) to safely induce blood brain barrier opening in a mouse model of Alzheimer’s disease (AD). The current understanding of how this method works is to vibrate the bubbles using ultrasound, causing them to expand and contract. This action displaces the vessel wall and transiently releases the tight junctions between blood vessel cells (endothelial cells) that keep a tight lock on the barrier.

Interestingly, they found that in doing so, they could reduce the number of the amyloid-beta plaques that are a characteristic (and defining) pathology of the disease. In the Amyloid Cascade Hypothesis, a dogmatic theory that has dominated the field for several decades, the aggregation of the amyloid-beta protein (a metabolite of the Amyloid Precursor Protein) into plaques results in neuronal death. This hypothesis has been continually revised and now blames the pre-aggregated, lower molecular weight amyloid oligomers for the ensuing destruction.

We will first begin by looking at the animal model in use. The APP23 mouse model of AD is widely used, but like any model has its weaknesses. The problem with many of the AD mouse models is that they do not necessarily recapitulate the most common human form of AD: sporadic. These models rely on genetic manipulation of the amyloid precursor protein to generate more amyloid-beta than physiologic levels seen in humans with the disease — basically, way too much. Right off the bat, they are biasing their results.

But let’s say, hypothetically, that the APP23 model is sufficient to answer questions about the causes of AD. The experimental groups consist of a wild type (normal) mouse, APP23 mouse without ultrasound (also called the “sham” group) and APP23 mouse with ultrasound (the treated group). Both of the APP23 groups received the microbubble injection. The authors show visually and biochemically that they can reduce levels of many forms of amyloid-beta with microbubble-coupled ultrasound compared to the sham group at no ancillary risk to the animal.

When they look at the behavioral manifestations of this, however, the effects are not as robust. In fact, it looks as if the sham group may get worse, which is difficult to explain. This is evident when the APP23 mouse (without ultrasound or microbubbles) is compared to the sham group. The treated group does show a slight restoration in the behavioral outcome, but considering the variable overlap with the sham group, it is difficult to interpret this as clinically significant.

Now, how is this working? Enter, microglia — maybe. As the innate immune cells of the brain, they have previously been implicated in the removal of amyloid-beta from the brain (including in my own work). The authors do show, rather convincingly, that microglia are eating up a lot of amyloid-beta in the treated group; almost twice as much as microglia in the sham group. That is a lot of amyloid for microglia, as they have occasionally been accused of inefficiency in this context.

The authors try to tie this in with the opening of the blood brain barrier by suggesting that albumin (the most common fluid-phase blood protein) leaks into the brain and somehow helps microglia gobble up higher levels of amyloid in the treated group. Not only is this ambiguously demonstrated in a cell culture model (in vitro) without any mechanistic evidence, but it sounds more like magic. Thankfully the authors concede that additional work should be completed on this front.

In looking for other mechanisms in which this reduction in amyloid could be occurring, the authors investigated microglial activation state, in which the only altered characteristics were those suggestive of phagocytosis (i.e. microglial cell morphology was more rounded and they expressed higher levels the phagocytic marker CD68). This does lend credence to their amyloid phagocytosis hypothesis, and they suggest this is somehow a direct result from the ultrasound stimulation. This seems a bit specious, so let us look at this scenario a different way.

The results of this study are predicated on opening of the blood brain barrier and the influx of albumin, which magically helps microglia consume more amyloid. Why the authors went with this explanation of amyloid-beta reduction and not the low hanging fruit of infiltrating peripheral monocytes, is beyond me.

It has been shown in many studies that peripheral monocytes (which become macrophages when they enter into a tissue) are much more efficient at engulfing amyloid-beta than endogenous brain microglia. Unfortunately, unless molecularly scrutinized, the differences between microglia and monocytes are nuanced, and the methods used in this study fail to differentiate them. The lack of distinction between the two cell types is a hot topic in neurodegenerative diseases at the moment and it surprises me that the reviewers and editors of the journal overlooked this obvious weakness in the study.

To be fair, the ability to open up the blood brain barrier non-invasively is a huge strength of this study and despite the lack of evidence for how it works to reduce levels of amyloid-beta, there are other important potential uses for this technique. One such example would be to enhance delivery of drugs to the central nervous system, a feat by which the neuroscience, neurology and neurosurgery communities are currently baffled. Once an effective drug is developed, ultrasound could be the means for targeted delivery to the brains of Alzheimer’s patients.

References

Hardy J (2006). Alzheimer’s disease: the amyloid cascade hypothesis: an update and reappraisal. Journal of Alzheimer’s disease : JAD, 9 (3 Suppl), 151-3 PMID: 16914853

Leinenga G, & Götz J (2015). Scanning ultrasound removes amyloid-? and restores memory in an Alzheimer’s disease mouse model. Science translational medicine, 7 (278) PMID: 25761889

Town T, Laouar Y, Pittenger C, Mori T, Szekely CA, Tan J, Duman RS, & Flavell RA (2008). Blocking TGF-beta-Smad2/3 innate immune signaling mitigates Alzheimer-like pathology. Nature medicine, 14 (6), 681-7 PMID: 18516051

Zabel M, Schrag M, Crofton A, Tung S, Beaufond P, Van Ornam J, Dininni A, Vinters HV, Coppola G, & Kirsch WM (2013). A shift in microglial ?-amyloid binding in Alzheimer’s disease is associated with cerebral amyloid angiopathy. Brain pathology (Zurich, Switzerland), 23 (4), 390-401 PMID: 23134465

Image via Akbudak Rimma / Shutterstock.

Shutterstock ID: 65181187