Category: Filipa M Ferreira

Neuralink, Elon Musk’s backed Brain Computer Interface company, is looking for a clinical trial director. Indeed, they have dozens of open positions and a very attractive website. This company aims at building a BCI system that can be used to help patients communicating through text and daily life devices. Initially, Musk has proposed that with this device, one could top up one’s own brain capacity. However, this seems to be more suited as the topic of science fiction than of real-life. Hence, the goal has been tweaked down a notch.

As with any other medical component, medical devices must go through highly regulated control and scrutiny. Among others, it must be shown that the device does not disturb brain homeostasis. This is quite difficult to prove, as (per figure below) it is quite easy to disturb the brain. Imagining a surgery to implant a device like the “link”, neuronal death and blood-brain-barrier dysfunction (opening) starts within minutes of the 1^st procedure. Moreover, all other cells are affected within hours, with the activation of astrocytes and microglia, and scarization being the most important aspects. Not surprisingly, the immune system gets activated very early on as well, with macrophages and neutrophils being on the first line of response. A consequence of normal process of macrophage is that these cells become tissue resident macrophages. In the brain, it means that they become “microglia-like”, which makes a chronological timeline of events difficult to understand. As an example, if a device that has been implanted for four months has triggered an immune response at three months, when you collect the samples at 3.5 or 4 months, most macrophages that were triggered with the first event will now have differentiated into microglia – a typical brain cell. Hence, based on this observation alone, one would not be able to detect a problem. 

Even if most of these processes tend to stabilize over-time, they stabilize away from their basal level. Hence, appropriate control groups and time points must be chosen, so to understand the impact of a new device in brain homeostasis.

References

DOI: 10.3389/fnins.2019.00689

DOI: 10.1021/acschemneuro.7b00403

DOI: 10.1021/cn500256e

Over the years I have been in a couple of seminars in which the wonders IBM Watson could provide were highlighted. I remember two of them: the ability to process large amounts of imaging data and the ability to analyse large amounts of “written” data, from patient records and from published papers.
By having access to these three types of data, it would be possible to diagnose a given patient, and compare their case to that of other patients that have gone through a similar disease progression, to better optimize their treatment. Moreover, by having access to new developments, an alternative or complementary treatment could have been found.

Almost 10 years later, IBM gave up on Watson and has just sold it to a private equity fund, along to all the medical data it has collected.
Interestingly, it seems that Watson suffered from the ailments of digital health:

• Lack of focus and clear goals
• Lack of interoperability of the different data sets
• The difficulty associated with processing natural language
• Lack of clear regulations and regulatory bodies

Luckily, these are all solvable problems and with the space and momentum created by IBM Watson, better and more on point AI and digital health platforms have appeared on the market.

Further read on IBM Watson sale:
https://medcitynews.com/2021/04/why-ibm-watson-health-could-never-live-up-to-the-promises/
https://www.mmm-online.com/home/channel/features/where-watson-went-wrong/
https://www.advisory.com/daily-briefing/2021/07/21/ibm-watson

And yet again the Multiple Sclerosis (MS) worlds is hyping another paper. This time around, the authors have shown that a protein of the Epstein-Barr virus resembles at the molecular level an endogenous protein present in neurons. This would explain why in MS B cells generate antibodies (and contribute to T-cell response) against one’s own neuronal cells.

The study is behind paywall but here is the link for a nice lay summary of the findings: https://multiplesclerosisnewstoday.com/news-posts/2022/01/25/study-explains-how-epstein-barr-virus-infection-could-cause-ms/

From my side just to say that when I first entered the MS field there were still doubts regarding the autoimmune component of MS. Was the immune response against one’s own brain cells due to molecular mimicry or due to bystander activation?

• Molecular mimicry: an adequate immune response against a foreign pathogen (virus, bacteria, fungi, etc) generates B-cells and T-cells that have the capacity to recognise that pathogen and components of the human body that are molecularly similar.
• Bystander activation: an adequate response against a foreign pathogen takes place in an area with some self-damage (or causes it) and in the “heat of the battle”, the immune system cannot distinguish what is foreign and a treat from what is self and harmless. Hence, it specifically creates cells to attack that part of the body.

Both become a problem when the immune cells remain active after the pathogen has been cleared “by seeing” the body part as a foreign treat, hence causing some destructive friendly fire.

Recently there was an hype on (my) LinkedIn about a new paper published in Science that claims a causal link between Epstein-Barr-Virus (EBV) infection and development of Multiple Sclerosis (MS).

What is so extraordinary about this paper that one can claim causation instead of correlation?

Long story short, and in my opinion, the most extraordinary is the number of samples studied. The scientists had access to samples collected by the US-military who, on average, collected 62 blood samples per person, for each of the 10 million people enlisted over a period of 20 years. In context, scientists had access to a study group the size of Portugal’s population.

Where these samples representative?

From the enlisted population, about 95% had previously contracted an EBV infection (in line with the prevalence of EBV in the general population) and in total 955 people went on to develop Multiple Sclerosis. If we simplify and assume 48 new MS patients per year, this is lower than the average annual incidence in the UK. Moreover, in the general population, 1 in 3 MS patients are men whereas in this study the ratio is the exact opposite (see study design below).

Is this a good paper?

Technically, the paper is not very elaborated (note that I did not look at the supplementary material), it checks serum levels of anti-EBV antibodies, antibodies against several other viruses for control, the presence of soluble Neurofilament L (sNfL), and EBV antigens. But the techniques applied are adequate.

What are the main conclusions?

The results are discussed assuming that sNfL is the earliest molecular marker of MS. Because, in these patients, detection of antibodies against EBV in serum (seroconversion) occurs before detection of sNfL, it is concluded that first there is an EBV infection and only then MS development.

The authors check the levels of antibodies against other viruses and see no differences between MS patients and the control group before and after disease onset, suggesting that no other viral infection precedes disease onset.

Importantly, the authors say that the absence of differences, comparing to the control group, in the response against other viruses before and after disease onset suggests that MS does not cause a susceptibility to viral infections; and specifically, that MS does not cause susceptibility to EBV infection, as the responses against this virus does not change before and after onset in the patient group.

Figure 1 from the paper, with the study design

REFERENCES

K. Bjornevik et al., Science 10.1126/science.abj8222 (2022).