Using machine learning to diagnose disease

Profiling the immune system in paediatric arthritis patients offers hope for improved diagnosis and treatment

A team of scientists from VIB and KU Leuven has developed a machine learning algorithm that identifies children with juvenile arthritis with almost 90% accuracy from a simple blood test. The new findings, published this week in Annals of the Rheumatic Diseases, pave the way for the use of machine learning to improve diagnosis and to predict which juvenile arthritis patients may respond best to different treatment options. The work was led by Professor Adrian Liston, a group leader at the Babraham Institute in Cambridge, UK and at VIB and KU Leuven in Leuven, Belgium.

Juvenile idiopathic arthritis is the most common rheumatic disease in children, but it presents in many different severities and forms. This diversity makes clinical assessment and patient classification difficult.

A team of researchers at Belgian research organisations VIB, KU Leuven and UZ Leuven undertook a detailed biological characterisation of the immune system of hundreds of children with and without juvenile arthritis to help the diagnosis or treatment decisions for this disease.

“Essentially, we took blood samples from more than 100 children, two thirds of whom had childhood arthritis,” explains Erika Van Nieuwenhove (VIB-KU Leuven), and first author of the study. “We analysed their immune system at a greater level of detail than was ever done before for this disease, and simply using this data we then used machine learning to see if we could tell which children had arthritis.”

The results were quite remarkable: the algorithm was about 90% accurate at identifying the children with the disease. “Using only information on the immune system, and no clinical data at all, we could design a machine learning algorithm that was about 90% accurate at spotting which kids had arthritis,” says Professor Adrian Liston (Babraham Institute, Cambridge, UK and VIB-KU Leuven). “This result is a proof-of-principle demonstration that immune phenotyping combined with machine learning holds huge potential to diagnose disease. Similar approaches could be applied to improve patient selection for treatments and clinical trials.”

The researchers are hopeful about the impact of this research in improving patient outcomes. “The tool needs further validation but otherwise there are no scientific barriers to this approach being quickly translated to the clinic,” comments Professor Carine Wouters (UZ Leuven), who was the clinical lead for this study. “Down the line, we could use this kind of detailed classification information—and machine learning analysis—to identify which patients will respond best to specific treatment options.”

Interview with "The Optimist" magazine

Read the article at The Optimist

Thinking back to your PhD, how would you describe this experience.

I quite enjoyed my PhD. The key success in a PhD is to find a match between supervisor and student. I only spoke to my supervisor every 3-4 months, and it was always about concepts and strategy rather than trouble-shooting. For me, I loved the independence that this gave me, and the amazing post-docs in the lab gave me more than enough technical advice. However, some of the other students around me did not like this approach to mentoring, and would have preferred weekly meetings going into the detail of their experiments. This was pure luck on my behalf - I could easily have ended up in a lab that I found stifling, because I didn't ask the right questions going in. This independence let me mould my PhD to my strengths. I learned just a few basic techniques and then applied them to novel questions. It was an approach that let me generate data and papers quickly, and led to an "easy PhD". 

To ask the famous question, is there anything you would like to change retrospectively regarding your PhD?

The flip side of having an "easy PhD" is that I never really had to leave my comfort zone. Since I didn't spend months (or years) painfully learning and optimising new techniques, I never became as technically skilled as the other PhD students around me. Science is so fast moving, that the best strategy is to learn how to learn, which you only get the hard way. Instead, I had my couple of techniques and I had learned how to plan experiments and writing papers. This made my post-doc really difficult - I didn't have the versatility or skill of other post-docs around me, who were picking up and using the latest techniques with trained ease, earned by blood, sweat and tears earlier on. Now as a PI, this deficit is not so important, since my job is all planning and writing, but even now I regret never learning to become a great experimentalist.   

 Which advice or tips would you give us PhDs on our way?

1. Analyse experiments as you go. I started the habit very early on of always analysing experiments once I finished them. By this I mean a full analysis, including a publication level graph, a figure legend and a few lines of text describing the result. It takes a little time, but it means you get real-time feedback on the quality of your experiments - give you have all the right controls, were the numbers high enough to make conclusions, are my conclusions solid enough to plan the next stage, etc. It also made writing papers and a thesis very simple - I just cut and paste my analysed data in, and I was half-way there. Since editing is much less intimidating than writing, I never developed that writing paralysis that some students get.

2. Don't stress about careers! The infamous "bottleneck" in the academic career is mostly illusional. In Flanders, perhaps 15-20% of PhD students go on to a long-term academic career, but even in countries with lower rates (2-5% would be normal) this this not due to a bottleneck. A PhD in biomedical science is one of the most desirable training programs possible for a modern career. The vast majority of people who leave academia are not pushed out; instead, biomed PhDs are leaving academic because of pull-factors - they find highly desirable jobs that they believe they will enjoy more. When I look around at the PhDs that I trained with or that I have trained, I can honestly say that not one has had a career failure. Yes, very few are now research professors, but that is because almost all of them found something else they liked more. Doctor, CEO, start-up company, scientific writer, senior public servant - all great jobs. Very, very few of the 100+ academic careers that I have followed have ended with someone getting pushed out of academia (i.e., timing out of the post-doc fellowship system), and those that did landed on their feet and found a great career that they now say is better suited to them. So.... don't stress about your future career. Concentrate on doing well in your PhD, and start planning your career a year in advance of any decision, but don't make yourself unhappy about uncertainty in which successful career path you will end up taking.

EMBO Young Investigator meeting

Great meeting with great people

Punting on the Cam

Visiting the original lab books of Rosalind Franklin

Front cover of Blood

Our study on myeloproliferative disorder is on the front cover of Blood

Farewell to John Barber

Sad to see John Barber leaving us to go back to his medical degree. He has spent the last year uncovering a novel genetic cause of neutropenia - details to follow soon!

Best of luck John, you'll be missed!

Congratulations to Prof Humblet-Baron!

Congratulations to Dr Stéphanie Humblet-Baron, who was just awarded the prestigious (and highly competitive) BOF-ZAP award. With this award Stéphanie starts a tenure-track research professorship and her own independent group.

The success of both Prof Humblet-Baron and Prof Schlenner at the BOF-ZAP awards puts the Translational Immunology laboratory in a great position. Going forward from my move to the Babraham, Prof Schlenner is leading the mouse immunology research and Prof Humblet-Baron is leading the clinical immunology research program.

Leaving two such talented and determined women to take over my lab, and push it to new heights, is my proudest legacy of 10 years in Leuven.

The Humblet-Baron team will develop and use cutting-edge systems immunology approach to study the many diseases in which the immune system places a key role, from primary immunodeficiency to infections to cancer to neurodegeneration. Watch out for great things new Prof Humblet-Baron is here!

Identical twins light the way for new genetic cause of arthritis

Identical twin girls who presented with severe arthritis helped scientists to identify the first gene mutation that can single-handedly cause a juvenile form of this inflammatory joint disease. By investigating the DNA of individual blood cells of both children and then modelling the genetic defect in a mouse model, the research team led by Adrian Liston (VIB-KU Leuven) was able to unravel the disease mechanism. The findings will help to develop an appropriate treatment as well.

Juvenile idiopathic arthritis is the most common form of all childhood rheumatic diseases. It is defined as arthritis that starts at a young age and persists throughout adulthood, but which does not have a defined cause. Patients present with a highly variable clinical picture, and scientists have long suspected that different combinations of specific genetic susceptibilities and environmental triggers drive the disease.

A single gene mutation

In a new study by researchers at VIB, KU Leuven and UZ Leuven, the cause of juvenile arthritis in a young pair of identical twins was traced back to a single genetic mutation.

"Single-cell sequencing let us track what was going wrong in every cell type in the twin’s blood, creating a link from genetic mutation to disease onset,” explains Dr. Stephanie Humblet-Baron, one of the researchers involved in the study. “It was the combination of next generation genetics and immunology approaches that allowed us to find out why these patients were developing arthritis at such a young age.”

Of mice and men

Parallel studies in mice confirmed that the gene defect found in the patients’ blood cells indeed led to an enhanced susceptibility to arthritis. Prof. Susan Schlenner, first author of the study, stresses the relevance of this approach: "New genetic editing approaches bring mouse research much closer to the patient. We can now rapidly produce new mouse models that reproduce human mutations in mice, allowing us to model the disease of individual patients."

According to immunology prof. Adrian Liston such insights prove invaluable in biomedical research: “Understanding the cause of the disease unlocks the key to treating the patient.”

From cause to cure

Liston’s team collaborated closely with prof. Carine Wouters, who coordinated the clinical aspect of the research: "The identification of a single gene that can cause juvenile idiopathic arthritis is an important milestone. A parallel mouse model with the same genetic mutation is a great tool to dissect the disease mechanism in more detail and to develop more effective targeted therapies for this condition.”

And the little patients? They are relieved to know that scientists found the cause of their symptoms: "We are delighted to know that an explanation has been found for our illness and more so because we are sure it will help other children."

Thankfully, the children’s arthritis is under good control at the moment. Thanks to the new scientific findings, their doctors will be in a much better position to treat any future flare-ups.

NFIL3 mutations alter immune homeostasis and sensitise for arthritis pathology 

Schlenner et al. 2018 Annals of the Reumatic Diseases

Advice for international faculty entering the Belgian system

After 10 years in Belgium as a foreign academic, I was reflecting on what lessons I had learned about Belgian academia that I wish I had learned earlier. The key lesson I wish someone had taught me is that Belgian academia works via the patronage system. Having come from the Anglo meritocracy system, it took me a long time to distinguish the features of patronage, as opposed to nepotism or just plain corruption. The system is based on patrons and their protegés, and is so embedded throughout the system that nearly everyone is both a patron and a protegé, in one large network of obligations. One you understand this, many previously inexplicable behaviours and patterns make sense, and it is possible to thrive.

The patron is the "godfather". Patrons do not give orders. They make their wishes known, and they expect their protegés to follow them. In the worst cases, this is exploited simply to further the career of the patrons (by making them successful in the eyes of their own patron). It is expected that the rise of the patron trickles down to the protegé, however it is understood that years or even decades can elapse in between. The patron will also have multiple protegés at the same stage, competing for the spoils of the patron. A highly successful patron may pull up multiple protegés through the system with them, but often it is just the one leading protegé who benefits (and may end up becoming patron to the other protegés). This gives the scenario, rather perplexing to an outsider, of good people putting their patron ahead of themselves year after year, without any benefit. In the best cases, the system can work better than an open merit-based system. Merit-based system work based on track-record, and assume that someone with a good record and a winning interview style will be successful in the position. By contrast, the relationship between patron and protegé is far more intimate, and a patron can ideally identify the person who would be best at the job, beyond the confounding luck and personality issues that can sway a CV or interview. In practice, the worst case scenario seems rather more common than the best case scenario.

The protegé is expected to follow the lead of the patron throughout their career, even after they have reached equivalent levels. The level to which this deference is given is quite remarkable to an outsider. The protegé, even as an independent academic, will follow the guidance of their patron is terms of who they should collaborate with, which grants they should submit, authorship order on papers and key financial decisions. Even when they strongly disagree with the advice, the protegé is more likely to get frustrated in private than to argue (or even ignore) their patron. It would take a really gross violation by the patron for the protegé to break off the relationship. Protegés will take on protegés of their own, becoming a patron to others while still remaining a protegé to their own patron. This creates a "family", with the überpatron at the head. The überpatron respects the downwards hierarchy, however, working via their direct protegés only.

There are aspects of patronage to every academic system, however the key to understanding Belgian academia is to realise that patronage permeates every facet. The degree to which the rules of the patron-protegés relationship are embedded in the Belgian character is difficult to overstate, and even Belgians that have become internationalised may struggle to not default into the pattern. The breakdown of the patron-protegés relationship is very rare, and invokes a deep sense of betrayal.

In practical terms, what does this mean for a foreigner entering Belgian academia?

First, as a foreigner you are unlikely to have a strong patron. For myself, after 10 years in Belgian, I never had a patron.  Admittedly, I am very poor protegé material, but the strongest patronage relationships are formed very early, through family or family friends. Even without having your own patron, however, your Belgian students may instinctively fall into the protegé role, and expect the patron-protegé relationship to function. 

Second, to some extent you need to chose between looking crass or being overlooked. Belgians do not put themselves forward, and self-promotion is considered coarse. It is the role of your patron to promote your virtues or ideas. Without a patron, you can either push yourself forward or be ignored. Self-assertion can still work, and is tolerated more in foreigners than in Belgians, but it won't make friends.

Third, breaking into collaborative networks is going to be difficult to you. Many collaborative grants and projects are built around the patron-protegé networks. Trying to join can be looked upon as nearly akin to crashing a family reunion. Some überpatrons are highly xenophobic, and you will never be allowed in. This can lead to frustration, where a good relationship with a colleague never blossoms into a collaboration, not because they don't want to work with you, but because their own patron disapproves. The best path to breaking into the collaborative networks lies in identifying a more liberal überpatron, and getting adopted into the family. Offering up a new technology or the like can be sufficient to get you adopted, for the advantage that it brings in. Adoption does depend on at least tacitly respecting the hierarchy: the network is not a collection of independent equals, and directly approaching protegés can be taken as highly disrespectful by überpatrons.

Fourth, respecting the hierarchy does not just apply to collaborative networks. Belgian academics would never directly approach one of your students or staff with a question, and expect the same "courtesy" from you. Even students from neighbouring labs don't *officially* talk to each other, and even trivial requests for help are expected to go up to their promoters for consultation first. This feels needless and inefficient, but ignoring the protocol is worse: a faux pas can shutdown relationships with your neighbours for good.

Fifth, you need to understand that arguing in a meeting is normally fruitless. Despite being in hundreds of management meetings over the past 10 years, I've never seen a question go to an actual vote, or even anything approximating a vote. The überpatrons in the meeting know who dominates based on who is in the meeting. To force protegés to openly support their patrons would be the height of rudeness - the system is built on implicit support, not explicit support. Decisions are made before the meeting, either between the überpatrons directly, or via jockeying to stack the meeting with their protegés. I never managed to suppress my instinct to give my opinion at meetings, even when it was clearly against the position of the überpatrons, but even if others at the meeting secretly agree with you it just ends up making you look obnoxious. The more successful approach is to identify the dominant network in advance (not difficult, the chair is usually the leading überpatron), and discuss with them before the meeting. Überpatrons can be thoughtful and consultative, as long as you approach them in private: changing their mind after a public disagreement results in too much loss of face.

Sixth, be highly sensitive to suggestions to place specific people in a position. If you are trying to get a position opened up to fill an unmet need, the actual person who will end up being recruited is typically of secondary importance to you. Open the position, and then find the best person to fill it. For many Belgian academics, however, the actual person who will be recruited is the key point. Getting their protegés positions is the key function of a successful patron. If a Belgian academic listens to your proposal and then suggests one of their protegés for the position, what they are often proposing is a deal: I would support this proposal if my person gets the position. If you dismiss the proposal as downstream, then the patron will think that either you are trying to open the position for your own protegé, or you have an agreement with another patron. The strategic design of proposals to incorporate a protegé from a key überpatron can often lead to success (assuming the protegé would actually be suitable for the position).

Finally, this should be read not as an attack on the Belgian system, but rather as an explanation of how it functions, as discovered by an initially clueless outsider. For Belgian readers, while they would certainly recognise the behaviours, they would probably be rather appalled to see it written in this way. This is not a codified behavior or quid pro quo; it is just how to be polite and respectful to your colleagues and, especially, your seniors. While it can be infuriating at times, and I've occasionally come home fuming about corruption and xenophobia, that is entirely because I was not raised with these same cultural preconceptions. It is surely no worse than the cultural shock of entering the American system, where entirely different sets of rules are at play, and even though they are invisible to me, Australian academia surely has cultural quirks that offend the sensibilities of outsiders. You don't have to embrace the Belgian system, or become part of it, but at the very least you should understand it and respect it enough to not be *unintentionally* obnoxious.

Stephanie Humblet-Baron tackles one immune disease after the other

In two recent studies, the same team of scientists has uncovered the mechanisms underlying two distinct immunological disorders affecting both children and adults. Stephanie Humblet-Baron(VIB-KU Leuven) was the researcher at the helm of both projects.

A pediatrician by training, Stephanie Humblet-Baron is building a career unravelling immunological disorders that affect children. She divides her time between the clinic and the lab, where she is a senior team leader in the lab of Adrian Liston (VIB-KU Leuven).

From disease to biology and back again

Ever since the start of her medical training, Humblet-Baron developed a special interest in unraveling the biological mechanisms that cause immunological problems. Many immune diseases are poorly understood, and this lack of knowledge also limits treatment options and choices.

“People sometimes refer to these diseases as rare,” says Humblet-Baron, “but we all carry risk factors for many immunological diseases. Even if a given mutation is rare, the accumulated variation in immunological responses affect a broad set of outcomes, for example how someone responds to cancer treatment or drugs for cardiovascular conditions. That is why understanding the mechanisms underlying immune-dysregulation is so important.”

In her most recent work, Humblet-Baron, together with her colleagues in the lab of Adrian Liston (VIB-KU Leuven), focused on the mechanisms causing myeloproliferative disease and hemophagocytic lymphohistiocytosis, two diseases that are fatal unless given aggressive treatment.

Myeloproliferative disorder: a partner in crime for dendritic cells

Dendritic cells are specialized antigen-presenting cells that play a crucial role in coordinating innate and adaptive immune responses. In both patients and mice, depletion of dendritic cells leads to myeloproliferative disorder, but how or why—no one really knew.

“To understand what was going wrong, we created a mouse model where dendritic cells were present in normal numbers, but were functionally impaired,” explains Humblet-Baron. “We found that without the antigen-presenting capacity of dendritic cells, the mice developed myeloproliferative disorder.”

The team uncovered that it was not the number of dendritic cells, but their partnership with CD4 T cells of the immune system that was crucial for disease development. When CD4 T cells were absent as well, the mice showed no symptoms of myeloproliferative disease.

This has important implications for patients, where specific mutations also manifest both dendritic cell deficiency and myeloproliferative disorder. “Based on the original model disease model, the proposed line of treatment would be dendritic cell replacement, currently only possible through bone-marrow transplantation,” says prof. Adrian Liston. “But these new results indicate that attenuating the activatory signal from CD4 T cells could also reduce the development of myeloproliferative disorder.”

Hemophagocytic lymphohistiocytosis: New light on a deadly disease

Hemophagocytic lymphohistiocytosis, HLH for short, is a severe disease less than 2 out of 3 patients survive. It can be triggered by a variety of factors, including genetic defects, viral infections, anti-tumor responses or unchecked autoimmunity. Excessive production of interferon γ was assumed to be the key pathological step, but based on patient evidence and a pre-clinical mouse model of the disease, the Leuven research team has now found that there is much more to it.

Humblet-Baron: “We found that the production of interferon γ was only responsible for part of the features of the disease. Excessive consumption of the immune signaling molecule interleukin 2 by hyperactivated CD8 T cells, the suppressor cells of our immune system, had a much greater impact on the inflammation.”

This means that at least two different disease pathways are at play—knowledge that indicates that we could save the lives of more patients if we also targeted both pathways during treatment.

“This study not only provides a new paradigm for understanding HLH, with major implications for its treatment, but also gives us a broad insight into how hyperactivated CD8 T cells cause damage,” adds prof. Adrian Liston.

“We can learn so much from an in-depth analysis of the immune cells present in a simple sample of blood from patients,” concludes Humblet-Baron, who hopes to uncover the mechanisms underlying many more of these immunological problems. “Coupled with the power of biochemical and animal research, these insights are really changing how we diagnose and treat patients in the clinic.”

—
Humblet-Baron, et al 2018 Blood. "Murine myeloproliferative disorder as a consequence of impaired collaboration between dendritic cells and CD4 T cells"

Humblet-Baron et al. 2018 Journal of Allergy and Clinical Immunology. "IFN-γ and CD25 drive distinct pathological features during CD8 T cell hyperactivation in hemophagocytic lymphohistiocytosis"

---