From a curt question “What do you want to treat here?” asked by a famous fellow doctor to unique research financed by the European Union. With the discovery of professor Sergiusz Jóźwiak and his colleagues, doctors can now generate a report on the prognosis for hospitalized children by simply pressing a button
The Children’s Memorial Health Institute is a hospital which helps children from all around Poland, including those suffering from tuberous sclerosis. It’s a relatively rare disease affecting 1 per 6,000 infants. It is caused by a mutation of either of two genes, TSC1 or TSC2, which leads to non-cancerous tumors in different organs: brain, heart, lungs, kidneys, fundus, skin.
In 9 out of 10 cases the disease is accompanied by epilepsy. It is diagnosed in 70 percent of young patients aged two or less and causes severe mental development problems. Their degree depends on when the first symptoms were observed – the earlier, when the brain was more fragile and vulnerable, the worse and the bigger the damage. 82 percent of children showing symptoms of epilepsy in the first year of their life become mentally retarded, with their IQ dropping below 20 in many cases. Unfortunately, the condition is extremely resistant to drug treatment. It usually takes years to fight the disease, which results in children getting even more delayed in their development.
Until recently, children affected by the said gene mutation were treated in the Children’s Memorial Health Institute by professor Sergiusz Jóźwiak, a pediatric neurologist, the head of the Department of Neurology and Epileptology of the Children’s Memorial Health Institute in 1997-2015, and a national consultant for pediatric neurology in 2009-2014.
“Our patients were usually admitted to the hospital for the first time when they were five months old after being referred to us due to infantile seizures,” says doctor Jóźwiak, the head of the Department of Pediatric Neurology at the Medical University of Warsaw. “The seizures were accompanied by development inhibition. Children, who had once been able to sit up, would now lie down. Those, who had been able to speak, wouldn’t utter a word. I had been observing that process for years. I knew that most of them would suffer from severe mental development delays and that they would remain dependent on their caregivers until the end of their days. I knew what the families would have to go through and that many of them would fall apart. Although the children were physically healthy, often there was no emotional or intellectual contact with them,” he recollects. “One of my patients, who is now an adult, is under care of her grandma, a frail old woman. Her granddaughter is autistic and weighs 130 kilos. When she was hungry, she could throw a fridge out of a window, but she wasn’t able to grab some food,” he says.
Stopping the downward spiral
Tuberous sclerosis can be diagnosed very early – tumors are visible when the baby is still in the mother’s womb. But even in the case of children who were diagnosed with the disease at such an early stage, doctors waited with the treatment until the first seizures occurred. The approach sounds logical, because how can you treat something that doesn’t even exist?
“I couldn’t stop thinking about it,” says doctor Jóźwiak. “We didn’t want to start the treatment even though we knew that 70 percent of children born with tuberous sclerosis would suffer from epilepsy within the next 24 months, which would result in brain damage. In 2007 I started to perform regular EEG scans of newborn babies before they experienced any seizures – this is the only medical procedure that allows to diagnose epilepsy [Editor’s note: electroencephalographs are used to read brain’s electrical activity from the surface of the skull]. What were the results? It turned out that minor changes were noticeable in EEG scans of 70 percent of patients aged two and three months. Several months later those children started experiencing seizures,” says Jóźwiak.
“‘Wouldn’t it be possible to start the treatment earlier?’ I thought. ‘Maybe we don’t have to wait for the shaking? Maybe we should treat small changes we can see on EEG scans since we know that, in most cases, they inevitably lead to epilepsy?’ No one has ever done that before. I decided to break the rules and to diverge from what we all had been taught: ‘Do not treat non-diagnosed epilepsy’. There wasn’t even a name for what I did: the term ‘epilepsy prevention’ did not exist.
In 2011 the Children’s Memorial Health Institute team led by doctor Jóźwiak started to treat 14 young patients with a new method no one had ever tried before. Children, who were all diagnosed with tuberous sclerosis at an early stage, had an EEG scan performed every four weeks throughout the first 24 months of their life.
Four children from the group (30 percent) showed no changes in their EEG scans nor they were later diagnosed with epilepsy. In the case of the remaining ten children (70 percent) the results were positive. Immediately after changes in EEG scans were discovered, antiepileptic drugs were administered to the patients. In the case of four of them, EEG results stabilized after some time and no clinical epilepsy symptoms followed. Six of them did suffer from seizures but the attacks were easier to control and had no impact on children’s development.
“Preventative treatment resulted in some children not developing epilepsy at all, and some having epileptic seizures later than expected; they were much weaker and easier to treat, and the changes they caused receded more often,” explains doctor Jóźwiak. “The conclusion was clear: if we start the treatment before the symptoms occur, we can avoid the risk of downward spiral and severe complex changes.
Facing an icy reception of the community
Doctor Jóźwiak presented his study results for the first time in 2008 at the conference on neonatal seizures in Tokyo, after which he faced an icy reception of the medical community. When he showed his findings at the European Convention of Pediatric Neurologists in 2011, one of the world’s most highly acclaimed pediatric epileptologists asked him a curt question: “What do you want to treat here? Epilepsy hasn’t even been diagnosed!”
“He was right. According to the official definition, epilepsy was non-existent” says Jóźwiak. “The medical community is prudent. It’s perfectly understandable. Our precept is primum non nocere – “first, do no harm”. We have to be sure that a new method or a new drug is safe. In the past, enthusiastic opinions regarding a new treatment method often proved premature and new outstanding medicines turned out to produce serious side effects. This is why the medical community was skeptical about our theory too. Our courage came only from our strong belief that in the case of this particular disease the procedure was right. I was convinced I wasn’t mistaken. But I knew that it would be difficult for our concept to be wildly recognized.
“Epilepsia”, the most important medical journal focusing on all aspects of epilepsy, rejected their article, because the study was not performed according to the scientific study standards; it lacked randomization, two groups of patients: treated with the new and the traditional method.
“I couldn’t do that,” he explains. “I was sure it was the right way to treat the patients. I wouldn’t be able to tell their parents: ‘Your child is going to be treated with an old ineffective method while the children from the other group are going to be treated according to a new effective procedure’. But as a randomized study was a must, I decided to approach the issue differently. I knew experts specializing in tuberous sclerosis from around the world. We had been meeting at the same conferences for many years. My idea was to find those who supported my theory and to convince them to take part in a joint international project that would prove me right.
That was the origin of EPISTOP, an international project financed by the European Commission, which had two objectives: to prove that preventative treatment is effective and to find the roots of epilepsy. The project involved experts from 16 European research centers and clinical hospitals, e.g. from Tor Vergata University Hospital in Italy, Necker Enfants Malades Hospital in France, Vrije Universiteit Brussel in Belgium, and Charité – Universitätsmedizin Berlin in Germany.
The genetic part of the project was coordinated by professor Dawid Kwiatkowski, an oncologist and geneticist, the head of the Division of Translational Medicine and Genetics, Brigham and Women’s Hospital, Harvard Medical School, renowned for his discovery of the TSC1 gene, who acted in the capacity of a non-European partner. The project was led by professor Jóźwiak.
The project was carried out from 2013 to 2019 and supported the hypothesis about higher effectiveness of preventative treatment.
“We also had a unique opportunity to analyze how epilepsy developed,” says Jóźwiak. “It was the first study of this kind on people. A unique experience. We studied over one hundred children. Each of them was subject to all possible tests: EEG, clinical, and psychological tests. Whole genome sequencing DNA tests, RNA and miRNA expression, proteins. The project finished last year but the data we have gathered are still being analyzed and I hope that they will help to discover new things.”
Loads of data to process
Overall 33 terabytes of data were collected. The Children’s Health Memorial Institute had to install a big server to store them. With that amount of data algorithms and artificial intelligence was a must. IT solutions were provided by the Polish company called Transition Technologies.
“Our job was to find certain biomarkers that would allow to draw conclusions on the risk of epilepsy and that would support the theory of the effectiveness of preventative treatment,” says Konrad Wojdan from TT. “The fact that we had over 33 terabytes of data and only 100 children was quite a challenge. How not to get lost in this constellation of tens of thousands of genes and proteins? How to link everything together? How to find meaningful correlations in all this chaos?”
“We applied machine learning methods,” he explains. “We managed to identify the most important biomarkers and we used them to train a classifier capable of estimating the risk of epilepsy developing in the case of particular children. The key issue of this method is smart reduction of the feature space. Additionally, we have provided both the eCRF system, i.e. the database, and the graphic layer which can be used to input data. Doctors can now generate a report on the prognosis for hospitalized children by simply pressing a button.
It took them one year. They spent all this time on coding and finding solutions in collaboration with doctors and researchers.
“We have recently launched a new project connected with artificial intelligence techniques called ‘Epimaker’, which uses our database,” adds Jóźwiak. “At the moment decisions about when to start the treatment are made somewhat arbitrarily. Basing only on the changes in EEG scans it is hard to tell whether a patient should be given their drugs now or in several days. It would be fantastic if we could determine the right moment. I hope that AI will help us to do that too.”
A pioneer who discovered nothing
In 2020 all children diagnosed with tuberous sclerosis at an early stage referred to the Children’s Health Memorial Institute have their EEG scans performed every four weeks. If any changes are discovered, the drug is administered. The procedure is now officially recommended by EU authorities.
“We have recently sent a survey to about 70 centers around the world in which we asked how they treated tuberous sclerosis,” says doctor Jóźwiak. “It turns out that most centers perform regular EEG scans and half of them administer antiepileptic drugs before actual clinical seizures. We’ve done it!” he rejoices.
The innovative method was first recognized by the patients themselves. In 2009 the American Tuberous Sclerosis Alliance honored doctor Jóźwiak with the Manuel Gomez Award for conducting some of the pioneering research in the field.
“Scientifically speaking, it’s not a big deal,” he says. “I didn’t discover any new drug or any new technology. I simply decided to administer the drug earlier. But from a children’s perspective it’s important. About two years ago I was in the American Epilepsy Academy. I was sitting at the table with an epileptologist from Lima. We were talking shop and at some point he asked: ‘Have you heard that you can treat epilepsy before the first symptoms? I’ve tried it and the results are excellent!’ “That’s nice…” I smiled. When he learnt who I was, he gave me a hug and said: “God bless you!”.
Przeczytaj polską wersję tego tekstu TUTAJ