)
Being young is not a barrier to making an impact in medical researchGerman version/Deutsche Version
The consequences of a point mutation
“>Alpha-1-antitrypsin (AAT) is an enzyme whose main function is to inhibit the enzyme elastase in order to protect us from it. Neutrophil cells of our innate immune system secrete elastase in response to inflammatory processes. Primarily to destroy bacteria. However, if it is not kept in check by AAT, it floods through the body in excess and breaks down the protein elastin, which is important for the elasticity of the lungs. An AAT deficiency would therefore usually lead to incurably progressive chronic obstructive pulmonary disorder (COPD) if it could not be substituted with AAT preparations obtained from donor blood plasma. A lack of AAT is caused by mutations in the gene that contains the blueprint for its production. This production mainly takes place in the liver cells. However, only correctly produced AAT enzymes can leave them without any problems. In some people, the mutation-related waste builds up to such an extent that it increasingly disrupts the function of the liver and can lead to cirrhosis via fibrosis. The replacement of a single amino acid at position 342 is particularly common. This mutation, known as Pi*Z, was the focus of Carolin Schneider’s doctoral thesis.
Through Europe with the Fibroscan
“That was an exciting doctoral thesis,” she recalls. Together with other doctoral students, her task was to recruit around 1,000 test persons across Europe in which either both alleles carried the Pi*Z mutation or only one or none at all. Equipped with a Peter Scriba doctoral scholarship from the German Society of Internal Medicine, she was able to afford a semester off for this. “We drove across Europe in an RWTH car. I visited clinics and self-aid groups in Göttingen, Odense, Bern and Vienna, for example.” In their luggage, they had all the equipment they needed to take blood samples to determine the Pi*Z genotype and a Fibroscan – a special sonography device to measure liver stiffness. The cardinal question that Schneider helped to answer was: How high is the risk of developing severe liver disease for carriers of just one mutant allele, i.e. for the heterozygous genotype of the co-dominantly inherited Pi*Z mutation? The homozygous genotype occurs in only one in 3000 Europeans, while the heterozygous genotype occurs in one in 25. “Liver disease caused by AAT deficiency is more frequently overlooked in homozygotes than lung disease,” explains Schneider. “This is why it is so important to identify risk groups that are hardly clinically conspicuous in good time and to inform them about possible preventive measures.” She herself had originally wanted to be assigned to the control group in the study, until it turned out that she also carries one allele with a Pi*Z mutation. “But fortunately, our study has shown that there are very simple ways to prevent this gene from triggering liver disease.”
Validation of almost 500,000 laboratory values
It was not only the largest study of its kind, praised the journal Gastroenterology in an editorial when it was published there in August 2020 with Carolin Schneider and her colleague Karim Hamesch as first authors[i] . The study also gave a clear signal that a heterozygous Pi*Z genotype in people with obesity, diabetes and alcohol abuse promotes the development of liver fibrosis, which called for appropriate preventive medical advice. The study was already large because of its basic cohort of 1,012 personally examined test persons, but above all because it validated the results obtained using the liver values (transaminases) of 444,642 test subjects from the UK Biobank. This is a non-profit limited company near Manchester that has been providing research since 2012 for a relatively small fee with anonymized health data, blood and tissue samples and gene profiles from 500,000 volunteers, which it examined and recorded up to 2010, and will follow their development for at least 30 years. “It was great that we were given access to this,” says Schneider. “Because every clinical study has certain biases.” In the case of the Liver Study Group, for example, they would have resulted from the focus on self-aid groups, in which by no means everyone affected participates. “But if we see similar effects in a huge crowd from the UK Biobank, then we have examined two independent cohorts and can say that this appears to be an association that can be confirmed.”
Open doors in Philadelphia
To her delight, Schneider found it easy to process the big data from the UK. She rediscovered her enthusiasm for tables, which she had once felt in her advanced math course. “We developed good algorithms for data analysis.” She wanted to try them out in other databases as well. Which is why she contacted Professor Daniel Rader in Philadelphia in 2019 when she learned that he had started to set up the Penn Medicine Biobank at the Perelman School of Medicine. At his large Department of Genetics, the cardiologist had also just discovered a new AAT mutation that protects against liver disease. He was so interested in the implications of Schneider’s research for lipid metabolism that, after a brief phone call, he invited her to Philadelphia to present her doctoral thesis. She had already submitted it for publication, but still had to wait for the paper to be revised before she could defend it. Rader said “no problem”, she could start working for him as a research assistant in December and submit her doctoral certificate later. Funded by the Walter Benjamin Program of the German Research Foundation, Carolin Schneider remained at the University of Pennsylvania until mid-2022, focusing more and more on research into non-alcoholic fatty liver disease (NAFLD).
Computational linguistics for fatty liver diagnostics
This is a silent epidemic in which fat droplets are deposited in the liver cells without alcohol or a chronic viral infection being the cause. Instead, it is primarily caused by genetic factors and obesity. Around two billion seemingly healthy people worldwide are affected by it, mostly without knowing it. In the affluent societies of the West, the NAFLD rate is now approaching 30 percent, with an upward trend, even among slim people. In just under a quarter of all those affected, this sooner or later leads to inflammation, fatty liver hepatitis, from which it is not far for some to progress to fibrosis, cirrhosis, and cancer. This risk could not be directly deduced from the UK Biobank in the AAT study because account data and not free texts are stored there for patient tracking. For example, only 6,000 of the 500,000 patients followed up were diagnosed with fatty liver, which cannot correspond to reality in view of its frequency. According to Schneider, this disproportion can also be explained by the fact that the diagnosis of fatty liver is irrelevant for accounting purposes because it is not reimbursed. Together with Daniel Rader, she therefore set about using Natural Language Processing (NLP) to comb through more than two million reports on biopsies and imaging examinations of the liver that were available in the Penn Medicine Biobank. She ran algorithms over these texts that can understand and evaluate language. And in this way, she was able to identify thousands of patients with fatty liver hepatitis who had not previously been diagnosed.[i] “They could all be recommended lifestyle changes to prevent the progression of their disease.”
A bridge between IT and the clinic
Carolin Schneider used the pandemic-related restrictions of her post-doc period to train in biostatistics and NLP to such an extent that she can now program herself. She now also uses machine learning and artificial intelligence methods – for example when analyzing MRI images of the liver – in order to eliminate subjective bias when selecting analysis criteria. Impressed by her achievements, the state of North Rhine-Westphalia brought her back to RWTH Aachen University in 2022 with a grant of 1.25 million euros to set up her own research group. Her group, whose diversity and interdisciplinarity she is proud of, even includes an astrophysicist who is transferring his experience from deciphering cosmological data floods to pattern recognition in medical databases. This is all the more important as Schneider has started to combine liver-related information from databases in China, Germany, the Netherlands, the UK, and the USA in order to predict who will eventually develop liver cancer. “We often hear the criticism that we are only presenting associations,” says Schneider. “And that’s true at first. But we are working on testing these associations in clinical pilot studies. My group forms a bridge between computer scientists and clinicians.” Her aim is to offer general practitioners in particular easy-to-use procedures, such as a score card with risk factors that go beyond standard laboratory diagnostics, so that they no longer overlook fatty liver so often and can make evidence-based prevention recommendations, including with regard to exercise and diet. For example, based on her analyses, she can clearly recommend foods that contain a lot of vitamin E[i] .
Mentor to her employees
Even though the liver remains her main interest and this “control center of metabolism” offered her a good start to her academic career, she will expand the horizons of her prevention-focused group to other organs and types of cancer in the future. “There are very urgent issues to whose solution we can contribute.” In addition to her work in patient care and research, teaching is also important to Schneider. Promoting the education of women in STEM subjects is close to her heart. Her first doctoral student already has a promising post-doctoral position. At the most recent congress of the American Society for Liver Diseases in Boston, three of her students were awarded prizes for their presentations. “Being young is not a barrier to making an impact in medical research,” says Carolin Schneider. The Life Sciences Bridge Award from the Aventis Foundation is intended to support her on her way to a permanent professorship.
Author: Joachim Pietzsch, Wissenswort; Photos: © Uwe Dettmar
[1] Schneider CV*, Hamesch K*, Gross A, Mandorfer M, Moeller LS, Pereira V, Pons M, Kuca, P , Reichert MC, Benini F, Burbaum B, Voss J, Gutberlet M, Woditsch V, Lindhauer C, Fromme M, Kümpers J, Bewersdorf L, Schäfer B, Eslam M, Bals R, Janciauskiene S, Carvão J, Neureiter D, Zhou B, Wöran K, Bantel H, Geier A, Dirrichs T, Stickel F, Teumer A, Verbeek J, Nevens F, Govaere O, Krawczyk M, Roskams T, Haybaeck J, Lurje G, Chorostowska-Wynimko J, Genesca J, Reiberger T, Lammert F, Krag A, George J, Anstee QM, Trauner M, Datz C, Gaisa NT, Denk H, Trautwein C, Aigner E, Strnad P,# On behalf of European Alpha- Liver Study Group; Liver Phenotypes of European Adults Heterozygous or Homozygous for Pi*Z Variant of alpha-1 antitrypsin (Pi*MZ vs Pi*ZZ genotype) and Non-carriers. Gastroenterology 2020 https://doi.org/10.1053/j.gastro.2020.04.058
[2] CV Schneider*, T Li*, D Zhang, AI Mezina, P Rattan, H Huang, KT Creasy, E Scorletti, I Zandvakili, M Vujkovic, L Hehl, J Fiksel, J Park, K Wangensteen, M Risman, KM Chang, M Serper, RM Carr, KM Schneider, J Chen*, DJ Rader*. Large-scale identification of undiagnosed hepatic steatosis using natural language processing EClinMed 2023. https://doi.org/10.1016/j.eclinm.2023.102149