iL‒CURRENT
February 18, 2026
Professor Haikun Liu from the DKFZ: "Publications don't save lives—products do."
Glioblastome zählen zu den aggressivsten und therapieresistentesten Krebsarten – trotz jahrzehntelanger Forschung bleiben echte Durchbrüche selten. Professor Dr. Haikun Liu erklärt im iL-Experteninterview, warum prädiktive Modelle, patientenindividuelle Tumororganoide und KI-gestützte Analysen den entscheidenden Wendepunkt markieren könnten – und wie Translation, Unternehmertum und interdisziplinäre Zusammenarbeit die Hirnkrebstherapie grundlegend verändern sollen.
Bei GENESIS 26 – The LIFE SCIENCE TECH DAY am 12. März wird Haikun Liu, der am DKFZ die Abteilung Molekulare Neurogenetik leitet und im letzten Jahr das Spin-off AIPTO TechBio GmbH gegründet hat, einen Vortrag halten und eine Experten-Session moderieren.
Professor Dr. Haikun Liu, Head of the Molecular Neurogenetics Department at the DKFZ and founder of AIPTO TechBio GmbH. Image: Haikun Liu/DKFZ
Five questions for ... Haikun Liu
Mr. Liu, glioblastomas - highly malignant brain tumors - are considered among the most complex and therapy-resistant forms of cancer. What is the key biological bottleneck that must be overcome to make a cure possible?
Haikun Liu: Over the past 15 to 20 years, there has been enormous effort in glioblastoma, including many clinical trials—yet none has delivered a clearly effective new drug for patients. The long absence of truly impactful new drug approvals suggests the key bottleneck is not a lack of efforts but lack of predictive models. Too many candidates look promising in cells and animal models, but those results fail to predict patient response. We therefore need preclinical systems that are genuinely predictive for clinical efficacy and can substantially raise the success rate from discovery to patient impact.
A second major bottleneck is that many commonly used models (inbred animals) are overly uniform and do not reflect real-world patient diversity. GBM patients are profoundly heterogeneous, and that heterogeneity determines who responds to a drug and who does not. We need models that capture patient-to-patient diversity so we can identify responders versus non-responders and design truly effective, stratified therapies.
Just over a year ago, you founded AIPTO TechBio. What triggered this step, and which gap between the laboratory and clinical application are you aiming to close with this company?
Liu: I founded AIPTO precisely because we recognized these limitations early on and decided to address them with a technology-driven solution. We developed an individualized, patient-derived tumor organoid platform that models each patient’s tumor biology in the lab while preserving clinically relevant heterogeneity. This makes the model far more patient-relevant and, importantly, supports prediction of therapy response in a clinically meaningful way. <br><br>What is transformative is that this kind of platform can connect the entire R&D pipeline in one coherent system - from target discovery and validation, to drug candidate evaluation, and ultimately to patient stratification and smarter clinical trial design. We developed it with patient benefit in mind, and translating it from academia into an industrial setting is the most direct way to bring it into real-world use and maximize its impact on patient care. As I often say: papers don’t save lives - products do.
Broader Paradigm Shift in Oncology
What future potential do AI-driven analyses, functional genomics, and novel model systems hold for deciphering glioblastomas - or brain tumors more broadly?
Liu: We are witnessing a powerful convergence of three developments: rapid advances in genomics (especially single-cell and spatial profiling), major progress in AI-driven revolution, and the increasing ability to culture and perturb human tumors in patient-derived systems. Platforms like individualized patient-derived tumor organoids can act as true data generators: we can apply defined drug perturbations in a human-relevant context and read out responses at single-cell resolution, producing extremely rich datasets - thousands of genes across many millions of cells, across many patients and conditions.
These datasets enable predictive AI models that learn the rules of response and resistance, accelerating target discovery, mechanism-of-action studies, drug development, and – crucially - patient stratification. In the end, this will not only impact glioblastoma or brain tumors. It has the potential to drive a broader paradigm shift in oncology: moving from trial-and-error approaches toward data-driven, patient-specific prediction and truly rational, stratified cancer treatment.
These datasets enable predictive AI models that learn the rules of response and resistance, accelerating target discovery, mechanism-of-action studies, drug development, and – crucially - patient stratification. In the end, this will not only impact glioblastoma or brain tumors. It has the potential to drive a broader paradigm shift in oncology: moving from trial-and-error approaches toward data-driven, patient-specific prediction and truly rational, stratified cancer treatment.
Where do you see the greatest challenges, but also the greatest opportunities, in translation—particularly in collaboration between research institutions, start-ups, industry, and clinical practice?
Liu: The biggest challenge in translation is that excellent science alone is not enough. To create real-world patient impact, you need an exceptional team, strong partners, and the persistence to navigate skepticism and setbacks - because bringing a technology from the lab into practice is always difficult, even when the concept is strong. You must convince collaborators, industry partners, often your own organizations and overcome all structure limitations.
However, we do not focus on complaining - we focus on how to make it work! By building the best possible conditions: assembling a strong team, building extensive networks, and making intelligent use of the Heidelberg environment, including the DKFZ, local innovation structures like the Innovation Lab, and the world-class neuro-oncology landscape on the Heidelberg campus.
We do that because the opportunity is enormous. Europe’s innovation ecosystem is steadily improving, and places like Heidelberg offer unique advantages: world-class cancer research, strong clinical networks, and a rapidly growing innovation environment. Historically, this region produced outstanding companies and impactful technologies - driven by talented, determined people. I believe we are again at an inflection point, we have the chance to shape what cancer research and treatment can look like over the next several decades.
However, we do not focus on complaining - we focus on how to make it work! By building the best possible conditions: assembling a strong team, building extensive networks, and making intelligent use of the Heidelberg environment, including the DKFZ, local innovation structures like the Innovation Lab, and the world-class neuro-oncology landscape on the Heidelberg campus.
We do that because the opportunity is enormous. Europe’s innovation ecosystem is steadily improving, and places like Heidelberg offer unique advantages: world-class cancer research, strong clinical networks, and a rapidly growing innovation environment. Historically, this region produced outstanding companies and impactful technologies - driven by talented, determined people. I believe we are again at an inflection point, we have the chance to shape what cancer research and treatment can look like over the next several decades.
The AIPTO approach, in which drug response data is transferred to DELPHAI (predictive AI agent for the treatment of glioblastomas). Graphic: Haikun Liu/DKFZ
As chairman of the first workshop session and a speaker at “GENESIS 26 – the LIFE SCIENCE TECH DAY”: What impulses do you hope to give participants in this dual role, and what kinds of collaborations do you hope the event will initiate?
Liu: GENESIS 26 is exceptionally timely because it brings together the exact ingredients needed for real impact: cutting-edge research in bioengineering and AI, strong technology and industry partners, and a shared commitment to translation. In my dual role, I want to encourage participants to think beyond disciplinary boundaries and to actively pursue collaborations that connect discovery science with practical implementation.
From my experience last year, meetings like this can translate very directly into new partnerships - I personally initiated several valuable collaborations after attending. This year, I hope the event will catalyze joint projects between academia and industry, between cancer researchers and engineers, and with pharma and biotech partners. Ultimately, the goal is to help people find the collaborators who make ambitious ideas feasible - and to turn promising science into measurable real-world outcomes for patients.
From my experience last year, meetings like this can translate very directly into new partnerships - I personally initiated several valuable collaborations after attending. This year, I hope the event will catalyze joint projects between academia and industry, between cancer researchers and engineers, and with pharma and biotech partners. Ultimately, the goal is to help people find the collaborators who make ambitious ideas feasible - and to turn promising science into measurable real-world outcomes for patients.
About the person
Prof. Dr. Haikun Liu heads department of Molecular Neurogenetics at DKFZ in Heidelberg. His lab looks for cures for glioblastoma, and he is the scientific founder of AIPTO.
About the DKFZ and AIPTO TechBio GmbH
The German Cancer Research Center (DKFZ) is Germany's largest biomedical research institution and part of the Helmholtz Association and the German Centers for Health Research.
AIPTO TechBio GmbH, a spin-off of the DKFZ, addresses the major challenge of brain cancer therapy with a disruptive technology platform that integrates next-generation patient avatars and generative, agentic AI. AIPTO's mission is to revolutionize brain cancer therapy and provide personalized, effective treatments.
The interview was conducted by Joachim Klaehn.
This website is not part of the Facebook website or Facebook Inc. Furthermore, this website is not supported by Facebook in any way. Facebook is a trademark of Facebook, Inc. We use Google remarketing pixels/cookies on this website to communicate with visitors to our website again and ensure that we can reach them with relevant news and information in the future. Google places our ads on third-party websites on the internet to communicate our message and reach the right people who have shown interest in our information in the past.