Researchers at UCLA have designed a novel artificial intelligence system that can discover promising drug combinations to combat treatment resistance in cancer, according to a university announcement today.
uncertain
that → design → announcement
The computational platform, named MAESTRO and detailed in the journal Advanced Science, uses large language models, or LLMs, to process established scientific knowledge regarding drug mechanisms and the biological pathways malignant cells exploit to survive treatment.
asserted
cells → name → treatment
The system evaluates potential therapeutic pairings, ranks the most promising candidates, and generates biological explanations that researchers can subsequently test in the laboratory.
”Developing drug combinations currently relies on either brute-force screening or researchers hypothesizing potential candidates based on existing literature,” said Neil Lin, an associate professor of mechanical and aerospace engineering and bioengineering at the UCLA Samueli School of Engineering and the study's senior author.
asserted
Lin → evaluate → Engineering
”It is time-consuming, and it can still be difficult to capture the complex interactions among multiple biological pathways.
asserted
it → consume → pathways
MAESTRO could substantially accelerate this process.
uncertain
MAESTRO → accelerate → process
By integrating existing biomedical knowledge, it can identify drug combinations that have a clear biological rationale and are most appropriate for experimental testing.
asserted
that → integrate → testing
To uncover therapeutic options that can overcome treatment resistance in prostate cancer, the researchers deployed MAESTRO—short for Mechanistic Augmentation and Ensemble Selection for Therapeutic Ranking Optimization—to evaluate thousands of naturally occurring compounds.
asserted
researchers → uncover → compounds
The platform highlighted several compounds that could potentially enhance the effectiveness of enzalutamide, an advanced prostate cancer medication originally discovered and developed at UCLA.
uncertain
that → highlight → UCLA
The research team focused its investigation on prostate cancer, a disease where resistance to therapy remains a formidable clinical challenge.
asserted
resistance → focus → therapy
Prostate cancer cells frequently depend on androgen receptor signaling to grow and proliferate.
asserted
cells → depend → signaling
Although targeted drugs like enzalutamide block this critical signaling, malignant cells can eventually adapt by activating alternative pathways that allow them to survive, ultimately leading to treatment resistance.
asserted
them → target → resistance
Combining therapeutic agents that target these alternative escape routes offers a potential strategy to overcome resistance.
asserted
that → combine → resistance
However, with thousands of potential compounds available and numerous interconnected biological pathways involved, identifying truly effective drug pairings remains an immense challenge.
asserted
identifying → involve → pairings
While artificial intelligence has shown promise in predicting successful drug combinations, many existing computational approaches require vast amounts of experimental training data and offer little insight into why specific drugs might work together.
uncertain
drugs → show → insight
The UCLA team engineered MAESTRO specifically to overcome these limitations.
asserted
team → engineer → limitations
Researchers supplied the LLM with scientific data directly relevant to the project, including prostate cancer biology, mechanisms of drug resistance, and known drug actions.
asserted
Researchers → supply → resistance
The computational model then analyzed how those cellular mechanisms might interact alongside androgen receptor blockade and ranked prospective compounds as potential therapeutic partners for enzalutamide.
uncertain
mechanisms → analyze → enzalutamide
Rather than depending on a single prediction, the researchers instructed the model to evaluate each candidate compound more than 100 times.
asserted
researchers → depend → compound
They then analyzed how highly each compound ranked and assessed the consistency of those rankings across repeated computational runs.
asserted
compound → analyze → runs
They categorized compounds that consistently earned high rankings as the most promising candidates for physical testing.
asserted
that → categorize → testing
The team utilized the platform to screen 3,592 natural products for their potential to work synergistically with enzalutamide.
asserted
team → utilize → enzalutamide
The system successfully identified 150 high-confidence candidates for laboratory testing in 3D prostate cancer spheroids.
asserted
system → identify → spheroids
Of the 150 compounds evaluated in vitro, 23 showed clear coupling with enzalutamide, meaning the combination produced a significantly greater anti-cancer effect than the individual treatments alone.
asserted
combination → evaluate → treatments
Subsequent testing identified berberine, a naturally occurring plant compound, as an exceptionally promising candidate.
asserted
testing → identify → candidate
The combination of berberine and enzalutamide demonstrated consistent coupling across four distinct prostate cancer models featuring different underlying mechanisms of treatment resistance.
asserted
combination → demonstrate → resistance
“Berberine was not an obvious candidate because it affects several biological pathways, making it difficult to predict how it might work with other drugs,” said James C.H. Lo, a postdoctoral student in the bioengineering department at UCLA when the work was conducted and the study's first author.
uncertain
work → affect → UCLA
“What surprised us was that the combination showed consistent synergy across all of the prostate cancer models, despite their different resistance mechanisms.
asserted
combination → surprise → mechanisms
This is where MAESTRO demonstrated its strength, helping us identify a promising combination that might have been overlooked by examining each pathway separately.”
uncertain
that → demonstrate → pathway
The researchers also tested whether the drug pairing's biological effects matched the precise cellular mechanism MAESTRO proposed.
asserted
MAESTRO → test → mechanism
They found that adding berberine to enzalutamide increased AMPK activity, a pathway vital to regulating cellular energy and stress, while suppressing mTORC1 signaling, which plays a critical role in cell growth and survival.
asserted
which → find → growth
The drug combination was also linked to increased cellular stress, along with reduced DNA replication and cell-cycle progression.
asserted
combination → link → replication
These findings indicate that berberine and enzalutamide represent a promising combination for further investigation, particularly in prostate tumors that have acquired resistance to standard care.
asserted
that → indicate → care
The authors also noted that the specific AMPK-mTORC1 mechanism highlighted in the study could serve as a valuable framework to guide the development of future drug combinations intended to bypass resistance mechanisms.
uncertain
mechanism → note → mechanisms
Moving forward, the research team plans to further assess the drug combination in patient-derived tumor models and subsequent animal studies.
asserted
team → move → models
They also aim to expand MAESTRO's capabilities by integrating additional biological data streams and applying the platform to explore new drug combinations, novel resistance mechanisms, and other cancer types.
asserted
They → aim → combinations
The Aporia
