NEW YORK – Using artificial intelligence-based drug development strategies with an emphasis on biomarker stratification, Lantern Pharma has rapidly advanced precision oncology candidates for treating brain, lung, and blood cancers and will be reporting initial clinical trial data on these programs this year.
Dallas-based Lantern has three targeted therapy programs in Phase I and Phase II studies. The firm uses its machine learning-based Response Algorithm for Drug Positioning and Rescue (RADR) platform to identify best responders to the therapeutic candidates it is developing internally and to drugs it is advancing in collaboration with other firms.
RADR uses machine learning, data analytics, and system biology approaches to analyze transcriptomic, genomic, and drug sensitivity data from public and private oncology databases and identify biomarkers associated with drug activity.
Using RADR, Lantern earlier this month said it had identified a key biomarker, PTGR1, that could identify patients likely to benefit from treatment with one of its drug candidates, LP-184. Lantern is now developing a real-time PCR-based assay to quantify the amount of PTGR1 RNA in patients' tumor samples and assess sensitivity to LP-184. The firm is studying LP-184 in a Phase I trial in solid tumors harboring DNA damage response deficiencies and plans to implement the PTGR1 diagnostic into upcoming trials of the drug.
The firm is also developing LP-300 in never-smoker non-small cell lung cancer patients harboring actionable genomic alterations such as MET exon 14, ALK, EGFR, or NTRK mutations, and it is testing LP-284 in non-Hodgkin's lymphomas. LP-284 has received orphan drug designation from the US Food and Drug Administration as a potential treatment for high-grade B-cell lymphoma with MYC and BCL2 rearrangements.
Lantern CEO Panna Sharma said the company aims to integrate biomarkers for patient selection as early as possible in its drug development programs, which can reduce the time and cost of bringing therapies to market. Many companies often initially bring drug candidates into clinical trials in all-comer settings without biomarkers while they're exploring its preliminary safety and efficacy and only use them prospectively to stratify patients in later studies, Sharma said.
"If we can use AI to identify and validate biomarkers while we're also developing the molecule, that saves a lot of time and money," Sharma said. "That's the challenge with precision oncology drug development, doing it in a way where you can really parallel track [biomarker validation] during your development of the molecule."
AI guiding LP-184 development
In line with the company's strategy, Lantern identified PTGR1 expression as a potential biomarker for patient selection early in LP-184's development process. According to Sharma, when researchers at Lantern used RADR to explore which cancer types had PTGR1 expression, they landed on an unexpected answer: central nervous system tumors.
"We had a biological bias that PTGR1 is not expressed in glioblastoma tissue," Sharm explained. "We also had a bias that it was going to be overexpressed in ovarian, liver, and prostate cancers where there was historical literature supporting it."
The team initially tested LP-184's activity in bladder, ovarian, pancreatic, and prostate tumors. But as they continued in silico screenings and RADR analysis, central nervous system (CNS) tumors emerged again as a potential target for LP-184. Sharma said the second time around, the research team validated that LP-184 could cross the blood-brain barrier. They then focused on analyzing more than a billion data points of the CNS cancer data.
"We really validated this drug would work in a wide range of brain cancers, not just glioblastoma," he said. "As we were learning more about the molecule, we also knew it seemed to be very active when there is DNA damage repair deficiency in tumors, which is something many pediatric brain tumors have. They have challenges in epigenetic regulation, which is linked to DNA repair, so these pediatric tumors were among the most sensitive to this drug."
However, the potential activity in adult and pediatric brain tumors presented a challenge for the small biotech. In 2021, Lantern's development programs for LP-184 in pancreatic cancer and glioblastoma multiforme and other malignant gliomas received orphan drug designations from the FDA, but recognizing that brain tumor and pediatric drug trials are often more expensive to conduct than trials in other cancer types, Lantern, in 2023, moved LP-184's brain tumor development programs to a separate, wholly owned subsidiary, called Starlight Therapeutics.
Lantern kept using the LP-184 name and developing the drug candidate in solid tumors, which since has led to a Phase I trial, while Starlight has advanced the molecule as STAR-001 in brain and CNS tumor indications.
Starlight is aiming to explore STAR-001 as a treatment for adult patients with glioblastoma and breast and lung cancer patients with brain metastases and for pediatric patients with atypical teratoid rhabdoid tumors, diffuse midline glioma, and high-grade hemispheric glioma. Researchers will still stratify patients based on PTGR1, which Sharma estimated is overexpressed in around 70 percent of glioblastoma patients.
Now that STAR-001 is at a company prioritizing its development in CNS tumor indications, it may attract the interest of other pharma companies. Sharma, who also leads Starlight alongside Chief Medical Officer Marc Chamberlain, said his team is open to partnering with drugmakers on such programs in the future.
"If [a big pharma] sees a company [like Starlight] and its sole focus is CNS oncology, it makes it very easy for them to say, 'Yes, we need that program,'" Sharma said.
Meanwhile, Lantern is still figuring out how to move forward LP-184's solid tumor indications. Now that the companion diagnostic to identify PTGR1 overexpression is in development, the firm is deciding which tumor types to focus on. Sharma said PTGR1 overexpression occurs in about 30 percent of recurrent bladder and pancreatic tumors and in as much as 50 percent of metastatic prostate and ovarian cancers.
But as prostate cancer has become a crowded market for precision medicines, especially after the US approval of Novartis' radiopharmaceutical Pluvicto (lutetium vipivotide tetraxetan), ovarian, pancreatic, or bladder cancer may offer more market opportunities for LP-184. "We're trying to go after tumors where there is no good therapy or options available," Sharma said.
AI for rescuing drugs
Lantern has used insights from RADR to also inform the development of its other drug candidates, LP-300 and LP-284. Lantern bought the rights to LP-300 from BioNumerik Pharmaceuticals in 2018, with the goal of rescuing it. BioNumerik had studied LP-300, previously dubbed dimensa, in more than 2,000 patients across many clinical trials, but when it failed to meaningfully improve survival in NSCLC patients in a Phase III study, the firm shelved it.
In a separate retrospective analysis of the failed Phase III trial of LP-300 in NSCLC, researchers sought to determine if the drug had greater efficacy in a particular subpopulation. While there was no genomic information available for analysis at this point, Lantern researchers found that never-smoker NSCLC patients treated with LP-300 and chemo had median overall survival of 25.2 months compared to 13.2 months on placebo plus chemo.
Lantern applied RADR to the available data on LP-300 and identified a biomarker profile associated with never-smoker NSCLC patients, comprising mutations in MET, ALK, EGFR, and NTRK. The firm began the Phase I/II HARMONIC trial to test LP-300 plus chemotherapy in this population in 2022, and earlier this year, began a Phase II expansion portion in that trial.
LP-284, which also has FDA orphan drug designation as a potential treatment for high-grade B-cell lymphoma with MYC and BCL2 rearrangements, is the latest drug in Lantern's pipeline to enter a Phase I clinical trial, which it did in March. This "drug is borne entirely from AI," Sharma said, noting that with the help of RADR, this program "went from an idea on a whiteboard to a clinical trial in less than two-and-a-half years."
"That's the power of using AI to discover and develop molecules, [but] it's not enough just to say you can design a molecule with AI. You still have to get into a human," Sharma said. "It was always part of our road map that we want to try to give each molecule as much of a chance at winning as possible."
For the rest of this year, Lantern expects data readouts from its ongoing trials of LP-184 and LP-300. The data forthcoming for LP-184 will provide insights into its safety and initial efficacy in solid tumors, which may inform Lantern's decisions on what indications to pursue. The data on LP-300 will provide an initial view of its efficacy from the first cohort of never-smoker NSCLC patients in the HARMONIC trial.