Seoul Asan Hospital team links RB1 deletion to poor liver cancer outcomes
In short
A joint research team led by Seoul Asan Medical Center identified complete loss of the RB1 gene as a biomarker of poor prognosis in liver cancer, found in about 14.6% of patients, who faced 3.32 times the death risk and 3.15 times the recurrence risk of other patients. The team also developed a deep-learning pathology model called FR-MIL to screen for these patients and proposed a combination of mitosis-inhibiting and PARP-inhibiting drugs that showed stronger tumour suppression in cell and animal experiments. The results were published in Signal Transduction and Targeted Therapy, and the hospital said clinical adoption remains a future step.
Read the full story 2 min read
A joint research team led by Seoul Asan Medical Center identified complete loss of the tumour suppressor gene RB1 as a new biomarker of poor prognosis in liver cancer, and developed an AI-based pathology diagnostic model plus a combination treatment strategy, the hospital said. The team comprised Shim Ju-hyun of Seoul Asan Medical Center's gastroenterology division, Sung Chang-ok of its pathology department, Sim Jung-seop of the University of Macau, Park Sang-hyun of POSTECH and Ahn Ji-hyun of Hanyang University Guri Hospital. [ 1 , 2 , 3 , 4 ]
The team analysed 561 liver cancer patients — 206 treated at Seoul Asan Medical Center and 355 from the US National Cancer Institute's TCGA public database — using whole-exome and RNA sequencing and other multi-omics profiling, then validated the findings in a separate group of 450 patients using genomic, single-cell and spatial transcriptome data. Patients with complete loss or inactivation of both RB1 genes, termed RB1-Bi, made up about 14.6% of all liver cancer cases. Their tumours were less differentiated and progressed faster, and the group had 3.32 times the risk of death and 3.15 times the risk of recurrence of other patients, leading the team to conclude that RB1 complete deletion is an independent poor-prognosis factor. [ 1 , 3 , 4 ]
To identify these patients without costly genomic analysis, the team developed a deep-learning pathology model called FR-MIL that predicts RB1 complete deletion from ordinary stained pathology slide images. It scored an F1 of 84.39% to 91.58% in an external validation cohort. Testing 876 drugs on RB1-deleted liver cancer cells showed a selective response to drugs that inhibit cell division or repair of damaged DNA, with synthetic lethality in which RB1-deleted cancer cells died selectively under PARP inhibitors. In cell and animal experiments, combining a cell-division inhibitor with a PARP inhibitor increased tumour suppression without notable systemic side effects. [ 1 , 4 ]
Newsis reported that the atezolizumab and bevacizumab combination used for advanced liver cancer responds in only some patients and eventually produces resistance, and that only a small number of patients have genetic mutations responsive to approved targeted therapies, making biomarker discovery urgent. Shim was quoted as saying it is encouraging that a new path tailored to gene characteristics has been opened for hard-to-treat liver cancer patients with limited treatment options and easy resistance, and that if the AI diagnostic model and combination strategy are introduced into clinical practice they are expected to improve survival and speed up precision medicine. The study was published in the international journal Signal Transduction and Targeted Therapy. [ 1 , 3 , 4 ]
Why it matters
Liver cancer with complete RB1 loss is described as aggressive and hard to treat, so a marker that identifies these patients plus a low-cost AI screening route could matter for precision treatment planning if it reaches clinics. The documents present the diagnostic model and drug combination as research results, not as approved tests or therapies, so any patient impact depends on later clinical introduction.
Key facts
- A joint research team led by Seoul Asan Medical Center identified complete loss or inactivation of the tumour suppressor gene RB1 (RB1-Bi) as a new biomarker of poor prognosis in liver cancer. [ 1 , 2 , 3 , 4 ]
- RB1 complete deletion accounted for about 14.6% of liver cancer patients, who had an estimated 3.32 times higher risk of death and 3.15 times higher risk of recurrence than other patients. [ 1 , 3 , 4 ]
- The team analysed 561 liver cancer patients — 206 from Seoul Asan Medical Center and 355 from the US National Cancer Institute's TCGA database — and validated the results in a separate group of 450 patients. [ 1 , 3 , 4 ]
- The deep-learning pathology model FR-MIL predicts RB1 complete deletion from ordinary stained pathology slide images, scoring an F1 of 84.39% to 91.58% in external validation. [ 1 , 4 ]
- Testing 876 drugs on RB1-deleted liver cancer cells showed a selective response to drugs that inhibit cell division or repair damaged DNA, including PARP inhibitors, with synthetic lethality observed. [ 1 , 4 ]
- In cell and animal experiments, combining a cell-division inhibitor with a PARP inhibitor increased tumour suppression without notable systemic side effects. [ 1 , 4 ]
- The study was published in the international journal Signal Transduction and Targeted Therapy. [ 1 , 4 ]
Confirmed by several sources
- Complete loss of the RB1 gene is a biomarker associated with poor prognosis in liver cancer, per the joint research team. [ 1 , 2 , 3 , 4 ]
- RB1 complete deletion was found in about 14.6% of liver cancer patients, with death risk 3.32 times and recurrence risk 3.15 times that of other patients. [ 1 , 3 , 4 ]
- The team built an AI pathology model to select RB1 complete deletion patients and proposed a combination therapy strategy. [ 1 , 2 , 3 , 4 ]
Still unclear
- Whether the FR-MIL diagnostic model and the combination therapy have been adopted in clinical practice. The documents describe cell and animal experiments and a pathology AI model, and quote a researcher saying adoption in clinical settings is expected in the future; no document states clinical use.
- The date the findings were announced differs between outlets. Asia Economy reports the hospital announced them on October 6, while Kyunghyang Shinmun says the research team announced them on October 7; no document reconciles the two.
- The full contents of Newsis's second report. That document is available only as a feed summary that is cut off mid-sentence, so its later details about the AI model cannot be read.
What local media are saying
Timeline, local time
- Asia Economy publishes a report on the findings, citing a Seoul Asan Medical Center announcement dated October 6. [ 1 ]
- Newsis publishes an item stating the joint research team identified RB1 complete deletion as a new liver cancer biomarker. [ 2 ]
- Newsis publishes a second report describing limited response and resistance to existing advanced liver cancer therapy and the need for biomarkers. [ 3 ]
- The following day, Kyunghyang Shinmun reports the study and the combination treatment strategy, saying the team announced it on October 7. [ 4 ]