A New Score Could Help Doctors Predict Which Liver Cancer Patients Will Benefit from a Three-Pronged Treatment
A new clinical scoring system may give doctors a clearer way to estimate how patients with unresectable hepatocellular carcinoma will respond to an increasingly powerful combination of treatments. The model, called the ATC score, was developed for people receiving transarterial chemoembolization, lenvatinib and a programmed cell death protein 1, or PD-1, inhibitor. In a retrospective study of 154 patients, the score separated individuals into groups with sharply different chances of long-term survival and disease control. The findings suggest that a small set of routinely measured clinical features could help turn a broadly applied treatment strategy into a more personalized one.
Hepatocellular carcinoma, or HCC, is the most common primary cancer arising in the liver. When tumors cannot be surgically removed, treatment becomes substantially more difficult because the disease may involve major blood vessels, occupy large portions of the liver or have spread beyond the organ. One approach is transarterial chemoembolization, known as TACE, in which physicians guide a catheter through the blood vessels to arteries feeding a tumor. Chemotherapy is delivered locally, while embolic material blocks the tumor’s blood supply. The procedure can expose cancer cells to high drug concentrations and deprive them of oxygen, but it does not always eliminate all malignant tissue.
The other components of the combination attack the disease through different biological pathways. Lenvatinib is a tyrosine kinase inhibitor that blocks signaling proteins involved in blood-vessel formation and tumor growth. HCC tumors often stimulate the development of new vessels to secure oxygen and nutrients; inhibiting pathways involving vascular endothelial growth factor receptors can restrict that support. PD-1 inhibitors work through the immune system. PD-1 is a checkpoint protein on immune cells that can be exploited by tumors to suppress T-cell activity. Blocking the PD-1 pathway may restore part of the immune response against cancer. In principle, TACE can release tumor antigens and alter the tumor environment, lenvatinib can restrain angiogenesis, and immunotherapy can mobilize immune cells against remaining cancer cells.
Although this three-part treatment, abbreviated T-L-P by the researchers, has shown encouraging activity, its effects vary considerably from one patient to another. Some patients experience substantial tumor shrinkage and prolonged disease control, while others progress despite treatment. Clinicians therefore need practical ways to identify likely responders and recognize patients whose disease or underlying liver dysfunction may limit the benefit. The research team, led by Pengpeng Zhu and Kaile Tian, examined medical records from patients who began T-L-P therapy between 2020 and 2022. The investigators randomly or administratively divided the study population into a training cohort of 90 patients, used to develop the model, and a validation cohort of 64 patients, used to test whether it retained its predictive value.
The researchers first assessed objective response rate, or ORR, which measures the proportion of patients whose tumors met predefined criteria for complete or partial shrinkage. They also examined disease control rate, which includes responses as well as stable disease. In the training cohort, 44 of 90 patients, or 48.9 percent, achieved an objective response, while 72 patients, or 80 percent, achieved disease control. In the validation cohort, the corresponding figures were 38 of 64 patients, or 59.4 percent, for objective response and 52 patients, or 81.3 percent, for disease control. These results describe outcomes in the selected study population and should not be interpreted as proof that the treatment will produce the same response rates in every clinical setting.
Statistical analysis identified three independent factors associated with objective response. The first was an alpha-fetoprotein concentration above 100 nanograms per milliliter. AFP is a protein produced during fetal development that can reappear at high levels in some people with HCC, although it is neither present in every case nor specific to this cancer. The second factor was a tumor burden score above 8. TBS combines tumor size and number into a single measure, generally increasing as tumors become larger or more numerous. The third was Child-Pugh class B liver function, a category indicating more substantial impairment than class A. The Child-Pugh system incorporates measures such as bilirubin, albumin, blood clotting, fluid accumulation and encephalopathy to estimate the liver’s functional reserve.
Each of these features was assigned one point in the ATC score, producing a range from zero to three. The name reflects the principal variables used in the model: AFP, tumor burden and Child-Pugh classification. Patients with no risk factors had a score of zero and were designated low risk; those with one factor had an intermediate-risk score of one; and those with two or three factors were placed in the high-risk group. The score is not a molecular test and does not directly measure immune activity, drug concentration or genetic mutations. Instead, it combines readily available indicators of tumor biology and the liver’s ability to withstand both cancer and treatment.
The differences between the groups were striking. In the training cohort, the estimated two-year overall survival rate was 83.57 percent for patients with an ATC score of zero, 64.13 percent for those scoring one and 20.69 percent for those scoring at least two. Two-year progression-free survival, which measures the length of time before the cancer worsens or the patient dies, was 66.48 percent, 46.55 percent and 9.33 percent in the same groups. The validation cohort showed a similar pattern: two-year overall survival was 94.12 percent for score zero, 64.71 percent for score one and 25.93 percent for scores of two or three. Two-year progression-free survival was 82.35 percent, 47.06 percent and 16.67 percent, respectively.
To evaluate discrimination, the investigators used the area under the receiver operating characteristic curve, or AUC. This statistic summarizes how well a model distinguishes between patients who do and do not experience an outcome, with a value of 0.5 representing chance performance and values closer to 1 indicating stronger discrimination. The ATC score achieved an AUC of 0.826 in the training cohort and 0.811 in the validation cohort. According to the study, this performance was better than that of any individual component alone. The validation result is particularly important because a model can appear highly accurate in the same data used to build it but lose performance in a separate group.
The researchers say the score could support treatment planning by providing an early estimate of expected benefit from T-L-P therapy. A patient with a low score might be considered a strong candidate for continuing an intensive combination approach, assuming treatment is otherwise safe and appropriate. A high score could prompt closer monitoring, more detailed discussion of alternatives or consideration of clinical trials designed for patients with difficult-to-treat disease. The model might also help researchers balance participants across risk categories in future studies, making it easier to determine whether a new therapy helps patients with poor baseline prognoses rather than simply reflecting differences between trial populations.
The results nevertheless come with important limitations. The study was retrospective, meaning the investigators analyzed existing records rather than assigning treatment prospectively under a controlled protocol. All 154 patients received the same general treatment combination, but unmeasured differences in disease characteristics, supportive care or clinical decision-making could have influenced the outcomes. The sample was also relatively small and drawn from institutions in China, so the model requires testing in larger, geographically diverse populations before it can be considered broadly reliable. In addition, a score that predicts response at the group level cannot determine an individual patient’s fate. Some people with high scores may respond exceptionally well, while some with low scores may not benefit.
The findings add to a growing effort to make combination therapy for advanced liver cancer more precise. TACE, targeted therapy and immune checkpoint blockade each affect a different part of the tumor ecosystem, but the same complexity that creates therapeutic potential also makes outcomes difficult to anticipate. The ATC score offers a deliberately simple framework built from clinical information already collected in routine care. Its next test will be prospective validation: researchers will need to apply the score before treatment begins, follow patients under standardized conditions and determine whether it improves decisions rather than merely describing prognosis. Until then, the model is best viewed as a promising research tool—one that could help clinicians move closer to matching the right treatment intensity to the biology of an individual patient.
Subject of Research: Risk stratification and personalized treatment management for patients with unresectable hepatocellular carcinoma receiving TACE, lenvatinib and PD-1 inhibitor therapy
Subject of Research: Cancer
Article Title: The ATC score: a novel efficacy scoring model for risk stratification and personalized management of unresectable HCC patients receiving TACE-lenvatinib-PD-1 inhibitor therapy
Article References: Zhu, P., Tian, K., Liu, X. et al., “The ATC score: a novel efficacy scoring model for risk stratification and personalized management of unresectable HCC patients receiving TACE-lenvatinib-PD-1 inhibitor therapy,” Cancer Cell International
Image Credits: AI Generated
DOI: 10.1186/s12935-026-04451-8
Keywords: unresectable hepatocellular carcinoma, TACE, lenvatinib, PD-1 inhibitor, ATC score, tumor burden score, alpha-fetoprotein, personalized cancer treatment
Tags: ATC score for liver cancerATC scoring system for HCCclinical features for liver cancer prognosisclinical features in liver cancer managementcombination therapy for liver cancercombination treatment for liver cancerHepatocellular carcinoma treatment strategieslenvatinib and PD-1 inhibitor therapyliver cancer patient stratificationliver cancer prognosisliver cancer survival predictionlong-term survival prediction in liver cancerPD-1 inhibitors in liver cancerpersonalized liver cancer managementpersonalized liver cancer therapypredicting liver cancer treatment responsepredictive scoring system in oncologyrisk stratification in liver cancertailored treatment strategies for hepatocellular carcinomatransarterial chemoembolizationtransarterial chemoembolization outcomesunresectable hepatocellular carcinoma treatment
