NEW YORK – Investigators in China and Singapore have identified a set of immune-related spatial transcriptomic features in the tumor microenvironment that predict cancer recurrence in individuals with the liver cancer hepatocellular carcinoma (HCC) and may help stratify patients for different treatments.
"Our team believes that spatial immune profiling represents a transformative approach for understanding tumor-immune interactions and developing precision medicine strategies in HCC management," Cheng Sun, director of the organ transplantation and immunology lab at First Affiliated Hospital of the University of Science and Technology of China, explained in an email.
As they reported in a study published in Nature on Wednesday, Sun and his colleagues began by using 10x Visium spatial transcriptomic profiling to assess almost 76,600 areas in formalin-fixed, paraffin-embedded tumor samples from 17 individuals who had liver resection surgery for HCC between 2018 and 2023.
The cases included 11 individuals with nonrecurrent HCC over five years of follow-up, they explained, along with half a dozen individuals who did experience HCC recurrence during that time.
By integrating multiplex immunohistochemistry, whole-slide imaging, and mass spectrometry-based proteomic insights across 30 recurrent and 31 nonrecurrent HCC cases, the team was able to spell out immune cell features in the tumor microenvironment that appeared to portend HCC recurrence.
In particular, the investigators flagged three natural killer cell subtype populations found at the tumor invasive front that appeared to coincide with lower risk of HCC recurrence and stretched out disease-free survival times.
Likewise, the team's analysis of the invasive front, adjacent stroma, and tumor center sites in eight HCC tumor samples led to more than 200 differentially expressed genes, including 23 genes with differential expression in single-cell RNA-seq data from natural killer cells.
Together, these findings prompted the team to develop a so-called tumor immune microenvironment spatial (TIMES) score for HCC recurrence that uses machine learning to combine clinicopathological data with spatial expression data for five proposed prediction biomarkers: SPON2, ZFP36L2, ZFP36, VIM, and HLA-DRB1.
Rather than relying on individual biomarkers or expression signatures generated from bulk tissue samples, Sun explained, the TIMES approach "systematically quantifies the spatial distribution patterns of key immune cell populations, capturing regional immune gradients that have proven highly predictive of recurrence risk."
The quantitative spatial expression score was further validated in another 231 participants from five prior multicenter cohorts, the authors reported, where TIMES had 82.2 percent accuracy and a specificity of nearly 86 percent.
"The predictive power of these biomarkers emerged through the integration of their spatial distributions, rather than individual marker expression levels alone," the authors explained.
Their subsequent gene knockout experiments in mouse models of HCC suggested that risk of the liver cancer's recurrence is bumped up in animals missing the SPON2 gene in their natural killer cells, apparently by altering interferon gamma cytokine activity, CD8-positive T-cell activation, and natural killer cell infiltration at the tumor's invasive front.
"The elucidation of SPON2's role in promoting NK cell mobilization and anti-tumor activity provides a mechanistic foundation for developing immunotherapeutic strategies aimed at enhancing SPON2-mediated NK cell responses," Sun explained, noting that SPON2-positive natural killer cells "represent a previously uncharacterized subset with enhanced cytolytic capacity and superior tumor-infiltrating properties."
More generally, he explained, the results suggest that it may be possible to use TIMES scores to stratify HCC patients with diminished or enhanced immune surveillance features in their tumor microenvironments to come up with patient-tailored treatment and monitoring plans.
"TIMES enables refined stratification of patients at high risk for recurrence, potentially guiding adjuvant treatment selection," Sun suggested. "Patients with low TIMES scores (indicating suboptimal immune surveillance) might benefit from more intensive monitoring protocols or consideration for adjuvant immunotherapeutic interventions."
Following on from their findings so far, members of the team are planning to assess the TIMES scoring approach in a prospective, multicenter study of diverse HCC patients treated in different clinical settings, with a focus on understanding the utility of the spatial expression stratification strategy for early-stage HCC cases.
On the treatment side, the team plans to dig into the possibility of boosting anti-tumor immune activity in the HCC tumor microenvironment with pharmacological treatments that bump up SPON2 expression or related signaling in natural killer cells.
The investigators are also studying potential relationships between the TIMES score and responses to existing treatments, particularly immune checkpoint immunotherapy. In parallel, they plan to search for blood-based biomarkers coinciding with the TIMES score, with an eye to developing noninvasive recurrence surveillance approaches for monitoring HCC recurrence.
"We are actively pursuing strategic collaborations with biotechnology and pharmaceutical companies to accelerate the translation of our spatial immune profiling platform into standardized clinical applications," Sun said. "This includes developing optimized workflow protocols, refining machine learning algorithms for diverse clinical settings, and integrating the TIMES system with existing diagnostic paradigms."