Cold-inducible RNA-binding protein (CIRBP) as a biomarker to predict recurrence of brain metastases

See the article by Dankner et al, pp. 1470–1480.

Key Points.

1. Invasion into the brain is associated with recurrence in surgically resected brain metastases (BrM). HI histopathology is associated with poor patient outcomes.

2. scRNA-seq reveals distinct transcriptional profiles in metastatic cells that have invaded into the brain.

3. CIRBP is overexpressed in HI BrM and is associated with efficient invasive growth within the brain.

4. CIRBP may license BrM invasion and recurrence and may foster invasion within the brain, specifically.

Brain metastases (BrM) continue to pose a growing problem in oncologic care. As patients survive longer with their primary cancers, BrM have evolved into one of the leading causes of patient morbidity and mortality. Although generally viewed as well-demarcated masses with a clear separation from the adjacent brain parenchyma,¹ there remain appreciable local recurrence rates in patients with gross total resections and/or stereotactic radiosurgery (SRS), the current mainstays of treatment. Histopathological analysis reveals microinvasion of BrM cancer cells beyond resection cavities, which contributes to recurrence.² Generally, then resected BrM require adjuvant postoperative SRS beyond the cavity margins to target invasive cancer cells and maximize disease control.³ The molecular landscape of BrM cell invasion is a growing topic of research interest. In this issue of Neuro-Oncology, Dankner et al investigate the expression of cold-inducible RNA-binding protein (CIRBP) in human BrM specimens and its relationship to the local invasion patterns and postsurgical recurrence.⁴ Additionally, they develop patient-derived xenograft (PDX) murine models to demonstrate the importance of CIRBP in BrM invasive growth.

The authors categorized the histopathologic invasion pattern of resected BrM from lung, breast, skin (melanoma), and gastrointestinal cancers into highly invasive (HI) and minimally invasive (MI). Of note, BrM of breast cancer origin were more likely to have an HI pattern relative to those originating from non-breast primary tumors. The HI pattern was associated with poor patient outcomes, including local recurrence, leptomeningeal metastasis, and decreased overall survival (OS) when compared to the MI pattern. The authors utilized single-cell RNA sequencing (scRNA-seq) to obtain transcriptional profiles of the BrM and paired surrounding brain. The results confirmed the presence of metastatic cancer cell microinvasion outside of the resection cavity.

To identify the genes functionally related to BrM invasive growth, gene set enrichment analysis (GSEA) was conducted and revealed alterations in 24 individual genes of interest. One of the altered genes was that encoding CIRBP, an mRNA-binding protein that plays a critical role in controlling the types of cellular adaptation to stressors that might be necessary for invading into adjacent brain.⁵ To further study the role of CIRBP in HI BrM, 30 PDX models were established. The invasion patterns when these PDXs were implanted intracranially into immunocompromised mice matched the invasion patterns of the patients from which they were derived. Interestingly, PDXs with an HI pattern were more likely to form leptomeningeal dissemination along the spinal cord. Tissue microarrays from both human and murine specimens were created, and immunohistochemistry (IHC) staining was performed to measure CIRBP expression. As predicted, specimens with the HI pattern demonstrated higher CIRBP expression when compared to tissues with an MI pattern. Interestingly, CIRBP expression was higher in tumor cells at the leading edge confronting the normal brain parenchyma relative to tumor cells in the center of the lesion.

To assess a causative role for CIRBP in licensing tumor invasion, short hairpin RNAs (shRNA) were used to knockdown CIRBP in an HI PDX model. Such knockdown produced suppression of tumor growth, reduced invasive capacity, and prolonged OS in the relevant mice. Furthermore, the CIRBP knockdown model showed a decrease in the proliferation marker Ki67, as well as in cancer-promoting extracellular signal-regulated kinase (ERK) signaling. The benefit of shRNA CIRBP knockdown was recapitulated in a triple-negative breast cancer cell line, which showed reduced invasive capacity. Tumor invasion was in turn rescued with overexpression of wild-type CIRBP. Of note, the effects of CIRBP manipulation were only seen when tumor cells were implanted into the mouse brain, and not when they were introduced into the mammary fat pad. These data imply that the role of CIRBP in tumor invasion may be specific to the brain.

These findings suggest tumor invasion patterns and their transcriptional profiles as important prognostic factors for patients with surgically resected BrM. They also highlight a potential therapeutic focus on countering the pathways facilitating invasion at the metastatic leading edge. The identification of a possibly “brain-specific” gene set requirement for local invasion further highlights what we increasingly recognize as important distinctions and mechanistic considerations when targeting tumors within the intracranial environs. Other recent high-profile work has also uncovered unique proclivities of the brain for fostering tumor invasion and immune escape.^6–8 These translational questions that result from such studies may help pioneer the utilization of invasion patterns and molecular data in the management of brain metastatic patients, specifically. Regarding this study, in particular, probing the downstream signaling pathways from CIRBP such as Cystatin C (CST3), Src-associated during mitosis of 68 kDa (Sam68), or ERK signaling pathways, may help add to the landscape of molecular factors controlling brain metastatic invasion.^9,10

In conclusion, high expression of CIRBP at the metastatic leading edge, beyond those areas subject to resection and local therapy, may indeed play a significant role in licensing BrM invasion and postoperative recurrence. Further studied into the causative role in brain invasion that CIRBP plays and its utility as a predictive biomarker for BrM recurrence are in order. Likewise, incorporating Dankner et al’s findings into the growing understanding of brain-unique treatment considerations will aid in providing novel and perhaps primary disease-agnostic therapeutic strategies that can move the needle for patients suffering from BrM.