Invasive growth associated with cold-inducible RNA-binding protein expression drives recurrence of surgically resected brain metastases

Abstract

Background Sixty percent of surgically resected brain metastases (BrM) recur within 1 year. These recurrences have long been thought to result from the dispersion of cancer cells during surgery. We tested the alternative hypothesis that invasion of cancer cells into the adjacent brain plays a significant role in local recurrence and shortened overall survival. Methods We determined the invasion pattern of 164 surgically resected BrM and correlated with local recurrence and overall survival. We performed single-cell RNA sequencing (scRNAseq) of >15,000 cells from BrM and adjacent brain tissue. Validation of targets was performed with a novel cohort of BrM patient-derived xenografts (PDX) and patient tissues. Results We demonstrate that invasion of metastatic cancer cells into the adjacent brain is associated with local recurrence and shortened overall survival. scRNAseq of paired tumor and adjacent brain samples confirmed the existence of invasive cancer cells in the tumor-adjacent brain. Analysis of these cells identified cold-inducible RNA-binding protein (CIRBP) overexpression in invasive cancer cells compared to cancer cells located within the metastases. Applying PDX models that recapitulate the invasion pattern observed in patients, we show that CIRBP is overexpressed in highly invasive BrM and is required for efficient invasive growth in the brain. Conclusions These data demonstrate peritumoral invasion as a driver of treatment failure in BrM that is functionally mediated by CIRBP. These findings improve our understanding of the biology underlying postoperative treatment failure and lay the groundwork for rational clinical trial development based upon invasion pattern in surgically resected BrM.

The Underlying Biology and Therapeutic Vulnerabilities of Leptomeningeal Metastases in Adult Solid Cancers

“…The ability of MDA-MB-231 cells to form LM upon intracranial injection was confirmed by Allen et al, who observed leptomeningeal dissemination when MDA-MB-231, but not MDA-MB-468 cells, were implanted [44]. These findings were expanded upon using a cohort of patient-derived xenograft models of brain metastases injected intracranially, which show differential propensity to form LM based upon the invasion pattern in the brain metastatic lesion (minimally versus highly invasive) and expression of CIRBP [45]. This suggests that cellular processes and molecular mediators that are present in only a subset of cancers may be involved in leading established parenchymal brain metastases to subsequently colonize the leptomeninges.…”

“…The converse may also be true, where leptomeningeal deposits can be observed invading into the brain parenchyma ( Figure 2). Metastatic cancer cells have been shown to cycle between the brain parenchyma and leptomeningeal compartments in rodent models [44], with cold-inducible RNA-binding protein (CIRBP) and cyclooxygenase-2 representing candidate genes implicated in this process [44,45]. It is possible that shuttling between the brain and leptomeningeal anatomical compartments is uniquely driven by invasion across the glia limitans, however, other possibilities remain.…”

Brain Tumor Biobank Development for Precision Medicine: Role of the Neurosurgeon

“…Mechanical dissociation in the operating room decreases processing time in the laboratory and aids in streamlining experimentation (Figure 2E) Dependent of the location of the brain metastasis, often a rim of the surrounding tissue within the parenchyma can be resected as well. Recent data also demonstrates heterogeneity between regions of metastatic tumors and invasion into the peritumoral border (33,34). Therefore, even with metastatic tumor regional information is routinely annotated with the tumor tissue.…”

Animal models of brain metastasis

“…Compatibility with in vitro passage is not granted and efficacy has been estimated between 24% and 48.8% depending on the amount of viable cancer cells in the surgical sample, ability of the cells to form colonies in the first culture, and the degree of senescence affecting cancer cells. 24 Most brain metastatic PDX models are derived from either systemic 25 or intracranial 26 inoculation, but few models have been reported to spontaneously metastasize from orthotopic (subdermal for melanoma 21 or mammary fat-pad for breast cancer 27 ) injection. Of note, a brain-metastatic SCLC PDX model has recently been established from subcutaneous injection.…”