Post-therapeutic microRNA-146a in liquid biopsies may determine prognosis in metastatic gastrointestinal cancer patients receiving ⁹⁰Y-radioembolization

In patients with liver metastases (metastatic liver cancer, mLC), loco-regional treatment by transarterial yttrium-90 (⁹⁰Y)-radioembolization (RE) is an established treatment (Vogel et al. 2021; Reig et al. 2021; Helmberger et al. 2021; Karanicolas et al. 2021).⁹⁰Y-RE has proven to be effective both in first (Salem et al. 2016) and second-line settings (Mulcahy et al. 2021), being able to boost immune response and indirectly to modulate the tumor immune microenvironment (TIME) composition in distant, non-treated yet, metastases. However, clinical outcomes vary, with a significant proportion of patients not achieving durable responses (Lau et al. 2013; Ricke et al. 2019). In addition, the low tolerance of the liver to irradiation, characterized by a rapid deterioration of hepatic function, can be a limiting factor (Sangro et al. 2008; Seidensticker et al. 2021; Ricke et al. 2021). Therefore biomarkers for treatment efficacy prediction are a highly needed tool. In the context of liquid biopsy, microRNAs (miRNAs) occupy a prominent role as non-invasive biomarkers in several types of diseases, including cancer. miRNAs are highly conserved, ~ 21–25 nucleotides long, non-coding RNA released from cells by active secretion (Kosaka et al. 2010) or after cell death (Turchinovich and Cho 2014). They are post-transcriptional repressors, inhibiting translation through transcript degradation (Di Leva et al. 2014; Hanahan and Weinberg 2000). Several studies, mainly in vitro and in animal models, have evaluated the modification of miRNA levels in response to radiation, confirming miRNAs as useful biomarkers for radio sensitivity (Drula et al. 2020). miRNAs play an important function in forging the tumor immune microenvironment (TIME), modulating the infiltration of immune cells and their function (Gagnon and Ansel 2019). Among others, miR-146a is a well-characterized miRNA taking part in multiple pro-inflammatory pathways (Boldin et al. 2011; Stickel et al. 2014, 2017). Functionally, miR-146a regulates the immune response and controls excessive inflammation, mainly through the repression of the NK-κB pathway (Taganov et al. 2006) and the polarization of macrophages towards the immunosuppressive, M2-type (Li et al. 2015). miR-146a displays multiple roles in the modulation of different pathways in different cancer types. According to the Kaplan–Meier plotter database (Nagy et al. 2018, 2021), a higher expression level of miR-146a is associated with a shorter overall survival (OS) in renal clear cell carcinoma (p = 0.04) and in hepatocellular carcinoma (p = 0.032) and with longer OS in bladder carcinoma (p = 0.0041), head-neck squamous cell carcinoma (p = 0.00044), breast cancer (p = 0.038), ovarian cancer (p = 0.0036) and lung adenocarcinoma (p = 0.012). However its expression level in plasma and the corresponding prognostic value remains to be clarified in HCC. The aim of this study was to investigate if miR-146a level might serve as prognostic factor in metastatic liver cancer of gastrointestinal origin undergoing sequential ablative therapy. Moreover, the relationship between miR-146a expression levels in plasma and immune infiltration of distant untreated metastatic TIME was analysed.

A growing body of evidence points to the role of ⁹⁰Y-RE to stimulate an antitumor immune response, through the induction of cellular destruction and release of tumor-associated antigens (Chew et al. 2019). Nevertheless, ⁹⁰Y-RE-induced, clinically measurable abscopal effects are rare (Powerski et al. 2020). Moreover, optimization of therapeutic strategies certainly requires the as of yet strong unmet medical need for predictive biomarkers. In the last decade, the discovery of minimally invasive, liquid biopsy-based biomarkers has shown large potential. Previous studies have addressed the effect of radiation on miRNA levels, as well as the correlation of specific miRNA with therapy response. In a lung cancer model, ionizing radiation induces the upregulation of miR-21 with inhibition of apoptosis (Wang et al. 2013), induction of cellular proliferation and therapy resistance (Wang et al. 2019a, b; Stechele et al. 2022). On the contrary, in a colorectal cancer model, the downregulation of miR-221 and miR-222 enhances radiation sensitivity (Zhang et al. 2011). Furthermore, in HCC and metastatic liver treated by local or locoregional therapies, the levels of miR-21, miR-210 and miR-122 increase and show predictive value (Andrasina et al. 2021).

In our own study, we aimed to investigate the significance of miR-146a as a non-invasive biomarker of TIME infiltration in distant metastases and its prognostic value. miR-146a did not correlate with any sign of deterioration in the liver function as a consequence of irradiation. Nevertheless, the univariable analysis indicated a trend toward an association between miR-146a and OS, while no association between miR-146a levels and therapy response was found. The histological TIME inspection after ⁹⁰Y -RE showed an association of M2 macrophage infiltration with higher expression of Tim3 and a lower infiltration of CD8 + T cells in those patients with high levels of miR-146a. miR-146a belongs to a class of miRNA known as inflamma-miR, due to their multiple roles in inflammation. mR-146a inhibits the pro-inflammatory, anti-tumorigenic nuclear factor-κB (NF-κB) (Wang et al. 2019a, b) by targeting IRAK1 and TRAF6, both proteins being part of the NF-κB pathway. miR-146a modulates also the macrophage M1–M2 polarization (Takeuchi and Akira 2010). M2 polarized macrophages mediate an anti-inflammatory and, therefore, pro-tumorigenic, response. In our study, immuno-histological analysis of distant metastases evidenced an increased M2 macrophage infiltration in those patients with high levels of miR-146a and low M2 macrophage infiltration in patients with low levels of miR-146a. This finding suggests that miR-146a can function as a time dependant biomarker of an anti-inflammatory, pro-tumorigenic status in liver metastases of GI cancer. Consequently, with further validation miR-146a might be considered an early marker for ⁹⁰Y-RE efficacy: patients displayed benefit from ⁹⁰Y -radioembolization when reduction of miR-146a levels were observed 48 h post therapy. The lack of correlation between survival and radiological response after ⁹⁰Y -RE may likely be attributed to advanced disease stages in our cohort.

Furthermore, we did not find any direct correlation between plasma miR-146a and T-cell infiltration; however, we discovered a significant association between miR-146a and the exhaustion marker Tim-3 expressed by immune cells. Tim-3 demonstrated a trend to decrease if CD8 + T cells infiltration was increased in the same region. It should be noted that immunophenotyping presently does not address the co-expression of the respective markers. Co-staining of tissues with these markers will be necessary to confirm our results. Immune exhaustion is one of the hallmarks of cancer. Tim-3 is an immuno-regulator that negatively modulates T-cell response and T-cell apoptosis (Wolf et al. 2020; Sakaguchi et al. 1995; Sakuishi et al. 2010). miRNAs play a role in CD8⁺ T cell exhaustion, as demonstrated previously with miR-31 (Bhela and Rouse 2017) and miR-155 (Stelekati et al. 2018). We hypothesize that miR-146a might modulate the expression of Tim-3 in the TIME, leading to CD8⁺ T cell exhaustion. Future work will be necessary to characterize the CD8⁺ T cells present in the lesions. Our results suggest that patients displaying an increase of miR-146a after ⁹⁰Y-RE may benefit from immunotherapy targeting Tim-3. Targeting miR-146a could be a novel strategy to improve the effect of radioembolization in patients receiving this type of therapy, alone or in combination with a downstream ablative therapy. Manipulation of miRNA levels by mimicking or inhibiting their expression represents an attractive strategy to sensitize tumours to radiation with a growing body of evidence supporting this approach (Chen et al. 2021). For example, the overexpression of tumor suppressor miRNAs such as miR-34a (Cortez et al. 2019) or members of the Let-7 family (Zhou et al. 2015) have been shown to sensitize lung and uveal melanoma tumor cells, respectively, to radiation. The downregulation of onco-miR such as miR-21 has been shown to arrest cell proliferation and angiogenesis (Tang et al. 2019). Accordingly, the downregulation of miR-146a in patients receiving radioembolization might therefore be a novel approach to improve therapeutic outcomes.

Limitations of this study must be acknowledged. The size of the cohort was small and inhomogeneous with respect to biology. Validation studies are necessary to confirm these preliminary data. miR analyses were conducted using core needle biopsies instead of larger resection material. For this reason, based on sampling error alone, the heterogeneity of the biomarker expression in the tumor may not be fully accurate. We were not able to define the origin of miR-146a, since the measurement of miR-146a was performed on plasma and we do not have paired PBMC available to exclude them as a major source. We do, however, intend to address this point in the future. Moreover, only a limited number of patients, tumor blocks collected before ⁹⁰Y-RE were available. Finally, the time interval between ⁹⁰Y-RE and sample acquisition was different throughout the study cohort.

In conclusion, miR-146a shows potential as a novel prognostic biomarker. A modulation of miR-146a levels in plasma might increase responses to ⁹⁰Y-RE and could represent a supporting strategy in personalized therapy of metastatic liver.