Identification of NINJ1 as a novel prognostic predictor for retroperitoneal liposarcoma

Soft-tissue sarcomas are a heterogeneous group of malignant mesenchymal tumors mostly arising from the embryonic mesoderm, among which liposarcoma is the most common pathological type in adults [1]. Liposarcomas are mainly located on the extremities (60%) and retroperitoneally (40%) [2]. Significant differences exist between retroperitoneal liposarcoma (RPLS) and liposarcoma of the extremities in both histological origin and biological behavior [2, 3]. However, most biomedical investigations have focused on tumors of the extremities, even though the treatment of RPLS is more problematic due to their complex anatomy, organ invasiveness, and high bleeding risk [4, 5]. Compared to liposarcomas of the extremities, RPLS is more prone to local recurrences (19%–44%) and shorter survival times [6]. Thus, there is an urgent need for a comprehensive understanding of the molecular mechanisms underlying RPLS and the identification of novel biomarkers specific to RPLS.

Previous reports provide a very limited understanding of the molecular basis of RPLS. Some well-known oncogenes, such as MDM2 [7], FGF-21 [8], HMGA1 [9], Calreticulin [10], NNAT [11], CCDC180 [12], and many miRNAs [13‐15] are established as crucial for liposarcoma growth and closely correlate with an unfavorable prognosis. In RPLS, Siglec-15 and Tsp2 were also reported as showing a correlation with tumor progression and poor disease-free survival [16, 17]. However, because most studies are based on tumors of the extremities, few druggable targets or widely accepted molecular prognostic factors are available for treating RPLS. Additionally, most liposarcoma biomarker studies have been based on a subjective and arbitrary selection of candidate genes without a systematic biomarker screening process, which has resulted in most of the proposed biomarkers being classical epithelial cancer biomarkers [18, 19]. Meanwhile, most systematic biomarker screens have been performed on mixed sarcoma cohorts of various pathological types. Considering the unique histological origin of RPLS, biomarkers specific to RPLS have potentially been overlooked. Here, we aimed to identify a credible and specific prognostic biomarker for RPLS.

The literature concerning prognostic biomarkers for RPLS is extremely limited, with only a few have been reported [21]. The main problems in this field have been poor consensus due to the subjectivity of biomarker selection and small sample sizes, which have impeded the development of clinically available biomarkers for RPLS. Here, we addressed these problems by building a TCGA-based classification and screening process for prognostic biomarkers, and by using our large samples to validate the screening results. The widespread recognition of the TCGA dataset and our own large RPLS population with relatively high homogeneity should contribute to the identification of a consensus biomarker for RPLS.

Our results delineated the functional landscape of the genes involved in one module, which partially explained its close correlation with the prognosis of patients with liposarcoma. In this module, NINJ1, SLC15A3, and nearly all the hub genes are reportedly involved in inflammation responses/regulation, signal transduction, and angiogenesis [22‐24]. It is worth noting that the neutrophil-to-lymphocyte ratio, an indicator of neutrophil activation, is an important unfavorable predictor in sarcoma patients [25]. A meta-analysis has also quantified an association between elevated neutrophil-to-lymphocyte ratio and poor OS (hazard ratio, 1.59; p < 0.001) and DFS (hazard ratio, 1.28; p < 0.001) in sarcoma patients [26]. Our omics data also supported the opinion that neutrophil activation is a crucial biological process in liposarcoma progression [23]. Furthermore, it has been reported that NINJ1 has two distinct functions in p53-dependent tumorigenesis [27, 28]. NINJ1 represses the translation of both wild-type and mutated p53 and acts as an oncogene and a tumor-suppressor gene, respectively, in cells with wild-type and mutant p53. According to the TCGA database, p53 is the first-ranked mutated gene among all liposarcoma genes. Thus, it is reasonable that NINJ1 may be a good prognostic factor for liposarcomas.

Our study has some limitations. First, all liposarcomas were included in the gene screening process because of the lack of RPLS-specific samples in the TCGA sarcoma dataset. However, the strict inclusion and exclusion criteria of our validation cohort largely compensate for this deficiency. Second, the size of the TCGA liposarcoma cohort was relatively small considering the large number of potential predictors. Further genomic and transcriptomic datasets from larger numbers of RPLS patients are urgently required. We believe that a larger and more complete RPLS database for biomarker discovery would greatly promote progress in this field.

In conclusion, we established a step-wise screening pipeline to identify credible and specific prognostic biomarkers for RPLS from the TCGA dataset and obtained a series of potential prognostic biomarkers. Notably, NINJ1 emerged as the top-ranked biomarker, significantly associated with both patient prognosis and biological importance. Further verification by IHC with our own RESAR cohort confirmed that NINJ1 was an independent prognostic predictor for RPLS. The identification of NINJ1 as a prognostic factor represents a significant advance in the management of RPLS, with the potential to guide treatment decisions, improve patient outcomes, and inspire the development of novel targeted drugs.