Archives

  • 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2020-03
  • 2020-07
  • 2020-08
  • br The gene signature outperformed

    2019-10-07


    3.3. The 15-gene signature outperformed conventional diagnostic approaches for LN metastasis detection in gastric cancer patients
    Using multivariate analysis, we demonstrated that our 15-gene sig-nature was able to successfully detect LN metastasis, independent of pre-operative clinical factors such as age, gender, tumor markers and clinical LN status determined by computed tomography (Table 3). To further evaluate the significance of this signature, we next compared the diagnostic potential of our gene signature versus various pre-operative clinical factors including tumor markers and clinical LN status. We found the diagnostic value of our 15-gene signature was signifi-cantly superior via-a-vis conventional tumor markers including the levels of circulating CEA (AUC = 0.520, 95% CI, 0.429–0.610; P = .044 [DeLong]) and CA19–9 (AUC = 0.518, 95% CI, 0.427–0.608; P = .0047 [DeLong]; Fig. 2b and Table 2).
    In addition, to evaluate the performance of the 15-gene signature, we compared its performance against the clinical N stage determined by preoperative diagnostic CT scans in the patients from the validation cohort. Interestingly, our gene AS1517499 signature demonstrated a sig-nificantly superior accuracy (0.803, 95% CI, 0.510–0.905) compared to CT imaging data for identifying the presence of LN metastasis (AUC = 0.742; P = .038 [DeLong]; Fig. 2c and Table 2), which only achieved an AUC of 0.595 (95% CI, 0.511–0.675).
    3.4. A combination of the 15-gene signature together with other clinico-pathological features further improves the diagnostic accuracy for LN me-tastasis detection in gastric cancer patients
    We next asked whether a combination of our 15-gene signature to-gether with currently used clinicopathological factors (e.g. age, gender, tumor markers, and clinical N stage using multivariate logistic regres-sion analysis) might further enhance the diagnostic accuracy of our panel. It was interesting to observe that indeed integration of our gene expression signature with the clinical N stage, significantly improved the discriminative accuracy of our biomarker panel further in
    4. Discussion
    As we usher into the new era of precision-medicine, tailoring indi-vidualized treatments are definitely going to serve as cornerstones for a more effective cancer care. Currently early stage GC patients, which are deemed to be high-risk for lymph node (LN) metastasis based upon various pre-surgical histopathological features are frequently over-treated, due to the lack of availability of adequate molecular markers that can more robustly identify such metastasis prior to the surgery. In this study, we undertook a systematic and comprehensive, genome-wide transcriptomic biomarker discovery, and developed a panel of genes for the identification of LN metastasis in patients with early stage gastric cancers (T1 and T2), using independent, publicly-available gene expression datasets. Subsequently, a 15-gene signature was optimized for qRT-PCR based analysis using the clinical testing cohort-1 by logistic regression analysis, followed by validation in an in-dependent patient cohort. Finally, we performed a head-to-head com-parison between the 15-gene signature with the conventional tumor markers (CEA and CA19–9) as well as CT-based imaging, and demon-strated its superiority for identifying GC patients with LN metastasis.
    Recent advancements in high-throughput sequencing technologies have resulted in comprehensive molecular characterization of GC [16]. Similar to other major malignancies, multiple molecular subtypes of GC have been proposed based on integrative analysis of transcriptome wide gene expression profiles [17]. Accordingly, several gene expression-based cancer biomarkers utilizing multiple genes have been suggested over the years [7,9]. Because RNA-sequencing provides molecular insights into tumor heterogeneity and the disease process, in this study we focused on establishing a gene expression based-signature for the diagnosis of LN metastasis in early stage GC patients using a transcriptomic-wide analysis of T1 tumors from GC patients. We identified a cluster of 15 highly expressed genes, several of which are functionally relevant and GC-associated genes including C5AR1, CD83, NR4A2, ETV4, and TRPV4. In gastric cancer, C5AR1 has been shown to promote motility and invasiveness of cancer by activating RhoA, and its expression is reported to be associated with prognosis of GC patients [18]. CD83 is a molecular marker for mature dendritic cells. In gastric cancer, decreased density of CD83 (+) dendritic cells and increased density of FOXP3 (+) regulatory T cells, are observed in the primary tumor and metastatic lymph nodes of GC, and has been shown to inversely correlate with prognosis of GC patients [19]. An in vitro study has shown that the expression of orphan nuclear receptor NR4A2 in GC cells attenuates 5-fluorouracil-induced apoptosis and af-fect chemoresistance, and predicts an unfavorable postoperative sur-vival of GC patients with chemotherapy [20]. Another putative oncogene, PEA3/ETV4 has been shown to be upregulated at both mRNA and protein levels in GC tissue and the increased expression cor-relates with the expression of their downstream metastasis associated target gene, MMP-1 and high expression of PEA3/ETV4 was associated with poor prognosis in GC [21]. Similarly, TRPV4 was shown to be a gene required for cancer cell invasion and trans-endothelial migration and its expression in GC correlated with poor clinical outcomes [22]. Furthermore, as a gene signature, three genes, NR4A2, FAM13A and PRSS21 had a significant contribution to our gene signature, suggesting that these genes play an important mechanistic role in GC LN metastasis.