Scholarlink

Background <p>Nasopharyngeal carcinoma (NPC) represents a highly prevalent and aggressive malignancy endemic to Southeast Asia. Early and accurate diagnosis is critical to improving survival outcomes; however, the absence of robust, stage-specific biomarkers remains a key obstacle to clinical implementation of early screening strategies.</p> Methods <p>We performed untargeted serum proteomic profiling using mass spectrometry in 15 treatment-naïve early-stage NPC patients and 15 VCA-IgA-positive healthy controls. Bioinformatics analyses were conducted to identify differentially expressed proteins (DEPs). Machine learning (random forest combined with recursive feature elimination) was employed to prioritize candidate biomarkers, which were subsequently verified using enzyme-linked immunosorbent assay (ELISA) in independent sample cohorts.</p> Results <p>In total, 1,428 serum proteins were identified, among which 1,410 were reliably quantified. We observed 31 upregulated and 189 downregulated proteins in NPC patients relative to controls. Spearman correlation analysis revealed significant associations: LTA4H (leukotriene A4 hydrolase) levels correlated with serum cell infiltration (<i>r</i> = 0.383, <i>p</i> = 0.032) and CD8 + T-cell abundance (<i>r</i> = 0.408, <i>p</i> = 0.021); both SULT1A3 (sulfotransferase family 1&#xa0;A member 3) and FGL1 (fibrinogen-like protein 1) levels were positively associated with M1 macrophage infiltration (<i>r</i> = 0.510, <i>p</i> = 0.003 and <i>r</i> = 0.430, <i>p</i> = 0.015, respectively).</p> <p>In a preliminary validation cohort (<i>n</i> = 80), ELISA yielded AUC values of 0.631 (95% CI: 0.515–0.736, <i>p</i> = 0.04) for LTA4H, 0.787 (95% CI: 0.681–0.871, <i>p</i> &lt; 0.001) for SULT1A3, and 0.688 (95% CI: 0.575–0.787, <i>p</i> = 0.002) for FGL1. In large-scale independent validation, SULT1A3 achieved an AUC of 0.826 (95% CI: 0.766–0.876; sensitivity = 78.89%, specificity = 75.47%) in cohort 1 (<i>n</i> = 196) and 0.796 (95% CI: 0.723–0.857; sensitivity = 76.67%, specificity = 76.67%) in cohort 2 (<i>n</i> = 150).</p> Conclusions <p>Through an integrated workflow combining proteomic screening, machine learning prioritization, and multi-stage ELISA validation, we identified SULT1A3 as a candidate serum-based biomarker for early detection of NPC. Preliminary findings suggest that SULT1A3 may have potential utility in clinical screening, though further validation in independent, multi‑center cohorts is required.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Potential evaluation of SULT1A3 as an early diagnostic marker for nasopharyngeal carcinoma: a study based on serum proteomics screening and ELISA validation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月27日

查看全文
  • Zhifang Lin,
  • Wei Song,
  • Wenhai Liang,
  • Jian Liu,
  • Ruiwei Gan,
  • Yueying Wang

摘要

Background

Nasopharyngeal carcinoma (NPC) represents a highly prevalent and aggressive malignancy endemic to Southeast Asia. Early and accurate diagnosis is critical to improving survival outcomes; however, the absence of robust, stage-specific biomarkers remains a key obstacle to clinical implementation of early screening strategies.

Methods

We performed untargeted serum proteomic profiling using mass spectrometry in 15 treatment-naïve early-stage NPC patients and 15 VCA-IgA-positive healthy controls. Bioinformatics analyses were conducted to identify differentially expressed proteins (DEPs). Machine learning (random forest combined with recursive feature elimination) was employed to prioritize candidate biomarkers, which were subsequently verified using enzyme-linked immunosorbent assay (ELISA) in independent sample cohorts.

Results

In total, 1,428 serum proteins were identified, among which 1,410 were reliably quantified. We observed 31 upregulated and 189 downregulated proteins in NPC patients relative to controls. Spearman correlation analysis revealed significant associations: LTA4H (leukotriene A4 hydrolase) levels correlated with serum cell infiltration (r = 0.383, p = 0.032) and CD8 + T-cell abundance (r = 0.408, p = 0.021); both SULT1A3 (sulfotransferase family 1 A member 3) and FGL1 (fibrinogen-like protein 1) levels were positively associated with M1 macrophage infiltration (r = 0.510, p = 0.003 and r = 0.430, p = 0.015, respectively).

In a preliminary validation cohort (n = 80), ELISA yielded AUC values of 0.631 (95% CI: 0.515–0.736, p = 0.04) for LTA4H, 0.787 (95% CI: 0.681–0.871, p < 0.001) for SULT1A3, and 0.688 (95% CI: 0.575–0.787, p = 0.002) for FGL1. In large-scale independent validation, SULT1A3 achieved an AUC of 0.826 (95% CI: 0.766–0.876; sensitivity = 78.89%, specificity = 75.47%) in cohort 1 (n = 196) and 0.796 (95% CI: 0.723–0.857; sensitivity = 76.67%, specificity = 76.67%) in cohort 2 (n = 150).

Conclusions

Through an integrated workflow combining proteomic screening, machine learning prioritization, and multi-stage ELISA validation, we identified SULT1A3 as a candidate serum-based biomarker for early detection of NPC. Preliminary findings suggest that SULT1A3 may have potential utility in clinical screening, though further validation in independent, multi‑center cohorts is required.