Development and validation of a model and nomogram for breast cancer diagnosis based on quantitative analysis of serum disease-specific haptoglobin N-glycosylation

Abstract

Abstract Background A better diagnostic marker is in need to distinguish breast cancer from suspicious breast lesions. The abnormal glycosylation of haptoglobin has been documented to assist cancer diagnosis. This study aims to evaluate disease-specific haptoglobin (DSHp)-β N-glycosylation as a potential biomarker for breast cancer diagnosis. Methods DSHp-β chains of 497 patients with suspicious breast lesions who underwent breast surgery were separated from serum immunoinflammatory-related protein complexes. DSHp-β N-glycosylation was quantified by mass spectrometric analysis. After missing data imputation and propensity score matching, patients were randomly assigned to the training set (n = 269) and validation set (n = 113). Logistic regression analysis was employed in model and nomogram construction. The diagnostic performance was analyzed with receiver operating characteristic and calibration curves. Results 95 N-glycopeptides at glycosylation sites N207/N211, N241, and N184 were identified in 235 patients with benign breast diseases and 262 patients with breast cancer. DSHp-β N-tetrafucosyl and hexafucosyl were significantly increased in breast cancer compared with benign diseases (p < 0.001 and p = 0.001, respectively). The new diagnostic model and nomogram included GN2F2, G6N3F6, GN2FS at N184, G-N&G2S2, G2&G3NFS, G2N3F, GN3 at N207/N211, CEA, CA153, and could reliably distinguish breast cancer from benign diseases. For the training set, validation set, and training and validation sets, the area under the curves (AUCs) were 0.80 (95% CI: 0.75–0.86, specificity: 87%, sensitivity: 62%), 0.77 (95% CI:0.69–0.86, specificity: 75%, sensitivity: 69%), and 0.80 (95% CI:0.76–0.84, specificity: 77%, sensitivity: 68%), respectively. CEA, CA153, and their combination yielded AUCs of 0.62 (95% CI: 0.56–0.67, specificity: 29%, sensitivity: 90%), 0.65 (95% CI: 0.60–0.71, specificity: 74%, sensitivity: 51%), and 0.67 (95% CI: 0.62–0.73, specificity: 60%, sensitivity: 68%), respectively. Conclusions The combination of DSHp-β N-glycopeptides, CEA, and CA153 might be a better serologic marker to differentiate between breast cancer and benign breast diseases. The dysregulated N-glycosylation of serum DSHp-β could provide insights into breast tumorigenesis.

Keywords

• Glycosylation
• Breast cancer
• Diagnosis
• Model
• Nomogram