Estimation of ALU Repetitive Elements in Plasma as a Cost-Effective Liquid Biopsy Tool for Disease Prognosis in Breast Cancer

Abstract

Background: Liquid biopsy is widely recognized as an efficient diagnostic method in oncology for disease detection and monitoring. Though the examination of circulating tumor cells (CTC) is mostly implemented for the assessment of genomic aberrations, the need of complex methodologies for their detection has impeded its acceptance in low-resource settings. We evaluated cell-free DNA (cfDNA) as a liquid biopsy tool and investigated its utility in breast cancer patients.

Methods: Total cell-free DNA was extracted from the plasma of breast cancer patients (n = 167) with a median follow-up of more than 5 years, at various stages of the disease. Quantitative PCR was performed to estimate the copy numbers of two fractions of ALU repetitive elements (ALU 115 and ALU 247), and DNA integrity (DI) was calculated as the ratio of ALU 247/115. Mutations in TP53 and PIK3CA in the cfDNA were estimated by next-gen sequencing (NGS) in a subset of samples. Associations of the levels of both the ALU fragments with various clinico-pathological factors and disease-free survival at various stages were examined. Nomogram models were constructed with clinical variables and ALU 247 levels to predict disease-free survival and the best performing model was evaluated by decision curve analysis.

Results: DI and ALU 247 levels were significantly lower (p < 0.0001) in the post-operative plasma when compared to their pre-surgery levels. DI and ALU 247 were found to be significantly higher in patients with metastasis (p < 0.05). Patients with higher levels of ALU 247 in their post-operative plasma had significant poor disease-free survival (p = 0.005). Higher levels of ALU 247 in the circulation also correlated with low tumor-infiltrating lymphocytes (TIL) within their primary tumors in the ER-negative breast cancer subtype (p = 0.01). Cox proportional hazard analysis confirmed ALU 247 as an independent variable of disease-free survival both in univariate and multivariate analysis [HR 1.3 (95% CI 1.047 to 1.613, p = 0.017)]. The nomogram model showed that the addition of ALU 247 with other variables significantly improved (C-index 0.823) the predictive ability of the model.

Conclusion: Our results confirm the utility of cfDNA as an evolving liquid biopsy tool for molecular analysis. Evaluation of larger fragments of cfDNA estimated through ALU 247 can provide vital information concurrent with the pathological process of disease evolution in breast cancer and warrants expansion to other cancer types.

Keywords: ALU 247; breast cancer; ctDNA; disease progression; liquid biopsy; prognosis.

Figure 1

(A) Heat map depicting the levels of ALU 115 and ALU 247, and the DI index among post-operative and pre-operative samples. (B–D) Univariate analysis depicting levels of ALU 115 and ALU 247, and the DI ratio, respectively, among matched/paired post-operative and pre-operative samples (n = 88 each).

Figure 2

(A–C) Univariate analysis depicting levels of ALU 115 and ALU 247 and the DI ratio, respectively, among the metastatic (n = 20), pre-operative (n = 111), remission (n = 12), and post-operative samples (n = 108). (D) Comparative analysis of ALU 247 levels among IHC groups Her2 (n = 16), ER+ (n = 57), and TNBC (n = 30) and (E) between dense/moderate (n = 16) and mild (n = 23) immune infiltrate groups of the ER-negative subtype. NS: not significant.

Figure 3

(A): disease-free survival between ALU 247 low (n = 27) and ALU 247 high (n = 9) samples and (B): between DI low (n = 37) and DI high (n = 13) samples.

Figure 4

(A) ROC curve for ALU 247 with other parameters with AUC analysis derived by logistic regression. (B) Nomogram analysis with the M1 model depicting disease-free survival based on conventional parameters. (C) Nomogram analysis with the M2 model depicting disease-free survival based on conventional parameters along with post-operative ALU 247 levels. (D) Decision curve analysis derived for the nomogram M1 and M2 depicting the net benefit of the model across different threshold probabilities.