Ultra-Sensitive Detection of Circulating Serum microRNAs (miRNAs) in Diffuse Large B-Cell Lymphoma (DLBCL) Patient-Derived Xenograft (PDX) Models and Correlation with Disease Status in DLBCL Patient

Background: MiRNAs are small non-coding RNAs that regulate post-transcriptional gene expression, contribute to various facets of cancer pathogenesis, and may serve as a sensitive diagnostic platform as well as potential novel therapeutic targets. We recently identified a 9 miRNA (miR-15a, let-7c, let-7b, miR-27a, miR10b, miR-18a, miR130a, miR24, and miR155) signature in serum of Smurf2^T/T mice that was detectable many months prior to the formation of visible DLBCL (Beheshti A et al PLOS One, 2017). We hypothesized that this circulating miRNA signature would also correlate with DLBCL status in PDX models and ultimately patients (pts).

Methods: We performed droplet digital PCR (ddPCR), which allows for rapid quantification of single miRNA molecules. First, we examined the 9 aforementioned miRNAs in serum collected from mice xenografted with 5 DLBCL PDXs (Townsend et al. Cancer Cell 2016) and age-matched NSG mice without xenografts as controls. We also quantified the amount of miRNAs directly from all PDX cell lines utilized and 2 commercially available DLBCL cell lines for comparison. Next, we collected and analyzed serum from 86 pts with DLBCL in the following conditions: pre-treatment with first-line therapy (n=11); progression after treatment (n=7); during treatment (n=16); and in complete remission (n=52). Seventeen healthy non-tumor bearing individuals were used as controls. Median age of healthy controls was 52 years (range, 29-70) vs 64 years (range, 21-87) for DLBCL pts (p<0.001; Wilcoxon rank-sum test). 69% of DLBCL pts were male vs 35% of healthy controls (p=0.012; Fisher exact test).

Results: The miRNA signature was enriched in DLBCL PDX mice with high expression confirmed in the 9 miRNAs, including MYC single and double hit models. The overall mean amounts of the miRNAs present in the serum of PDX bearing mice were mostly equivalent to the amount present in the original PDX cell line (Fig 1), with exception of one PDX cell line (DFBL-75549) that consistently had 100x more miRNA present in the serum of the mice vs associated cell line. In addition, we identified significantly increased circulating levels of the same miRNA signature in the serum of DLBCL pts (Table 1) (two-sided Wilcoxon rank-sum test, p<0.05) with similar patterns of change as seen in the murine models. Interestingly, higher circulating levels of let-7b were associated with a higher stage at diagnosis (stage I-II vs III-IV, median 54.0 vs. 163.2; P=0.007) and higher levels of both miR-27a and miR-24 were associated with having a MYC rearrangement (median 20.0 vs. 4.8; P=0.003; median 58.4 vs. 24.4; P=0.046). Higher levels of miR-18a were associated with Myc positivity by immunohistochemistry (0-9% of cells vs. 10-59% vs. ≥60%, median 3.3 vs. 2.8 vs. 8.8; P=0.030) as well as having received a prior therapy for DLBCL (median 4.8 for previously treated vs. 1.1 for untreated; P=0.016). Using the 23 on-treatment and progression samples compared with healthy samples, we selected cut-points based on the Youden Index from receiver operating curve (ROC) analysis and were able to classify the remission samples with an accuracy of 46%-88%. miR-24 performed the best in classification with a sensitivity of 85% and a specificity of 100%, while 6 of the 9 miRs had a classification rate >80%. Using a 5 miR signature with cut-points selected from recursive partitioning, we were able to classify remission samples with an accuracy of 91% (sensitivity 90%, specificity 94%).

Conclusions: Altogether, ultrasensitive detection of circulating miRNAs originally identified in a lymphoma xenograft knockout model was readily detectable and highly elevated in DLBCL PDX models. Additionally, there were significantly increased circulating levels of the miRNA signature from the serum of DLBCL pts. Particular miRs were associated with pt stage and the presence of MYC overexpression or rearrangement in pts with DLBCL. Furthermore, circulating miRNAs were able to reliably distinguish DLBCL pts in remission from healthy controls based on a novel 5 miR signature. Validation in additional cohorts is needed to confirm whether miRNA quantification from serum may be a broadly applicable strategy for diagnosis, classification and response assessment among pts with DLBCL.

View largeDownload slide

View largeDownload slide

Close modal