Expression pattern and role of circular RNAs in the prognosis of mantle cell lymphoma in younger patients

Treatment with cytarabine-containing chemo-immunotherapy followed by autologous stem cell transplant (ASCT) has improved the outcomes for patients with mantle cell lymphoma (MCL). Despite such advances in treatment, a long-term relapse pattern is observed in patients, highlighting the heterogeneity in disease progression. Younger patients with MCL have improved long-term outcomes; however, there are limited studies on prognostic biomarkers of MCL. The MCL International Prognostic Index (MIPI) is the most widely used prognostic measure for newly diagnosed patients with MCL, but it has not been validated prospectively.

Circular ribonucleic acids (circRNAs) are endogenous non-coding RNA molecules that are very stable and have longer half-lives than messenger RNAs, indicating that they may have potential as prognostic biomarkers in cancer. Dahl, et al.¹ recently published a study in Leukemia investigating the expression patterns and prognostic potential of circRNAs as novel biomarkers in younger patients with MCL. The key findings from the study are summarized here.

Methods

The exploratory cohort included diagnostic tumor tissues from 14 patients with MCL and six healthy controls (four samples of naïve B cells and two samples from reactive lymph node tissue). The MCL samples were grouped into three categories based on MIPI-combined (MIPI-C): a high-risk group (n = 5), a high-intermediate-risk group (n = 5), and a combined low and low-intermediate-risk group (n = 4).

The training (n = 74) and validation (n = 89) cohorts were formed of patients with MCL from previous phase II trials—MCL2 and MCL3, respectively—who were <66 years of age and had circRNA data available. All 163 patients in both cohorts had received rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-maxi-CHOP), alternating with rituximab + high-dose cytarabine followed by high-dose chemotherapy and ASCT.

RNA-sequencing was used in the exploratory cohort to explore the expression of circRNAs. Least absolute shrinkage and selection operator was used to identify predictive variables among the circRNAs from the training cohort (MCL2). A risk score (circSCORE) was calculated for each patient in the training cohort based on the circRNAs with prognostic impact.

Baseline characteristics of the training and validation cohorts

The median age of patients in both the training and validation cohorts was 58 years (range, 53–61 years) and 75% of patients were male (Table 1).

Table 1. Baseline characteristics of training and validation cohort*

Results

CircRNAs are widely expressed in MCL

In total, 80% of unique circRNAs were expressed in MCL samples in the exploratory cohort; interestingly, circRNA expression in tissue from reactive lymph nodes was found to more closely resemble MCL tissue than naïve B cells. CircZNF91 was the circRNA with highest expression in all three MCL groups, and the median circRNA expression was significantly lower in the control samples of the exploratory cohort compared with the three MCL groups (p < 0.0001).

There were significant differences in the circRNA expression between the three MIPI-C risk groups. The highest expression of circRNAs was seen in the combined low and low-intermediate-risk group (0.143; 95% confidence interval [CI], 0.130–0.155) and the lowest was in the high-risk group (0.085; 95% CI, 0.079–0.094), while the high-intermediate-risk group displayed intermediate circRNA expression (0.112; 95% CI, 0.102–0.125).

CircRNA expression is inversely associated with the cell proliferation marker Ki67

With an inverse correlation between circRNAs and Ki67 (r = 0.733; p = 0.004), the median expression of high-abundance (cut-off at 0.05, average read per million) circRNAs in patients with normal Ki67 was significantly higher (0.125; 95% CI, 0.114–0.146) compared with patients who had high Ki67 (0.090; 95% CI, 0.079–0.995; P < 0.0001). Samples from the exploratory cohort showed that normal Ki67 clustered more uniformly than samples with high Ki67.

Evaluation of 40 circRNAs as potential biomarkers in MCL

From the high-abundance dataset of 40 unique circRNAs from the training cohort, nine were incorporated into a single model predictive of time to progression (TTP). These nine circRNAs showed that patients with low levels of the unique circRNAs displayed high-risk features (high Ki67, high MIPI score, and TP53 mutations) more frequently compared with patients who had higher circRNA expression levels. A similar trend was observed for blastoid morphology and MIPI-C high or high-intermediate-risk groups, but it was not as pronounced.

A circRNA expression signature predicts outcome in MCL

Patients in the training cohort MCL2 with a high circSCORE (n = 29) showed significantly shorter TTP compared with patients who had a low circSCORE (n = 45) (hazard ratio [HR], 6.0; p < 0.0001). This result was even more noticeable for lymphoma-specific survival (LSS) (HR 12.1; p < 0.0001). Patients with high circSCORE (n = 28) had a shorter median progression-free survival of 4.5 vs 7.7 years in the circSCORE low risk group (n = 61).

Patients in the circSCORE high-risk group demonstrated characteristics associated with aggressive disease more commonly compared with circSCORE low-risk patients in both cohorts (Table 2).

Table 2. High-risk characteristics of MCL2 and MLC3 patients based on circSCORE risk group*

CircSCORE is an independent prognostic biomarker in MCL

The prognostic impact of circSCORE was maintained independently for both TTP (HR, 3.2; p = 0.01) and LSS (4.6; p = 0.02) in the validation cohort (n = 59).

MIPI high-risk and presence of TP53 were the only high-risk features that also had a prognostic effect, suggesting that high circSCORE may be as important as MIPI high-risk and presence of TP53 mutations as a prognostic indicator. 

CircSCORE might also be a useful biomarker in patients with TP53^wt MCL

Pooled data of patients from the training (n = 50) and validation (n = 66) cohort with TP53 mutation showed a significantly shorter TTP (HR, 2.0; p = 0.02) in the circSCORE high-risk group (n = 26) vs low-risk group (n = 68). Patients with TP53 wild type (TP53^WT) MCL in the circSCORE high-risk group also showed significantly worse LSS compared with patients in the low-risk group (HR, 3.6; p = 0.003).

Conclusion

This study validates that circSCORE is a novel prognostic biomarker in MCL and that it has the potential to enhance the identification of high-risk disease among younger patients treated with cytarabine-containing chemoimmunotherapy and ASCT. Further research is warranted to investigate the prognostic potential of circSCORE in patients treated with other MCL regimens and in a relapse setting. Further investigation of circSCORE in a larger cohort of patients with TP53^wt MCL is also needed.