Overexpression of HMGB1 Receptor RAGE Is Associated with Worse Clinical Outcome in Patients with Chronic Lymphocytic Leukemia

Background: Receptor for advanced glycation end-products (RAGE) is the major receptor for inflammatory molecule HMGB1. Increased levels of RAGE are associated with poor prognosis in many types of tumours. Following engagement with extracellular HMGB1, RAGE internalises and promotes the survival and migration of cancer cells through activation of NF-kB. We have previously determined significantly increased HMGB1 levels in the plasma of CLL patients compared to healthy controls, associated with poorer clinical outcomes. HMGB1 induces differentiation of nurse like cells via RAGE and/or Toll-like receptor 9 (TLR9) harboring CLL progression and maintenance. However, it is unknown whether HMGB1/RAGE/TLR9 signaling axis can directly promote CLL B-cell survival and contribute to disease progression.

We hypothesise that HMGB1 could contribute to CLL progression via direct activation of RAGE/TLR9 pathways. The aim of this study is to characterise whether activation of HMGB1-mediated pathways is associated with poorer clinical outcome in patients with CLL.

Methods: HMGB1-mediated colocalisation of TLR9 and RAGE proteins was assessed in fresh primary CLL cells using the ImageStream X Mark II flow cytometer and fluorescent microscopy. Expression of RAGE and TLR9 were determined by immunohistochemistry on tissue microarrays (TMAs), containing CLL lymph node samples (n=91). The correlation between RAGE and TLR9 with other established CLL markers were analysed by both Spearman's multiple correlation and chi-square analyses. The association between protein expression and overall survival (OS) was analysed by uni- and multivariate cox regression analysis on both continuous and categorical variables to increase statistical robustness. The test/validation set-defined data cut points were determined by X-tile software.

Results: HMGB1-mediated internalisation of RAGE and its colocalisation with TLR9 in primary CLL-B-cells were detected qualitatively by fluorescent microscopy and quantitatively by ImageStream X Mark II flow cytometry, demonstrating that HMGB1 directly activates both RAGE and TLR9 in CLL B-cells.

RAGE and TLR9 proteins are heterogeneously expressed in CLL lymph nodes. Multiple correlation analysis identified a significant correlation between RAGE and TLR9 receptor expression (p<0.0001) suggesting they have co-linked functions. RAGE expression positively correlates with levels of Ki67 and positivity of ZAP70 (p<0.0001; p<0.05), and negatively correlates with the median OS rates, suggesting that higher expression of RAGE is associated with CLL cell proliferation and poor OS. However, RAGE expression is not correlated with CD68, a marker of tumor-associated macrophages, suggesting that increased RAGE protein may be mainly expressed by CLL B-cells.

We analysed the prognostic effect of RAGE, TLR9, Ki67 and age individually by continuous univariate cox regression analysis and found that only the age of CLL patients is an independent prognostic marker for this cohort of CLL patients. Expression cut points were statistically generated using the X-tile software to define RAGE expression as high/medium/low. Increased RAGE expression (n=13) confers inferior OS (median 3 years) compared to patients with medium/low RAGE expression (n=29 and 48 respectively; median OS 7.5 years; log rank test p<0.05). There is a significant prognostic interaction between high RAGE, high Ki67 and ZAP70 categorical variables (p<0.0001), highlighting the inferior effect of overexpression of RAGE on disease outcome.

Conclusions: Our data demonstrates for the first time that overexpression of HMGB1 receptor RAGE in CLL lymph nodes is associated with a worse clinical outcome in patients with CLL and positively correlated with higher expression of Ki67 and positivity of ZAP70. HMGB1 can directly activate both RAGE and TLR9 in primary CLL B-cells. We therefore propose that HMGB1 may play an important role in promoting CLL cell proliferation and survival and could be a potential target for the treatment of CLL. However, we need to further confirm our findings by combining other independent validation sets and intensive functional studies on the role of HMGB1 in CLL.