Investigations of circulating microRNAs in malignant diseases is currently a very active field and a large amount of data have already been published about this subject [8, 11–13]. One remarkable characteristic of circulating microRNAs is their stability. To a large extent, it results from their association with various types of carriers. Some of these carriers are vesicular with a relatively large size; the two main categories are microvesicles (100 nm to 1 μm in diameter) and exosomes (30 to 100 nm). There are also non-vesicular carriers of smaller size like the HDL lipoproteins and non-lipid ribonucleoprotein complexes which, to a large extent, remain to be characterized [10, 11, 14, 15].
NPC appears as a privileged model for investigations of circulating tumor microRNAs for two reasons: 1) malignant cells are latently infected by EBV in virtually all NPCs; 2) NPC cells have intense production of microRNAs from the BART cluster in the absence of production from the BHRF1 cluster [2, 4, 16]. It is not yet clear whether mir-BARTs can be produced in the healthy EBV-carrier or outside tumor tissues in NPC patients. According to in vitro models, latently infected B-cells are not expected to produce miR-BARTs but rather BHRF1 microRNAs [2, 4, 17]. It is known that the EBV lytic-replicative cycle is consistently taking place in the epithelial cells of the oral cavity (including tonsils and may be salivary glands) . We do not know yet whether these lytically infected epithelial cells produce and release EBV miR-BARTs.
Our data were obtained from NPC plasma samples mainly collected from patients born in France and North Africa. They are consistent with the data reported by Wong et al. in a series of serum samples from 15 Chinese patients . Presence of circulating miR-BART17 thus appears as a consistent feature of the disease, regardless of patient origins. Although there are discussions on whether use of serum or plasma samples is optimal for recovery of circulating microRNAs, our detection of miR-BART17 from plasma samples was apparently as efficient as the detection reported by Wong et al. who worked with serum samples [7, 19, 20]. In contrast with Wong et al., we could detect small amounts of miR-BART17 in plasma samples from non-NPC donors. This might be due to the fact that we took in account samples with very small concentrations of microRNAs, namely those giving Ct values higher than 35, whereas they are automatically classified as negative by many investigators. The presence of miR-BART17 in plasma samples from a few control donors might reflect low-level production of miR-BARTs by non-malignant cells for example in the oral cavity. This should not undermine the main conclusion of our study: detection of the miR-BART17-5p above the threshold of 506 copies/mL appears as a marker of NPC plasma samples with good sensitivity (77%) and high specificity (90%).
To further improve the sensitivity and specificity of miR-BART detection in plasma samples, we attempted to characterize their carriers. On the basis of previous results, we were assuming that there were carried by NPC tumor exosomes . However, fractionation of plasma elements on a KBr gradient showed that the cellular micro-RNA, miR-16, partially co-purified with exosomes whereas miR-BART17 was recovered in a completely distinct fraction. Although it has only been observed for our two samples subjected to gradient fractionation, this observation may be relevant to the design of future studies. On the one hand, the lack of co-purification of miR-BART17 with exosomes is surprising since we and others have shown that the BART microRNAs are abundant in exosomes released by EBV-infected cells in vitro[6, 21]. On the other hand, it appears to be consistent with a recent report about liver microRNAs . This report shows that, depending on physiological or pathological conditions, the same micro-RNAs co-purify either with an exosome-rich or a protein-rich fraction of the plasma. Therefore we should consider the hypothesis that the miR-BARTs are secreted in association with exosomes by NPC cells in vitro but not in the tumor context in vivo. The concentration of miR-BART17 at the bottom of the KBr gradient suggests that it is associated with non-floating elements, probably ribonucleoprotein complexes of small size and/or devoid of lipids which remained to be characterized. Other groups have reported incorporation of plasma microRNAs in non-vesicular complexes containing the Agonaute 2 (ago2) protein . In future investigations, it will be interesting to know whether plasma ribonucleoprotein complexes containing miR-BART17 can be immunoprecipitated with anti-ago2 antibodies.
Our series of NPC patients was obviously too heterogeneous to attempt correlations between the concentration of plasma miR-BART17 and clinical parameters. However, we were able to collect sequential plasma samples for one patient (Patient n°5) who had a rapidly progressive disease. We observed a dramatic increase in the concentration of miR-BART17 which was parallel to tumor progression. This suggests that the plasma concentration of miR-BART17 is related, at least in part, to the tumor mass. That point will require further investigations. In this small series, we have found no obvious correlations between the concentrations of plasma miR-BART17 and the plasma EBV DNA loads. One can speculate that both miR-BART and EBV DNA plasma copy numbers are affected by multiple factors: the total tumor burden, the apoptotic index of the malignant cells, their rate of necrosis, the vascularization of tumor lesions and the rate of degradation of microRNAs and DNA by peripheral nucleases. However, the relative impact of each of these factors is probably different for the miR-BARTs and for the EBV-DNA. This could explain why their plasma concentrations might evolve independently at various stages of the disease. On the other hand, if the concentrations of miR-BARTs and EBV DNA are at least partially independent, they have more chances to provide non-redundant information for assessment of the tumor status and/or prediction of the tumor evolution. In order to substantiate this hypothesis, we currently investigate: 1) the relationships between the abundance of circulating miR-BARTs and the total tumor mass in a series of non-metastatic NPC patients; 2) the longitudinal evolution of circulating miR-BARTs in connection with the tumor response to the initial treatment. While this study was near completion, Chan et al. have reported a consistent detection of miR-BART7 in the plasma of NPC patients . Both studies converge on two important points: 1) EBV BART microRNAs are detectable in the plasma of some non-NPC donors but at a low level in contrast with most NPC patients; 2) the concentration of circulating miR-BARTs seems to be relatively independent of the plasma viral DNA load.