Undifferentiated nasopharyngeal carcinoma (UNPC) constitutes the most common epithelial malignancy occurring in the nasopharynx, and is characterized by peculiar epidemiologic and clinicopathologic features [1, 2]. UNPC has a high incidence in Southern China and Southeast Asia, whereas it is a rare disease in Europe and North America. Moreover, compared to other head and neck carcinomas, UNPC shows higher invasive and metastatic potential, being frequently diagnosed only after dissemination [3–5]. The presence of a highly cellular lymphoid stroma, admixed with neoplastic cells, represents a peculiar morphologic feature of UNPC. The majority of infiltrating cells is constituted by non-neoplastic T lymphocytes, but other reactive elements such as macrophages, plasma cells, eosinophils, and neutrophils, are also present in varying proportions [6, 7]. Although the role of lymphoid stroma in UNPC is still poorly defined, immunophenotypic analyses have demonstrated that infiltrating T lymphocytes and tumor cells express several immune regulatory receptor/ligand pairs [8, 9], suggesting that biologically relevant interactions between the two cell components probably occur in vivo.

Multiple factors contribute to the pathogenesis of UNPC, including genetic predisposition, environmental factors, and Epstein-Barr virus (EBV) infection [1, 2, 10, 11]. In particular, EBV genomes are detected in virtually all UNPC worldwide and are monoclonal, consistent with a causal role for the virus [1, 2, 10]. EBV infection in UNPC cells is characterized by the expression of a limited set of latent proteins, including EBNA-1, LMP-1, and LMP-2 [1, 2, 10]. The LMP-1 protein transforms B both lymphocytes and rodent fibroblasts ad may contribute to the invasive and metastatic potential of carcinoma cells [12, 13]. Recent evidence indicates that LMP-1 functions as a constitutively active tumor necrosis factor (TNF) receptor being able to recruit cellular signal transducing molecules in the C-terminus of the protein, resulting in the activation of several signaling pathways in a ligand-independent manner [14]. In particular, LMP-1 shares several functional properties with activated CD40, a member of the TNF receptor superfamily, being able, in certain instances, to substitute for CD40 in vivo [15, 16].

CD40 binds to a ligand (CD40L) which is an ~35 kDa transmembrane protein expressed on activated T cells, mast cells, basophilis, eosinophils, and activated platelets [17, 18]. The interaction between CD40 and CD40L is crucial for B-cell activation, antibody production, isotype switching and cell-mediated immune responses [17, 19, 20]. Moreover, the CD40 pathway is also involved in regulating the growth and survival of normal and transformed epithelial cells [21, 22]. In particular, CD40 triggering on UNPC cells may directly convey survival signals, being able to prevent Fas-mediated apoptosis [23]. Like other members of the TNF family, CD40L exists also as a soluble isoform (sCD40L) produced by enzymatic cleavage at a metalloproteinase-sensitive site in the membrane-proximal region of the extracellular domain of the molecule [24, 25]. sCD40L is biologically active and can replace, at least in vitro, the normal T-cell-derived CD40L signal to CD40-bearing cells [24, 25]. sCD40L can be detected in the serum of healthy individuals and elevated levels have been reported in patients with autoimmune diseases [26–28], hepatitis B virus infection [26], lymphoproliferative disorders [29], and lung cancer [30]. Nevertheless, no information is currently available on the presence and amount of sCD40L in the serum of UNPC patients.

We therefore sought to determine the serum levels of sCD40L in a series of UNPC patients from Italy, a non-endemic area for this disease. The study was undertaken with the aim to identify potential associations between circulating levels of sCD40L and the virologic and clinico-pathologic parameters of UNPC at diagnosis, and to define the predictive and prognostic value of serum sCD40L.

The CD40-CD40L interaction plays a central role in the regulation of immune responses and mediates diverse biological responses that may contribute to the growth and survival of tumor cells. The relevance of CD40-dependent signaling in tumor development is also emphasized by the observation that the oncogenic properties of the EBV protein LMP-1 are due to its ability to hijack part of the CD40 cascade [14–16]. In the present study, we demonstrate that UNPC patients have significantly increased serum levels of sCD40L compared with healthy donors of similar sex and age. Interestingly, UNPC patients 40 years-old or younger carried higher amounts of serum sCD40L, an observation that further supports the hypothesis that UNPC of young people has distinctive epidemiologic and clinico-pathologic features [1, 2]. In fact, while in endemic areas UNPC does not usually occur before 45 years of age [34], in North Africa this tumor also affects young people [1, 2, 7, 35–37], representing 25% of all UNPC-affected patients [38]. Italian UNPC patients have similar epidemiologic features, as shown by a previous retrospective study [39] and by the finding that 28% of patients from the present series were 40 years-old or younger.

Several lines of evidence indicates that tumor cells may shed pro-coagulant factors that can be responsible for the generation of active thrombin [40]. This, in turn, may lead to platelet activation and the consequent release of various factors, including sCD40L [3, 41]. In UNPC patients, however, the mean plasma levels of F1+2 were within the normal range, excluding thus an increased generation of active thrombin. These findings, together with the lack of correlation between the F1+2 levels and the amount of serum sCD40L, rule out that sCD40L is mainly released from platelets in UNPC patients. Moreover, the lack of membrane CD40L in circulating T cells also excludes that a systemic activation of the immune system may be responsible for enhanced release of sCD40L. Interestingly, upon activation, the percentage of CD40L+ T helper cells was significantly reduced in UNPC patients as compared to controls. Similar findings were also observed in HIV-infected children [42]. This effect is probably not due to the loss of distinct subpopulations of T helper lymphocytes, since Italian UNPC patients carry percentages and absolute numbers of CD4+ memory (CD45R0+), CD4+ naive (CD45RA+/CD62L+) and activated (HLA-DR+) CD4+ lymphocytes similar to those of healthy donors [[43], our unpublished results]. Our findings rather support the hypothesis of an underlying functional defect of T cell responses in these patients, as also suggested by previous reports [44, 45]. In particular, some of us have recently demonstrated that circulating CD4+ T lymphocytes from Italian UNPC patients showed impaired IL-2 secretion and increased IL-10 production, consistently with a Th1/Th2 dysregulation [43]. Nevertheless, the nature and the mechanisms responsible for these functional abnormalities are not yet established and warrants further investigation.

Unlike what observed in circulating T cells, tumor infiltrating lymphocytes expressed membrane CD40L in all cases investigated, consistently with previous findings [8]. Besides strengthening the pathogenic relevance of the interactions occurring between UNPC cells and infiltrating lymphocytes, these results also suggest that sCD40L may be released within tumor microenvironment by activated T cells. This possibility is also supported by the finding that the amount of sCD40L in the serum may decrease in patients treated with local radiotherapy alone. Intriguingly, serum levels of sCD40L were inversely correlated with LMP-1 expression by tumor cells, suggesting that LMP-1-negative UNPC cells may require CD40 activation for growth and/or survival. These findings are in line with the known ability of LMP-1 to usurp part of CD40-dependent signalling and reinforce the notion that CD40 activation is of pathogenic relevance for UNPC [8, 15, 16, 23].

Within tumor tissue, CD40 triggering by both CD40L+ lymphocytes and sCD40L may directly activate signalling pathways contributing to the transformed phenotype of UNPC cells, particularly in the LMP-1-negative cases. It is worth considering that CD40 activation may enhance the invasive potential of a variety of cells through the induction of matrix metalloproteinases [46–49], enzymes that efficiently degrade extracellular matrix proteins [50, 51], favoring thus the metastatic process. Indeed, in our UNPC series, serum sCD40L levels were not correlated with T stage or regional lymph node involvement, ruling out any relationship with loco-regional tumor burden. Conversely, a significantly higher proportion of UNPC patients with distant metastases at presentation carried high levels of sCD40L, suggesting a possible contribution of this factor to an early hematogenous spreading of UNPC cells. The high levels of sCD40L detected in the serum of UNPC patients may also prevent apoptosis of UNPC cells circulating in the blood and favour their survival and/or growth after localization at distant sites. Moreover, considering that matrix metalloproteinases are involved in the generation of sCD40L [24, 24], local production of these enzymes induced by CD40 triggering may in turn enhance the release of sCD40L.

While sCD40L levels may be useful to identify UNPC patients with occult distant metastasis at presentation, our study indicate that the amount of serum sCD40L does not provide a reliable estimate of tumor burden. This is in keeping with the hypothesis that sCD40L is not a product of tumor cells, being rather part of a still poorly defined anti-tumor immune response. On these grounds, additional studies are required to better understand the role played by sCD40L within the frames of the complex immune response mounted by UNPC patients against tumour cells. Moreover, no significant prognostic role on clinical outcome was found for serum sCD40L in our series. It should be considered, however, that 5-year DFS (81%) and OS rates (62%) of our patients compare favorably with the results of other UNPC series [52–54]. These data suggest that, if active treatment strategies are used, the prognostic impact of variables such as sCD40L may be measurable only in larger series. On these grounds, it would be relevant to include the evaluation of serum sCD40L among the parameters of possible prognostic value in studies enrolling high number of UNPC patients from both endemic and non-endemic areas.

The results of the present study indicate that sCD40L may contribute to CD40 activation in UNPC cells, particularly among LMP-1-negative cases, further supporting the crucial role of CD40 signalling in the pathogenesis of UNPC. sCD40L levels may be useful to identify UNPC patients with occult metastases at presentation.