Background
Different biological and clinicopathological parameters next to the commonly used tumor and nodal stage have been proposed to help stratifying patients to certain treatment protocols. Specifically, molecular properties of the tumor microenvironment such as hypoxia, inflammation and angiogenesis are currently investigated to serve as a target for therapy[
1].
The aim of the present research was to evaluate potential prognostic markers in a retrospective series of HNSCC treated by radiotherapy ± chemotherapy. A recently published report on a retrospective series of patients with a primary HNSCC treated by radiotherapy ± chemotherapy and investigated by dynamic contrast-enhanced perfusion-computer tomography (perfusion-CT), indicated that low tumoral perfusion predicted poor local control (LC), when stratified according to the median perfusion value [
2]. This patient group was now used for further exploration.
Our first hypothesis was that CT-determined tumoral perfusion could be a surrogate marker for hypoxia, which is an important negative prognostic factor in radiation treatment of HNSCC [
3,
4].
For this purpose, we evaluated the correlation of tumoral perfusion with the expression of carbonic anhydrase IX (CA IX) and glucose transporter-1 protein (GLUT-1) using immunohistochemistry. These antigens are possible intrinsic markers of hypoxia, both mainly controlled by the hypoxia inducible factor-1 (HIF-1) pathway. Their clinical relevance to assess the expression pattern and the relation with hypoxia has been suggested by others for several tumors [
5‐
8].
CA IX is a transmembrane carbonic anhydrase involved in the reversible hydration of carbon dioxide to carbonic acid. It is a widely investigated protein, overexpressed in many tumor types [
9]. In HNSCC, its expression is associated with worse prognosis [
6,
10].
GLUT-1 mediates cellular glucose uptake and thus facilitates anaerobic glycolysis. This protein is largely undetectable in normal epithelium and benign tumors, but is expressed in several types of human carcinomas such as HNSCC and associated with poor prognosis [
5,
7,
8,
11].
Correlation was assessed between tumoral perfusion and each of these intrinsic markers, as well as between CA IX and GLUT-1 mutually. The prognostic value of these immunohistochemical parameters was also evaluated.
Our second hypothesis was that the CT-determined tumoral perfusion could be linked with the measurement of vascular endothelial growth factor (VEGF) and interleukin-6 (IL-6).
VEGF is a pro-angiogenic cytokine known to stimulate the proliferation and migration of endothelial cells and to promote vessel permeability. It is one of the hypoxia-responsive genes, upregulated by increased levels of HIF-1 [
12,
13]. VEGF serum levels are elevated in patients with HNSCC compared to healthy controls [
12,
14,
15]. A negative prognostic role for circulating VEGF serum levels has been indicated for laryngeal carcinoma [
15].
IL-6 is a proinflammatory cytokine produced by inflammatory cells as well as tumor cells [
14,
16]. In cancer patients, it has a multifunctional role concerning systemic alterations like cachexia as well a loco-regional impact on for example tumor cell motility and intercellular adhesion. Although mainly involved in inflammatory reactions, its expression can be induced by the transcription factor nuclear factor-kappaB (NF-κB) under hypoxic conditions [
17]. Increased serum IL-6 levels have been detected in several tumor types, having a prognostic meaning for some of them [
18‐
20]. The possible involvement of IL-6 in radioresistance was mentioned in several in vitro publications [
21‐
23].
Available serum samples were analysed to evaluate VEGF and IL-6 at pretreatment time using an enzyme linked immunosorbent assay (ELISA).
We intended to evaluate the prognostic value of these serum cytokines, as well as the possible link between tumoral perfusion and the cytokine levels. This link could be established by the hypoxia mediated HIF-1 or NF-κB pathways.
The present work results on the possible prognostic significance of different biological and clinicopathological parameters studied in a retrospective study of patients with HNSCC. Furthermore, the relationship with CT-determined tumoral perfusion was investigated.
Discussion
The tumor microenvironment plays a critical role in tumor aggressiveness and treatment response. Several microregional features have been shown to be relevant for treatment outcome, such as hypoxia, inflammation and angiogenesis. The major question stays whether these markers can be used in an integrated way to stratify patients to certain treatment options [
1].
Starting from a retrospective series of patients with HNSCC treated by radiotherapy ± chemotherapy, we investigated the prognostic significance of several biological and clinicopathological parameters. Earlier established results showed that for this whole patient group, CT-determined tumoral perfusion was correlated with poorer LC [
2].
Two subgroups were now used for further analysis. In the first subgroup of 67 patients, immunohistochemistry for CA IX and GLUT-1 was performed on a representative paraffin-embedded pretreatment biopsy. In the second subgroup of 34 patients, ELISA was used to determine pretreatment VEGF and IL-6 serum levels. The correlation between these parameters mutually, as well as their prognostic value, was investigated.
Hypothesizing that CT-determined tumoral perfusion could be an inverse marker for hypoxia (less perfusion relates to more hypoxia), we assessed the correlation between tumoral perfusion, the intrinsic hypoxia markers CA IX and GLUT-1, and the serum levels of VEGF and IL-6, known to be upregulated under certain microenvironmental conditions like hypoxia.
Besides a borderline inverse association between perfusion and VEGF, no significant correlation was found in our study. There are three possible explanations in support of this finding.
First, it is likely that the dynamic CT measures rather acute hypoxia, while CA IX and GLUT-1, as well as VEGF and IL-6, are more indicative for chronic hypoxia.
Second, the CT-determined tumoral perfusion in this patient group was a measurement that took place through one region of interest, i.e. the maximum diameter of the tumoral region. On the other hand, CA IX and GLUT-1 immunohistochemistry was performed on one tumor biopsy taken at the periphery of the tumor field, and VEGF and IL-6 values were measured as systemic serum levels.
Third, the relationship between perfusion and oxygenation has shown to be in a precious balance [
1,
25].
The borderline association found between perfusion and VEGF can be explained through the hypoxia-hypothesis. Nevertheless, an opposite finding would also have been possible, associating the angiogenetic property directly with perfusion.
The further analysis of the biological and clinicopathological parameters showed that patients with higher nodal stage expressed significantly higher IL-6 serum levels. This experience is in accordance with previous reports linking elevated IL-6 levels to extensive disease [
20], and more specifically, to lymph node involvement [
26]. It is not surprising since this multifunctional cytokine plays an important role in promoting tumorigenesis.
Concerning the prognostic value of the clinicopathological parameters, we found a correlation with outcome for known factors like tumor stage (RC and OS) and N3 stage (LC, RC, DFS, OS). Of these, only the nodal stage appeared to behave as an independent prognostic factor for RC and DFS.
Concerning the prognostic value of the biological parameters, poorer tumor perfusion predicted poorer LC and DFS, as expected [
2]. This correlation was lost in the smallest subgroup, probably due to a too small patient number in this group. Furthermore, in multivariate analysis, the perfusion only kept a borderline prognostic value for outcome.
For CA IX and GLUT-1, the combined assessment of patients with both markers expressed above the median did indicate an independent correlation with worse LC, RC and DFS. This result demonstrates that a combined assessment gives additive information, though we have to note that the prognostic value of each of both markers is not clearly established, given prognostic [
5‐
8,
27] as well as non-prognostic [
28] publications.
In the second subgroup, IL-6 pretreatment serum level stratified according to the median seemed to be the only independent predictor of LC, DFS and OS. VEGF pretreatment levels were not correlated with prognosis.
The prognostic value of IL-6 has been described in many tumor types [
18], but not yet in HNSCC treated by radiotherapy ± chemotherapy. In our multivariate analysis, we found a strong prognostic impact of the IL-6 pretreatment serum level on local control. This might indicate that there is indeed a link between IL-6 and radioresistance, as already shown
in vitro by others [
21‐
23].
VEGF- reports in literature are a bit conflicting, with papers demonstrating [
15] and denying [
29] a prognostic role for VEGF serum levels at pretreatment time.
In this study, due to methodological reasons, we performed serum and immunohistochemical analyses for two subgroups instead of the whole patient group.
The advantage of using serum samples over immunohistochemistry on biopsies is that the ELISA test is quickly and simple to perform. Furthermore, the technique is not too invasive and quite established.
On the other hand we must be aware that HNSCC cells may not be the only source of the elevated serum cytokine levels in patients. The serum levels may also depend in part upon individual host inflammatory responses and conditions like cachexia, and certainly not upon hypoxia only. In our patient group however, we did not see a correlation between VEGF or IL-6 serum levels and sedimentation rates, determined at the same pretreatment time. Furthermore, there are some pitfalls in the measurement of these cytokines [
30]. Concerning VEGF, the isoform specificity and platelet interaction may be of influence. The use of banked frozen serum for analysis may also be perceived as a shortcoming. Because of the retrospective character of this study, we could not control the methods used for taking and storing the serum samples.
Conclusion
In conclusion, we found a prognostic value in this series of patients with HNSCC treated by radiotherapy ± chemotherapy for some of the investigated parameters, with a major role for IL-6. To our knowledge, we are the first to report a link in HNSCC between IL-6 pretreatment serum levels and radioresistance in vivo. This link is supported by the strong prognostic association of pretreatment IL-6 with local control, known to be the most important parameter to judge radiotherapy responses. This finding has to be validated in other, preferably prospective, studies.
Furthermore, the combined assessment of CA IX and GLUT-1 correlated independently with prognosis. This is a valuable indication that a combined approach is important in the investigation of prognostic markers.
We believe that additional to the standard TNM classification this combined assessment of several biological and clinicopathological parameters will show most appropriate to stratify patients to certain treatment protocols. The efforts made to search for prognostic factors in radiotherapy settings will stretch towards a more efficient combination of radiotherapy with other treatment strategies like chemotherapy, radiosensitizers, bioreductive drugs and biological targeting. Therefore, further research will be necessary.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HDS carried out the immunohistochemistry, ELISA's and the acquisition and analysis of data. She also drafted the manuscript. WL participated in the conception of the study and helped to draft the manuscript. EV assisted in the immunohistochemistry performance and scoring technique. LG participated in the scoring process, and the analysis of data. RH performed perfusion-CT and collected data. SN conceived of the study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.