Purpose: To investigate the impact of hypoxia-induced gene expression and cancer stem cell (CSC) marker expression on outcome of postoperative cisplatin-based radiochemotherapy (PORT-C) in patients with locally advanced head and neck squamous cell carcinoma (HNSCC).

Experimental Design: Expression of the CSC markers CD44, MET, and SLC3A2, and hypoxia gene signatures were analyzed in the resected primary tumors using RT-PCR and nanoString technology in a multicenter retrospective cohort of 195 patients. CD44 protein expression was further analyzed in tissue microarrays. Primary endpoint was locoregional tumor control.

Results: Univariate analysis showed that hypoxia-induced gene expression was significantly associated with a high risk of locoregional recurrence using the 15-gene signature (P = 0.010) or the 26-gene signature (P = 0.002). In multivariate analyses, in patients with HPV16 DNA–negative but not with HPV16 DNA–positive tumors the effect of hypoxia-induced genes on locoregional control was apparent (15-gene signature: HR 4.54, P = 0.006; 26-gene signature: HR 10.27, P = 0.024). Furthermore, MET, SLC3A2, CD44, and CD44 protein showed an association with locoregional tumor control in multivariate analyses (MET: HR 3.71, P = 0.016; SLC3A2: HR 8.54, P = 0.037; CD44: HR 3.36, P = 0.054; CD44 protein n/a because of no event in the CD44-negative group) in the HPV16 DNA–negative subgroup.

Conclusions: We have shown for the first time that high hypoxia-induced gene expression and high CSC marker expression levels correlate with tumor recurrence after PORT-C in patients with HPV16 DNA–negative HNSCC. After validation in a currently ongoing prospective trial, these parameters may help to further stratify patients for individualized treatment de-escalation or intensification strategies. Clin Cancer Res; 22(11); 2639–49. ©2016 AACR.

Translational Relevance

Locally advanced, HPV-positive head and neck squamous cell carcinomas (HNSCC) are responding very well to postoperative radiochemotherapy with virtually no recurrences detected in our recent study. To define patient groups for treatment individualization, it is important to establish further markers that could help to identify patients with a favorable prognosis but also with a very high risk of tumor recurrence within the HPV-negative group. In resected tumor specimens obtained prior to radiochemotherapy, we show for the first time that HPV16 DNA–negative, hypoxic tumors are significantly associated with an increased expression of cancer stem cell (CSC) markers and poor locoregional tumor control. In contrast, HPV16 DNA–negative HNSCC with low hypoxia and low CSC marker expression have a favorable locoregional tumor control rate. After validation in prospective trials, these markers may be used for further patient stratification for treatment intensification or de-escalation strategies.

Although treatment efficacy has been improved for patients with locally advanced head and neck squamous cell carcinomas (HNSCC), the overall 5-year survival rate is still about 50% (1, 2). Resected locally advanced HNSCC with high-risk features are routinely treated with postoperative radiochemotherapy (PORT-C) after three randomized independent trials have shown an improved locoregional control and overall survival compared with postoperative radiotherapy alone (3–5). However, tumors respond very heterogeneously to this treatment and biomarkers are needed to identify patient groups who require de-escalated or intensified treatment schedules.

Recently, we have shown in a systematic multicentric retrospective evaluation that the human papillomavirus type 16 deoxyribonucleic acid (HPV16 DNA) status is a strong prognosticator for locoregional control after PORT-C in HNSCC, especially in patients with locally advanced oropharyngeal squamous cell carcinomas (6). HPV-positive tumors are more radio(chemo)sensitive than HPV-negative tumors (7, 8) and locoregional control after PORT-C was achieved in our analysis in virtually all patients with HPV16 DNA–positive tumors (6). In contrast, in patients with HPV16 DNA–negative tumors locoregional control rates of approximately 80% were reached, implying that the HPV16 DNA infection status is not suitable as a sole biomarker for the prediction of locoregional tumor recurrences in these patients. Additional biomarkers are needed for the stratification of HPV16 DNA–negative patients into two groups: those, who are very likely to develop tumor recurrence and may profit from treatment intensification and into those, who are not expected to develop tumor recurrences.

Tumor hypoxia has been shown to be negatively associated with radiotherapy treatment response (9, 10). Nordsmark and colleagues showed that pretreatment tumor hypoxia in HNSCC with partial oxygen pressure (pO2) values of less than 2.5 mmHg is associated with significantly lower locoregional tumor control rates compared with patients with higher tumor oxygenation values (11). In a multi-institutional study on 397 patients with locally advanced HNSCC, Nordsmark and colleagues found that pretreatment tumor oxygenation is a highly significant prognostic factor for survival after primary radiotherapy applied alone or combined with chemotherapy, surgery, or radiation sensitizers (12). In recent years, hypoxia detection using [(18F)]-fluoroazomycin arabinoside (FAZA) PET/CT and [(18F)]-fluoromisonidazole (FMISO) PET/CT imaging have been found to have a prognostic potential in HNSCC (13, 14).

The molecular response of the tumor to hypoxia can also be analyzed on the transcriptional level. Several gene classifiers have been developed as a measure of tumor hypoxia in biopsies routinely taken for diagnosis (15–18). Toustrup and colleagues developed a 15-gene classifier for patients with HNSCC who received primary radiotherapy (15). This gene classifier is based on studies using oxygen electrode measurements in xenograft models and in an independent training set of 58 patients with HNSCC. It is therefore associated with the oxygen levels in tumors and has been shown to correlate with the benefit from hypoxic cell sensitization during radiotherapy. Another highly prognostic hypoxia metagene signature was developed by Buffa and colleagues for patients with head and neck, breast, and lung cancers (18). The 26-gene classifier by Eustace and colleagues, that is based on this metagene signature (18), has been shown to predict the benefit from hypoxia-modifying treatment in laryngeal cancer (16). However, as the gene classifiers are based on the measurement of hypoxia-induced gene expression and not on hypoxia itself, the classification by the gene signature does not necessarily correlate with hypoxia assessed by PET (13, 19).

To date, the potential of hypoxia gene signatures to discriminate pretreatment tumor specimen regarding their hypoxia-induced gene expression levels has not been tested in the postoperative setting (15, 16, 18). However, there is some evidence from an experimental study of our laboratory, which showed that pretreatment hypoxia impacts local tumor control after radiotherapy also when radiotherapy was applied under homogenous anoxic conditions (20). This would suggest that hypoxia impacts the outcome of radiotherapy not only by a direct biochemical or radiobiologic effect on cellular radioresistance, but that also other biologic mechanisms are involved. Hypoxia, as an external factor, might also contribute to increased radioresistance of cancer stem cells (CSC). According to the CSC hypothesis, CSCs are the only cells, which are capable of self-renewal and of tumor recurrence after treatment (21). Therefore, the number of CSCs is an important determinant for the radiation dose needed to achieve local tumor control (reviewed in ref. 22) and this may also differ between tumors of the same histopathologic type (23). In addition, CSC-related radiobiologic mechanisms such as repopulation during treatment and recovery from radiation-induced damage between the single fractions have been shown to increase tumor resistance against radiotherapy (24–26).

HPV16 DNA positivity may be used in future prospective clinical trials as a stratification marker for a very good patient outcome after PORT-C in high-risk HNSCC and thereby as a basis for radiotherapy de-escalation strategies. An important question is whether further markers could help to identify patients with a favorable prognosis within the HPV16 DNA–negative group, which would therefore also be candidates for treatment de-escalation strategies. At the same time, additional markers could help to detect a subgroup within the patients with HPV-negative tumors with a very high risk of recurrence, who would need to be included into treatment intensification trials. To further stratify patients with HPV16 DNA–negative HNSCC, we assessed the effects of pretreatment tumor hypoxia and the expression of potential CSC markers (CD44 and CD44 protein, MET and SLC3A2) on the primary endpoint locoregional tumor control in the multicenter retrospective patient cohort of the German Cancer Consortium Radiation Oncology Group (DKTK-ROG).

Patients

The patient eligibility criteria have been described previously (6). Briefly, patients with histologically proven squamous cell carcinoma arising from the oral cavity, oropharynx, or hypopharynx who were treated between 2004 and 2012 in eight DKTK partner sites were included in this study. They all received curatively intended cisplatin-based PORT-C according to standard protocols with a minimum follow-up of 24 months. All patients had a high risk for locoregional tumor recurrence with a tumor stage pT4 and/or >3 positive lymph nodes and/or positive microscopic resection margins and/or extracapsular spread. In total, 221 patients were included in this study, with a median follow-up of 47.3 months. The patient characteristics are shown in Supplementary Table S1A.

Processing of FFPE material

From all DKTK partner sites, formalin-fixed paraffin-embedded (FFPE) material of primary tumor specimens (removed by surgery) were collected centrally and processed under standardized procedures at the DKTK partner site Dresden. All FFPE tissue specimens were subjected to hematoxylin and eosin staining in order to confirm the presence of squamous cell carcinoma.

DNA extraction and PCR array-based analysis of HPV status

Genomic DNA was extracted from 5-μm FFPE tissue sections using the QIAamp DNA FFPE tissue kit (Qiagen GmbH) according to the manufacturer's instructions and stored at −20°C until required. HPV DNA analyses including genotyping were performed using the LCD-Array HPV 3.5 kit (CHIPRON GmbH) according to the manufacturer's instructions. Briefly, PCR was performed using the Primer Mix A (My 11/09) and B (“125”) provided with the LCD-Array HPV 3.5 kit and the HotStarTaq Plus Master Mix (Qiagen GmbH). Hybridization mix including 5 μL of each amplified PCR product A and B were added to each field of the LCD-Array. After staining and washing, the hybridization spots were scanned and analyzed using the SlideReader Software (CHIPRON GmbH). For internal quality control purposes, a positive control (HPV33 DNA, UT-SCC-45 xenograft) and a negative control (RNase free water; Qiagen GmbH) were included in each array. Six tissue samples had to be omitted from HPV DNA analysis due to too low DNA yield.

RNA extraction

Total RNA was extracted from 5-μm FFPE tissue sections (about 400 mm2 tissue area per sample) centrally at the DKTK partner site Dresden using the fully automated Tissue Preparation System (TPS; Siemens Healthcare Diagnostics; for in vivo diagnostic use) according to the manufacturer's instructions. RNA quantity and purity were estimated using the Qubit fluorometer (Life Technologies GmbH), and 25 samples were omitted due to limited tissue material or poor RNA quality. Control RNA (positive control) and sections of tissue-free paraffin blocks (negative control) were included in each RNA extraction set. Their RNA was analyzed in parallel for quality control purposes. For RNA extractions and subsequent analyses by real-time PCR (RT-PCR) and nanoString technology, FFPE samples from all partner sites were randomized. The technical assistants and scientists involved in processing and analyses of the biomaterial were blinded to the clinical data.

nanoString analyses

Gene expression analyses were performed using a custom designed code set and nanoString Elements reagents (nanoString Technologies) including the genes of two hypoxia gene signatures (Supplementary Table S1B) as well as the potential stem cell markers MET, SLC3A2, and CD44. As the CD44 probe design was found to be incorrect, this marker had to be omitted from the analysis. CD44 expression was therefore analyzed by RT-PCR (see: cDNA generation and RT-PCR). Total RNA as well as reporter and capture probes specific to the genes of interest were mixed and incubated at 62°C for 22 hours. Samples were then kept at 4°C for a maximum of 18 hours and subjected to the nCounter system and processed as described elsewhere (27). Raw counts were logarithmized and then normalized to the mean of the internal level of reference genes ACTR3, NDFIP1, RPL37A, B2M, GNB2L1, RPL11, POLR2A or to the reference genes of the corresponding hypoxia gene signatures (Supplementary Table S1B), respectively (15, 16).

cDNA generation and RT-PCR

RNA was converted to cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Life Technologies). Because of low RNA yield, each cDNA was preamplified for genes of interest using the pooled TaqMan gene expression assays according to the manufacturer's protocol (10 cycles). TaqMan gene expression assays were used for both hypoxia classifier and reference genes (15, 16) as well as for CD44. RT-PCR was performed on an ABI StepOne Plus Real-Time PCR System (Life Technologies). For every patient, all RT-PCR reactions were performed on a single PCR plate. Gene expression values were logarithmized and then normalized to the mean of the reference genes ACTR3, NDFIP1, RPL37A, B2M, GNB2L1, RPL11, or the reference genes of each hypoxia gene classifier.

Immunohistochemical analysis of CD44

For immunohistochemical analysis, tissue microarrays (TMA) of primary HNSCC specimen were generated. They consisted of up to 3 cores (1-mm diameter each) per tumor representing the tumor heterogeneity. Cores of HNSCC of 195 patients were analyzable for CD44 protein expression. Immunohistochemical staining was performed as described previously (6), with minor modifications. After antigen retrieval (pH 6; Dako) for 35 minutes at 630 W, sections were incubated with the monoclonal mouse anti-human CD44 antibody (dilution 1:500; clone DF1485; Dako) for 30 minutes at room temperature. Negative control slides were incubated with corresponding mouse IgG1 antibody control (Dako). Blinded samples were evaluated semiquantitatively by two independent observers (A. Linge and C. von Neubeck) with an interobserver variability of <5%. CD44 staining intensity was scored (0, +, ++, +++) and tumors with a minimum of one positive core (+, ++, +++) were considered as positive.

Clinical endpoints and statistical analysis

The primary endpoint was locoregional tumor control (LRC). As secondary endpoints, freedom from distant metastases (DM) and overall survival (OS) were evaluated. All endpoints were calculated from the first day of radiotherapy to the date of event or censoring.

The corresponding survival curves were estimated by the Kaplan–Meier method. The impact of potential prognostic variables on the endpoints was evaluated using the Cox regression model in which the respective gene expressions were included as binarized parameters. Parameters found to be significant in univariate analysis were included in a multivariate Cox model. Statistical analyses were performed for all patients and for the subgroup of patients with HPV16 DNA–negative tumors. In the subgroup of patients with HPV16 DNA–positive tumors only two recurrences occurred. Therefore, it was not possible to detect any significant differences in LRC for this subgroup. To obtain the optimal cutoffs for the binarization of the CSC marker expression, for every CSC marker, each possible cutoff leading to a different patient stratification was considered. For each of these cutoffs, univariate Cox regressions with the binarized CSC marker were performed for the subgroup of HPV16 DNA–negative tumors using 10.000 bootstrap samples of the patient cohort stratified for tumor localization. Afterwards, the fraction of significant results (power) was calculated for each cutoff, leading to the optimal cutoff, which has the largest power. For the CSC markers CD44, MET, and SLC3A2 this results in the cut-off values 0.2 (power 73%), −4.135 (power 77%), and −3.135 (power 66%), respectively. For stratification with respect to hypoxia-induced gene expression, tumors were assigned to a less hypoxic or a more hypoxic class, according to low or high expression levels of the corresponding hypoxia gene signature, using two-class k-means clustering based on the Euclidian distance. To compare the survival curves of patients stratified by HPV16 DNA status, CSC marker expression and hypoxia status, log-rank tests were employed. To assess correlations between continuous variables the Pearson correlation coefficient was used. Between binary parameters the mean square contingency coefficient (phi coefficient) was employed. For all analyses, two-sided tests were performed and P values <0.05 were considered statistically significant. All analyses were performed by the SPSS 21 software (IBM Corporation) except for the bootstrapping procedure, which was performed by STATA 11 (StataCorp LP).

Hypoxia gene expression

In this study, tumor hypoxia was assessed using the 15-gene signature developed by Toustrup and colleagues (15) and the 26-gene signature by Eustace and colleagues (16) in parallel. Using nanoString analyses, univariate analysis showed that hypoxic tumors are significantly associated with a high risk of loco-regional tumor recurrence after PORT-C using the 15-gene signature (HR, 3.55; P = 0.010) and the 26-gene signature (HR, 9.37; P = 0.002; Fig. 1; Supplementary Table S2). The 26-gene signature was also significant for the secondary endpoints DM (HR, 2.15; P = 0.046) and OS (HR, 2.48; P = 0.002). For the subgroup of patients with HPV16 DNA –negative tumors, the effect of hypoxia was also significant for both locoregional control and overall survival (15-gene signature: LRC, HR 4.66, P = 0.005; OS, HR 1.88, P = 0.031; 26-gene signature: LRC, HR 11.3, P = 0.017; OS, HR 2.05, P = 0.040; Fig. 1; Supplementary Table S2), whereas no significance has been revealed for patients with HPV-positive tumors (not shown).

Figure 1.

(A–D) Kaplan-Meier estimates of locoregional tumor control of (A and C) all patients or (B and D) patients with HPV16 DNA–negative HNSCC. Patients with tumors expressing low hypoxia gene levels showed significantly better locoregional tumor control rates compared to patients with highly hypoxic tumors.

Figure 1.

(A–D) Kaplan-Meier estimates of locoregional tumor control of (A and C) all patients or (B and D) patients with HPV16 DNA–negative HNSCC. Patients with tumors expressing low hypoxia gene levels showed significantly better locoregional tumor control rates compared to patients with highly hypoxic tumors.

Close modal

The results of the multivariate analyses of the total patient population are summarized in Supplementary Table S3. The multivariate analyses also included extracapsular extension (ECE) status (significant for secondary endpoints) and tumor localization, because these variables were significant prognostic factors in our previous analysis (6). Both hypoxia gene classifiers revealed a significant association of tumors classified as “hypoxic” with poor locoregional control (15-gene classifier: HR 3.73, P = 0.008; 26-gene classifier: HR 6.00, P = 0.017; Supplementary Table S3).

In patients with HPV16 DNA–negative tumors, the effect of hypoxia on locoregional control was apparent in multivariate analysis (15-gene classifier: HR 4.54, P = 0.006; 26-gene classifier: HR 10.27, P = 0.024; Table 1). Comparing the two hypoxia classifiers, the 26-gene signature was stronger correlated with HPV status (R = −0.271, P < 0.001) and therefore may have less power regarding patient stratification compared with the 15-gene signature (R = 0.007, P = 0.92). As nanoString analysis is a relatively new technique for analyzing gene expression levels, hypoxia gene expression levels of both classifiers were validated by RT-PCR and revealed similar results (not shown). The results of both techniques were highly correlated as expected (Supplementary Table S4).

Table 1.

Multivariate analyses of hypoxia gene signatures and additional prognostic factors for patients with HPV16 DNA–negative tumors only

Locoregional controlDistant metastasesOverall survival
HR (95% CI)PHR (95% CI)PHR (95% CI)P
15-gene signature 4.54 (1.56–13.2) 0.006 1.42 (0.67–3.01) 0.356 1.80 (1.01–3.21) 0.047 
ECE 1.18 (0.52–2.69) 0.694 2.77 (1.22–6.28) 0.015 1.81 (1.02–3.20) 0.042 
Oral cavitya 1.81 (0.76–4.27) 0.178 1.63 (0.71–3.74) 0.251 1.54 (0.86–2.75) 0.144 
Hypopharynxa 0.69 (0.19–2.54) 0.581 2.05 (0.78–5.38) 0.146 0.63 (0.26–1.55) 0.317 
26-gene signature 10.27 (1.36–77.2) 0.024 1.53 (0.63–3.69) 0.347 1.84 (0.90–3.77) 0.095 
ECE 1.39 (0.62–3.15) 0.428 2.83 (1.25–6.39) 0.013 1.93 (1.09–3.42) 0.025 
Oral cavitya 1.51 (0.65–3.52) 0.337 1.59 (0.69–3.66) 0.278 1.46 (0.81–2.62) 0.209 
Hypopharynxa 0.95 (0.26–3.49) 0.944 2.17 (0.82–5.75) 0.118 0.71 (0.29–1.76) 0.462 
Locoregional controlDistant metastasesOverall survival
HR (95% CI)PHR (95% CI)PHR (95% CI)P
15-gene signature 4.54 (1.56–13.2) 0.006 1.42 (0.67–3.01) 0.356 1.80 (1.01–3.21) 0.047 
ECE 1.18 (0.52–2.69) 0.694 2.77 (1.22–6.28) 0.015 1.81 (1.02–3.20) 0.042 
Oral cavitya 1.81 (0.76–4.27) 0.178 1.63 (0.71–3.74) 0.251 1.54 (0.86–2.75) 0.144 
Hypopharynxa 0.69 (0.19–2.54) 0.581 2.05 (0.78–5.38) 0.146 0.63 (0.26–1.55) 0.317 
26-gene signature 10.27 (1.36–77.2) 0.024 1.53 (0.63–3.69) 0.347 1.84 (0.90–3.77) 0.095 
ECE 1.39 (0.62–3.15) 0.428 2.83 (1.25–6.39) 0.013 1.93 (1.09–3.42) 0.025 
Oral cavitya 1.51 (0.65–3.52) 0.337 1.59 (0.69–3.66) 0.278 1.46 (0.81–2.62) 0.209 
Hypopharynxa 0.95 (0.26–3.49) 0.944 2.17 (0.82–5.75) 0.118 0.71 (0.29–1.76) 0.462 

NOTE: P values <0.05 were considered statistically significant.

Abbreviation: ECE, extracapsular extension.

aBaseline oropharynx.

CSC marker expression

The potential CSC markers MET and SLC3A2 were included in the nanoString gene panel. CD44 was analyzed using RT-PCR and its protein expression was assessed by IHC. In univariate analysis, MET, SLC3A2, CD44, and CD44 protein were significantly associated with the primary endpoint locoregional tumor control (MET: HR 5.19, P = 0.001; SLC3A2: HR 6.54, P = 0.002; CD44 (RT-PCR): HR 3.56, P = 0.01; CD44 (IHC): HR 9.09, P = 0.03; Fig. 2, Supplementary Table S5). In contrast to CD44 and CD44 protein, MET and SLC3A2 were also significantly associated with increased distant metastases and decreased overall survival (Supplementary Table S5). Similar effects for locoregional control were observed in the subgroup of HPV16 DNA–negative tumors for MET, SLC3A2, CD44, and CD44 protein (Supplementary Table S5; Fig. 2). As the CSC markers are correlated (Supplementary Table S6A), four Cox models, each including one CSC marker, were performed for multivariate analyses in the total patient cohort (Supplementary Table S6B). In this multivariate analysis, MET and SLC3A2 showed a strong association with locoregional control (MET: HR 3.61, P = 0.011; SLC3A2: HR 3.61, P = 0.045) and with distant metastases (MET: HR 2.78, P = 0.01; SLC3A2: HR 3.75, P = 0.006). Both CD44 and CD44 protein did not reveal a significant association with any of the three endpoints in the total patient cohort. However, in patients with HPV16 DNA–negative tumors, MET, SLC3A2, CD44, and CD44 protein showed an association with the primary endpoint locoregional tumor control (MET: HR 3.71, P = 0.016; SLC3A2: HR 8.54, P = 0.037; CD44: HR 3.36, P = 0.054; CD44 protein: n/a because of no event in the CD44-negative group; Table 2).

Figure 2.

A, immunohistochemical staining of CD44 showing different staining intensities. All staining intensities (+, ++, +++) were considered as positive staining. B–I, Kaplan–Meier estimates of locoregional tumor control of CD44 protein (assessed by IHC) and CD44, MET, and SLC3A2 gene expression of (B, D, F, H) all patients or (C, E, G, I) patients with HPV16 DNA–negative HNSCC. Patients with little or no CSC marker expression showed better locoregional tumor control compared with patients with stronger CSC marker expression.

Figure 2.

A, immunohistochemical staining of CD44 showing different staining intensities. All staining intensities (+, ++, +++) were considered as positive staining. B–I, Kaplan–Meier estimates of locoregional tumor control of CD44 protein (assessed by IHC) and CD44, MET, and SLC3A2 gene expression of (B, D, F, H) all patients or (C, E, G, I) patients with HPV16 DNA–negative HNSCC. Patients with little or no CSC marker expression showed better locoregional tumor control compared with patients with stronger CSC marker expression.

Close modal
Table 2.

Multivariate analyses of CSC markers and additional prognostic factors for patients with HPV16 DNA–negative tumors only

Locoregional controlDistant metastasesOverall survival
HR (95% CI)PHR (95% CI)PHR (95% CI)P
MET 3.71 (1.28–10.8) 0.016 2.83 (1.16–6.92) 0.022 2.52 (1.32–4.81) 0.005 
ECE 1.30 (0.58–2.94) 0.524 2.83 (1.25–6.42) 0.013 1.96 (1.10–3.49) 0.022 
Oral cavitya 1.81 (0.78–4.20) 0.169 1.65 (0.73–3.77) 0.232 1.52 (0.86–2.71) 0.152 
Hypopharynxa 0.70 (0.19–2.57) 0.594 2.06 (0.78–5.41) 0.144 0.61 (0.25–1.51) 0.285 
SLC3A2 8.54 (1.14–63.9) 0.037 4.61 (1.35–15.8) 0.015 1.44 (0.72–2.88) 0.297 
ECE 1.43 (0.64–3.23) 0.386 3.23 (1.41–7.39) 0.005 1.93 (1.08–3.42) 0.025 
Oral cavitya 1.60 (0.69–3.74) 0.274 1.43 (0.63–3.27) 0.397 1.55 (0.87–2.78) 0.141 
Hypopharynxa 0.86 (0.24–3.17) 0.823 2.66 (1.00–7.10) 0.051 0.69 (0.28–1.70) 0.419 
CD44 3.36 (0.98–11.5) 0.054 1.62 (0.65–4.02) 0.300 1.11 (0.58–2.11) 0.757 
ECE 1.41 (0.64–3.10) 0.398 2.84 (1.25–6.48) 0.013 1.87 (1.05–3.31) 0.033 
Oral cavitya 1.71 (0.74–3.94) 0.211 1.57 (0.68–3.64) 0.295 1.57 (0.86–2.86) 0.140 
Hypopharynxa 0.81 (0.22–2.97) 0.751 1.79 (0.64–5.02) 0.267 0.52 (0.20–1.36) 0.181 
CD44 — ** 2.43 (0.57–10.4) 0.231 1.30 (0.51–3.31) 0.584 
ECE — — 3.09 (1.40–6.82) 0.005 1.86 (1.07–3.24) 0.029 
Oral cavitya — — 1.63 (0.72–3.70) 0.246 1.55 (0.87–2.76) 0.136 
Hypopharynxa — — 2.84 (1.14–7.04) 0.025 0.65 (0.28–1.51) 0.315 
Locoregional controlDistant metastasesOverall survival
HR (95% CI)PHR (95% CI)PHR (95% CI)P
MET 3.71 (1.28–10.8) 0.016 2.83 (1.16–6.92) 0.022 2.52 (1.32–4.81) 0.005 
ECE 1.30 (0.58–2.94) 0.524 2.83 (1.25–6.42) 0.013 1.96 (1.10–3.49) 0.022 
Oral cavitya 1.81 (0.78–4.20) 0.169 1.65 (0.73–3.77) 0.232 1.52 (0.86–2.71) 0.152 
Hypopharynxa 0.70 (0.19–2.57) 0.594 2.06 (0.78–5.41) 0.144 0.61 (0.25–1.51) 0.285 
SLC3A2 8.54 (1.14–63.9) 0.037 4.61 (1.35–15.8) 0.015 1.44 (0.72–2.88) 0.297 
ECE 1.43 (0.64–3.23) 0.386 3.23 (1.41–7.39) 0.005 1.93 (1.08–3.42) 0.025 
Oral cavitya 1.60 (0.69–3.74) 0.274 1.43 (0.63–3.27) 0.397 1.55 (0.87–2.78) 0.141 
Hypopharynxa 0.86 (0.24–3.17) 0.823 2.66 (1.00–7.10) 0.051 0.69 (0.28–1.70) 0.419 
CD44 3.36 (0.98–11.5) 0.054 1.62 (0.65–4.02) 0.300 1.11 (0.58–2.11) 0.757 
ECE 1.41 (0.64–3.10) 0.398 2.84 (1.25–6.48) 0.013 1.87 (1.05–3.31) 0.033 
Oral cavitya 1.71 (0.74–3.94) 0.211 1.57 (0.68–3.64) 0.295 1.57 (0.86–2.86) 0.140 
Hypopharynxa 0.81 (0.22–2.97) 0.751 1.79 (0.64–5.02) 0.267 0.52 (0.20–1.36) 0.181 
CD44 — ** 2.43 (0.57–10.4) 0.231 1.30 (0.51–3.31) 0.584 
ECE — — 3.09 (1.40–6.82) 0.005 1.86 (1.07–3.24) 0.029 
Oral cavitya — — 1.63 (0.72–3.70) 0.246 1.55 (0.87–2.76) 0.136 
Hypopharynxa — — 2.84 (1.14–7.04) 0.025 0.65 (0.28–1.51) 0.315 

NOTE: P values <0.05 were considered statistically significant.

Abbreviation: ECE, extracapsular extension.

aBaseline oropharynx.

**, as there were no events in the CD44 protein negative group, the Cox model did not converge.

In addition, patients were stratified by the smoking status during therapy. Except for the potential CSC marker MET (R = 0.21, P = 0.004) there were no significant correlations between smoking and CSC marker expression or hypoxia. As in univariate analysis smoking had no impact on the clinical endpoints, this factor was not considered for multivariate analyses.

Association of tumor hypoxia and cancer stem cells

To assess whether the hypoxic microenvironment is contributing to the radio(chemo)resistance associated with CSC marker expression, the correlation between hypoxia-induced gene expression and CSCs was analyzed. Interestingly, hypoxic tumors showed a strong positive association with CSC marker expression, whereas tumors with low hypoxia showed much more heterogeneous CSC marker expression in the total patient cohort (Table 3). Similar correlation coefficients between hypoxia and CSC marker expression were also obtained within the subgroups of patients with HPV16 DNA–positive or negative tumors (not shown).

Table 3.

Correlation of CSC marker expression and tumor hypoxia for all patients

15-gene signature26-gene signature
HypoxiaHypoxia
LowHighLowHigh
CD44 
 Low 21 16 23 14 
 High 48 96 42 102 
 R = 0.194 P = 0.009 R = 0.277 P < 0.001 
CD44 
 Low 39 37 49 27 
 High 38 74 19 93 
 R = 0.174 P = 0.017 R = 0.485 P < 0.001 
MET 
 Low 52 42 59 35 
 High 27 75 13 89 
 R = 0.294 P < 0.001 R = 0.518 P < 0.001 
SLC3A2 
 Low 44 33 55 22 
 High 35 84 17 102 
 R = 0.276 P < 0.001 R = 0.579 P < 0.001 
15-gene signature26-gene signature
HypoxiaHypoxia
LowHighLowHigh
CD44 
 Low 21 16 23 14 
 High 48 96 42 102 
 R = 0.194 P = 0.009 R = 0.277 P < 0.001 
CD44 
 Low 39 37 49 27 
 High 38 74 19 93 
 R = 0.174 P = 0.017 R = 0.485 P < 0.001 
MET 
 Low 52 42 59 35 
 High 27 75 13 89 
 R = 0.294 P < 0.001 R = 0.518 P < 0.001 
SLC3A2 
 Low 44 33 55 22 
 High 35 84 17 102 
 R = 0.276 P < 0.001 R = 0.579 P < 0.001 

The 15-gene but not the 26-gene signature was able to further stratify patients with HPV16 DNA–negative/high CSC marker–expressing tumors into a group with significantly lower locoregional tumor control rate and a more favorable group. In univariate analysis, this was true for the CSC markers SLC3A2 (P = 0.03, Fig. 3A) and MET (P = 0.046, Fig. 3B), but not for CD44 or CD44 protein. Multivariate analyses including HPV16 DNA, hypoxia, and CSC markers showed a significant independent correlation of MET and the 15-gene signature with locoregional tumor control, while the other CSC markers CD44, CD44 protein and SLC3A2 were not significant (Supplementary Table S7). The Cox model including hypoxia and MET fitted the data significantly better than the model with hypoxia alone (P = 0.026, likelihood-ratio test of nested models). Combining patients with HPV-negative tumors, which simultaneously are hypoxic and have a high CSC marker expression, gives a patient subgroup with the highest risk of recurrence (Fig. 3C and D).

Figure 3.

Kaplan–Meier estimates of locoregional tumor control of patients with (A and B) HPV16 DNA–negative HNSCC and high expression levels of SLC3A2 (A) or MET (B). Patients with low hypoxic tumors have better locoregional control rates compared with patients with highly hypoxic tumors. C and D, patients with high risk of tumor recurrence (HPV16 DNA–negative, CSC marker positive, high hypoxia) show significantly poor locoregional control rates than the patients with HPV16 DNA positive, low CSC marker expressing and low hypoxic tumors.

Figure 3.

Kaplan–Meier estimates of locoregional tumor control of patients with (A and B) HPV16 DNA–negative HNSCC and high expression levels of SLC3A2 (A) or MET (B). Patients with low hypoxic tumors have better locoregional control rates compared with patients with highly hypoxic tumors. C and D, patients with high risk of tumor recurrence (HPV16 DNA–negative, CSC marker positive, high hypoxia) show significantly poor locoregional control rates than the patients with HPV16 DNA positive, low CSC marker expressing and low hypoxic tumors.

Close modal

Association of tumor hypoxia and CSCs with HPV16 DNA status

To assess the impact of the HPV16 DNA status on tumor hypoxia and CSC marker expression, their correlation was analyzed. A significant correlation of the HPV16 DNA status with tumor hypoxia was found for the 26-gene signature (R = −0.271, P < 0.001) but not for the 15-gene signature (R = 0.007, P = 0.918; Supplementary Table S8A). Specifically, in the group of HPV16 DNA–negative HNSCC, the majority of the tumors were found to be hypoxic, whereas for HPV16 DNA–positive tumors similar numbers of more hypoxic and less hypoxic tumors were seen.

The expression of the potential CSC markers CD44, SLC3A2, and MET was also significantly correlated with HPV16 DNA status (CD44: R = −0.399, P < 0.001; MET: R = −0.319, P < 0.001; SLC3A2: R = −0.430, P < 0.001; Supplementary Table S8B).

Previously, we and others have shown that the HPV infection status is a strong prognosticator for locoregional control in patients with locally advanced HNSCC who received postoperative radio(chemo)therapy (6, 28, 29) with an improved locoregional control and radiosensitivity of HPV16 DNA positive in comparison with HPV16 DNA–negative tumors. In our previous multicenter evaluation, we found virtually no recurrences in the HPV16 DNA–positive group among high-risk HNSCC patients who received PORT-C (6). Patients in this highly advantageous group may therefore be candidates for potential radiotherapy de-escalation trials. However, for further stratification of patients with HPV-negative HNSCC, additional biomarkers are required to predict locoregional control to identify additional patients who could be assigned to the very good prognosis group and to define a patient group with very unfavorable outcome that could be a candidate for treatment intensification strategies.

The oxygenation status of the tumors is one of the known biomarkers for outcome of primary radiotherapy in HNSCC, that is, of macroscopic tumors, which have not undergone surgical removal. Here, we aimed to determine the relevance of the hypoxia status using the hypoxia gene signatures by Toustrup and colleagues (15) and Eustace and colleagues (16) in the tumor specimen obtained by surgical resection before PORT-C. We found that the pretreatment hypoxia status was prognostic in patients with PORT-C. Patients with high hypoxia levels in the surgical specimen obtained prior to PORT-C had a poor locoregional tumor control compared with patients with low hypoxic tumors. This effect mainly attributes to the subgroup of HPV-negative HNSCC. Taking into account that hypoxia was evaluated within the surgical specimen and radiochemotherapy was applied to the potential residual tumor cells after surgery, the finding of a significant association of hypoxia within the resected primary tumor with outcome of PORT-C at a first glance is surprising. As it is unlikely that the low number of residual tumor cells after surgery also differs in hypoxia, this finding suggests that hypoxia impacts outcome of radiotherapy not only by a direct biochemical or radiobiologic effect on cellular radioresistance of the tumor but also by other radiobiologic mechanisms. This would be in line with previous experimental data from our laboratory, showing that pretreatment hypoxia impacts local tumor control after radiotherapy also when radiotherapy was applied under homogeneous anoxic conditions (20). As recent studies suggested that tumor hypoxia also favors stemness and invasive growth as an external factor (reviewed in refs. 30–33), we explored the role of potential CSC markers in our cohort. CSCs are known to play a major role in radioresistance (reviewed in ref. 34), and a highly hypoxic tumor microenvironment contributes to an increased clonogenic potential (20, 35).

CD44 is a widely explored CSC marker in HNSCC (30, 36, 37). A study by de Jong and colleagues showed that CD44 and CD44 protein levels significantly predict local recurrence after radiotherapy in patients with early-stage laryngeal cancers (38). Here, we showed that patients with surgically resected tumors with no detectable CD44 protein expression have increased locoregional tumor control rates compared with those with CD44 protein–expressing tumors.

Recently, CD98 has been established experimentally as a putative CSC marker (39). In a study on 711 patients with oropharyngeal squamous cell carcinoma, HPV-positive tumors have been shown to express less CSC markers such as CD44 and CD98, whereas CD44 and CD98 positivity was associated with significantly lower progression-free and OS (40). In our patient cohort, we found that patients with high expression of SLC3A2, which is encoding for one of the CD98 heterodimers, show a poor locoregional tumor control and an increased risk for distant metastases also after PORT-C.

The MET pathway has been shown to promote self-renewal and tumorigenicity in HNSCC stem-like cells (41) and pharmacologic selective inhibition of MET has been shown to lead to elimination of CSCs (42). Furthermore, MET was found to be associated with poor prognosis in patients with locally advanced p16-negative HNSCC who were treated with primary radiochemotherapy (43). In the current study, we have shown that MET overexpressing HPV16 DNA–negative HNSCC are associated with poor locoregional tumor control and increased distant metastases after PORT-C. Interestingly, Pennacchietti and colleagues showed that hypoxia activates the transcription of the MET proto-oncogene, that MET overexpression is associated with hypoxic areas of tumors and that MET inhibition prevents hypoxia-induced cell growth (44). Our data, showing a positive correlation of MET with tumor hypoxia, support this association of both factors. Still, the relatively low number of low hypoxic tumors among the HPV16 DNA–negative/MET positive group showed a significantly higher locoregional control rate compared with the highly hypoxic MET–positive tumors. In addition, our data indicate that, in patients with HPV16 DNA–negative tumors, SLC3A2 and MET may be suitable markers to determine a group of patients who have a very poor prognosis and therefore may profit from treatment intensification.

While the hypoxia profiles used in this study were prognostic for locoregional tumor control, they were not indicative for an increased risk of distant metastases (Table 1). This appears counterintuitive in light of experimental and clinical data showing that hypoxia may drive distant metastases (31, 45–47). However, to our best knowledge, this so far has not been investigated in HNSCC in the postoperative setting. Some of the genes in the hypoxia profiles utilized in our study might also be associated with other biologic phenomenons, including stemness of cancer cells. The coefficients of correlation between the different CSC markers and hypoxia profiles (Table 3) indicate that, although a significant correlation exists, both parameters to a large extent are expressed independent of each other. In contrast to the hypoxia profiles, CSC marker expression correlated in our study not only with locoregional control but also with distant metastases. Expression of CSC markers has been suggested as a potential surrogate of CSC density, that is, of the number of cells at risk to metastasize per given volume of tumor tissue (30, 34, 36, 46, 48, 49). Further mechanistic investigations into the relationship of hypoxia, stemness, and metastatic risk in HNSCC, in correlation with clinical data, appears therefore to be an interesting avenue for better stratification of patients for systemic therapies and for discovery of novel targets. An important particularity of our study is that biomarkers have been investigated in a cohort of patients treated postoperatively by radiochemotherapy, with overall higher locoregional control rates compared with studies evaluating primary radiochemotherapy in HPV-negative or -positive HNSCC (50). Currently, it is unknown whether the predictive potential of biomarkers differs for the different clinical risk groups receiving postoperative or primary radiochemotherapy. To approach this question, the latter risk group of patients is currently evaluated using the same biomarkers studied here by the DKTK-ROG.

Taken together, this is the first systematic multicentric analysis showing an association of high levels of tumor hypoxia and CSC marker expression within the surgically removed primary tumor with impaired locoregional tumor control after PORT-C in HNSCC. In HPV16 DNA–negative HNSCC, negativity for these parameters may help to identify a subgroup of patients with locoregional control rates as high as observed in HPV16 DNA–positive tumors. In addition, hypoxia and CSC marker positive tumors seem to constitute a subgroup of patients that is undertreated with current standard PORT-C. After validation in a currently ongoing prospective study, these parameters may help to further stratify patients for individualized treatment de-escalation or intensification strategies.

I. Tinhofer reports receiving commercial research grants from Merck Serono and Pfizer and is a consultant/advisory board member for Merck Serono. C. Belka reports receiving a commercial research grant and speakers bureau honoraria from Merck Darmstadt. J. Alsner and J. Overgaard are listed as a co-inventors on a provisional patent application on a method for determining clinically relevant hypoxia in cancer that is owned by Aarhus University, Aarhus, Denmark, and the part concerning prediction of benefit from Nimorazole is licensed to Azanta Denmark A/S, Hellerup, Denmark. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Linge, A. Abdollahi, J. Debus, I. Tinhofer, C. Rödel, C. Belka, S.E. Combs, D. Zips, M. Baumann, M. Krause

Development of methodology: A. Linge, S. Löck, F. Lohaus, C. von Neubeck, A. Abdollahi, J. Debus, J. Alsner, J. Overgaard, M. Baumann, M. Krause

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Linge, M. Baumann, M. Krause, V. Gudziol, A. Nowak, F. Lohaus, C. von Neubeck, A. Abdollahi, J. Debus, I. Tinhofer, V. Budach, A. Sak, M. Stuschke, P. Balermpas, C. Rödell, M. Avlar, A.L. Grosu, C. Bayer, C. Belka, S. Pigorsch, S.E. Combs, S. Welz, D. Zips, D.E. Aust, G.B. Baretton, J. Overgaard

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Linge, S. Löck, M. Jütz, J. Debus, I. Tinhofer, C. Rödel, S.E. Combs, D. Zips, F. Buchholz, H.D. Thames, M. Baumann, M. Krause

Writing, review, and/or revision of the manuscript: A. Linge, S. Löck, V. Gudziol, F. Lohaus, M. Jütz, J. Debus, I. Tinhofer, V. Budach, P. Balermpas, C. Rödel, M. Avlar, C. Belka, S.E. Combs, S. Welz, D. Zips, D.E. Aust, G.B. Baretton, A. Dubrovska, J. Alsner, J. Overgaard, M. Baumann, M. Krause

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Linge, V. Gudziol, F. Lohaus, A. Abdollahi, A. Sak, M. Stuschke, P. Balermpas, M. Avlar, A.L. Grosu, C. Bayer, S. Pigorsch, D. Zips, F. Buchholz, D.E. Aust, G.B. Baretton

Study supervision: J. Debus, M. Baumann, M. Krause

Other (wet lab work): A. Linge, M. Jutz

Other (surgical treatment): A. Nowak

The authors gratefully acknowledge the excellent technical assistance by Mrs. Sigrid Balschukat and Mrs. Liane Stolz-Kieslich. The authors also thank all pathologists, head and neck surgeons, and maxillofacial surgeons at the eight treatment centers who provided tumor material and data for this study.

The study was financed by a Joint Funding Grant within the German Cancer Consortium (DKTK) awarded to the DKTK-ROG (principle investigator: Michael Baumann). The DKTK is funded as one of the National German Health Centers by the Federal German Ministry of Education and Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Ferlay
J
,
Soerjomataram
I
,
Dikshit
R
,
Eser
S
,
Mathers
C
,
Rebelo
M
, et al
Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012
.
Int J Cancer
2015
;
136
:
E359
86
.
2.
Leemans
CR
,
Braakhuis
BJ
,
Brakenhoff
RH
. 
The molecular biology of head and neck cancer
.
Nat Rev Cancer
2011
;
11
:
9
22
.
3.
Bernier
J
,
Domenge
C
,
Ozsahin
M
,
Matuszewska
K
,
Lefebvre
JL
,
Greiner
RH
, et al
Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer
.
N Engl J Med
2004
;
350
:
1945
52
.
4.
Cooper
JS
,
Pajak
TF
,
Forastiere
AA
,
Jacobs
J
,
Campbell
BH
,
Saxman
SB
, et al
Postoperative concurrent radiotherapy and chemotherapy for high-risk squamous-cell carcinoma of the head and neck
.
N Engl J Med
2004
;
350
:
1937
44
.
5.
Fietkau
R
,
Lautenschläger
C
,
Sauer
R
,
Dunst
J
,
Becker
A
,
Baumann
M
, et al
Postoperative concurrent radiochemotherapy versus radiotherapy in high-risk SCCA of the head and neck: results of the German phase III trial ARO 96-3
.
J Clin Oncol
2006
;
24
(
suppl 18
):
5507
.
6.
Lohaus
F
,
Linge
A
,
Tinhofer
I
,
Budach
V
,
Gkika
E
,
Stuschke
M
, et al
HPV16 DNA status is a strong prognosticator of loco-regional control after postoperative radiochemotherapy of locally advanced oropharyngeal carcinoma: results from a multicentre explorative study of the German Cancer Consortium Radiation Oncology Group (DKTK-ROG)
.
Radiother Oncol
2014
;
113
:
317
23
.
7.
Kimple
RJ
,
Smith
MA
,
Blitzer
GC
,
Torres
AD
,
Martin
JA
,
Yang
RZ
, et al
Enhanced radiation sensitivity in HPV-positive head and neck cancer
.
Cancer Res
2013
;
73
:
4791
800
.
8.
Gupta
AK
,
Lee
JH
,
Wilke
WW
,
Quon
H
,
Smith
G
,
Maity
A
, et al
Radiation response in two HPV-infected head-and-neck cancer cell lines in comparison to a non-HPV-infected cell line and relationship to signaling through AKT
.
Int J Radiat Oncol Biol Phys
2009
;
74
:
928
33
.
9.
Fyles
AW
,
Milosevic
M
,
Wong
R
,
Kavanagh
MC
,
Pintilie
M
,
Sun
A
, et al
Oxygenation predicts radiation response and survival in patients with cervix cancer
.
Radiother Oncol
1998
;
48
:
149
56
.
10.
Overgaard
J
. 
Hypoxic radiosensitization: adored and ignored
.
J Clin Oncol
2007
;
25
:
4066
74
.
11.
Nordsmark
M
,
Overgaard
M
,
Overgaard
J
. 
Pretreatment oxygenation predicts radiation response in advanced squamous cell carcinoma of the head and neck
.
Radiother Oncol
1996
;
41
:
31
9
.
12.
Nordsmark
M
,
Bentzen
SM
,
Rudat
V
,
Brizel
D
,
Lartigau
E
,
Stadler
P
, et al
Prognostic value of tumor oxygenation in 397 head and neck tumors after primary radiation therapy. An international multi-center study
.
Radiother Oncol
2005
;
77
:
18
24
.
13.
Mortensen
LS
,
Johansen
J
,
Kallehauge
J
,
Primdahl
H
,
Busk
M
,
Lassen
P
, et al
FAZA PET/CT hypoxia imaging in patients with squamous cell carcinoma of the head and neck treated with radiotherapy: results from the DAHANCA 24 trial
.
Radiother Oncol
2012
;
105
:
14
20
.
14.
Zips
D
,
Zophel
K
,
Abolmaali
N
,
Perrin
R
,
Abramyuk
A
,
Haase
R
, et al
Exploratory prospective trial of hypoxia-specific PET imaging during radiochemotherapy in patients with locally advanced head-and-neck cancer
.
Radiother Oncol
2012
;
105
:
21
8
.
15.
Toustrup
K
,
Sorensen
BS
,
Nordsmark
M
,
Busk
M
,
Wiuf
C
,
Alsner
J
, et al
Development of a hypoxia gene expression classifier with predictive impact for hypoxic modification of radiotherapy in head and neck cancer
.
Cancer Res
2011
;
71
:
5923
31
.
16.
Eustace
A
,
Mani
N
,
Span
PN
,
Irlam
JJ
,
Taylor
J
,
Betts
GN
, et al
A 26-gene hypoxia signature predicts benefit from hypoxia-modifying therapy in laryngeal cancer but not bladder cancer
.
Clin Cancer Res
2013
;
19
:
4879
88
.
17.
Lendahl
U
,
Lee
KL
,
Yang
H
,
Poellinger
L
. 
Generating specificity and diversity in the transcriptional response to hypoxia
.
Nat Rev Genet
2009
;
10
:
821
32
.
18.
Buffa
FM
,
Harris
AL
,
West
CM
,
Miller
CJ
. 
Large meta-analysis of multiple cancers reveals a common, compact and highly prognostic hypoxia metagene
.
Br J Cancer
2010
;
102
:
428
35
.
19.
Gronroos
TJ
,
Lehtio
K
,
Soderstrom
KO
,
Kronqvist
P
,
Laine
J
,
Eskola
O
, et al
Hypoxia, blood flow and metabolism in squamous-cell carcinoma of the head and neck: correlations between multiple immunohistochemical parameters and PET
.
BMC Cancer
2014
;
14
:
876
.
20.
Yaromina
A
,
Thames
H
,
Zhou
X
,
Hering
S
,
Eicheler
W
,
Dorfler
A
, et al
Radiobiological hypoxia, histological parameters of tumour microenvironment and local tumour control after fractionated irradiation
.
Radiother Oncol
2010
;
96
:
116
22
.
21.
Clarke
MF
,
Dick
JE
,
Dirks
PB
,
Eaves
CJ
,
Jamieson
CH
,
Jones
DL
, et al
Cancer stem cells–perspectives on current status and future directions: AACR Workshop on cancer stem cells
.
Cancer Res
2006
;
66
:
9339
44
.
22.
Baumann
M
,
Krause
M
,
Thames
H
,
Trott
K
,
Zips
D
. 
Cancer stem cells and radiotherapy
.
Int J Radiat Biol
2009
;
85
:
391
402
.
23.
Hill
RP
,
Milas
L
. 
The proportion of stem cells in murine tumors
.
Int J Radiat Oncol Biol Phys
1989
;
16
:
513
8
.
24.
Petersen
C
,
Zips
D
,
Krause
M
,
Schone
K
,
Eicheler
W
,
Hoinkis
C
, et al
Repopulation of FaDu human squamous cell carcinoma during fractionated radiotherapy correlates with reoxygenation
.
Int J Radiat Oncol Biol Phys
2001
;
51
:
483
93
.
25.
Petersen
C
,
Zips
D
,
Krause
M
,
Volkel
W
,
Thames
HD
,
Baumann
M
. 
Recovery from sublethal damage during fractionated irradiation of human FaDu SCC
.
Radiother Oncol
2005
;
74
:
331
6
.
26.
Thames
HD
 Jr
,
Withers
HR
,
Peters
LJ
,
Fletcher
GH
. 
Changes in early and late radiation responses with altered dose fractionation: implications for dose-survival relationships
.
Int J Radiat Oncol Biol Phys
1982
;
8
:
219
26
.
27.
Geiss
GK
,
Bumgarner
RE
,
Birditt
B
,
Dahl
T
,
Dowidar
N
,
Dunaway
DL
, et al
Direct multiplexed measurement of gene expression with color-coded probe pairs
.
Nat Biotechnol
2008
;
26
:
317
25
.
28.
Snietura
M
,
Piglowski
W
,
Jaworska
M
,
Mucha-Malecka
A
,
Wozniak
G
,
Lange
D
, et al
Impact of HPV infection on the clinical outcome of p-CAIR trial in head and neck cancer
.
Eur Arch Otorhinolaryngol
2011
;
268
:
721
6
.
29.
Heiduschka
G
,
Grah
A
,
Oberndorfer
F
,
Kadletz
L
,
Altorjai
G
,
Kornek
G
, et al
Improved survival in HPV/p16-positive oropharyngeal cancer patients treated with postoperative radiotherapy
.
Strahlenther Onkol
2015
;
191
:
209
16
.
30.
Bütof
R
,
Dubrovska
A
,
Baumann
M
. 
Clinical perspectives of cancer stem cell research in radiation oncology
.
Radiother Oncol
2013
;
108
:
388
96
.
31.
Hill
RP
,
Marie-Egyptienne
DT
,
Hedley
DW
. 
Cancer stem cells, hypoxia and metastasis
.
Semin Radiat Oncol
2009
;
19
:
106
11
.
32.
Mohyeldin
A
,
Garzon-Muvdi
T
,
Quinones-Hinojosa
A
. 
Oxygen in stem cell biology: a critical component of the stem cell niche
.
Cell Stem Cell
2010
;
7
:
150
61
.
33.
Peitzsch
C
,
Perrin
R
,
Hill
RP
,
Dubrovska
A
,
Kurth
I
. 
Hypoxia as a biomarker for radioresistant cancer stem cells
.
Int J Radiat Biol
2014
;
90
:
636
52
.
34.
Baumann
M
,
Krause
M
,
Hill
R
. 
Exploring the role of cancer stem cells in radioresistance
.
Nat Rev Cancer
2008
;
8
:
545
54
.
35.
Kim
Y
,
Lin
Q
,
Zelterman
D
,
Yun
Z
. 
Hypoxia-regulated delta-like 1 homologue enhances cancer cell stemness and tumorigenicity
.
Cancer Res
2009
;
69
:
9271
80
.
36.
Brunner
TB
,
Kunz-Schughart
LA
,
Grosse-Gehling
P
,
Baumann
M
. 
Cancer stem cells as a predictive factor in radiotherapy
.
Semin Radiat Oncol
2012
;
22
:
151
74
.
37.
Faber
A
,
Barth
C
,
Hormann
K
,
Kassner
S
,
Schultz
JD
,
Sommer
U
, et al
CD44 as a stem cell marker in head and neck squamous cell carcinoma
.
Oncol Rep
2011
;
26
:
321
6
.
38.
de Jong
MC
,
Pramana
J
,
van der Wal
JE
,
Lacko
M
,
Peutz-Kootstra
CJ
,
de Jong
JM
, et al
CD44 expression predicts local recurrence after radiotherapy in larynx cancer
.
Clin Cancer Res
2010
;
16
:
5329
38
.
39.
Martens-de Kemp
SR
,
Brink
A
,
Stigter-van Walsum
M
,
Damen
JM
,
Rustenburg
F
,
Wu
T
, et al
CD98 marks a subpopulation of head and neck squamous cell carcinoma cells with stem cell properties
.
Stem Cell Res
2013
;
10
:
477
88
.
40.
Rietbergen
MM
,
Martens-de Kemp
SR
,
Bloemena
E
,
Witte
BI
,
Brink
A
,
Baatenburg de Jong
RJ
, et al
Cancer stem cell enrichment marker CD98: a prognostic factor for survival in patients with human papillomavirus-positive oropharyngeal cancer
.
Eur J Cancer
2014
;
50
:
765
73
.
41.
Lim
YC
,
Kang
HJ
,
Moon
JH
. 
C-Met pathway promotes self-renewal and tumorigenecity of head and neck squamous cell carcinoma stem-like cell
.
Oral Oncol
2014
;
50
:
633
9
.
42.
Sun
S
,
Liu
S
,
Duan
SZ
,
Zhang
L
,
Zhou
H
,
Hu
Y
, et al
Targeting the c-Met/FZD8 signaling axis eliminates patient-derived cancer stem-like cells in head and neck squamous carcinomas
.
Cancer Res
2014
;
74
:
7546
59
.
43.
Baschnagel
AM
,
Williams
L
,
Hanna
A
,
Chen
PY
,
Krauss
DJ
,
Pruetz
BL
, et al
c-Met expression is a marker of poor prognosis in patients with locally advanced head and neck squamous cell carcinoma treated with chemoradiation
.
Int J Radiat Oncol Biol Phys
2014
;
88
:
701
7
.
44.
Pennacchietti
S
,
Michieli
P
,
Galluzzo
M
,
Mazzone
M
,
Giordano
S
,
Comoglio
PM
. 
Hypoxia promotes invasive growth by transcriptional activation of the met protooncogene
.
Cancer Cell
2003
;
3
:
347
61
.
45.
Clatot
F
,
Gouerant
S
,
Mareschal
S
,
Cornic
M
,
Berghian
A
,
Choussy
O
, et al
The gene expression profile of inflammatory, hypoxic and metabolic genes predicts the metastatic spread of human head and neck squamous cell carcinoma
.
Oral Oncol
2014
;
50
:
200
7
.
46.
Sullivan
R
,
Graham
CH
. 
Hypoxia-driven selection of the metastatic phenotype
.
Cancer Metastasis Rev
2007
;
26
:
319
31
.
47.
Chan
DA
,
Giaccia
AJ
. 
Hypoxia, gene expression, and metastasis
.
Cancer Metastasis Rev
2007
;
26
:
333
9
.
48.
Krause
M
,
Yaromina
A
,
Eicheler
W
,
Koch
U
,
Baumann
M
. 
Cancer stem cells: targets and potential biomarkers for radiotherapy
.
Clin Cancer Res
2011
;
17
:
7224
9
.
49.
Baumann
M
,
Krause
M
. 
CD44: a cancer stem cell-related biomarker with predictive potential for radiotherapy
.
Clin Cancer Res
2010
;
16
:
5091
3
.
50.
Kimple
RJ
,
Harari
PM
. 
The prognostic value of HPV in head and neck cancer patients undergoing postoperative chemoradiotherapy
.
Ann Transl Med
2015
;
3
:
S14
.