Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
The Laryngoscope C 2013 The American Laryngological, V Rhinological and Otological Society, Inc. Prognostic Markers in Stage I Oral Cavity Squamous Cell Carcinoma Johannes Dunkel, MD; Samuli Vaittinen, PhD, MD; Reidar Grenman, PhD, MD; Ilpo Kinnunen, PhD, MD; Heikki Irjala, PhD, MD Objectives/Hypothesis: Early-stage oral squamous cell carcinoma (OSCC) treatment is based on anatomic location, clinical TNM staging, and histological grade. It is a heterogeneous disease group. Classification of patients with OSCC by immunohistochemical analysis of established oncoproteins and evaluate disease course was our primary objective. Characterization of stage I OSCC patients in Southwest Finland was our secondary objective. Study Design: Immunohistochemical analysis of tumor specimens and retrospective analysis of patient data of the patient treated in Turku University Hospital for T1N0M0 OSCC during the years 2000-2004. Methods: Paraffin-embedded tumor specimens from 35 OSCC patients were collected and analyzed for HIF-1a, CD44, p16, Ki67, and podoplanin by immunohistochemistry and correlated with clinical findings. Results: Tumoral CD44 and HIF1-a expression levels, in combination, predicted 5-year disease-free survival. Reduced expression of CD44 and elevated expression of HIF1-a is associated with the lowest probability of disease-free survival compared to the population as a whole (P < .001 in Kaplan-Meier analysis). Patients with grade I tumors demonstrated improved disease-specific survival compared to those with grade II tumors (P ¼ .027). No association was seen between p16 expression, Ki67 labeling index, or podoplanin expression and prognosis in our 35 specimens. Conclusions: HIF-1a and CD44 immunohistochemical detection could potentially serve as a prognostic tool in therapy selection for early-stage OSCC. Key Words: HIF-1a, podoplanin, p16, CD44, immunohistochemistry, head and neck cancer, head and neck squamous cell carcinoma, biomarker, prognosis. Level of Evidence: 2b. Laryngoscope, 123:2435–2441, 2013 INTRODUCTION Approximately 275,000 new patients are diagnosed with oral cavity cancer yearly, and it is estimated to cause 1.7% of all cancer-related deaths worldwide.1,2 The clinical behavior of early-stage oral squamous cell carcinoma (OSCC) is highly variable. Treatment of early stage OSCC has varied from combined surgery with or without neck dissection to different combinations of radiotherapy. Local surgery, usually laser assisted, combined with sentinel lymph node biopsy or selective neck dissection is the modality currently used by many centers. However, when our samples were collected, sentinel lymph node biopsies had yet to become general practice. A recent Cochrane review on surgical treatFrom the Medicity Research Laboratory (J.D., H.I.), Department of Pathology (S.V.) and Turku Doctoral Programme of Biomedical Sciences (J.D.), University of Turku; and the Department of Pathology (S.V.) and Department of Otorhinolaryngology–Head and Neck Surgery (R.G., I.K., H.I.), Turku University Hospital, Turku, Finland. Editor’s Note: This Manuscript was accepted for publication October 22, 2012. This work was supported by Turku Doctoral Programme of Biomedical Sciences, The Finnish Medical Foundation, Turku University Foundation, The Finnish Otolaryngology Foundation, Turku University Hospital, and Kirsti and Tor Johansson Cancer and Heart Foundation. The authors have no other funding, financial relationships, or conflicts of interest to disclose. Send correspondence to Reidar Gr enman, PhD, MD, Department of Otorhinolaryngology, Turku University Hospital, Kiinamyllynkatu 4-8, P.O. Box 52, 20521 Turku, Finland. E-mail: [email protected] ment of oral and oropharyngeal cancer showed no advantage in elective neck dissection in the clinically negative neck compared to therapeutic neck dissection when indicated.3 In general, local surgery is a sufficient treatment, but occasionally the disease behaves aggressively and currently no reliable markers predicting the course of the disease exist. Implementation of prognostic biomarkers associated with head and neck squamous cell carcinoma (HNSCC) is needed. The TNM classification system is the most used and reliable factor predicting prognosis in HNSCC. Tumor grade may not hold prognostic value according to previous studies.4 Several biomarkers have been studied, but none are routinely used in clinical practice. Candidate biomarkers exist. Ki67 is a proliferation marker expressed by cells that have entered the cell cycle. It is widely used in breast cancer, but its prognostic role in HNSCC has remained controversial.5,6 Hypoxia-related markers, such as HIF-1a, have been investigated, but their roles in the prognoses of HNSCC remain disputable.7 The hyaluronic acid receptor, CD44, is a cell adhesion and homing receptor known to play multiple roles in cancer development.8 Podoplanin is a lymphatic endothelial marker also expressed by the basal membrane of the epithelium. It has been linked to poor outcome in HNSCC.9 The cell cycle progression inhibitor, p16, is a protein indicating human papilloma virus (HPV) infection in HNSCC.10,11 DOI: 10.1002/lary.23888 Laryngoscope 123: October 2013 Dunkel et al.: Prognostic Markers in Oral Cavity Cancer 2435 In Finland, OSCC is often diagnosed relatively early due to a comprehensive local basic health care system.12 In the absence of prognostic biomarkers, patients in most centers undergo diagnostic imaging and surgical treatment, including a sentinel lymph node biopsy. To investigate efficient methods of predicting disease course, we screened candidate prognostic markers by immunohistochemistry in OSCC biopsy samples. Additionally, we wanted to elucidate the incidence and natural course of stage I oral carcinoma in patients treated with only local surgery in an indiscriminate population of 700,000 in Southwestern Finland. MATERIALS AND METHODS Patients This population-based study was conducted in the area of Turku University Hospital with a population base of approximately 700,000. OSCC samples were collected in the Department of Otorhinolaryngology–Head and Neck Surgery at Turku University Hospital during the years 2000 to 2004. Diagnoses were confirmed from hematoxylin and eosin-stained sections by the same experienced pathologist at Turku University Hospital. TNM status was based on clinical examination, including panendoscopy and a contrast-enhanced computed tomography scan of the head and neck. All samples were collected prior to treatment, and the primary treatment consisted exclusively of surgical resection of the tumor. Clinical outcome was retrospectively traced from the hospital’s records. Local recurrent disease, cervical lymph node metastasis, distant metastasis, and diseaserelated death were considered as end points. Antibodies Monoclonal primary antibodies used in this study were Hermes-3 for CD4413 at 0.20 lg/mL, D2-40 for podoplanin (Covance, Princeton, NJ) 1:100; Ki-67 MIB-1 clone (Dako, Glostrup, Denmark) 1:25; HIF-1a (Thermo Fisher Scientific, Waltham, MA), 1:100; p16 (BD Pharmingen, Franklin Lakes, NJ), 1:25; and 3G614, an antibody targeting a chicken T-cell antigen, serving as a negative control. All of the antibodies were diluted in phosphate-buffered saline (PBS) at the concentration indicated above, and they were of the mouse antihuman immunoglobulin G (IgG) subtype. Microscopic Evaluation D2-40 staining was graded as follows: grade 1, positive reaction in the mucosal epithelial cells limited to only basal cells; grade 2, positive in the basal and upper cell layers, but lower than that in the fourth layer; grade 3: positive from the basal to the fifth layer and beyond. Intra- and peritumoral D240þ vessels were counted with a 0.0625-mm2 grid under a magnification of 400. The CD44 staining was graded as follows: grade 1, <10% of the tumor cells stained positively; grade 2, 10% to 50% of the cells stained positively; grade 3, >50% of the cells stained positively. The Ki67 labeling index was evaluated by counting the percentage of positive cells in five different locations within the tumor. Both HIF-1a staining and p16 staining were evaluated as low, moderate, or strong in the area of the highest grade within each tumor. All of the slides were evaluated by a single blind approach by two independent observers (J.D., H.I.) under an Olympus BX40 (Olympus, Tokyo, Japan) microscope. In cases where the examiners’ results differed, the samples were re-examined until consensus was reached. Statistical Analysis Relationships among clinicopathological variables, immunohistochemical staining patterns, and prognoses were evaluated. To study the combined effect of CD44 and HIF-1a, CD44 expression of above 50% was considered as high and that of below 50% as low. Moderate and strong HIF-1a staining patterns were both considered as strong. These two variables were then cross-tabulated to divide patients into three groups (CD44 low/HIF-1a high; CD44 low/HIF-1a low or CD44 high/HIF-1a high; and CD44 high/HIF-1a low). Kaplan-Meier survival curves were constructed to compare both disease-specific 5-year survival as well as 5-year disease-free survival, where disease-related death was used as an end point in overall survival and recurrent disease or metastasis as an end point in disease-free survival; significance levels were calculated with the log-rank test. Correlation coefficients were determined by Spearman’s rho. Incidences of end points among groups were cross-tabulated. Significance values were calculated with the Mann-Whitney U test or Fisher exact test where applicable. IBM SPSS Statistics version 19.0 (IBM, Armonk, NY) was used for all statistical analyses. RESULTS Patient Data Immunohistochemistry Samples were fixed in formalin, embedded in paraffin, and cut into 5 lm sections. Deparaffinization was performed with xylene and rehydration with descending alcohol. Boiling in 0.01% citric acid at a pH of 8.0 in a microwave oven for 6 minutes was done for antigen retrieval, and samples were cooled down. Hydrogen peroxide (0.3%) in PBS was used to block endogenous peroxidase activity. Samples were washed, and unspecific binding was blocked with 1.5% normal serum in PBS for 60 minutes. Primary antibody was added to the samples and incubated at room temperature for 60 minutes. Slides were washed in PBS, and secondary antibody (Elite anti-mouse IgG ABC kit; Vector Laboratories, Burlingame, CA) was used according to the manufacturer’s recommendations. Diaminobenzidine was used for visualization for 3 minutes at room temperature, and Mayer’s hematoxylin was used for background staining. Laryngoscope 123: October 2013 2436 The material consists of 44 clinical T1N0M0 previously untreated OSCCs diagnosed during the selected 5year period of 2000 to 2004. Sixteen (36%) of the patients were males. The ages ranged from 34 to 87 years, with a mean age of 65 years and median age of 63 years. Eight patients were smokers, six had smoked earlier, and 30 had never smoked. Thirty-four of the tumors were located in the mobile tongue. Other sites were floor of the mouth (n ¼ 6), mandibular gingiva (n ¼ 3), and maxillary gingiva (n ¼ 1). Six were excluded from immunohistochemical analyses because we were not able to obtain the primary surgical sample and three because the samples available were deemed insufficient. Thus, 35 samples of the 44 were immunohistochemically analyzed. Dunkel et al.: Prognostic Markers in Oral Cavity Cancer had no statistical significance in clinical outcome (Table 1) or immunohistochemical staining patterns (Table 2). Eleven patients had a normal tissue margin of <10 mm in the primary surgical specimen after primary treatment. A close follow-up only was chosen for eight of these patients, mostly due to other severe diseases affecting the general health of the patient, two were reoperated, and one received postoperative radiotherapy. Of these 11 patients, two were diagnosed with a locally recurrent disease, two with local recurrences and neck metastases, one metastatic disease, and one patient had a second primary tumor 71 months postoperatively. The lack of sufficient margins had no impact on disease-specific or disease-free survival in this patient material. Follow-Up of the Patients In this retrospective study, 11 (25%) patients relapsed and six (14%) died of the disease within the follow-up time of 5 years (Fig. 1A). The mean follow-up time until recurrence was 14 months after primary treatment. Local recurrence occurred in eight patients (18%), five of whom had additional metastases (supplementary data). Exclusively local diseases were treated with surgery alone or in combination with radiotherapy; two patients are alive and one died of a disease-related cause. Three patients with metastases, in addition to local recurrence, were reoperated, and two patients only received palliative treatment; all of these patients died of a disease-related cause. Three patients had neck metastases, one of whom had additional distant metastases during the follow-up. One of these was reoperated successfully, one died despite surgical treatment and radiotherapy, and one received only palliative treatment. Four patients had a second primary disease more than 5 years after primary treatment. Three of these patients with second primary carcinomas were operated, one of whom received additional radiotherapy, and one received radiotherapy alone. The 5year disease-specific survival for all patients was 86%, and 5-year disease-free survival was 75%. Fig. 1. Kaplan-Meier curves for disease-free survival during the 5year follow-up. (A) All patients pooled. (B) Patients grouped by CD44 and HIF1-a status. Results After Primary Surgery Local surgery was the selected curative treatment modality for all patients in this patient population. Histologically, the tumors were of squamous cell origin and the mean size was 15 mm (4 mm–35 mm). Although clinically T1, the histological examination revealed five of the tumors to be over 20 millimeters in size (2035mm), indicating a pT2 disease. The tumor size had no prognostic significance. Twenty-nine tumors were grade I, 10 were grade II, and five were not originally classified. Higher tumor grade predicted poorer diseasespecific survival (P ¼ .027), and a strong tendency was seen in disease-free survival (P ¼ .059). Smoking history Laryngoscope 123: October 2013 Role of Human Papillomavirus in the Pathogenesis of OSCC Two (5.7%) of the 35 tumors analyzed stained positive for p16 indicating a human papillomavirus (HPV) infection in the tumor (Fig. 2A). Expression of p16 had no impact on disease-specific survival in this limited material of 35 samples. Course of the Disease in Relation to Podoplanin Expression and Ki67 Labeling Index Podoplanin staining was analyzed on the basal cell layer of the tumor. Podoplanin-positive lymphatic vessels were counted both intratumorally and in the peritumoral area (Fig. 2B). None of these correlated with the course of the disease (Table 1). The proliferation marker, Ki67, was stained, and the proportion of the positive cells in the tumor was counted (Fig. 2C). The Ki67 labeling index did not correlate with the course of the disease (Table 1). Dunkel et al.: Prognostic Markers in Oral Cavity Cancer 2437 Fig. 2. Examples of images of the immunostained tumor samples for all markers. (A) p16 staining. (B) Podoplanin staining. (C) Ki67 staining. (D) CD44 staining, low expression. (E) CD44 staining, high expression. (F) HIF1-a staining, low expression. (G) HIF1-a staining, high expression. (H) Negative control. Laryngoscope 123: October 2013 2438 Dunkel et al.: Prognostic Markers in Oral Cavity Cancer TABLE 1. lll. Disease-related death within follow-up Yes No Local recurrence or metastasis within follow-up Yes No p-value Yes No p-value 1 5 7 25 n.s. 1 9 7 21 n.s. Earlier 0 6 1 5 Male Female 3 3 13 25 n.s. 3 8 13 20 1 1 28 <0.001* 5 24 <0.05* 2 Yes 4 1 6 1 <0.03** n.s. 5 1 5 1 n.s.** n.s. No 29 4 9 24 <25% 25% 2 4 19 13 n.s. 4 7 17 10 n.s. <10% 0 3 n.s. 2 1 n.s. 10-50% >50% 4 1 12 15 6 2 10 14 Low 1 18 2 17 <0.02* Intermediate High 2 2 6 6 5 3 3 5 <0.03** Low 1 4 1 4 n.s. Intermediate High 3 1 17 9 6 3 14 7 Neither 0 10 <0.004* 0 10 <0.002* One of two Both 2 5 15 3 n.s.** 4 6 13 2 <0.001** n.s. n.s. n.s. p-values were considered statistically significant if p<0.05. *p-values from cross tabulatation of events over five years, calculated with Fisher’s exact test. **p-values from Kaplan-Meier curves, calculated with the log rank test. Course of the Disease in Relation to HIF-1a and CD44 Expression The tumors were stained for the hypoxia-related marker HIF-1a and a cell adhesion protein CD44 (Fig. 2D–G). Strong HIF-1a expression correlated negatively with the 5-year disease-free survival (P ¼ .022) in the Kaplan-Meier analysis. A negative trend was seen in the connection between HIF-1a expression and 5-year disease-specific survival, but the differences among groups failed to reach statistical significance (P ¼ .29). Positive trends were seen between CD44 expression and both disease-specific and disease-free survival in Kaplan-Meier analyses. The combination of high CD44 expression and low HIF-1a expression had a strong positive correlation with the 5-year disease-free survival in the KaplanMeier analysis (P < .001, Fig. 1B). None of the 10 patients with low HIF-1a- and high CD44-expressing tumors had disease recurrence in contrast to 4/17 (24%) for patients with one and 6/8 (75%) with both risk factors (P < .002, Table 1). The same trend appeared in disease-specific survival. DISCUSSION Due to the difficulty in predicting the outcome of oral cavity cancer, currently all patients receive surgical Laryngoscope 123: October 2013 treatments combined with sentinel lymph node biopsy, radiotherapy, or both. Our objective was to discover an efficient method of selecting patients needing local resection alone, through detection of biomarkers in the cancer biopsy. Patients with low expression levels of HIF-1a and high expression levels of CD44 all stayed in remission, whereas those with high expression levels of HIF-1a and low expression levels of CD44 had a 75% recurrence rate during their 5-year follow-up. Independently, these markers are weak predictors of 5year disease-free survival. With the combination of these markers we were able to reliably predict disease course even in a reasonably small sample size of 35 OSCC samples. Our findings could suggest that using CD44 and HIF-1a immunohistochemistry, in combination, would be an effective way to select patients to be potentially treated by one treatment modality alone to reduce morbidity. The extent of the disease and patient-related factors are the key elements influencing the choice of treatment modality in OSCC today. Surgery ranges from limited surgery to extensive operations requiring major tissue reconstruction. Radiotherapy, especially when combined with chemotherapy, increases the risk of adverse events (e.g., xerostomia, osteoradionecrosis, and difficulties in swallowing and speech). Nevertheless, even T1 tumors, Dunkel et al.: Prognostic Markers in Oral Cavity Cancer 2439 TABLE 2. xxxx. Smoking Ki67 LI <25% 25% Yes No Earlier p-value 2 5 15 10 4 2 n.s. HPV infection Yes 1 1 0 n.s. Podoplanin expression at the basal membrane No Low 5 0 22 5 6 0 n.s. Intermediate 3 12 5 High Low 2 4 7 11 1 4 Intermediate 0 8 0 High <10 % 1 1 5 2 2 0 10-50 % 2 11 3 >50 % Neither 2 3 11 5 3 2 One of two 1 12 4 Both risk factors 1 7 0 HIF-1a grade CD44 grade Combination of high HIF-1a and low CD44 expression n.s. n.s. n.s. p-values were considered statistically significant if p<0.05. p-values were calculated with Fisher’s exact test. which have been resected with sufficient margins, may recur. Thus, the means of selecting patients requiring more than one treatment modality are needed. Our data show that local resection alone can yield a 25% recurrence rate of small local tumors during the routine follow-up time of 5 years. This is unacceptable by the current high standards of oral cancer treatment. This would obviously indicate that either all oral cancer patients need extensive cancer therapy or that the means of selecting patients for less extensive therapy are needed. The significance of CD44 in OSCC remains unclear. Overexpression has previously been linked to poor outcome in squamous cell carcinomas of the oropharynx, hypopharynx, and larynx, whereas evidence concerning the oral cavity is slightly less assuring.15 Elevated expression of CD44 has also been linked to poor outcome in other malignancies such as hematological malignancies,16 colorectal cancer,17 and gastric cancer.18 It has previously been speculated that CD44 can be used as a cancer stem cell marker, meaning that tumor cells expressing CD44 are capable of forming new tumors.19 Hypoxic tumors have poorer response to radiotherapy as reviewed by Silva et al.20 HIF-1a plays a role in cancer development by activating the transcription of several different genes under hypoxic conditions.21 Many of these genes are crucial in angiogenesis, apoptosis, and cell metabolism. The same genes have been shown to be upregulated by hypoxia and HIF-1a in vitro.22 Interestingly, taking into account the combined expression levels of CD44 and HIF-1a proved to be effective in predicting disease-free survival of OSCC patients. This combination is effective because both molecules participate in a broad spectrum of cellular functions. Thus, evaluating both of their expression levels gives us Laryngoscope 123: October 2013 2440 information of several aspects of the tumor microenvironment. There is emerging evidence that podoplanin could serve as a potential prognostic marker in oral cancer as podoplanin is overexpressed by metastatic tumors.9 The tumors we analyzed were stage I tumors, which have not metastasized. The density of lymphatic vasculature has been shown to correlate positively with lymph node metastases at the time of diagnosis of HNSCC.23 Our study showed no significance between either podoplanin expression of the basal membrane or lymphatic vasculature and clinical outcome in early stage OSCC. The biomarker p16 is a valid biomarker in screening for a transcriptionally active HPV infection.24 Tumors of the oropharynx can be screened for HPV by p16 immunohistochemistry to find patients with improved survival, which is due to higher radiosensitivity.25,26 The role of HPV infection in oral cavity cancer initiation and progression remains disputable, and we were not able to address this issue with only two HPVpositive samples. The Ki67 labeling index had no impact on prognosis in these early-stage forms of disease, even though some studies have indicated otherwise in HNSCC.27 Demographically, the early-stage patients analyzed here did not have the typical profile or risk factors for HNSCC: 64% were female and 68% were life-long nonsmokers. However, increasing numbers of nonsmoking female patients with oral cancer have been previously reported.28 The median age of 63 years corresponded well with previously reported oral cancer patients. Evaluating the prognoses of HNSCC patients is difficult. HNSCC may have different characteristics and behave in different ways according to the site of origin.29–31 This suggests that OSCC is in fact a Dunkel et al.: Prognostic Markers in Oral Cavity Cancer heterogeneous group of malignancies occurring in the same region. Our findings may simplify the categorization process. CONCLUSION The outcome data we collected of 44 patients show that treating even low-stage OSCC aggressively is required if no method of grouping the patients is used. We think that developing an efficient clinical chemistry method that assesses CD44 and HIF-1a coexpression patterns in OSCC biopsies could provide us with a method of predicting the disease course. This method could serve as an interesting yet easy and cost-effective way to select the patients for whom one treatment modality would be sufficient, and thereby reduce the morbidity caused by the therapy. BIBLIOGRAPHY 1. Warnakulasuriya S. Global epidemiology of oral and oropharyngeal cancer. Oral Oncol 2009;45:309–316. 2. Ferlay J, Shin H-R, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer 2010;127:2893–2917. 3. Bessell A, Glenny A-M, Furness S, et al. Interventions for the treatment of oral and oropharyngeal cancers: surgical treatment. Cochrane Database Syst Rev 2011;(9):CD006205. 4. Zatterstrom UK, Wennerberg J, Ewers SB, Willen R, Attewell R. Prognostic factors in head and neck cancer: histologic grading, DNA ploidy, and nodal status. Head Neck 1991;13:477–487. 5. Schliephake H. Prognostic relevance of molecular markers of oral cancer— a review. Int J Oral Maxillofac Surg 2003;32:233–245. 6. Weigel MT, Dowsett M. Current and emerging biomarkers in breast cancer: prognosis and prediction. Endocr Relat Cancer 2010;17:R245–R262. 7. Hoogsteen IJ, Marres HA, Bussink J, van der Kogel AJ, Kaanders JH. Tumor microenvironment in head and neck squamous cell carcinomas: predictive value and clinical relevance of hypoxic markers. A review. Head Neck 2007;29:591–604. 8. Wang SJ, Bourguignon LY. Role of hyaluronan-mediated CD44 signaling in head and neck squamous cell carcinoma progression and chemoresistance. Am J Pathol 2010;178:956–963. 9. Yuan P, Temam S, El-Naggar A, et al. Overexpression of podoplanin in oral cancer and its association with poor clinical outcome. Cancer 2006; 107:563–569. 10. Nakao Y, Yang X, Yokoyama M, et al. Induction of p16 during immortalization by HPV 16 and 18 and not during malignant transformation. Br J Cancer 1997;75:1410–1416. 11. Breiteneder-Geleff S, Soleiman A, Kowalski H, et al. Angiosarcomas express mixed endothelial phenotypes of blood and lymphatic capillaries: podoplanin as a specific marker for lymphatic endothelium. Am J Pathol 1999;154:385–394. Laryngoscope 123: October 2013 12. Makitie AA, Koivunen P, Keski-Santti H, et al. Oral tongue carcinoma and its treatment in Finland. Eur Arch Otorhinolaryngol 2007;264:263–267. 13. Jalkanen S, Bargatze RF, de los Toyos J, Butcher EC. Lymphocyte recognition of high endothelium: antibodies to distinct epitopes of an 85–95-kD glycoprotein antigen differentially inhibit lymphocyte binding to lymph node, mucosal, or synovial endothelial cells. J Cell Biol 1987;105: 983–990. 14. Salmi M, Jalkanen S. A 90-kilodalton endothelial cell molecule mediating lymphocyte binding in humans. Science 1992;257:1407–1409. 15. Kokko L-L, Hurme S, Maula S-M, et al. Significance of site-specific prognosis of cancer stem cell marker CD44 in head and neck squamous-cell carcinoma. Oral Oncol 2011;47:510–516. 16. Liu J, Jiang G. CD44 and hematologic malignancies. Cell Mol Immunol 2006;3:359–365. 17. Mulder JW, Kruyt PM, Sewnath M, et al. Colorectal cancer prognosis and expression of exon-v6-containing CD44 proteins. Lancet 1994;344: 1470–1472. 18. Mayer B, Jauch KW, Gunthert U, et al. De-novo expression of CD44 and survival in gastric cancer. Lancet 1993;342:1019–1022. 19. Prince ME, Sivanandan R, Kaczorowski A, et al. Identification of a subpopulation of cells with cancer stem cell properties in head and neck squamous cell carcinoma. Proc Natl Acad Sci U S A 2007;104:973–978. 20. Silva P, Homer JJ, Slevin NJ, et al. Clinical and biological factors affecting response to radiotherapy in patients with head and neck cancer: a review. Clin Otolaryngol 2007;32:337–345. 21. Pugh CW, Ratcliffe PJ. Regulation of angiogenesis by hypoxia: role of the HIF system. Nat Med 2003;9:677–684. 22. Greijer AE, van der Groep P, Kemming D, et al. Up-regulation of gene expression by hypoxia is mediated predominantly by hypoxia-inducible factor 1 (HIF-1). J Pathol 2005;206:291–304. 23. Kyzas PA, Geleff S, Batistatou A, Agnantis NJ, Stefanou D. Evidence for lymphangiogenesis and its prognostic implications in head and neck squamous cell carcinoma. J Pathol 2005;206:170–177. 24. Thomas J, Primeaux T. Is p16 immunohistochemistry a more cost-effective method for identification of human papilloma virus-associated head and neck squamous cell carcinoma? Ann Diagn Pathol 2012;16:91–99. 25. Reimers N, Kasper HU, Weissenborn SJ, et al. Combined analysis of HPVDNA, p16 and EGFR expression to predict prognosis in oropharyngeal cancer. Int J Cancer 2007;120:1731–1738. 26. Attner P, NAsman A, Du J, et al. Survival in patients with human papillomavirus positive tonsillar cancer in relation to treatment. Int J Cancer 2012;131:1124–1130. 27. Sittel C, Ruiz S, Volling P, et al. Prognostic significance of Ki-67 (MIB1), PCNA and p53 in cancer of the oropharynx and oral cavity. Oral Oncol 1999;35:583–589. 28. Hakulinen T, TryggvadOttir L, Gislum M, et al. Trends in the survival of patients diagnosed with cancers of the lip, oral cavity, and pharynx in the Nordic countries 1964–2003 followed up to the end of 2006. Acta Oncol 2010;49:561–577. 29. Jones AS, Husband D, Rowley H. Radical radiotherapy for squamous cell carcinoma of the larynx, oropharynx and hypopharynx: patterns of recurrence, treatment and survival. Clin Otolaryngol Allied Sci 1998;23: 496–511. 30. Kreimer AR, Clifford GM, Boyle P, Franceschi S. Human papillomavirus types in head and neck squamous cell carcinomas worldwide: a systematic review. Cancer Epidemiol Biomarkers Prev 2005;14:467–475. 31. Chung CH, Ely K, McGavran L, et al. Increased epidermal growth factor receptor gene copy number is associated with poor prognosis in head and neck squamous cell carcinomas. J Clin Oncol 2006;24:4170–4176. Dunkel et al.: Prognostic Markers in Oral Cavity Cancer 2441