|Home | About | Journals | Submit | Contact Us | Français|
Lymphatic metastasis is associated with up to a 50% decrease in survival, yet the molecular mechanisms driving their establishment remain poorly understood. This study assessed clinicopathological characteristics correlated to nodal metastasis among patients with head and neck squamous cell carcinoma for the identification of pathways on which to focus molecular studies. Pathology records were queried for cases diagnosed with invasive squamous cell cancer of the upper aerodigestive tract between 1993 and 2003. Charts and pathology reports were scored for 16 characteristics. The univariate association of each variable with lymph node status was assessed. Based on the univariate analysis, a multiple logistic regression model was developed to assess the simultaneous association of variables with lymph node status. Of the 644 cases identified, 234 had a surgical specimen analyzed. All variables were scored for 185 of the 234 cases. Multivariate stepwise regression analysis identified clinical stage (p = 0.0269), pathologic stage (p = 0.0162), grade (p = 0.0094), lymphovascular invasion (p = 0.0393), and family history of cancer (p = 0.0079) as independently predictive of lymphatic metastases. Our study confirms that grade, pathologic stage, clinical stage, and lymphovascular invasion are predictors of regional metastasis. These correlations suggest that studying the molecular mechanisms of differentiation, interstitial pressure at the primary tumor site, and peritumoral lymphangiogenesis may provide insight into lymphatic metastasis. Additionally, we identified family history of cancer as a new predictor of lymphatic metastasis. Thus, genetic analysis of families with cancer, irrespective of type, may identify genes important for regional metastasis.
Worldwide, there are approximately 640,000 new cases of head and neck squamous cell carcinoma (HNSCC) each year, which result in about 350 000 deaths (Parkin, Bray, Ferlay & Pisani, 2005). In 2008, there were 48,000 new cases and 11,000 deaths in the United States (Jemal, Siegel, Ward, Hao, Xu, Murray & Thun, 2008). There have been significant improvements in diagnosis and local management over the past 20 years; yet the 5-year survival for HNSCC patients remains among the lowest of the major cancers (Hardisson, 2003). The majority of cancer mortality is due to metastatic disease rather than primary tumors (Schildberg, Meyer, Piltz & Koebe, 1995; Steeg & Theodorescu, 2008). Lymph node metastasis characterizes tumors that are more aggressive, predicts resistance to therapy, and indicates the likelihood of distant metastasis (Wissmann & Detmar, 2006). Although the regional lymph nodes are not essential organs, regional metastasis is associated with up to a 50% decrease in survival, irrespective of distant metastasis (Cady, 1984; Howell & Grandis, 2005; Dunne, Muller, Eisele, Kessel, Moll & Werner, 2006; Jemal et al., 2008). The presence of cervical lymph node metastasis is among the most important factors for HNSCC prognosis and for determination of appropriate treatment (Leong, Cady, Jablons, Garcia-Aguilar, Reintgen, Jakub, Pendas, Duhaime, Cassell, Gardner, Giuliano, Archie, Calvin, Mensha, Shivers, Cox, Werner, Kitagawa & Kitajima, 2006). Simply stated, we do not understand the causal nature of this association. More than 50% of patients with HNSCC present with regional lymph node disease (Ozdek, Sarac, Akyol, Unal & Sungur, 2000; Jemal et al., 2008).
The poor prognosis of advanced stage disease makes studying lymphatic metastasis very important. While numerous clinical studies have confirmed that regional metastasis is an important prognostic factor for HNSCC (Richard, Sancho-Garnier, Micheau, Saravane & Cachin, 1987), surprisingly few clinical studies have addressed factors that are predictive of lymphatic metastasis. The purpose of those few studies was to identify the clinical and pathological predictors of lymph node metastasis in order to help the clinician predict whether the patient without overt lymph node metastases was likely have occult metastases and should be subjected to elective neck dissection. To date, no attempts have been made to interpret any of these findings in the context of mechanisms important for this ominous association. Even in clinical studies that examined molecular markers (Takes, Baatenburg De Jong, Alles, Meeuwis, Marres, Knegt, De La Riviere, De Wilde, Mooi, Hermans & Van Krieken, 2002), no link was made to molecular cancer research. It is widely accepted that cancer cells acquire distinct capabilities, such as the ability to proliferate in excess, to avoid apoptosis, and to recruit new blood vessels, in order to become generally metastatic (Hanahan & Weinberg, 2000). Most basic science research efforts have focused on genetic and molecular changes that allow primary tumor cells to escape the primary site rather than the molecular mechanisms that allow metastatic tumor cells to colonize the lymph node. Molecular studies have elucidated the pathways for a number of genes that are involved in the development of malignancies. The molecular mechanisms driving the establishment of lymphatic metastasis, however, remain poorly understood.
A comprehensive understanding of the clinical and pathologic features of regional metastatic disease will allow molecular research to focus on the biological pathways important for lymphatic metastasis. Studies to date, however, have only reported on limited clinical and pathologic predictors of lymphatic spread and have been biased towards the identification of factors predictive of occult lymphatic spread in early stage tumors. Furthermore, identified factors vary between studies (Table 1). The only factor that has been reported repeatedly as correlating with the development of regional metastases is primary tumor invasion. The clinicopathological predictors of lymphatic metastasis can do more than aid in the assessment of patients’ nodal status; they can focus basic science research to address the questions that are relevant to the pathogenesis of lymphatic metastasis. Our goal was to assess comprehensively the pre-treatment clinical and pathologic features that predict lymphatic metastatic spread with the aim of identifying pathways on which to focus further molecular studies.
This retrospective study was based on a review of the hospital records of patients who were diagnosed with invasive squamous cell cancer of the upper aerodigestive tract between 1993 and 2003 at the University of Texas Medical Branch. There were 644 cases of invasive squamous cell cancer of the upper aerodigestive tract initially identified. Charts were retrospectively reviewed for age, gender, race, site of primary tumor, size of primary tumor, clinical stage of primary tumor, history of a second primary tumor, family history of cancer, family history of head and neck cancer, history of tobacco use, and history of alcohol use. From that group, 234 had a surgical resection specimen analyzed. Pathology reports were reviewed and scored for pathologic stage of primary tumor, tumor grade, lymph vascular invasion, perineural invasion, and lymph node status. Tumor staging was performed according to the criteria of the American Joint Committee on Cancer. Tumor grade was categorized as well-differentiated when good stratification, regular nuclei, abundant eosinophilic cytoplasm, intercellular bridges, and minimal mitosis (<10 mitoses/10 high power fields) were observed. Poor differentiation was marked by an undifferentiated cell pattern, marked nuclear atypia, high cellular pleomorphism, absence of keratinization, and high rate of mitosis (>20 mitoses/10 high power fields). Those whose characteristics fell in between were called moderate. Lymph vascular invasion was defined as positive when tumor cells were observed inside lymphatic, arterial or venous vessels. Perineural invasion was defined as positive when tumor cells were observed within a neural sheath. All variables were successfully scored for 185 of the 234 cases.
Descriptive statistics were calculated to summarize demographic information and distribution of predictor variables in the study population. The univariate association of each variable with lymph node status (positive vs. negative) was assessed using the chi-square test. Based on the univariate analysis, a multiple logistic regression model was developed to assess the simultaneous association of variables with lymph node status. Model-building was performed to determine the final model which includes the significant variables predictive of lymph node status. Odds ratios and 95% confidence intervals were calculated for each variable in the model. P-value < 0.05 was considered statistically significant.
The demographics and characteristics of the study population are summarized in Table 2. The patients were more likely to be white men who used alcohol and tobacco but had no family history of cancer. They ranged in age from 29 years to 90 years, with a median age of 56 years. There were 42 (22.7%) well differentiated tumors, 104 (56.2%) moderately differentiated tumors and 31 (16.8%) poorly differentiated tumors. Only 22 (11.9%) were noted to have lymphovascular invasion and 16 (8.6%) had perineural invasion on histological analysis.
We first aimed to define the association of each clinicopathological characteristic with lymph node status independently (Table 3). Using univariate analysis to determine the association of each patient characteristic to lymph node status, significant variables were pathological grade, clinical stage, lymph vascular invasion and tumor site. Tumors originating in the oral cavity or larynx were less likely to be lymph node positive while tumors originating in the oropharynx and hypopharynx were more likely to be lymph node positive. Tumors with poor pathological grade, advanced clinical stage or lymph vascular invasion were more likely to have regional involvement.
Multivariate stepwise regression analysis identified four variables as independently predictive of lymphatic metastases (Table 4). The identification of clinical tumor stage [Odds Ratio (OR) = 1.445], pathologic tumor stage (OR = 1.339), tumor grade (OR = 2.051), and lymphovascular invasion (OR = 3.226) confirm previous reports. We also identified family history of cancer (OR = 3.358) as a new predictor of lymphatic metastases.
For many cancers including HNSCC, the presence of lymph node metastasis is among the most important prognostic factors for staging and for determining appropriate treatment. Few clinical studies have addressed the clinical and pathological factors that are predictive of lymphatic metastasis (Kowalski, Franco & de Andrade Sobrinho, 1995; Martinez-Gimeno, Rodriguez, Vila & Varela, 1995; Woolgar & Scott, 1995; Fukano, Matsuura, Hasegawa & Nakamura, 1997; Ozdek et al., 2000; Sparano, Weinstein, Chalian, Yodul & Weber, 2004; Kane, Gupta, Kakade & A, 2006). Moreover, the data generated in these studies have not been applied to molecular lymphatic metastasis research. Even in clinical studies that examined molecular markers, no link was made to molecular cancer research (Takes et al., 2002). This study aims to further translational research efforts by using clinicopathological correlates of lymph node metastasis to identify pathways on which to focus molecular studies. Our study confirms that clinical tumor stage, pathologic tumor stage, grade, and lymphovascular invasion are clinicopathological predictors of regional tumor spread. Our study also identifies family history of cancer as a new predictor of lymphatic metastases.
The importance of T stage as a predictor of lymphatic metastasis suggests that size, the main characteristic determining T stage, is an important tumor feature that predicts tumor spread regionally. This finding is not intuitively obvious as it is independent of invasion (discussed below). An increase in tumor size, however, has been associated with increased intratumoral interstitial fluid pressure (IFP). In 1992, Gutmann, et al., reported that the IFP of HNSCC increased as tumor volume increased (p < 0.001) and that there was no correlation with systemic blood pressure (Gutmann, Leunig, Feyh, Goetz, Messmer, Kastenbauer & Jain, 1992). While normal tissues maintain an IFP near −1 mm Hg (Wiig, Rubin & Reed, 2003), the tumors ranged from 4 mm Hg to 33 mm Hg, with the highest pressure recorded in a tumor that measured 24 mL (Gutmann et al., 1992). The high IFP found in solid tumors has been reported to interfere with drug uptake, leading to resistance to chemotherapy (Heldin, Rubin, Pietras & Ostman, 2004; Lunt, Kalliomaki, Brown, Yang, Milosevic & Hill, 2008). Importantly, increased IFP has been shown to increase tumor cell egress of human melanoma xenografts into lymphatic vessels (Rofstad, Tunheim, Mathiesen, Graff, Halsor, Nilsen & Galappathi, 2002). In contrast, no correlation was found between increased tumor IFP and lymph node metastasis in a study of 25 breast cancer resections (Nathanson & Nelson, 1994). In a more recent study, the IFP was measured in lung tumors, including mainly adenocarcinomas and squamous cell carcinomas, of 22 patients prior to resection (Padera, Kadambi, di Tomaso, Carreira, Brown, Boucher, Choi, Mathisen, Wain, Mark, Munn & Jain, 2002). While they determined that the average IFP was raised, they did not find a correlation between the absolute IFP and lymph node metastasis (Padera et al., 2002). However, this group proposes that the IFP gradient rather than absolute tumor IFP affects lymphatic metastasis (Jain, Tong & Munn, 2007). These findings highlight the need for further study of primary tumor IFP, especially given that clinical observations suggest a link in HNSCC.
It has been proposed that the decrease in interstitial fluid pressure which follows inflammation is caused by a downregulation of β1-integrin interactions with extracellular matrix (Wiig et al., 2003). The pathway by which cells regulate the interstitial pressure may be controlled by the β1-integrin system (Reed, Berg, Gjerde & Rubin, 2001). Could an increase in β1-integrin activity contribute to the IFP found in HNSCC? The PDGF receptor has also been implicated in the regulation of IFP (Pietras, Ostman, Sjoquist, Buchdunger, Reed, Heldin & Rubin, 2001; Pietras, Sjoblom, Rubin, Heldin & Ostman, 2003). A recent study that examined PDGF-R in HNSCC determined that inhibition reduced autocrine stimulation of PDGF and paracrine stimulation of VEGF, but the IFP was not addressed (Bran, Bran, Hormann & Riedel, 2009). Understanding the molecular mechanisms of intratumoral IFP regulation in HNSCC may lead to a greater understanding of the pathways that promote lymph node metastasis.
The importance of lymphovascular invasion for predicting lymphatic metastasis indicates that tumor cell migration and invasion, as well as peritumoral lymphangiogenesis are important areas for further research. Migration and invasion have indeed been the focus of intense study in cancer research, however, studies focusing on peritumoral lymphangiogenesis and its interplay with tumor invasion into lymphatics has received comparatively little focus. Like interstitial pressure, the processes that control peritumoral lymphangiogenesis also involve PDGF and VEGF (Stacker, Achen, Jussila, Baldwin & Alitalo, 2002; Nathanson, 2003; Achen, Mann & Stacker, 2006). While there have been recent efforts to profile the expression of these and other angiogenic growth factors, (Montag, Dyckhoff, Lohr, Helmke, Herrmann, Plinkert & Herold-Mende, 2009) their molecular mechanisms in HNSCC need to be further studied. The expression of vascular endothelial growth factor C (VEGF-C) is higher in HNSCC compared to normal tissue (Neuchrist, Erovic, Handisurya, Fischer, Steiner, Hollemann, Gedlicka, Saaristo & Burian, 2003). In 2006, Franchi, et al., demonstrated that increased VEGF-C expression and inducible nitric oxide synthase activity correlates with increased microvascular density and increased lymph node metastasis in HNSCC (Franchi, Massi, Santucci, Masini, Degl’Innocenti, Magnelli, Fanti, Naldini, Ardinghi, Carraro & Gallo, 2006). Studies in prostate cancer (Tsurusaki, Kanda, Sakai, Kanetake, Saito, Alitalo & Koji, 1999), colorectal cancer (Akagi, Ikeda, Miyazaki, Abe, Kinoshita, Maehara & Sugimachi, 2000), thyroid cancer (Yu, Lo, Chan, Lam, Leung & Luk, 2005), breast cancer (Nakamura, Yasuoka, Tsujimoto, Imabun, Nakahara, Nakao, Nakamura, Mori & Kakudo, 2005), cervical cancer (Gombos, Xu, Chu, Zhang & Acs, 2005), and lung cancer (Renyi-Vamos, Tovari, Fillinger, Timar, Paku, Kenessey, Ostoros, Agocs, Soltesz & Dome, 2005) have also shown that VEGF-C biology plays an important role lymph node metastasis. The further elucidation of the molecular pathways which favor peritumoral lymphangiogenesis may aid the understanding of mechanisms that promote lymph node metastasis.
The importance of tumor grade for predicting lymphatic metastasis suggests that tumor cell differentiation should be the subject of further molecular research. A number of studies have identified modifiers of differentiation as prognostic markers for HNSCC (Ozdek et al., 2000; Sparano et al., 2004). These prognostic markers may also play an important role in pathways that enhance lymphatic metastasis. Some studies have suggested that HPV positivity is associated with both poor differentiation (Haraf, Nodzenski, Brachman, Mick, Montag, Graves, Vokes & Weichselbaum, 1996; Paz, Cook, Odom-Maryon, Xie & Wilczynski, 1997; Sugiyama, Bhawal, Dohmen, Ono, Miyauchi & Ishikawa, 2003) as well as high frequency of lymphatic metastasis (Haraf et al., 1996; Paz et al., 1997; Hoffmann, Gorogh, Gottschlich, Lohrey, Rittgen, Ambrosch, Schwarz & Kahn, 2005), suggesting that the same set of alterations may result in both phenotypes. An alternative interpretation suggests that HPV could drive the poor differentiation phenotype that in turn may drive an increased rate of lymphatic metastasis. Studying molecular changes associated with poor differentiation of tumors may lead to concomitant understanding of lymphatic tumor spread.
The identification of family history of cancer as a new predictor of lymphatic metastasis reveals that a genetic component may contribute to the development of regional metastasis in HNSCC. Previous reports have described family history as a risk factor for HNSCC, but there has been no investigation of familial cohorts that have lymphatic metastasis (Trizna & Schantz, 1992; Foulkes, Brunet, Kowalski, Narod & Franco, 1995). Application of classical genetic techniques for identification and characterization of family cohorts, as well as identification of the relevant underlying genetic alterations may facilitate the identification of genes whose function impacts the process of lymphatic spread of tumors. Furthermore, the genes identified in these familial cohorts can be used as a filter for large data sets, such as those generated by cDNA microarray analysis (Chen, Zhang, Zhang, Yuan, Yuan, Jin & Xiong, 2003; Dasgupta, Tripathi, Qin, Bhattacharya-Chatterjee, Valentino & Chatterjee, 2006; Hummerich, Muller, Hess, Kokocinski, Hahn, Fursten-berger, Mauch, Lichter & Angel, 2006; Roepman, Kemmeren, Wessels, Slootweg & Holstege, 2006; Mendez, Fan, Choi, Agoff, Whipple, Farwell, Futran, Weymuller, Zhao & Chen, 2007; Schlecht, Burk, Adrien, Dunne, Kawachi, Sarta, Chen, Brandwein-Gensler, Prystowsky, Childs, Smith & Belbin, 2007).
The use of clinical findings to direct biomedical research can aid in the understanding of the mechanisms that drive lymphatic metastasis. In his 2008 Nature article (Butler, 2008), Declan Butler said that “a chasm has opened up between biomedical researchers and the patients who need their discoveries” and that “neither basic researchers, busy with discoveries, nor physicians, busy with patients, are keen to venture there.” This approach, going from bedside to bench (Ledford, 2008), will maximally promote the reverse translation of relevant findings back to the bedside, bridging the “chasm” between physicians and researchers. Understanding the molecular mechanisms important for lymphatic metastasis may allow us to rationally design and apply drugs to interfere with this little understood, but profoundly important feature of cancer.
This work was supported by a Howard Hughes Medical Institute Early Career Physician Scientist Award to VAR and by NIH-NCI T32 CA 117834 to TMG. The authors have no conflicting financial interests.