1972 Correlation between Computed Tomographic Density of Lymph Node Metastases and Response to Cisplatin-Based Chemotherapy in Patients with Head and Neck Squamous Cell Carcinoma in an Area in which Betel Quid Chewing Is Prevalent Hung-Ming Wang, M.D.’ Shu-Hang Ng, M . D . ~ Cheng-Hsu Wang, M.D.’ Chuang-Chi Liaw, M.D.’ Ming-Hsui Tsai, M . D . ~ Gi-Ming Lai, M . D . ~ ’ Division of Hematology/Oncoiogy, Department of Internal Medicine, Chang Gung Memorial Hospital, Lin-Kou Medical Center, Taipei, Taiwan, Republic of China. * Department of Diagnostic Radiology, Chang Gung Memorial Hospital, Lin Kou Medical Center, Taipei, Taiwan, Republic of China. Department of Head and Neck Surgery, Chang Gung Memorial Hospital, Lin-Kou Medical Center, Taipei, Taiwan, Republic of China. The preliminary result (with fewer patients) of this study was presented at the Annual Meeting of The Chinese Oncology Society, Taipei, Taiwan, Republic of China, April 16, 1995. Address for reprints: Hung-Ming Wang M.D., Division of Hematology/Oncology, Department of Internal Medicine, Chang Gung Memorial Hospital, 199, Tun Hwa North Road, Taipei 105, Taiwan, Republic of China. Received February 7, 1996; revisions received May 21, 1996, and July 5, 1996; accepted July 5, 1996. 0 1996 American Cancer Society BACKGROUND. Identifying the factors predicting response to chemotherapy is imporlant for patients with head and neck squamous cell carcinoma (HNSCC). It allows more rational selection of subsets of patients who may benefit from multidisciplinary treatment. Correlation of lymph node density in contrast-enhanced computed tomographic (CT) scans of HNSCC with response to Chemotherapy was observed in the recent literature. This prospective study was designed to validate this clinical issue. METHODS. From January 1992 to March 1995, 71 patients with untreated HNSCC were included in this study in which the following criteria were met: 1) a lymph node > 3 cm by physical examination or z 2 cm by scanographic examination; 2) clinically evaluable disease treated by cisplatin-based neoadjuvant chemotherapy; and 3) availability of a pretherapeutic contrast-enhanced CT scan showing the cross-sections of relevant lymph node metastases. The density of the largest lymph node was compared with that of the nuchal muscles by a radiologist blinded to the patient’s therapeutic outcome. A lymph node was classified as hypodense if more than 33% of the lymph node surface area was comprised of a hypodense zone, and isodense if less than a third of the lymph node surface area was comprised of a hypodense zone. RESULTS. Fifty-one patients (72%) had the largest lymph node classified as hypodense, and 63 patients (89%) were found to have extranodal spread (ENS) in the relevant lymph nodes. Fifty-nine patients were betel quid chewers. Lymph node density was not related to T classification, primary site, or histologic differentiation of the primary tumor. There was also no correlation between lymph node density and lymph node N classification. The lymph node chemotherapy response rate was 35% (7 of 20) of the isodense group and 47.1% (24 of 51) of the hypodense group ( P = 0.36).ENS was found to have no impact on the chemotherapy response. CONCLUSIONS. Computed tomographic density of lymph node metastases did predict chemotherapy response in the HNSCC patients in the current study from an area in which betel quid chewing is prevalent. Cancer 1996; 78:1972-9. 0 1996 American Cancer Society. KEYWORDS: head and neck neoplasms, cervical lymph node, chemotherapy, betel quid. H ead and neck squamous cell carcinoma (HNSCC) (excludingnasopharyngeal carcinoma) accounts for 4-5% of the cancer incidence in Taiwan.’ The majority of the patients with HNSCC present with Lymph Node Density and Drug Response in HNSCCMlang et al. Stage 111 or IV disease. Despite optimal local therapy, long term survival rates for these patients remain less than 30%. During the last 10 years, chemotherapy has been actively used in conjunction with surgery and/ or radiotherapy in the management of patients with advanced stage head and neck carcinoma. Nonrandomized trials applying neoadjuvant chemotherapy in advanced stage HNSCC patients have demonstrated encouraging results with 60-90% tumor regression and 20-50% complete tumor regre~sion.~-~ However, none of the randomized trials using neoadjuvant chemotherapy has demonstrated increased survival rate^.^-^ Neoadjuvant chemotherapy would only benefit the responders, in terms of enhancement of locoregional control, and preservation of anatomic function in conjunction with c o s m e s i ~ . ~Therefore, f ~ ' ~ f ~ to identify the predicting factors for response to chemotherapy would be helpful in selecting suitable patients to receive neoadjuvant chemotherapy, which is considered to be relatively time-consuming and potentially toxic multidisciplinary management. The reported prognostic indicators for response to neoadjuvant chemotherapy included performance status (PS), primary tumor site, extent of lymph node involvement, and tumor size.l0-" The authors' previous study revealed that the tumor size and the presence of lymph node metastases were the most important predictors for response to neoadjuvant chem ~ t h e r a p y .Predictive '~ factors for chemosensitivity in patients with lymph node metastases are important for medical oncologists to justify the multimodality therapeutic sequence. A relationship between responses to chemotherapy and the density of lymph node metastases in contrast computed tomographic (CT) scans was demonstrated in recent trial^.'^.'^ Complete response rates from chemotherapy were worse if greater than 33% of the metastatic lymph node area was occupied by a hypodense zone. This phenomenon was attributed to the functional vascularization of the lymph nodes for efficient drug delivery. Epidemiologic studies have indicated that betel quid chewing appears to be an important factor in the development of HNSCC, especially oral cavity carcinoma, in Taiwan.l6-lEAccording to a recent survey, approximately 2.8 million people (approximately 13.3%of the Taiwanese population) are currently betel quid chewers. In Taiwan, betel quid chewers carry a relative risk 28 times higher than average for oral carcinoma development when compared with individuals who do not chew betel quid. If the betel quid users also consume cigarettes and alcohol, the relative risk increases 123 times. Approximately 86% of the Taiwanese betel quid users were also cigarette smokers 1973 and 75% also consumed alcohol habit~ally.'~ The responses to neoadjuvant chemotherapy of the authors' HNSCC patients (80-90% of whom were betel quid chewers) were not as good as those reported from the Western ~0untries.l~ The major characteristics of these patients that differed from those in the literature included the predominance of oral cavity carcinoma, a high incidence of severe mucositis, and bulky tumor burden. These characteristics were closely related to the culture of betel quid chewing, and may account for the lower response rates to chemotherapy of the authors' HNSCC patients. Therefore, this prospective study was designed to evaluate whether the correlation between the lymph node density in contrast CT scans and response to chemotherapy could be demonstrated in HNSCC patients from a betel quid chewing area. The histologic demonstration of extranodal spread (ENS) of the tumor in cervical metastasis has, according to many authors, a negative impact on the outcomes of HNSCC.20-2'However, nothing in the literature mentioned the relationship between ENS and chemotherapy response. In this article, ENS was identified on contrast CT scan and the possible role of ENS in the lymph node response to chemotherapy was discussed. PATIENTS AND METHODS From January 1992 to April 1995, 230 patients with HNSCC were included in prospective neoadjuvant chemotherapy trials. Patients eligible for these trials were required to have histologically confirmed squamous cell carcinoma; a measurable or evaluable locoregional disease with no distant metastasis: and no prior treatment. All patients had a World Health Organization (WHO) PS of 0-2, adequate renal function (serum creatinine < 1.5 mg/dL), leukocyte counts =4000/mL, and platelet counts 2 100,00O/mL.The pretreatment staging procedures were comprised of history and physical examination, complete blood count and biochemistry, chest roentgenography, bone scan, ultrasonography of liver, and CT scan of the head and neck. The chemotherapy regimens included 1) cisplatin (CDDP), 100 mg/m2 per day on Day 1; and 5fluorouracil (5-FU), 1 gm/m2 continuous intravenous infusion, on Day 1 to Day 4 or 5; or 2) CDDP, 100 mg/ m2, on Day 1: 5-FU, 1 gmlm' per day continuously intravenous infusion, on Day 1 to Day 4; and methotrexate (MTX), 200 mglm', on Days 15 and 22 with oral calcium folinate, 15 mg every 6 hours, rescue for 3 days (starting at 24 hours after MTX). Both regimens were given every 4 weeks. Patients who met the following criteria proposed by Janot et a1.15 were included in this study: clinically measurable lymph node > 3 cm or lymph node scano- 1974 CANCER November 1, 1996 / Volume 78 / Number 9 FIGURE 1. Axial postcontrast computed tomography scan shows an extremely large isodense lymph node involved the left upper jugular area with extranodal spread. The common carotid artery (large arrow) is encased by the tumor lymph node and the jugular vein is not visible. Foci of hypodense lymph nodes (arrows) are seen within the aggregation of lymph nodes. graphic greatest dimension 2 2 cm; clinically evaluable disease treated by CDDP-based neoadjuvant chemotherapy; and availability of a pretherapeutic contrast CT scan obtained within 2 weeks before the treatment initiation and showing cross-sections of relevant metastatic lymph nodes. The CT scans were performed with GE CT9800 or GE CT Prospeed Plus machines (GE Medical System, Milwaukee, WI). Iodine-based contrast material (Hypaque-76, a brand of diatrizoate meglumine and diatrizoate sodium, Sanofi Winthrop Pharmaceuticals, NY) was injected by intravenous bolus at a dosage of 100 mL. The primary tumor and the clinically relevant lymph nodes were scanned in the axial plane with 5mm thickness. The pretherapeutic contrast CT scans for staging purpose were analyzed by the radiologist in charge of the head and neck imaging section at the current study hospital, who was blinded to the treatment choice and therapeutic outcomes. The largest visible lymph node greater than 2 cm in greatest dimension was chosen for the grading purpose. The density of normal adjacent muscle tissue was chosen as the density reference. Lymph node density was graded according to the classification proposed by Munck et al.14A lymph node was classified as isodense if less than one-third of the lymph node was com- FIGURE 2. Axial postcontrast computed tomography scan shows an extremely large hypodense lymph node with extranodal spread on the right side. middle neck. prised of hypodense zones (Fig. 1) and as hypodense if one-third or more of the lymph node was comprised hypodense zones (Fig. 2). The lymph nodes with peripheral capsular enhancement and relative central hypodensity were graded as isodense if the central lymph node architecture was not more hypodense than the adjacent muscles (Fig. 3). When multiple lymph nodes were present and of a similar size with different density characteristics, the overall grading was considered as hypodense. The ENS was identified on the contrast CT scans by the radiologists as indistinct margins, thickened adjacent muscle or deep fasciae, an enhancing lymph node rim, or infiltration of the adjacent fat planes.23 Response was evaluated by clinical and radiologic studies after two cycles of chemotherapy. Nonresponders and those with progressive disease (PD) immediately underwent locoregional treatment. The responders were given a third cycle of chemotherapy and were then reevaluated for response before receiving definite locoregional treatment. Definition of response was defined by the WHO A complete response (CR) was defined if there was no evidence of tumor on gross palpation andlor by CT scan. A partial response was defined as 250% decrease in the product of the longest tumor dimension and its greatest perpendicular diameter and no increase in the size of any other known disease. Tumor regression < 50% of the Lymph Node Density and Drug Response in HNSCC/Wang et al. 1975 TABLE 1 TNM Classification"of the Patients N1 N2a N2b N2c N3 Total 2 4 2 5 6 13 17 14 27 1 T2 T3 2 2 8 10 T4 Totdl 3 5 12 12 1 12 50 71 'American Joint Committee on Cancer, 198R. TABLE 2 Lymph Node Density and Tumor Characteristics Isodense Hypodense P value 1 0.36 9 R 26: 16 8 10 26 16 0.21 5 4 47 0.10 15 11 9 33 18 0.45 (+I 15 5 1.15 I2.27 48 3 3.87 I1.70 0.03 (-1 ~ ~~~ FIGURE 3. Axial postcontrast computed tomography scan shows an extensively enlarged lower jugular lymph node with peripheral capsular enhancement and relative central hypodensity. It was graded as isodense because the central lyniph node parenchyma was not more hypodense than the adjacent muscles. It also demonstrated extranodal spread. initial tumor size was termed stable disease. PD was the product of the two greatest dimensions that showed 225% increase over the initial product, or evolution of a new lesion. Factors evaluated for their correlation with lymph node density included tumor site, histologic differentiation, 1' and N categories, and ENS. The T and N categories, lymph node density, and ENS were evaluated for their correlation with chemotherapy response. The chi-square tests and Fisher's exact test were used to analyze statisticall significance of qualitative parameters and the Student's t test of quantitative parameters. The level is accepted as significant if the P value < 0.05. RESULTS Seventy-one patients met the eligible criteria and were included in this study. The median age of the patients was 50 years (range, 30-74 years) and the sex ratio (maleifemale) was 70: 1. Approximately 86%were betel quid chewers, 90% were cigarette smokers, and 93% consumed alcohol. The primary tumor sites were the following: 9 in the oropharynx, 36 in the oral cavity, 25 in the hypophaxynx, and 1 in the larynx. TNM distribution is shown in Table 1 (according to the American Joint Committee for Cancer Staging 1988"). The neoadjuvant chemotherapy regimen comprised of CCDP + 5-FU was used in 44 patients, and CDDP + 5-FU + MI'X in 27 patients. - ~~~ ~ Primary site Oropharynx Oral cavity Hypopharynx Histologic differentiation Well Moderate-poor Primary tumor extent T1-2 T3-4 Lyniph node size N1-2 N3 10 Extranodal spread Mean size of lvmuh node icml 0.57 Fifty-one patients (72%) were classified as having hypodense lymph nodes and the remaining 20 patients as having isodense lymph nodes. The possible correlation between lymph node density and other factors such as primary site, histologic differentiation, T and N categories, and ENS was analyzed and is shown in Table 2. Eleven patients were referred from other hospitals without knowledge of the tumor differentiation. There was no statistically significant correlation between hypodensity grade and primary tumor site, histologic differentiation, or mean size of lymph nodes. The incidence of ENS was 89%, which was lower in the isodense group than in the hypodense group. The mean lymph node size was 4.16 2 1.87 cm and 2.25 2 0.35 cm in the lymph nodes with and without ENS, respectively ( P < 0.001). Correlation analysis of chemotherapy response with other factors was shown in Table 3 . There were no significant differences between T and N classification in correlation with chemotherapy response in 1976 CANCER November 1, 1996 / Volume 78 / Number 9 TABLE 3 Chemotherapy Response and Tumor Characteristics TABLE 4 Comparative Differences Between Studies Patient no. Responders ( W ) P value 21 50 16 (76.2%) 29 (58%) 0.14 44 27 21 (47.7%) 10 (37%) 0.37 20 51 7 (35%) 24 (47.1%) 63 26 (41.2%) 5 162.5%1 Janot et al. W a g et al. No. of patients 36a 71 1 13 4 22 2 0 9 36 25 1 0.36 Tumor site Nasopharynx Oropharynx Oral cavity WOPhW Larynx T stage 0.28 T1-2:T3-4 (%) N stage N1:?12:N3(%) Primary tumor TI -3 T4 Lymph node N1-2 N3 Lymph density Isodense Hypodense Extranodal spread (+I I-) a these patients with lymph node metastases. The response rate of the isodense lymph nodes (35%) was not significantly different from that of the hypodense lymph nodes (47.1%)( P = 0.36).Also, no difference in the response rate was observed between the lymph nodes with or without ENS. Hypodense lymph node % of all patients % of N1-2:% of N3 Mean lymph node size (cm) lsodense lymph node Hypodense lymph node Chemotherapy response Isodense lymph node (%) Hypodense lymph node (W) % Of N1-2:% of N3 a 13:a7 6:53:41 56 71:33 71 7567 3.3 ? 0.6 2.5 ? 0.5 4.2 ? 2.3 3.9 ? 1.7 69 25 33:60 35 47 48~30 Six patients had second primaries, total 42 primary sites. DISCUSSION Neoadjuvant chemotherapy has been widely accepted as an important part of the multidisciplinary management for patients with advanced stage HNSCC, particularly for the inoperable patients or for the purpose of organ p r e ~ e r v a t i o n . ~Because ~ ~ ’ ~ - ~only responders may benefit from this time-consuming and potentially toxic therapeutic sequence, identifymg any predictive factors of response to chemotherapy before treatment is important for more rational patient selection. In Taiwan, 80-90% of HNSCC patients chew betel quid habitually. The authors’ previous neoadjuvant chemotherapy trials revealed some characteristics that were different from the literature reports of the Western society. Their findings revealed that the primary tumor site was predominantly in the oral cavily, especially the buccal mucosa and the hard palate; there was a high prevalence of WHO Grade 3-4 chemotherapy-related mucositis (40% of patients) and more bulky tumor burden.’lz6The response rate to the CDDP and 5-FU combination regimen in the 96 patients was 56% with only 3% CR.13 The low response rate may be due to the inadequate dose intensity resulting from betel quid chewing-related mucosa damage and mucositis, and bulky tumor secondary to delayed management because of the low socioeconomic level of betel quid c h e ~ e r s . ’ ~ ~ ~ ~ ~ ~ ~ Iodinated contrasted CT scan demonstrated semiquantitative reflection of blood perfusion; therefore, the lymph node density is likely to represent distribu- tion of the contrast agent in the vasculature of the lymph node. Because an efficient vascular flow is necessary for homogeneous drug delivery in tumors, it appears reasonable to link chemotherapy response with lymph node density.I4,l5The patients with regional lymph node metastases demonstrated even worse chemotherapy response in the current study and other literature^.'^^'^^^^ It was interesting to learn whether lymph node density could be a predictive factor for chemotherapy response of metastatic lymph nodes in the study patients. Only the lymph nodes with a scanographic greatest dimension 2 2 cm were investigated, because the false-positive rate of metastatic lymph nodes at this The same grading size is considered low (4%-7%).31,32 method of lymph node density was used as that according to the report of Munck et al.14for the purpose of comparison. However, the current study failed to confirm the correlation between lymph node density and chemotherapy response. The patient characteristics from the current study were compared with those from Janot et al.15 and are presented in Table 4, both studies focused on HNSCC patients only. The current series had much more oral cavity HNSCC patients (51%) than those from the studies by Munck et al.’* (12%) and , (21 of 71) of the patients in Janot et al.I5 ( l l % )30% lymph Node Density and Drug Response in HNSCClWang et al. the current study had as primary sites the buccal mucosa and hard palate. In addition to the availability of drug delivery to the target, another factor affecting the capacity of a drug to kill the cancer cell is the cell's biochemical resistance to the The probability of a given tumor containing resistant clones, when a patient is newly diagnosed, would be a function of the mutation rate and the size of the tumor.34The incidences of lymph node metastasis at presentation were lower in patients with the buccal mucosa as the primary site of HNSCC (9-31%) and those whose primary site was the hard palate (13-24%) than those whose primary site was the pharynx (50-80%).35 Most of the oral cavity HNSCC patients enrolled in this study with lymph nodes 2 2 cm in greatest dimension had bulky tumor burden in their primary sites. Therefore, there may be more chemoresistant clones in oral cavity HNSCC and its metastatic lymph nodes. In addition, as previously described, the impact from betel quid chewing further negatively influenced lymph node chemotherapy response. These lower lymph node response rates to chemotherapy may overshadow the significant prognostic indicator of CT lymph node density. The other factor that may account for the different results may be the intrinsic defect of the grading system used, which was proposed by Munck et al.14Why was one-third of the lymph node's density considered the critical factor under this grading system, rather than one-fourth or one-fifth? A necrotic lymph node was a necrotic lymph node, and to classify them into different groups (isodense or hypodense) according to the density variations might be a false indicator and demonstrate biased data. In addition, using the largest visible lymph node greater than 2 cm in dimension for grading purposes may not correlate well with the outcome. Many HNSCC patients had multiple lymph nodes with various densities (Fig. l), and their outcomes might be affected largely by the smaller necrotic lymph nodes with poor response to treatment modalities rather than the large isodense lymph nodes. Another potential error was to grade a necrotic lymph node as isodense if the density of the central lymph node parenchyma was equivalent to adjacent muscles (Fig. 3). It was not only difficult to classify a necrotic lymph node into the isodense group, but in this situation it also might be related to the technique of CT scan. This type of grading system biased the data and demonstrated why the response of the patients in the current study with isodense lymph nodes was not better than those with hypodense lymph nodes. In addition to the issues mentioned above, the grading method used in these trials cannot reflect the truly viable tumor burden, which might be closely re- 1977 lated to the chemotherapy response within different lymph nodes.33The central lucent area of the CT scan in hypodense lymph nodes usually contains both tumor infiltration and necrosis. Whereas both of these pathologic changes cause decreased lymph attenuation at CT scan, they cause different signal intensities in magnetic resonance imaging. Most tumor cells have intermediate signal intensity on T1- and T2-weighted images. Necrotic areas have low T1-weighted and very high T2-weighted signal i n t e n ~ i t y . ~These ~ . ~ ' changes are admixed with the signal intensity of the uninvolved lymph node tissue, leading to a complex interplay of signal intensities. Because the CT scan cannot truly demonstrate the relative amount of lymph node necrosis that varies from lymph node to lymph node with a hypodense center, it is not surprising that the lymph node density of contrast CT scan correlated poorly with chemotherapy response. Also, the isodense lymph nodes may contain more viable tumor cells than hypodense lymph nodes with similar scanographic greatest dimension. This contradiction between lymph node density and viable tumor mass implies the over simplification to correlate lymph node density with chemotherapy response. Both the trial by Munck et al.I4 and the current study revealed that the hypodense aspect was independent of lymph node size, and also demonstrated that greater than 70% of lymph node metastases were graded as hypodense. However, the report by Janot et al.,15the isodense lymph nodes were shown to have a larger scanographic dimension, N3 disease had a significantly higher isodense rate than N1-2 disease. Consequently, the chemotherapy response rate of N3 disease was higher than that of N1-2 disease. This result was different from those in the literature and deserved to be ~ l a r i f i e d . " ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Although the histologic demonstration of ENS of tumor in cervical metastasis was associated with a poor outcome in patients with HNSCC, no correlation between the ENS and the lymph node response to chemotherapy was demonstrated in this trial. In lymph nodes larger than 2 cm in greatest dimension, ENS is present in greater than 50%.23The fact that 89% of the patients in the current study had ENS implied poor response to treatment modalities and poor outcomes. The great discrepancy between the numbers of patients with or without ENS may preclude a meaningful statistical analysis. CONCLUSION In conclusion, the current study failed to validate that lymph node density could be a predictor for chemotherapy response of metastatic lymph nodes in HNSCC patients. The variable response rates of lymph CANCER November 1, 1996 / Volume 78 / Number 9 1978 nodes were affected by multiple factors. Betel quid chewing behavior resulting in differences in primary tumor sites, dose intensity, and tumor burden may contribute to these conflicting results from different studies. There were also many debatable i,;(sues regarding the grading system proposed by Munck et al.,I4 and these might affect the eventual outcomes. Future trials with adequate patient population and a more clear method €or lymph node density grading may be needed before using lymph node density for patient selection in multidisciplinary managemenit of advanced HNSCC. REFERENCES 1. Department of Health, the Executive Yuan, R O C . Cancer Registry Annual Report in Taiwan Area, 1990. June 1994. Dimery IW, Hong WK. Overview of combined modality therapies for head a n d neck cancer. J Natl Cancer Inst 1993;8595-111. Schantz SP, Harrison I,B, Hong WK. Cancer of head and neck. In: Devita VT, IIellman S, Rosenberg SA, editors. Cancer: principles & practice of oncology. 4th edition. Philadelphia: J.B. Lippincott, 1993:574-630. Toohill RJ, Anderson T, Byhardt RW, Cox JD, Duncavage JA, Grossman TW, et al. Cisplatin and fluorouracil as neoadjuvant chemotherapy in head and neck cancer. Arch Otolaryngo/ Head Neck Surg 1987; 113:758-GI. The Department of Veterans Affairs Laryngeal Cancer Study Group. Induction chemotherapy plus radiation compared with surgeiy plus radiation in patients with advanced laryngeal cancer. iVEngl J Men 1991;324:1685-90. Paccagnella A, Orlando A, Marchiori C, Zorat PL, Cavaniglia G, Sileni VC, et al. Phase 111 trial of initial chemotherapy in stage 111 or IV head and neck cancers: a study by the Cruppo di Studio sui Turnori della Testa e del Collo. / Natl Cancer I~LsC1994;86:265-72. Martin M, Malaurie E, Langlet PM, Vergnes L, 1.elievre G, Mazeron JJ, et al. A randomized prospective study of CDDP and 5-FU as neoadjuvant chemotherapy in head a n d neck cancer: a final report. Proc A m Soc Clin O n c d 1995; 14:294. Jacobs C, Goffniet DR, Goffinet L, Kohler M, Fee WE. Chemotherapy as a substitute for surgery in the treatment of advanced resectable head and neck cancer. Cancer 1987; 60:1178-83. Pfister DG, Strong E, Harrison I., Haines IE, Pfister DA, Sessions 11, et al. Larynx preservation with combined chemotherapy and radiation therapy in advanced but resectable head and neck cancer. J Clin Oncol1991;9:850-9. Cognetti F, Pinnaro P, Ruggeri EM, Carlini P, Perrino A, Impiombato FA, et al. Prognostic factors for chemotherapy response and survival using combination chemotherapy as initial treatment of advanced head and neck squamous cell cancer. J Clitz Oncol 1989;7:829-37. Wolf G l , Makuch IIW, Raker SR. Predictive factors for tumor response to preoperative chemotherapy in patients with head and neck squainous carcinoma: the head and neck contracts program. Cuncer 1984;54:2869-77. Clavel M, Mansour AR. Head and neck cancer: prognostic factors for response to chemotherapy. Eur I Cuncer 1991;27:349-56. Wang HM, Wang CH, Chen JS, Chang HK, Kiu MC, h a w CC, et al. Cisplatin and 5-FU as neoadjuvant chemotherapy: Y 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. . 18. 19. predicting response in head anti neck squamous cell cancer. / Forrnos Med Assoc 1995;94:87-94. Munck JK, Cvitkovic E, Piekarski ID, Benhamou E, Recondo G, Bachouchi M, et al. Computed tomographic density of metastatic lymph nodes as a treatment-related prognostic factor in advanced head and neck cancer. / Nut1 Cancer lnst 1991;83:569-75. Janot F, Cvitkovic E, Piekarski JD, Sigal R, Armand JP, Bensm a h e A, et al. Correlation between nodal density in contrasted scans an d response to cisplatin-based chemotherapy in head and neck squamous cell cancer: a prospective validation. Head Neck 1993; 15222-9. Chong KM. Betel nut chewing and mouth cancer in Taiwan. Observation of the oral mucosa in the betel nut chewers. J Forrnos Med Assoc 1966;65:79-85. Kwan HW. A statistical study o n oral carcinomas in Taiwan with emphasis o n the relationship with betel nut chewing: a preliminary report. J Fornzos Med Assoc 1976;75:497505. Kuo MW, Jeng JH, Chiang CP, Hahn LJ. Mutations of Ki-ras oncogene codon. 12 in betel quid chewing-related human oral squamous cell carcinoma in Taiwan. J Orul Patho/ Med 1994;23:70-4. KO YC. A review of betel quid chewing and its health effects in Taiwan. Symposium on betel quid chewing and its health effect [in Chinese] [abstract]. Krrolisiiing R.O.C. 1993;11-2. 20. Johnson JT, Barnes L, Myers EN. The extracapsular spread of tumors in cervical node metastasis. ;Irclz Otolnryngol 1981;107:725-9. 21. Snow GB, Annyas AA, VanSlooten AE, Rartelink H, Hart AAM. Prognostic factors of neck node metastasis. Clin.O t o h vyngol 1982;7:185-92. 22. Johnson JT, Myers EN, Bedetti CD, Barnes EL, Schrainm VL, Thearle PB. Cervical lymph node metastases. Arch Otolar~mgo1 Head Neck Surg 1985; 11 1 :534-7. 23. Son1 PM. Lymph nodes. In: Som PM, Curtin HD, editors. Head and neck imaging. 3rd edition. St. Louis: Mosby-Year Book, Inc., 1996:772-93. 24. Miller AB, Hoogstraten B, Staquet M, Winkler A. Reporting results of cancer treatment. Cancer 1981;47:207-14. 25. American Joint Committee on Cancer. Manual for sraging of cancer. 3rd edition. Philadelphia: J.B. Lippincott, 1988. 26. Muir C, Weiland I.. Upper aerodigestive tract cancer. Cuncer 1995;75: 147-53. 27. Rooney M, Kish 1, Jacobs J, Kinzie J, Weaver A, Crissman J, et al. Improved complete response rate and survival in advanced head and neck cancer after three-course induction therapy with 120-hour 5 - F U infusion and cisplatin. Cancer 1985;55:112:%-8. 28. Ko YC, Chiand TA, Chang SJ, Hsieh SF. Prevalence of betel quid chewing habit in Taiwan and related sociodemographic factors. J Oral Putliol Med 1992;21:261-4. 29. Bachaud JM, David J M , Shubinski RF, Perineau D, Boussin G, Serrano E, et al. Predictive factors of a complete response to an d adverse effects of a CDDP-5-FU conibination as primary therapy for head an d neck squamous carcinomas. / I,aryngo/ O f o l 1993;107:924-30. 30. ‘l‘hyss A, Schneider M, Santini J , Caldani C, Vallicioni J, Chauvel P, et al. Induction chemotherapy with cisplatin a nd 5-fluorouracil for squamous cell carcinoma of the head and neck. Br J Cancer 1986;54:755-60. Lymph Node Density and Drug Response in HNSCCMlang et al. 31. Moreau P, Goffard Y, Collignon J. Computed tomography of metastatic lymph nodes: a clinical, computed tomographic, pathologic correlative study. Arch Otolaryngol Head Neck Surg 1990;116:1190-3. 32. Rainer T, Ofner G, Marckhgott E. Ultrasound diagnosis of regional lymph node metastasis of the neck in patients with head-neck neoplasm: sono-morphologic criteria and diagnostic accuracy. Laryngorhinootologie 1993;7273-7. 33. Devita VT. Principles of chemotherapy. In: Devita VT, Hellman S, Rosenberg SA, editors. Cancer: principles & practice of oncology. 4th edition. Philadelphia: J.B. Lippincott, 1993;236-92. 34. Goldie JH, Coldman AJ. A mathematic model for relating the drug sensitivity of tumors to the spontaneous mutation rate. Cancer Treat Rep 1979;63:1727-33. 1979 35. Million RR, Cassisi NJ, Mancuso AA, Stringer SP, Mendenhall WM, Parsons JT. Management of the neck for squamous cell carcinoma. In: Million RR, Cassisi NJ, editors. Management of head and neck cancer: a multidisciplinary approach. 2nd edition. Philadelphia: J.B. Lippincott, 1994:75- 142. 36. Mancuso AA, Dillon WP. The neck. Radio1 Clin North A m 1989;27:407-34. 37. Yousem DM, Sam PM, Hackney DB, Schwaibold F, Hendrix RA. Central nodal necrosis and extracapsular neoplastic spread in cervical lymph nodes: MR imaging versus CT. Radiology 1992;182:753-9. 38. Davis RK, Stoker K, Harker G, Davis K, Gibbs FA, Harnsberger HR, et al. Prognostic indicators in head and neck cancer patients receiving combined therapy. Arch Otolaryngol Head Neck Surg 1989;115:1443-6.