C in Spain, representing 5.6% of all malignant tumors,

Transcription

1 Prognostic Factors in Laryngeal Carcinoma A Multifactorial Study of 416 Cases EUGENIA PERA, MD,* ABELARDO MORENO, MD,t AND LORENZO GALINDO, MD,t The prognostic significance of 20 clinical and histologic parameters was evaluated in a series of 416 laryngeal carcinomas treated between 1969 and 1981 at the Hospital de Santa Cruz y San Pablo, Barcelona, Spain. Clinical follow-up ranged from 2 to 13 years. The variables considered were: age, smoking habits, previous laryngeal disease, duration of clinical symptoms, motility of vocal cords, interval between diagnosis and treatment, postoperative infection, topographic location of the tumor, previous tracheostomy, ulceration, tumor size, resection margins, histologic grading, patterns of growth, types of invasion, mitotic index, depth of invasion, positive lymph nodes, T categories and N categories. Fourteen parameters were found to be significantly related to survival. After a multifactorial study using Cox's life table regression model, 5 of those 14 parameters were found to influence survival independently. The five variables ordered according to prognostic significance were: (1) positive lymph nodes, (2) T categories, (3) histologic grading, (4) ulceration, and (5) location. Cancer 58: , ARCINOMA OF THE LARYNX is a common neoplasm C in Spain, representing 5.6% of all malignant tumors, according to data of 4 1 laboratories of Surgical Pathology.' The great majority are squamous cell carcinomas, with only infrequent examples of other histologic variants (small and large cell undifferentiated carcinomas and adenocystic and mucoepidermoid carcinomas). Verrucous car~inoma~,~ occurs rarely, and is associated with a favorable prognosis. Several studies have disclosed a group of major variables influencing survival rates, i.e., size of tumor, cord motility, histologic differentiation, and lymph node The impact on prognosis is less clear for a number of other clinical and morphologic parameters. Previous data from our series" demonstrate that many of the prognostic variables are interrelated. For instance, lymph node involvement is significantly related to histologic subtype (poorly differentiated squamous cell carcinoma and undifferentiated carcinomas), age (older patients), T parameter (T3 and T4) neoplasms), size (tumors greater than 2.5 cm.), and topographic location (supraglottic and hypopharyngeal cancers). For this reason, it is From the Departments of *Oto-rhino-laryngology and tpathology Hospital de la Santa Cruz y San Pablo, Universidad Autbnoma de Barcelona, Barcelona, Spain. Address for reprints: Abelardo Moreno, MD, Department of Pathology, Hospital de la Santa Cruz y San Pablo, Av. San Antonio Ma Claret 167, 025 Barcelona, Spain. Accepted for publication December 4, important to know the statistically independent significance of each prognostic variable under consideration. In this study, 20 clinical and histologic parameters were evaluated to assess their prognostic influence in survival. In the first part of this study, each parameter is considered independently. In the second part, variables with prognostic significance are compared to obtain a final set of independent factors influencing survival. Materials and Methods Four hundred forty-eight cases of carcinoma of the larynx diagnosed at our hospital from 1969 to 1981 were culled from our files. Thirty-two cases were excluded because of widespread metastatic disease, inadequate followup, or lack of histologic proof for verification of diagnosis. The remaining 416 cases were reviewed with regard to general clinical data, laryngoscopic findings, gross pathologic description, and microscopic findings. All patients were treated according to the same therapeutic protocol, namely: surgery followed by radiotherapy (50 to 60 rad). Clinical follow-up ranged from 2 to 13 years. Before the data were computerized, the follow-up was actualized by contact with patients. Tumor site, ulceration, and size were obtained from the gross pathologic description. Tumor location was classified into glottic, supraglottic, transglottic, and hypopharyngeal categories. Size of the tumor was determined by measuring the largest diameter of the lesion. 928

2 No. 4 LARYNGEAL CARCINOMA - Pera et al. 929 Motility of vocal cords was evaluated by laringoscopy. Histologic subtypes, margin of resection, pattern of growth and invasion (pushing and infiltrative), mitotic index, depth of invasion and lymph node involvement were evaluated by reviewing the microscopic slides in all cases. Mitotic index was recorded by counting mitosis in 10 high-power fields (4X) in areas of maximal mitotic activity. Assignment of T and N categories was made using the American Joint Committee definitions of TNM categories for laryngeal cancer. I Each parameter was evaluated for survival according to Gehan s actuarial survival method.12 Subgroups were compared by the method described by Lee and Desu.13 Multivariate analysis was performed using a Cox s life regression model.14 Results All patients in this series were males with a median age of 58.4 years (range, 2 1 to 87 years). Patients older than 60 years had a significantly worse prognosis (P = 0.02). Smoking habits were recorded in all but six patients. The median value was 48 packs/year. Patients smoking 42 packslyear or more had a significantly worse prognosis (P = ). A previous history of laryngeal disease was recorded in 7 I patients ( 17%). There were no differences in survival within patients with and without previous laryngeal diseases. The duration of clinical symptoms ranged from 1 to 180 months (median, 12.2 months). Duration of symptoms did not influence survival in our series. One hundred seventy-four patients (4 1.8%) had normal cordal motility; 101 patients (24.3%) had decreased cordal motility; and the remaining 141 patients (33.9%) showed absent cordal motility. When these three groups of patients were compared, statistical differences were obtained (P = 0.01). Patients with absent cordal motility had a worse prognosis as compared with the other two subgroups. There was no difference in survival between patients with normal and decreased cordal motility (P = ). The interval between diagnosis and treatment ranged from 1 day (intraoperatory diagnosis) to 83 days (median, 25.7 days). Length of this interval did not influence survival. Postoperative infection was recorded in 93 patients (22.4%). Patients with this complication had a significantly worse prognosis as compared to patients lacking postoperative infection (P = 0.0 1). Preoperative tracheostomy was performed in 24 patients (5.8%). Preoperative tracheostomy did not influence survival. The distribution of neoplasms according to anatomic location is shown in Table 1. In our series, the most com- TABLE I. Distribution of Neoplasms According to Anatomical Location No. Percent Supraglottis I Glottis Transglottis Hypopharynx mon location was the supraglottis (I95 cases, 46.9%), followed by glottis (1 37 cases, 32.9%). Transglottic and hypopharyngeal lesions were infrequent. Comparison of survival disclosed statistical differences in prognosis when cases were grouped by anatomic location (Fig. 1). Significative differences in survival were found between supraglottic and hypopharyngeal lesions (P = 0. 16), glottic and hypopharyngeal lesions (P = O.OOl), and glottic and transglottic lesions (P = ). Ulceration of the tumor was recorded in 63% of the cases. Comparative study of survival yielded significative differences between ulcerated and nonulcerated carcinomas (P = 0.03), the former having a poorer prognosis (Fig. 2). Tumor size ranged between 0.4 and 7.0 cm. Fifty percent of the neoplasms were smaller than 2.5 cm., but only 16% of the cases had a maximum diameter smaller than 2 cm. Lesions smaller than 2.5 cm. carried a significative better prognosis (P = 0.03). There was tumor infiltration at the resection margins in 89 cases (2 1.4%), with a better prognosis for patients with adequate surgical resection (P = 0. 19). Table 2 reflects the distribution of cases according to histologic classification. Since poorly differentiated squamous cell carcinomas and small and large cell undifferentiated carcinomas represent a small number, they have been grouped together. The infrequent histologic subtypes of carcinomas were eliminated from this study. When survival was compared, there were statistical differences between subgroups (Fig. 3). Moreover, undifferentiated and poorly differentiated carcinomas had a significantly worse prognosis when compared with well-differentiated and moderately differentiated squamous cell carcinomas (P = 0.57 and P = , respectively). There were no statistical differences between the latter two subgroups (P = ). The gross pattern of growth was interpreted as exophytic (polypoid) in 14 tumors (3.4%); endophytic in 377 tumors (90.6%), and mixed endo-exophytic in 2 1 tumors (5.0%). In the remaining cases, this parameter could not be evaluated. Comparative survival rates demonstrated a worse prognosis for endophytic neoplasms (P = ). Statistical differences were also found when endophytic and exophytic neoplasms were compared, but not between endophytic and mixed endo-exophytic tumors (P

3 930 CANCER August Vol e FIG. 1. Survival according to location. 1: Supraglottic carcinomas; 2: glottic carcinomas; 3: hypopharyngeal carcinomas; 4: transglottic carcinomas. (Vertical: proportion surviving; horizontal: months; P = 0.05) c = 0.14), or between mixed-pattern and exophytic neoplasms (P = ). According to the pattern of invasion, neoplasms with infiltrative growth (332 cases) predominated over tumors with expansive growth. In five cases, this parameter could not be evaluated. There were no statistical differences in survival between these two subgroups (P = ). Mitotic index ranged from 1 to 123 mitotic figures/lo high-power fields (median, 39.5) in the 402 cases in which this parameter could be evaluated. Four subgroups were OA FIG. 2. Survival according to tumor ulceration. 1 : Ulcerated neoplasms; 2: nonulcerated neoplasms. (Vertical: proportion surviving; horizontal: months; P = 0.03)

4 ~ No. 4 LARYNGEAL CARCINOMA - Peru ef ul. 93 I TABLE 2. Distribution of Cases According to Histologic Classification TABLE 3. Distribution of the Depth of Invasion* No. Percent No. Percent Carcinoma in srtu Well-differentiated SCC Moderately differentiated SCC Poorly differentiated SCC and undifferentiated SCC: squamous cell carcinoma made representing count values of less than 16, 16 to 30, 31 to 53 and more than 53. Cut-off numbers represents the upper limits in 25%, 50%, and 75% of the series, respectively. Number of mitosis did not influence survival (P = 0.75). Table 3 represents the frequence distribution of depth of invasion according to anatomic levels (mucosa, submucosa, cartilage). In one case, this parameter could not be evaluated. Two in situ carcinomas were also excluded because of its low incidence in our series. When the other three subgroups were evaluated, we observed statistical differences, with a poorer prognosis for neoplasms infiltrating the whole laryngeal wall (P = 0.01). Differences were also obtained when tumors infiltrating the whole laryngeal wall were compared to those reaching cartilage (P = 0.3). However, there were no significant differences between the other two subgroups. Lymph node metastasis was noted in 101 cases (24.3%). There was a significantly worse prognosis in cases with Mucosa Submucosa Cartilage All lavers * One case not evaluated. TABLE 4. Distribution of Cases According to T Categories No. TI 50 T2 I29 T3 191 T4 46 Percent positive lymph nodes (Fig. 4). Multiple metastatic lymph nodes carried also a poor prognosis (P = 0.62). Table 4 represents the distribution of cases according to T category. There was a significative difference in survival among the four subgroups (Fig. 5): between TI and T2 neoplasms (P = 0.56), between T2 and T3 neoplasms (P = 0.05), between T1 and T4 neoplasms (p = 0.01), and between T2 and T4 neoplasms (P = 0.0 1). When N subgroups were compared regarding prognosis (NO, N 1 +2, N3), statistically significative differences were found (P = 0.0 1). Significative differences were also obtained between NO and N3 (P = 0.01) and R FIG. 3. Survival according to histological classification. I : Well-differentiated SCC, 2: moderately 0.5 differentiated SCC; 3: poorly differentiated SCC and undifferentiated carcinomas. (Vertical: proportion surviving; horizontal: months; P = ) ao

5 ~ 932 CANCER August Vol FIG. 4. Survival according to lymph node metastasis. 1: Present; 2: absent. (Vertical: proportion surviving; horizontal: months; P = 0.) N 1 +2 versus N3 (P = ), but not between NO and N 1 +2 subgroups. Multivariate Studv When the previously obtained significative variables were entered into Cox s regression equation in a stepwise fashion (condition P < O.l), a final set of five indepen- dently significative prognostic variables was obtained (Table 5). These are: positive lymph nodes, T category, histologic grading, ulceration, and anatomic location. Discussion A set of 20 clinical and histologic parameters were evaluated to assess their independent prognostic significance FIG. 5. Survival according to T categories. I: TI; 2: T2; 3: T3; 4: T4. (Vertical: proportion surviving; horizontal: months; P = 0.). 03 m x

6 No. 4 LARYNGEAL CARCINOMA * Peru et al TABLE 5. Cox s Regression Model: Summary of Stepwise Results Improvement Global Step no. Variable entered DF Log likelihood Chi-square P value Chi-square P value 1 Lymph nodes I T category O.Oo0 3 H Classification , Ulceration Location DF: degrees of freedom in a series of 4 16 laryngeal carcinomas. After a multivariate analysis using Cox s life table regression model, five independent variables were obtained: positive lymph nodes, T category, histologic grading, ulceration, and anatomic location. Evidence of lymph node metastasis is the most important prognostic factor according to our results and to the data of Desaulty and Sancho Garnier. As shown by the multifactorial study, the number of involved lymph nodes and N category have no independent prognostic significance. In OUT study, only histologic evidence of lymph node involvement has been considered. According to our results, prophylactic lymphadenectomy for all T3 and T4 carcinomas seems advisable. The prognostic implications of T categories are of special significance for the clinical evaluation and preoperative prognostic prediction of patients harboring la0 ngeal carcinoma. A complete preoperative study, including laryngography and computerized axial tomography (CAT) should help in better assessment of T categories. The correct therapeutic approach to laryngeal carcinoma should be chosen after a complete clinical evaluation. We have not been able to find any report concerning the prognostic value of T categories using a multifactorial analysis. Histologic classification is important in the prediction of prognosis in laryngeal carcinoma. The two opposite poles in the spectrum of morphologic classification, namely verrucous carcinoma and small cell undifferentiated carcinoma. are histologic vanants with a well-defined clinical outcome, good prognosis for the former and aggressive behavior for the latter. Several studies in the literature have stressed the importance of histologic classification in predicting However, no multifactorial studies evaluating the independent significance of this parameter have been published. Prognostic differences in the spectrum of differentiation of squamous cell carcinoma are of practical significance. It is important to emphasize that prognostic differences exists only for poorly differentiated squamous cell carcinoma. The morphologic differentiation of squamous between well-differentiated and moderately differentiated squamous cell carcinoma is sometimes difficult because of sampling errors and interpretation of morphologic criteria. However, this differentiation is not of prognostic significance. Aggressive surgical procedures are indicated for poorly differentiated carcinomas independent of T categories. The prognostic value of tumor ulceration has not been evaluated in multifactorial studies of laryngeal carcinomas. Although ulcerated carcinomas are large invasive neoplasms, the prognostic significance of ulceratjon is independent of the level of invasion and the tumor size. There is a general agreement in the literature on the prognostic significance of anatomic location in laryngeal ~arcinoma..~~.~~ Our study demonstrates that this prognostic significance is independent from other prognostic factors. Differences in laryngeal wall thickness, presence of anatomic barriers.20~2 and distribution of lymphatic vas~ulature~~.~~ may explain this prognostic difference. As a final conclusion, independent prognostic factors derived from this study, should be used in future therapeutic protocols in order to increase their efficiency and reduce mortality due to laryngeal carcinoma. REFERENCES 1. Galindo L, Algaba F. Patologia GeogrLfica del Cancer en Espafia: Estudio global. VIII Congreso Nacional de Anatomia Patol6gica. Tenerife, Espaiia. Patologia 1981 (Suppl especial); Kraus FT, Perez Mesa C. Vermcous carcinoma: Clinical and pathological study of 105 cases involving oral cavity, larynx and genitalia. Cancer 1966; 19: Fisher R. Vemcous carcinoma of the larynx: A study of its pathologic anatomy. In: Workshops From the Centennial Conference on Laryngeal Cancer. Toronto, Canada: Appleton-Century-Croffs, 1974; Nahum A, Bone R, Davidson T. The case for elective prophylactic neck dissection. Laryngoscope 1977; Desaulty A, Sancho Garnier H. Etude prospective des kpithtliomes du larynx: (EPEL ). Rtsultats d une ttude multicentrique. Ann Otolaryizgol (Paris) 1983; 1: Ferlito A. Histological classification of larynx and hypopharynx cancers and their clinical implications. Acta OfolurJ3izgol (S/ock/z) 1976; (SUPPI) 342: Bennett S, Futrell J, Roth J, Hoye R, Ketchan D. Prognostic significance of histologic host response in cancer of the larynx or hypopharynx. Cancer 1971; 28: Schuller D, McGuirt W, McCabe B, Young D. The prognostic significance of metastatic cervical lymph nodes. Laryrigoscope 1980; 90: Mittal B. Marks J, Ogura J. Transglottic carcinoma. Conccr 1984; 53: Pera E. Carcinoma de laringe: Evaluaci6n pron6stica del interval0 libre de enfermedad y de la supervivencia. Tesis doctoral. Barcelona: Universidad Autonoma de Barcelona, American Joint Committee for Cancer Staining and End Results

7 934 CANCER August Vol. 58 Reporting. Staging ofcancer at Head and Neck Sites: Oral Cavity, Pharynx, Larynx and Paranasal Sinuses. Chicago: American Joint Committee. 1977; Gehan EA. Statistical methods for survival time studies. In Cancer therapy: Prognostic factors and criteria. New York Raven Press, 1975; Lee E, Desu M. A computer program for comparing K samples with right censored data. Compuf Progr Med 1975; Cox DR. Regression models for life tables. J Roy Stat Soc 1972; 34: Kashima HK. The characteristics of laryngeal cancer correlating with cervical lymph node metastasis. In: Workshops From the Centennial Conference on Laryngeal Cancer. Toronto, Canada: Appleton-Century- Crofts, 1974; Sala 0, Ferlito A. Morphological observations of immunisgiology of laryngeal cancer. Acta Otoluryngol (Stockh) 1976; 81: Shah J, Randall H. Epidermoid carcinoma of the supraglottic larynx. Am J Surg 1974; 128: Cachin Y, Richard J, Eschwege F, Micheau C. Les cancer du larynx: Collection de monographies de canctrologie. Pans: Masson, Lederman M. Radiotherapy of cancer of the larynx. J Laryngol ; 74: Bridger GP, Nassar VH. Cancer spread in the larynx. Arch Otolutyngol 1972; 95: Tucker G. The anatomy of laryngeal cancer. In: Workshops From the Centennial Conference on Laryngeal Cancer. Toronto, Canada: Appleton-Century-Crofts, 1974; El Badawi S, Goepfert H, Fletcher G, Herson J, Oswald M. Squamous cell carcinoma of the pyriform sinus. Laryngoscope 1982; 92: Welsh L, Rizzo T. Laryngeal spaces and lymphatic: Current anatomic concepts. Ann Otol Rhino1 Luryngol 1983; (Suppl) 92: