P the esophagus may differ from those of squamous

Transcription

1 Incidence of Cancer of the Esophagus in the US by Histologic Type PAUL c. YANG, MD, MPH, AND SCOTT DAVIS, PHD' Data from nine US population-based cancer registries participating in the Surveillance, Epidemiology, and End Results (SEER) program from 1973 through 1982 were analyzed to examine demographic characteristics related to the Occurrence of the two major types of cancer of the esophagus. The overall annual incidence rate per 100,000 persons was 2.6 for squamous cell carcinoma and 0.4 for adenocarcinoma. The sex ratio for adenocarcinoma varied from one age group to the next and was highest in the 50- to 59-year-old group. It was relatively the same for squamous cell carcinoma. The male-to-female ratio was higher for adenocarcinoma (seven in whites and 10 in blacks) than for squamous cell carcinoma (three and four, respectively). The highest sex-specific ratio for adenocarcinoma occurred in the lower third of the esophagus. Blacks had a fourfold to fivefold higher rate of squamous cell carcinoma than whites, but the rate of adenocarcinoma in blacks was 30% of the rate in whites. The incidence of squamous cell carcinoma in black men and women increased by approximately 30% between 1973 and 1982, and the rate of adenocarcinoma among white men increased 74%. Nearly half of the squamous cell carcinomas occurred in the middle of the esophagus, whereas the majority (79%) of the adenocarcinomas arose in the lower third. These data suggest that the two major histologic types of esophageal cancer may be of different etiologic origin. Cancer 61: , ATTERNS OF OCCURRENCE of adenocarcinoma of P the esophagus may differ from those of squamous cell carcinoma in terms of its association with mean age at diagnosis, sex, race, history of smoking and alcohol consumption, and hiatal hernia.'** We analyzed data from a US network of population-based cancer registries to identify and describe epidemiologic features of these two major histologic types of cancer of the esophagus. The study was specifically designed to evaluate incidence by age, sex, race, geographic area, year of diagnosis and anatomic subsite. The results of such an approach may be helpful in describing etiologically different classes of tumors and could serve to direct future studies of more specific etiologic hypotheses. Methods Persons diagnosed with cancer of the esophagus between 1973 and 1982 were identified through the nine From the Department of Epidemiology, SC-36, School of Public Health and Community Medicine, University of Washington, Seattle, Washington and the Division of Public Health Sciences, Fred Hutchinson Cancer Research Center. Seattle, Washington. * Leukemia Society of American Special Fellow. Address for reprints: Scott Davis, PhD. Fred Hutchinson Cancer Research Center, Division of Public Health Sciences, Columbia Street. Seattle, WA Accepted for publication August population-based cancer registries in the United States that comprise the National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) program. The areas served by SEER registries include the states of Connecticut, Hawaii, Iowa, New Mexico, Utah, and the urban areas surrounding San Francisco/Oakland, Detroit, Seattle (since 1974), and Atlanta (since 1975). Using information obtained primarily from the records of hospitals in these areas. the subsite, histology, and behavior of all neoplasms were coded according to the International Classification of Diseases for Oncology (ICD-0). Additional operational details of the SEER program are provided else~here.~ The series documented in this article includes all incident cases diagnosed with cancer of the esophagus (ICD-0 site codes of to 150.9, excluding gastroesophageal junction). Persons 80 years and older at diagnosis were excluded because the diagnosis of esophageal cancer and the specification of histologic type are more likely to be incomplete and less accurate in the very old. Four hundred and ninety-five cases (7.5%) whose diagnoses were not histologically confirmed were also excluded. There were 6044 cases available for analysis. Estimates of the populations covered by each of the SEER registries are based on data supplied to SEER by the US Bureau of the Census and were used to calculate 612

2 No. 3 CANCER OF THE ESOPHAGUS BY HISTOLOGY - Yang and Davis 613 TABLE 1. Age-Specific Incidence Rates* of Cancer of the Esophagus by Sex and Histologic Type for All Races Adenocarcinoma Squamous cell carcinoma Age Male Female M/Ft Male Female M/Ft (4)$ 0.0 (0) (3) 0.0 ( I) (11) 0.0 (1) (22) 0.1 (8) (62) 1.8(182) 0.1 (I 1) 0.1 (15) (237) 9.1 (933) 1.0 ( 1 11) 3.7 (409) (221) 0.4 (33) ( I 378) 6.2 (524) (144) 0.7 (49) (856) 6.1 (331) 3.7 * Annual rate per 100,000 persons. t Ratio of incidence rates based on three significant digits. $ Numbers in parentheses are number of cases. incidence rates in a standard manner. Where indicated, incidence rates were age-adjusted to the 1980 US population using the direct method of adjustment. Tests for linear trends in incidence rates over time were performed using the software developed by Rothman and B~ice.~ The classification of tumor origin within the esophagus is usually based on a constellation of roentgenologic, endoscopic, and surgical information, and/or the gross description of resected specimens. Because there is no anatomic landmark for a guide in classifying the tumor subsite, the following two systems have been adopted by the ICD-0: upper-middle-lower versus cervical-thoracic-abdominal. The former has been used by the majority of clinicians and was used to classify 87% of the cancers in this study. For analytic purposes, we grouped cancer of the cervical esophagus into the upper third; thoracic into the middle third; and abdominal into the lower third of the esophagus. Results The overall annual age adjusted incidence rate for malignant tumors of the esophagus in persons younger than 80, all histologies combined, was 3.2 per 100,000 persons. The rate was 2.6 for squamous cell carcinoma and 0.4 for adenocarcinoma. Tumors of other histologic types were rare. Table 1 shows the age-specific incidence rates by sex for the two major histologic types of esophageal cancer. For both types the rates were extremely low under age 40, beyond which they continued to rise with each increasing decade of age. The male-to-female ratio for adenocarcinoma vaned from one age group to the next and peaked at 13.1 during the fifth decade of life. The maleto-female ratio for squamous cell carcinoma was relatively the same from one age group to the next. The male-to-female ratio generally exhibited a similar pattern according to histologic type in whites and blacks. Males experienced an approximately threefold to fourfold greater risk than females for squamous cell carcinoma, and a sevenfold to tenfold higher risk for adenocarcinoma (Table 2). The incidence of squamous cell carcinoma was about fourfold to fivefold higher in blacks than in whites, but the incidence rate of adenocarcinoma in blacks was only one third of that observed in whites (Table 3). There was no noteworthy change in the incidence of squamous cell carcinoma in white men or women between 1973 and In contrast, the rate increased approximately 30% in both black men (P = 0.02) and women (P = 0.01) (Fig. 1). While the incidence of adenocarcinoma among white men increased by 74% (P < 0.001) during the ten-year study period, no comparable increase was seen in white women (Fig. 2). The inci- TABLE 2. Incidence Rates of Cancer of the Esophagus by Histologic Type and by Race and Sex White Black Cell type Male Female M/Ft Male Female M/Ft Squamous cell carcinoma 2.9 (2146)$ 1.1 (992) I (985) 4.1 (329) 3.7 Adenocarcinoma 0.8 (594) 0.1 (96) (I 5) 0.0 (2) 10.0 Other 0.2 (151) 0.1 (58) (40) 0.2 (20) 2.8 NOS 0.2 (1 19) 0.1 (61) (24) 0.2 (12) 2.4 Total 4.1 (3010) 1.4 (1207) (1064) 4.5 (363) 3.6 NOS histology not othenvise specified. * Annual rate per 100,OOO persons adjusted to the age distribution of the 1980 US standard population by 5-year age categories. t Ratio of incidence rates based on three significant digits. $ Numbers in parentheses are number of cases.

3 614 CANCER February I 1988 Vol. 6 I TABLE 3. Incidence Rates* of Cancer of the Esophagus by Histologic Type and by Sex and Race Male Female Cell type White Black B/Wt White Black BIWt Squamous cell carcinoma I I 3.7 Adencarcinoma Other NOS NOS histology not otherwise specified. * Annual rate per 100,OOO persons adjusted to the age distribution of dence of histology-not-specified tumors remained stable over time. Because the incidence of esophageal cancer varies among different racial groups, and the racial composition of the nine SEER areas differs, we compared the incidence among whites only in each of the participating SEER areas to examine the influence of geographic area on esophageal cancer. There was variation in the rates of each of the two histologic types across the nine participating areas (Table 4). The magnitude of the variation was greater for adenocarcinoma than for squamous cell carcinoma, but there was no clear geographic pattern apparent for either type. The distribution of tumor subsite within the esophagus differed considerably between the two histologic types (Table 5). Squamous cell carcinoma occurred most often (50%) in the middle third of the esophagus the I980 US standard population by 5-year age categories. t Ratio of incidence rates based on three significant digits. whereas adenocarcinoma arose primarily (79%) from the lower third. While the male-to-female ratio for squamous cell carcinoma showed no noteworthy variation along the esophagus, it was the highest in the lower third for adenocarcinoma (Table 6). Discussion At present there is little population-based information that describes the epidemiology of the separate histologic types of cancer of the esophagus. Nevertheless, the overwhelming predominance of adenocarcinoma in men and in whites observed here has been consistently reported from several case series'*296 and cancer registries3" The increase in the incidence of squamous cell carcinoma among black men and women observed here is consistent with that reported from Detroit, one of the v) z 8 14 W (1 P- BLACK MALES RG. 1. Age-adjusted annual incidence rates of squamous cell carcinoma of the esophagus by sex, race, and year of diagnosis. Rates are adjusted to age distribution of the 1980 US population by 5-year age categories WHITE I I I YEARS FWS

4 No. 3 CANCER OF THE ESOPHAGUS BY HISTOLOGY - Yang and Davis FIG. 2. Age-adjusted annual incidence rates of adenocarcinoma of : 0,5 - the esophagus by sex, race, and year of diagnosis. Rates are adjusted to age distribution of the 1980 US population by 5-year age u $ # 0,4 - - g 3 0,2-0,l -, 0,O /&-\ 4 / # WHITE FEMALES /) --*/ 3 I I I I YEARS participants in the SEER program,* and corresponds well with the recently reported trends of esophageal cancer mortality in the US.' The potential concerns in the interpretation of the reported site-specific incidence rates are three issues related to the classification and reporting of tumors by histologic type and primary site. First, it is probable that some degree of misclassification occurred as a result of combining two different classification systems. This is not likely to be of consequence for tumors of any subsite except those in the middle third of the esophagus. To the extent that tumors classified as being of thoracic origin actually arose from the lower third, our site-specific incidence rates would artificially inflate the rates for the middle third. However, such bias is likely to have been minimal because tumors coded as thoracic in origin comprised only 5% of the site-specified tumors in our series. Also, the distribution of histologic types for tumors coded as originating from this subsite was similar to that for tumors originating in the middle third (85% versus 89%, respectively, being squamous cell carcinoma; 6% versus 5%, respectively, being adenocarcinoma). Second, a sizable proportion (14%) of cancer of the esophagus in our series was coded as subsite-not-specified. If tumors of certain subsites were preferentially unclassified, the observed site-specific rates might be biased. This type of bias seems unlikely in the present dataset, however, because the proportional distribution of the separate histologic types within the esophagus presented here closely resembles that reported from several US cancer registries between 1955 and 1976,7 and from the US Third National Cancer Survey from 1969 to Third, there is a possibility that the proportion of tumors of certain subsites coded as site-not-specified has differed for men and women. However, this is not likely TABLE 4. Incidence. of Cancer of the Esophagus in Whites by Area and Histologic Type Squamous All histologic SEER area cell carcinoma Adenocarcinoma ty pes San Francisco 2.3 (604)t 0.3 (81) 2.9 (747) Connecticut 2.5 (751) 0.5 (139) 3.3 (985) Detroit 2.3 (684) 0.4(121) 3.0 (891) Hawaii 2.7 (51) 0.1 (3) 2.9 (56) Iowa New Mexico 1.4 (401) 0.9 (81) 0.4(121) 0.3 (24) 2.0 (575) 1.4 (126) Seattle 1.7 (342) 0.7 ( 134) 2.6 (521) Utah 0.8 (71) 0.4 (41) 1.3 (121) Atlanta 2.1 (153) 0.3 (26) 2.6 (195) All Areas 1.9 (3138) 0.4 (690) 2.6 (4217) SEER: Surveillance, Epidemiology, and End Results Program. * Annual rates per 100,OOO persons adjusted to the age distribution of 1980 US standard population by 5-year age categories. t Numbers in the parentheses are number of cases.

5 ~ ~ 616 CANCER February I 1988 Vol. 6 I TABLE 5. Incidence Rates* of Cancer of the Esophagus by Histologic Type and Subsite for Both Sexes and All Races Squamous cell carcinoma Adenocarcinoma Subsite No. Percent Rate" No. Percent Rate* Upper third I Middle third I Lower third Total subsite specified Subsite not specified I I I6-0.6 Total Annual rate per million persons adjusted to the age distribution of the 1980 US standard population by 5-year age categories. to account for the observed difference in rates between men and women along the esophagus for adenocarcinoma because the proportions of site-not-specified tumors were similar for the two sexes (15% versus 20%, respectively). It might also be postulated that the lower rates of adenocarcinoma in blacks could be largely explained by the higher rates of histology-not-specified tumors in blacks than in whites. It is unlikely, however, that the classification of tumors according to histologic type by community pathologists would be systematically biased in such a maner related to race. In addition, the lower risk of adenocarcinoma among blacks was observed in both sexes. The reported increase in the incidence of adenocarcinoma in white men during the decade under study must be interpreted with caution. Adenocarcinomas in the lower esophagus with involvement of the gastroesophageal junction were at one time classified as being of gastric Because ofthe recognition of an association of adenocarcinoma with Barrett's esophagus (esophageal mucosa lined by columnar epithelium), a great number of such tumors, especially those with concomitant Barrett's epithelium, are more likely to have recently been classified as arising from the esophagus.' The extent to which the observed rise is a result of the change in the interpretation of gastroesophageal junc- tion tumors by clinicians cannot be assessed given the dataset in this study. However, such a change in classification is not likely to totally explain the observed increase because no comparable increase was observed in white women. We found that a vast majority of adenocarcinoma arose from the lower third of the esophagus, the site at which Barrett's epithelium commonly OCCUTS.'~ Others have reported that a large proportion (approximately 90%) of cases of adenocarcinoma were found with concomitant Barrett's esophagus.'." These observations suggest that adenocarcinoma of the esophagus may have a distinct histogenetic origin, most likely deriving from glandular epithelium. The male-to-female risk ratio for adenocarcinoma (about 7 to 1) noted here, particularly in the lower third of the esophagus, is substantially higher than for complicated Barrett's esophagus (about 3 to 1 based on hospital This difference suggests that factors that are closely related to sex other than common risk factors for adenocarcinoma and complicated Barrett's esophagus may also play an important role in the genesis of adenocarcinoma of the esophagus. The epidemiologic differences presented here provide evidence that the two major types of esophageal cancer may reflect different etiologic influences. The variation in sex ratio by age is apparent for adenocarcinoma but not for squamous cell carcinoma. The male-to-female TABLE 6. Incidence Rates* of Cancer of the Esophagus by Histologic Type and Subsite and Sex for All Races Squamous cell carcinoma Adenocarcinoma Subsite Male Female M/Ft Male Female M/Ft Upper third 6.4 (553)$ 2.4 (228) (14) 0.0 (3) 5.5 Middle third 16.5 (1409) 6.5 (621) 2.5 I.O (89) 0.2 (20) 5.4 Lower third 10.4 (886) 3.6 (345) (424) 0.6 (58) 8.8 Subsite not specified 6.8 (581) 2.0 (190) (97) 0.2 (19) 6.1 Total 40.2 (3429) 14.4 (1384) (624) l.o(lo0) 7.4 * Annual rate per million persons adjusted to the age distribution of t Ratio of incidence rates based on three significant digits. the 1980 US standard population by 5-year age categories. $ Numbers in parentheses are number of cases.

6 No. 3 CANCER OF THE ESOPHAGUS BY HISTOLOGY * Yang and Davis 617 rate ratio is higher for adenocarcinoma than for squamous cell carcinoma. Blacks have a higher risk of developing squamous cell carcinoma, whereas the opposite is true for adenocarcinoma. Adenocarcinoma tends to arise from the lower third of the esophagus, but nearly half of the squamous cell carcinomas occur in the middle third. The rate of adenocarcinoma in white men has increased since 1973 in the population under study, whereas the rise in incidence for squamous cell carcinoma was largely explained by the increase in rates among black men and women. At present the most fully established risk factors for cancers of the esophagus are the use of alcohol and tobacco. Suspected risk factors include nutrition deficiencies and chronic es~phagitis. ~* Barrett s esophagus has also recently been associated with an increased risk of adenocarcinoma of esophagus. * There is very little information available, however, to indicate that the different histologic types of esophageal cancer exhibit different risk profiles. Wang ef al. have recently reported that smoking and alcohol consumption exerted a stronger influence on the development of esophageal squamous cell carcinoma than adenocarcinoma arising from the lower esophagus, gastroesophageal junction, and gastric cardia. Thus, it seems prudent to consider more carefully the possibility that these known and suspected risk factors for esophageal cancer exert different influences according to histologic type. Our results should serve to better direct future research designed to evaluate separately the etiologic factors for the two major histologic types of esophageal cancers. REFERENCES 1. Wang HH, Antonioli DA, Goldman H. Comparative features of esophageal and gastric adenocarcinomas: Recent changes in type and frequency. Hum Pathol 1986; 17: Rosenberg JC, Budev H, Edwards RC et a/. Analysis of adenocarcinoma in Barrett s esophagus utilizing a staging system. Cancer 1985; 55: Young IL, Percy CL, Asire AJ. Surveillance, Epidemiology, and End Results: Incidence and Mortality Data, NCI Monogr. 57, NIH Pub. No , Bethesda, MD: National Institutes of Health, 1981; Mausner J, Bahn AK. Measures of Morbidity and Mortality. In: Epidemiology: An Introductory Text. Philadelphia: WB Saunders 1974; Rothman KJ, Boice JD. Trend Analysis for Proportions. In: Epidemiology Analysis with a Programmable Calculator. Boston: Epidemiology Resources Inc., 1982; Sanfey H, Hamilton SR, Smith RRL et a/. Carcinoma arising in Barrett s esophagus. Surg Gynecol Obsfef 1985; Bosch A, Frias 2, Caldwell WL. Adenocarcinoma of the esophagus. Cancer 1979; 43: Blot WJ, Fraumeni JF. Trends in esophageal cancer mortality among US blacks and whites. Am J Public Health 1987; 77: Webb JN, Busuttil A. Adenocarcinoma of the esophagus and of the esophagogastric junction. Br J Surg 1978; 65: Spechler SJ, Goyal RK. Barrett s esophagus. N Engl J Med 1986; 3 I5: I. Spechler SJ, Robbins AH, Rubins HB et a/. Adenocarcinoma and Barrett s esophagus: An overrated risk? Gasrroenferology 1984; 87: Cameron AJ, Ott BJ, Payne WS. The incidence of adenocarcinoma in columnar-lined (Barrett s) esophagus. N Engl J Med 1985: Skinner DB, Walther BC, Riddell RH ef al. Barrett s esophagus: Comparison of benign and malignant cases. Ann Surg 1983; Schottenfeld D. Epidemiology of cancer of the esophagus. Semin Oncol 1984; 11: Kuylenstierna R, Munck-Wilkland E. Esophagitisand cancer of the esophagus. Cancer 1985;