Cancers of the Upper Aerodigestive Tract in Ontario, Canada, and the United States

Transcription

1 1728 Cancers of the Upper Aerodigestive Tract in Ontario, Canada, and the United States David P. Skarsgard, M.D. 1 Patti A. Groome, Ph.D. 1 William J. Mackillop, M.B. 1 Sam Zhou, M.A. 1 Deanna Rothwell, M.Sc. 1 Peter F. Dixon, M.B. 1 Brian O Sullivan, M.B. 2 Steven F. Hall, M.D. 3 Eric J. Holowaty, M.D. 4 1 Radiation Oncology Research Unit, Department of Oncology, Queen s University, Kingston Regional Cancer Center and Kingston General Hospital, Kingston, Ontario, Canada. 2 Department of Radiation, Oncology, Princess Margaret Hospital, Toronto, Ontario, Canada. 3 Department of Otolaryngology, Queen s University, Kingston, Ontario, Canada. 4 Ontario Cancer Registry, Toronto, Ontario, Canada. Presented in part at the Annual Meeting of the Canadian Association of Radiation Oncologists, Vancouver, British Columbia, Canada, September 26 28, Supported in part by a grant from the National Cancer Institute of Canada (to W.J.M.). The authors are grateful for the assistance of Ms. Beverley Shortt in the preparation of the article. Patti Groome is a Career Scientist of the Ministry of Health of Ontario. Deanna Rothwell was a Research Associate of the Radiation Oncology Research Unit Kingston. David P. Skarsgard s current address: Department of Radiation Oncology, Saskatoon Cancer Center, Saskatoon, Saskatchewan, Canada. BACKGROUND. Squamous cancers of the upper aerodigestive tract (UADT) are related to the use of tobacco and/or alcohol, and in North America they are more common among the poor. They are usually locoregionally confined at diagnosis, and local treatment with surgery and/or radiation therapy is often curative. This study compares the incidence and survival of this group of diseases in Canada and the U.S., two North American neighbors with many cultural similarities but significant differences in their health care and social programs. METHODS. To describe and compare the case mix, incidence, and outcome of squamous cancers of the UADT in Ontario, Canada, and the U.S., we used the Ontario Cancer Registry (OCR) and the Surveillance, Epidemiology, and End Results (SEER) registries in the U.S. to identify all cases of cancer with International Classification of Disease (ICD) codes 141, 143-9, 160-1, and a subset of 140, which were diagnosed between 1982 and ICD-O histology codes were placed into clinically relevant groupings, and ICD-9 site codes were grouped into sites as defined by the International Union Against Cancer and the American Joint Committee on Cancer. Age-adjusted incidence rates were calculated for each site. For the SEER registry, race specific incidence rates were also calculated. Observed and expected survival were plotted by site and registry, and from these, relative survival was calculated. Survival was compared during the first 5 years after diagnosis and during the next 5 years among patients who had survived the first 5 years. RESULTS. Of the 16,577 and 42,990 cases identified in the OCR and SEER registries, respectively, squamous cancer was by far the most common histology (94.1% in OCR, 94.6% in SEER) and will form the main subject of this report. The distribution of squamous cancers by site, subsite, age, and gender were remarkably similar in the two populations. Overall, the incidence was about 17% higher in the U.S. than in Ontario, and this difference was seen for all sites except the nasopharynx, which was more common in Ontario. The higher incidence in the U.S. in part reflects the much higher rate for African Americans than for Americans of other ethnic backgrounds. During the first 5 years after diagnosis, when most deaths from UADT cancer occur, there was a significant relative survival difference in favor of the U.S. for cancer of the supraglottis, and in favor of Ontario for cancer of the oral cavity. There was a nonsignificant trend in favor of Ontario for cancer of the nasopharynx. Within the SEER population, for all sites except the nasopharynx, 5-year relative survival was considerably worse for African Americans than for Americans of other ethnic backgrounds. Examination of survival beyond 5 years after diagnosis for patients who had survived the first 5 years revealed that for all sites, the observed survival continued to diverge markedly from the expected survival. The excess mortality ranged from less than 20% for glottic and nasopharyngeal cancers to about 30 40% for oropharyngeal and supraglottic cancers. Deanna Rothwell s current address: Institute for Clinical Investigative Sciences, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada. Eric Holowaty s current address: Surveillance Unit, Cancer Care Ontario, 610 University Avenue, Toronto, Ontario, Canada. Address for reprints: William J. Mackillop, M.B., Radiation Oncology Research Unit, Kingston General Hospital, Apps Level 4, Kingston, Ontario K7L 2V7 Canada. Received July 7, 1999; revisions received November 11, 1999; accepted November 11, American Cancer Society

2 Cancer of the Upper Aerodigestive Tract/Skarsgard et al Squamous cancers arising at different sites in the upper aerodigestive tract share a number of common characteristics. Most are limited to the primary site and the regional lymph nodes at the time of diagnosis, and local treatment with surgery or radiotherapy may be curative. 1,2 However, when local control is not achieved, these cancers usually progress rapidly, resulting in death within a few years of diagnosis. 3,4 Late deaths from a previous cancer of the upper aerodigestive tract are uncommon, and therefore 5-year survival provides a fairly good indicator of the outcome of curative treatment. Tobacco, and in some cases alcohol, are etiologic factors in most of these CONCLUSIONS. Despite remarkable similarities in case mix between the two countries, UADT cancers were more frequent in the SEER population of the U.S. than in Ontario, and this was partly attributable to the much higher incidence among African Americans. Significant differences between the registries in 5-year survival were seen for several sites. African Americans with UADT cancers had much worse prognoses than did Americans of other ethnic backgrounds. Patients who survive their UADT cancer remain at a higher-than-expected risk of death even after they have been cured. Cancer 2000;88: American Cancer Society. KEYWORDS: registry, incidence, survival, squamous cell carcinoma, oral cavity, oropharynx, nasopharynx, hypopharynx, supraglottis, glottis. diseases; 5,6 second primary cancers therefore continue to take a toll on these patients in later years, 7 as do other lifestyle-related illnesses. 5 Most of what is known about cancers of the upper aerodigestive tract comes from retrospective reports of the experience of individual hospitals or cancer treatment centers, often from the perspective of only one medical or surgical specialty. These clinical series may provide a very detailed picture of the behavior of a specific cancer in a carefully evaluated group of patients, but referral bias and selection bias influence their results, 8 and such studies therefore do not provide a valid picture of the impact of these diseases on the population as a whole. There have also been some prospective studies, including randomized trials, which provide an even clearer picture of the behavior of well defined subgroups of cases, but these suffer from the same limitations. In contrast, populationbased cancer registries provide a description of a given cancer or group of cancers that is free of these biases, although much less detailed. A previous analysis of information in the Surveillance, Epidemiology, and End Results (SEER) registries has provided a unique picture of the incidence and outcome of upper aerodigestive tract cancers in the U.S. 9 The goals of the current study were as follows: 1) to use registry data to provide a similarly unbiased but more detailed picture of the case mix of cancers of the upper aerodigestive tract, describing them as histologic and anatomic entities that are recognizable and relevant to the clinicians who treat them; and 2) to compare the incidence and outcome for patients with similar cancers in the U.S. and the Canadian province of Ontario. Although the two populations are similar in many respects, being predominantly Caucasian with similar lifestyle and standard of living, we hypothesize that some differences may be seen in incidence and outcome, perhaps as a result of differences in the span of socioeconomic strata, differences in treatment choice and effectiveness, and differences in the health care systems of the two countries. Cancers of the upper aerodigestive tract provide an ideal setting in which to make this comparison, because they occur more commonly in persons of lower socioeconomic status, 10 for whom the quality and availability of medical treatment may differ from that enjoyed by the rest of society. METHODS Sources of Data Information about all cancer patients whose cancer was diagnosed between January 1, 1982 and December 31, 1994 was provided by the Ontario Cancer Registry (OCR) in Canada and by the SEER program in the U.S. The OCR The OCR is a population-based registry operated by Cancer Care Ontario. 11 It covers the entire Canadian province of Ontario, with a population of approximately 11 million people. The registry obtains its information about cancer incidence and mortality from data that are collected routinely for purposes other than cancer registration, including hospital separation records, provincial cancer center records, pathology reports, and death certificates. A computerized system of probabilistic record linkage is used to reconcile information from these different data sources. The OCR is a passive registry that does not actively seek additional information on registered cases. Cancer is not a reportable disease in Ontario.

3 1730 CANCER April 1, 2000 / Volume 88 / Number 7 The SEER program The SEER program, operated by the National Cancer Institute of the National Institutes of Health, is made up of nine individual population-based registries operating in selected geographic areas of the U.S. 12 These areas include the entire states of Connecticut, Iowa, New Mexico, Utah, and Hawaii and the metropolitan areas of Detroit, San Francisco Oakland, Seattle Puget Sound, and Atlanta and represent approximately 10% of the U.S. population. The nine areas were selected for their ability to operate and maintain a population-based cancer reporting system and for their epidemiologically significant population subgroups. The SEER population is not, however, demographically representative of the entire U.S. population due to overrepresentation of minority groups. Unlike the OCR, SEER actively gathers data on all cancer patients within the areas of coverage. Data elements Both registries classified anatomic sites of cancer by using the ICD-9 coding system, 13 and both classified tumor morphology by using the International Classification of Disease for Oncology (ICD-O) system. 14 Both registries provided age at diagnosis, date of diagnosis, gender, date of last follow-up, and vital status at last follow-up. SEER provided information about race that was not available in the OCR. Although some information about the extent of disease was available in SEER, no information was available in the OCR; therefore, this variable was not evaluated in this comparative study. Data quality A study conducted in 6 of the 9 SEER areas has shown 97.7% completeness in case ascertainment, 15 and it has been estimated similarly that the OCR is 95% complete for all sites combined. 16 The accuracy of specific data elements recently was compared, and error rates in classifying the primary site were estimated at 2.5% for SEER and 8.3% for Ontario (Holowaty EJ, Moravan V, Lee G, et al., unpublished data). TABLE 1 Classification of Anatomic Sites for Upper Aerodigestive Tract Cancers According to ICD-9 Site Codes Anatomic site ICD-9 codes Description Exclusions Oral cavity Mucosal lining of lips Tongue (base of tongue) (lingual tonsil) 143 Gum 144 Floor of mouth 145 Other/unspecified (soft palate) mouth (uvula) Oropharynx Base of tongue Lingual tonsil Soft palate Uvula 146 Oropharynx (ant. aspect of epiglottis) Nasopharynx 147 Nasopharynx Hypopharynx 148 Hypopharynx Larynx Anterior aspect of epiglottis 161 Larynx SNCE 160 Other/NOS 149 Other/ill-defined site of origin in UADT ICD: International Classification of Disease; SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid; NOS: not otherwise specified; UADT: upper aerodigestive tract. Classification of Anatomic Sites For the purposes of this study, our definition of the upper aerodigestive tract included the oral cavity, oropharynx, nasopharynx, hypopharynx, larynx, paranasal sinuses, nasal cavity, middle ear, auditory tube, and mastoids. Those sites were defined using the criteria adopted by the American Joint Committee on Cancer(AJCC) and the International Union Against Cancer (UICC), as outlined in the TNM staging manual. 17 The major salivary glands were excluded because malignancies, which occur at that site, differ in their etiology, histology, and natural history, from those arising in the covering epithelium of the upper aerodigestive tract. 18 The ICD-9 classification of the anatomic location of tumors, 13 provided by the registries, was translated into the clinical groupings defined by the AJCC and UICC, by using the rules provided for this purpose in the TNM staging manual 17 (Table 1). Although the groupings are similar to the categories defined by the ICD-9 system, a few modifications were necessary to conform to the TNM rules and to more accurately describe disease entities that are familiar to clinicians. For example, the base of the tongue, lingual tonsil, soft palate, and uvula were considered part of the oropharynx rather than the oral cavity, in which they had been grouped by the ICD-9 system, and similarly, the anterior aspect of the epiglottis was considered a part of the larynx rather than the oropharynx. Information provided by the fourth digit of the ICD-9 code, which was present in both registries, was used to subdivide anatomic sites into clinically relevant subsites for further description and comparison of case mix between the two countries. For example, subsites of the oral cavity include the anterior two-

4 Cancer of the Upper Aerodigestive Tract/Skarsgard et al TABLE 2 Histology of Cancers of the Upper Aerodigestive Tract a in Ontario and the U.S., Histologic type Ontario (OCR) US (SEER) No. % No. % Squamous cancers 15, , Squamous cell and variants 14, , Unspecified carcinomas Unspecified neoplasms Glandular malignancies Sarcomas Rare tumors Esthesioneuroblastoma Melanoma Others Total 16, , OCR: Ontario Cancer Registry; SEER: Surveillance, Epidemiology, and End Results. a Kaposi sarcomas, lymphomas, leukemias, myelomas, and lymphoreticular and immunoproliferative diseases were excluded. thirds of the tongue, the floor of the mouth, the gums, and the hard palate. Subsites of the oropharynx include the tonsils, the base of the tongue, and the soft palate and uvula. Subsites of the hypopharynx include the pyriform sinus, the postcricoid region, and the posterior wall. Classification of Histologies ICD-O morphology codes also were grouped into categories more familiar to clinicians (Table 2). Our classification was intended to reflect the clinical approach to cancers of the upper aerodigestive tract but is similar to that of Berg. 19 For the purposes of this study, the category of squamous cancers was defined broadly to include known squamous cell carcinomas and their well defined variants, as well as unspecified malignancies that usually would be managed in the same way as squamous carcinomas. Well defined cancers that are clinically distinct entities, for example, adenocarcinomas, sarcomas, and melanomas, were excluded. Analysis of Incidence Census data and intercensal estimates for Ontario and the SEER areas of the U.S. were used to define the populations at risk in any given year during the period of the study. For each major cancer site, the incidence rate was calculated for each year of the study and averaged over all years. Crude incidence rates in both countries were adjusted by age and gender to the 1970 standard U.S. population. For the SEER registry, racespecific incidence rates also were calculated. Analysis of Survival Observed survival Observed survival was defined as the time from the date of diagnosis to the date of death. In both registries, the date of diagnosis is defined as the earliest date at which the diagnosis of cancer is mentioned in a report relating to the case, which may precede the final confirmation of the diagnosis. The date of death for Ontario patients is provided by death certification based on information provided by the Registrar General of Ontario. Information on deaths in SEER patients is obtained by a process of active follow-up, which is supplemented by passive sources such as Medicare data and death certificates. In SEER, death records were complete to December 31, 1994, whereas in the OCR, they were complete to December 31, To maximize comparability between the registries, we therefore censored all patients in both registries as of December 31, Whereas dates of diagnosis and death were provided to the day by the OCR, they were only provided to the month in SEER. To permit comparison between the registries, we rounded the dates of diagnosis and death for patients in the OCR to the nearest month, as in the SEER registry. The probability of survival as a function of time was calculated using the life table method. Differences in survival between the registries were evaluated using the log rank test. Relative survival To correct for differences in mortality from all causes between the U.S. and Canada, we calculated relative survival by using the method described by Ederer et al. 20 Briefly, Ontario and U.S. life tables were used to assign to each case an expected survival at 1 10 years, based on age at diagnosis, gender, and year of diagnosis. For SEER patients, race also was taken into account. The expected survival for any group of patients then was calculated by averaging the expected survival of its individual members. Relative survival was calculated as the observed divided by the expected survival at a given time after diagnosis. Relative survival point estimates at 5 years after diagnosis, and at 10 years in 5-year survivors, were compared between the registries by using the method of Ederer et al. RESULTS Distribution of Histologies Table 2 shows the distribution of histologies for cancers of the upper aerodigestive tract in the two registry populations. In both registries, the majority of cases were squamous cell carcinomas and known variants, but the proportion of cases with ill-defined cancers was higher in the OCR than in the SEER population. For the

5 1732 CANCER April 1, 2000 / Volume 88 / Number 7 TABLE 3 Anatomic Distribution of Squamous Cancers of the Upper Aerodigestive Tract in Ontario and the U.S., Ontario (OCR) U.S. (SEER) Subsite % total No. % site Median n age Male/ female % total No. % site Median age (yrs) Male/ female Oral Cavity , Anterior of tongue Floor of mouth Gums Hard palate Retromolar trigone Buccal mucosa Ill defined/nos Oropharynx Tonsil Base of tongue Posterior wall Soft palate/uvula Other/NOS Nasopharynx Hypopharynx Pyriform sinus Postcricoid Posterior wall Other/NOS Pharynx/NOS Larynx , Supraglottis Glottis Subglottis Ill defined/nos SNCE Maxillary sinus Other sinus/nos Nasal cavity Middle ear/mastoid Total , , OCR: Ontario Cancer Registry; SEER: Surveillance, Epidemiology, and End Results; SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid; NOS: not otherwise specified. purposes of this study, the ill-defined cancers were included with the squamous cell carcinomas, resulting in almost identical proportions of squamous cancers in the two populations (94.1% in the OCR vs. 94.6% in SEER). Glandular malignancies, sarcomas, and other rare histologies accounted for less than 6% of cases in each registry and were excluded from the analysis. Distribution of Anatomic Sites Table 3 shows the distribution of squamous cancers among the anatomic sites and subsites of the upper aerodigestive tract. Overall, the relative frequency of involvement of different sites was very similar in Ontario and the U.S., as was the gender ratio within each site. The median age at diagnosis was also very similar between the registries for all sites except the nasopharynx, in which the median age was 4 years younger in Ontario than in the U.S. In both registries, one-third of all cases involved the larynx, and the ratio of glottic to supraglottic cancers was about 2:1. Approximately one-quarter of cases involved the oral cavity, and most of these arose in the anterior two-thirds of the tongue or in the floor of the mouth. Approximately one-fifth of cases involved the oropharynx, and most of these arose in the tonsil or in the base of the tongue. Incidence Table 4 shows the gender- and age-adjusted incidence of squamous cancers of each of the major anatomic sites, and of all sites combined, in Ontario and in the U.S. The overall age-adjusted incidence of squamous cancers of the upper aerodigestive tract was 11.5 per 10 5 in Ontario and 13.4 per 10 5 in the U.S. In the U.S., the incidence in African Americans (16.1 per 10 5 ) was

6 Cancer of the Upper Aerodigestive Tract/Skarsgard et al TABLE 4 Age-Adjusted Incidence Rates a (per 10 5 ) for Squamous Cancers of the Major Upper Aerodigestive Tract Sites in Ontario and the U.S., Site Ontario (OCR) US (SEER) US (African Americans) US (Non-African Americans) Oral cavity 3.2 ( ) 3.6 ( ) 3.1 ( ) 3.7 ( ) Oropharynx 2.1 ( ) 2.8 ( ) 4.2 ( ) 2.6 ( ) Nasopharynx 0.6 ( ) 0.4 ( ) 0.5 ( ) 0.4 ( ) Hypopharynx 0.9 ( ) 1.1 ( ) 1.7 ( ) 1.0 ( ) Supraglottis 1.1 ( ) 1.6 ( ) 2.1 ( ) 1.5 ( ) Glottis 2.2 ( ) 2.5 ( ) 2.6 ( ) 2.5 ( ) All sites b 11.5 ( ) 13.4 ( ) 16.1 ( ) 13.0 ( ) OCR: Ontario Cancer Registry; SEER: Surveillance, Epidemiology, and End Results; CI: confidence interval. a Age-adjusted to standard U.S population. b Includes other upper aerodigestive tract sites in addition to those listed above (see Table 3). approximately 25% higher than that in the rest of the population (13.0 per 10 5 ). The excess of cases in the African American population was greatest in cancers of the oropharynx, hypopharynx, and supraglottis. When African Americans were excluded, the rates in the U.S. were more similar to those in Ontario. In both countries, the age-adjusted incidence of squamous cancers of the upper aerodigestive tract decreased during the period of the study. In Ontario, the rate for all sites combined fell from 11.9 per 10 5 (95% confidence interval [CI], ) in the period to 11.0 per 10 5 ( ) in the period, a decrease of 8%. In SEER, the rate for all sites combined fell from 13.9 per 10 5 ( ) in the period to 12.7 per 10 5 ( ) in the period, a decrease of 9%. The incidence rates for most individual sites also decreased during the period of the study in both registries (data not presented). Survival in the First 5 Years Figure 1 shows observed survival curves for the first 5 years after diagnosis for squamous cancers of each major anatomic site in Ontario and in the U.S. Table 5 provides point estimates of observed and relative survival at 5 years. In cancer of the oral cavity, observed survival at 5 years was 3.8% higher in Ontario than in the U.S. Expected survival for the normal population, adjusting for age and gender, is higher in Canada than in the U.S., but relative survival was still 4.0% higher in Ontario than in the U.S. Five-year observed and relative survival for cancer of the nasopharynx were, respectively, 6.5% and 5.3% higher in Ontario than in the U.S., but the relative survival difference did not reach statistical significance at the 5% level. For cancer of FIGURE 1. Observed and expected survival in the first 5 years after diagnosis of squamous cancers of the upper aerodigestive tract, by site. Statistics compare observed survival between the two registries (log rank test). (A) oral cavity (P ); (B) oropharynx (P ); (C) nasopharynx (P ); (D) hypopharynx (P ); (E) larynx, comprised of supraglottis (P ) and glottis (P ); (F) sinus, nasal cavity, middle ear, auditory tube, and mastoid (P ). In (E), the expected survival curves for supraglottic and glottic cancer were very similar; therefore, only one curve was shown for each registry (larynx cancer). In the calculation of relative survival (Table 5), however, the expected survival for each individual subsite was used. SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid. the supraglottis, 5-year observed and relative survival were, respectively, 2.3% and 3.1% higher in the U.S. than in Ontario. For other sites, there were no significant differences in relative survival between Ontario and the U.S. In both registries, the outcome was best in cancer of the glottic larynx and worst in cancer of the hypopharynx. Within the SEER population, 5-year observed and relative survival were evaluated separately in populations of African Americans and Americans of other ethnic backgrounds (Table 6). For all sites combined, 5-year relative survival was only 36.7% (95% CI, ) in African Americans, compared with 51.7%

7 1734 CANCER April 1, 2000 / Volume 88 / Number 7 TABLE 5 Five-Year Survival and Relative Survival (%) of Squamous Cancers of the Major Upper Aerodigestive Tract Sites in Ontario and the U.S., Site Country 5-yr survival (95% CI) 5-yr relative survival (95% CI) P value a Oral cavity U.S. (SEER) 43.9 ( ) 51.8 ( ) Ontario (OCR) 47.7 ( ) 55.8 ( ) Oropharynx U.S. (SEER) 33.0 ( ) 38.1 ( ) Ontario (OCR) 33.3 ( ) 37.8 ( ) Nasopharynx U.S. (SEER) 44.3 ( ) 49.2 ( ) Ontario (OCR) 50.8 ( ) 54.5 ( ) Hypopharynx U.S. (SEER) 21.7 ( ) 25.5 ( ) Ontario (OCR) 20.8 ( ) 23.9 ( ) Supraglottis U.S. (SEER) 42.4 ( ) 48.6 ( ) Ontario (OCR) 40.1 ( ) 45.5 ( ) Glottis U.S. (SEER) 66.1 ( ) 78.9 ( ) Ontario (OCR) 68.3 ( ) 79.5 ( ) SNCE U.S. (SEER) 41.4 ( ) 50.3 ( ) Ontario (OCR) 43.2 ( ) 52.4 ( ) All Sites b U.S. (SEER) 42.9 ( ) 50.3 ( ) Ontario (OCR) 45.9 ( ) 53.0 ( ) CI: confidence interval; SEER: Surveillance, Epidemiology, and End Results; OCR: Ontario Cancer Registry; SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid. a Compares 5-year point estimates of relative survival between the registries. b Includes other upper aerodigestive tract sites in addition to those listed above (see Table 3). ( ) in the rest of the SEER population. Similar wide discrepancies were seen for most individual sites. Survival after 5 Years The overwhelming majority of deaths due to squamous cancers of the upper aerodigestive tract occurs within 5 years of diagnosis. 3,4 We selected the subgroup of patients who survived for the first 5 years after diagnosis, who had most likely been cured of their initial cancer, and described their survival during the next 5 years. Figure 2 compares observed and expected survival between 5 and 10 years after diagnosis, in those patients who survived the first 5 years. Table 7 provides 10-year point estimates of observed and relative survival in this group of patients. At every site, the death rate in the cancer population continued to exceed the expected rate for the general population. The extent to which the observed mortality exceeded the expected varied with the site of the cancer. At each site, however, a similar excess of late deaths was seen in Ontario and the U.S. The excess mortality was greatest in cancers of the hypopharynx and oropharynx and least in cancers of the glottic larynx and nasopharynx. DISCUSSION The mix of histologies in cancers of the upper aerodigestive tract was similar in Ontario and the U.S. The TABLE 6 Five-Year Survival and Relative Survival (%) of Squamous Cancers of the Upper Aerodigestive Tract in the U.S. ( ): African Americans vs. Americans of Other Races in the U.S. African Americans 5-yr survival (95% CI) Other races 5-yr relative survival (95% CI) African Americans Other races Oral cavity 29.3 ( ) 45.2 ( ) 33.8 ( ) 53.5 ( ) Oropharynx 23.1 ( ) 34.9 ( ) 26.4 ( ) 40.2 ( ) Nasopharynx 41.0 ( ) 44.6 ( ) 44.9 ( ) 49.6 ( ) Hypopharynx 17.5 ( ) 22.6 ( ) 20.3 ( ) 26.5 ( ) Supraglottis 37.1 ( ) 43.3 ( ) 43.1 ( ) 49.5 ( ) Glottis 56.8 ( ) 67.1 ( ) 68.8 ( ) 80.0 ( ) SNCE 20.3 ( ) 43.2 ( ) 25.1 ( ) 52.3 ( ) All site 31.6 ( ) 44.1 ( ) 36.7 ( ) 51.7 ( ) CI: confidence interval; SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid. a Includes other upper aerodigestive tract sites in addition to those listed above (see Table 3). higher proportion of poorly defined cancers in Ontario probably reflects the finding that the OCR is a passive cancer registry, which does not actively pursue additional data on registered cases, as is the practice in the SEER registries in the U.S. We elected not to exclude these poorly defined cases because, in so doing, we might have biased our survival results in favor of Ontario. Instead, the indeterminate diagnoses were included with the squamous carcinomas, because the majority almost certainly belonged in that category. When that was done, the proportions within each of the major histologic groups were almost identical in the two populations (Table 2). The anatomic distribution and male female ratios of squamous cancers of the upper aerodigestive tract were also close to identical in Ontario and the U.S. Median age at diagnosis was very similar for every major site except nasopharynx, in which it was 4 years younger in Ontario than in the U.S. Cancer of the nasopharynx has a much higher incidence in certain portions of the Asian population, in which it typically occurs at a younger age. 21 This observation could be related to the higher proportion of persons of Asian origin in Ontario compared with the SEER population of the U.S. 22,23 The overwhelming similarity in case mix of upper aerodigestive tract cancers between the two countries is consistent with the hypothesis that case mix is governed largely by the underlying biologic tendency of specific cancers to occur at specific sites, and that this operates similarly in both countries. It also suggests that the detailed AJCC/UICC anatomic classification of this diverse group of diseases can be applied meaningfully at the level of the population and can be compared between populations, despite differences in the operation of their registries.

8 Cancer of the Upper Aerodigestive Tract/Skarsgard et al TABLE 7 Survival and Relative Survival (%) after 5 More Years among 5-Year Survivors a of Squamous Cancers of the Major Upper Aerodigestive Tract Sites in Ontario and the U.S., Site Country Survival after 5 more years (95% CI) Relative survival after 5 more years (95% CI) Excess mortality (%) Oral cavity U.S. (SEER) 63.0 ( ) 72.4 ( ) 27.6 Ontario (OCR) 69.1 ( ) 79.2 ( ) 20.8 Oropharynx U.S. (SEER) 57.4 ( ) 64.0 ( ) 36.0 Ontario (OCR) 65.5 ( ) 72.4 ( ) 27.6 Nasopharynx U.S. (SEER) 75.2 ( ) 81.8 ( ) 18.2 Ontario (OCR) 79.0 ( ) 84.8 ( ) 15.2 Hypopharynx U.S. (SEER) 51.0 ( ) 59.1 ( ) 40.9 Ontario (OCR) 54.2 ( ) 62.0 ( ) 38.0 Supraglottis U.S. (SEER) 61.5 ( ) 69.2 ( ) 30.8 Ontario (OCR) 68.1 ( ) 76.7 ( ) 23.3 Glottis U.S. (SEER) 70.2 ( ) 81.4 ( ) 18.6 Ontario (OCR) 73.6 ( ) 84.6 ( ) 15.4 SNCE U.S. (SEER) 68.0 ( ) 80.4 ( ) 19.6 Ontario (OCR) 71.2 ( ) 83.7 ( ) 16.3 All sites b U.S. (SEER) 64.1 ( ) 73.1 ( ) 26.9 Ontario (OCR) 69.6 ( ) 79.1 ( ) 20.9 CI: confidence interval; SEER: Surveillance, Epidemiology, and End Results; OCR: Ontario Cancer Registry; SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid. a Survival at 5 years has been normalised to 100%. b Includes other upper aerodigestive tract sites in addition to those listed above (see Table 3). FIGURE 2. Observed and expected survival beyond 5 years, in patients who have survived the first 5 years after diagnosis, for squamous cancers of the upper aerodigestive tract, by site. Statistics compare observed survival between the two registries (log rank test): (A) oral cavity (P ); (B) oropharynx (P ); (C) nasopharynx (P ); (D) hypopharynx (P ); (E) larynx, comprised of supraglottis (P ) and glottis (P ); (F) sinus, nasal cavity, middle ear, auditory tube, and mastoid (P ). In (E), the expected survival curves for supraglottic and glottic cancers were very similar; therefore only one curve was shown for each registry (larynx cancer). In the calculation of relative survival (Table 7), however, the expected survival for each individual subsite was used. SNCE: sinus, nasal cavity, middle ear, auditory tube, mastoid. The overall incidence of squamous cancers of the upper aerodigestive tract was approximately 17% higher in the U.S. than in Ontario. The excess incidence in the U.S. appears to be partly explained by the higher incidence of these diseases in African Americans compared with the rest of the U.S. population, and when the SEER population of Americans of other ethnic backgrounds was compared with the Ontario population, the rates were more similar. This is not to suggest, however, that the Ontario population is predominantly white, or that its ethnic composition is similar to that of the non African American portion of SEER. A recent Canadian census showed that 17.1% of the Ontario population was nonwhite, including 3.7% each for South Asians and Chinese, 3.3% blacks, and 1.3% aboriginals. 22 This is not dissimilar to the 19.7% of the non African American SEER population that is nonwhite, 23 although the distribution of non-white groups differs between the two populations. It is known that the upper aerodigestive tract cancers are more common among the economically poorer segments of the population in Canada and the U.S. than among the more affluent (Mackillop WJ, unpublished data).it has been suggested that apparent associations between ethnicity and cancer incidence are due to differences in socioeconomic status among ethnic groups rather than to ethnicity in itself. 24 Indeed, whereas African Americans make up only 12% of the U.S. population, they account for almost one-third of Americans living below the poverty line. 24 The observed decrease in the incidence of squamous cancers of the upper aerodigestive tract during the study period, which was of similar magnitude in both registries, is noteworthy. It is unlikely that this is artifactual, because during the same period, the incidence of some other types of cancer, e.g., female breast, prostate, and lung) increased in both the U.S. 25 and Canada. 26 This observation may be the result of

9 1736 CANCER April 1, 2000 / Volume 88 / Number 7 changes in exposure to known and/or unknown risk factors. Differences in observed and relative survival among the major sites were qualitatively, and in some cases, quantitatively similar in the U.S. and Ontario. In both countries, for example, patients with cancers of the glottic larynx had the best prognoses, whereas patients with cancers of the hypopharynx had the worst prognoses. For a few sites, however, there were significant quantitative differences in outcome between the two countries. For cancers of the oral cavity, 5-year observed survival was better in Ontario than in the U.S. When the difference in all-cause mortality between Ontario and the U.S. was taken into account, the relative survival in Ontario remained significantly higher. For cancers of the nasopharynx, 5-year observed survival was also significantly better in Ontario than in the U.S., but relative survival differences were not statistically significant despite a large numeric difference in favor of Ontario, due to the small number of cases. For cancers of the supraglottis, 5-year observed and relative survival were both significantly better in the U.S. than in Ontario. Cancer survival may be affected by patient-related, disease-related, and/or treatment-related variables. This study is admittedly weakened by the lack of any information in the OCR on extent of disease and race, which prevents us from controlling for these prognostically important variables in our comparisons of outcome between the two populations. Nevertheless, even if such information were available, it still would be difficult to attribute differences in outcome to differing treatment strategies because of numerous other differences between the two countries. Thus, our suggested explanations for the observed survival differences are largely speculative. We believe, however, that this study is uniquely able to evaluate the overall impact of particular cancers on different populations as a whole, for reasons that may include any known and unknown patient-, disease-, and treatment-related factors. In this study, African Americans with upper aerodigestive tract cancers had a markedly worse outcome than the rest of the U.S. population. Large discrepancies were seen for almost every major site, with the possible exception of nasopharynx. This is at least partly due to increased stage at diagnosis for African Americans, 3,27 and if this variable were controlled for, then these interracial differences in outcome would undoubtedly be smaller. However, such an observation would obscure the finding that the overall impact of most types of upper aerodigestive tract cancers is much greater in African Americans than in the rest of the U.S. population. The association between race, as a surrogate measure of socioeconomic status, and cancer survival has been described previously; 28,29 although cultural and/or genetic factors cannot be ruled out, these data support the contention that cancer patients who are of low socioeconomic status have worse prognoses than the rest of the population. Although no firm conclusions can be reached about the causes of any survival differences between the U.S. and Ontario, some possible explanations can be suggested. The relatively better outcome of cancer of the nasopharynx in Ontario, compared with the U.S., may be attributable to the higher proportion of the Ontario population that is of Asian origin, 22,23 in whom this disease has a different natural history, and these patients have better prognoses than other racial groups. 30 The finding that the survival differential between African Americans and Americans of other ethnic backgrounds was most narrow in cancer of the nasopharynx is likely due to the finding that smoking and alcohol, which are commonly associated with low socioeconomic status, are not etiologic factors in this disease. 6 The interregistry differences in the outcome of cancers of the oral cavity and supraglottis are more difficult to explain. They could be due to systematic differences in stage at presentation and/or general health of Canadian and American patients, given the fundamental differences in the health care systems of the two countries. Ontario, like the rest of Canada, has a publicly funded system that aims to provide a single level of care to all persons regardless of financial means, although treatment may be delayed because of waiting lists for treatment. 31,32 In contrast, in the U.S., quality treatment is more immediately available to those who can afford it but may be inaccessible to those who cannot, and cancer outcome varies much more widely across socioeconomic strata in the U.S. than in Canada. 33 Differences in the availability of medical care between the two countries may be especially relevant in cancers of the upper aerodigestive tract, which more often affect persons of lower socioeconomic status. 10 Although intriguing, this is unlikely to be a large part of the explanation for the survival differences between the two countries, however, because in oral cavity cancers, the Canadians fared better, whereas in supraglottic cancers, they fared worse. It is unlikely that Canadians with oral cavity cancers would be in better general health and/or present at earlier stages than their American counterparts, whereas the opposite would be true for patients with supraglottic cancers. Differences in treatment between Canada and the U.S. may provide more plausible explanations for the observed differences in outcome for cancers of the

10 Cancer of the Upper Aerodigestive Tract/Skarsgard et al oral cavity and supraglottis. What little is known about patterns of treatment of upper aerodigestive tract cancers in Canada and the U.S. suggests that there are systematic differences between the countries that may be large enough to cause measurable differences in survival. A registry-based observational study recently has shown that total laryngectomy is used much more frequently in the initial management of both glottic and supraglottic cancers in the U.S. compared with Canada. 34,35 These differences in practice reflect differences of opinion among leading specialists in Canada and the U.S. 36 Canadians believe that primary radiotherapy, reserving surgery for salvage, is the treatment of choice even for patients who have advanced laryngeal cancers, because it conserves natural voice in those in whom radiotherapy is successful, and because delay in laryngectomy does not carry any risk to the patient s life. American specialists believe that initial laryngectomy is the treatment of choice in patients with advanced laryngeal cancer, because a trial of radiotherapy may permit the disease to progress to the point of incurability before the surgery can be performed. If the Canadian view is correct for glottic cancer but incorrect for supraglottic cancer, this might explain the similar outcome in both countries in the former and the American advantage in the latter. The two diseases have distinct natural histories that makes this plausible. There is not enough known about differences in the treatment of oral cavity cancers between Canada and the U.S. to permit any speculation about the relation between treatment and outcome of that disease, but further comparative studies could readily provide such information. In preparing our comparison of survival between Ontario and the U.S., we calculated relative survival in the conventional way, by using population life tables to adjust for differences in all-cause mortality between the two countries. This approach must be used with caution in upper aerodigestive tract cancer patients, however, because these lifestyle-related diseases occur in a segment of the population that is subject to a much higher-than-expected mortality rate from other causes. To obtain some estimate of the excess mortality in upper aerodigestive tract cancer patients that is due to causes other than the primary malignancy, we selected patients who survived the first 5 years and evaluated their survival during the subsequent 5 years. By 5 years after diagnosis, almost all of the patients who have not been cured of their primary squamous cancer of the upper aerodigestive tract already had died of their disease. 3,4 This analysis revealed that 5- year survivors from primary squamous cancers of the upper aerodigestive tract face a huge risk of death in the following 5 years, and there is little reason to believe that this risk would diminish later. These excess deaths may be due to second primary cancers, to ischemic heart disease, to chronic obstructive lung disease, or to any of the other smokingrelated diseases. 5 The magnitude of the excess risk of death is similar in Ontario and the U.S. but varies from one anatomic site to another. It is smaller in cancers of the glottic larynx, in which smoking is the sole etiologic factor, than in the other parts of the upper aerodigestive tract, in which alcohol is also involved. It is also small in cancer of the nasopharynx, in which etiologic factors other than smoking are important. 6 These observations confirm that the successful treatment of cancers arising in the upper aerodigestive tract in no way restores this patient population to normal health. This registry-based comparative study shows that, despite remarkable similarities in the case mix of cancers of the upper aerodigestive tract between two superficially similar countries, Canada and the U.S., there are some important differences in incidence and outcome. It also demonstrates the potential usefulness of cancer registry data in providing a complete and unbiased picture of the expression and impact of a particular group of cancers on a population. REFERENCES 1. Schantz SP, Harrison LB, Hong WK. Cancer of the head and neck. In: DeVita VT, Hellman S, editors. Cancer: principles and practice of oncology. 4th ed. Philadelphia: JB Lippincott, 1993: Sloan D, Goepfert H. Conventional therapy of head and neck cancer. Hematol Oncol Clin North Am 1991;5: Hoffman HT, Karnell LH, Funk GF, Robinson RA, Menck HR. The national cancer data base report on cancer of the head and neck. Arch Otolaryngol Head Neck Surg 1998;124: Boysen M, Natvig K, Winther FO, Tausjo J. Value of routine follow-up in patients treated for squamous cell carcinoma of the head and neck. J Otolaryngol 1985;14: Wald NJ, Hackshaw AK. Cigarette smoking: an epidemiological overview. Br Med Bull 1996;52: Decker J, Goldstein JC. Risk factors in head and neck cancer. N Engl J Med 1982;306: Cooper JS, Pajak TF, Rubin P, Tupchong L, Brady LW, Leibel SA, et al. Second malignancies in patients who have head and neck cancer: incidence, effect on survival and implications based on the RTOG experience. Int J Radiat Oncol Biol Phys 1989;17: Feinstein AR. Clinical biostatistics. St. Louis, MO: The C.F. Mosby Co., Muir C, Weiland L. Upper aerodigestive tract cancer. Cancer 1995;75: Faggiano F, Partanen T, Kogevinas M, Boffetta P. Socioeconomic differences in cancer incidence and mortality. IARC Sci. Publ. 1997;(138):

11 1738 CANCER April 1, 2000 / Volume 88 / Number Clarke EA, Marrett LD, Krieger N. Cancer registration in Ontario: a computer approach. In: Jensen OM, Parkin DM, MacLennan R, Muir CS, Skeet RG editors. Cancer registration principles and methods. Lyon, France: IARC Pub. No. 95;1991: Miller BA, Ries LAG, Hankey BF, Kosary CL, Harras A, Devesa SS, et al SEER cancer statistics review, Bethesda, MD: National Institutes of Health, National Cancer Institute. NIH Pub. No International classification of diseases revision. Geneva: World Health Organization, Percy C, Van Holten V, Muir C, editors. International classification of diseases for oncology. 2nd ed. Geneva: World Health Organization, Zippin C, Lum D, Hankey BF. Completeness of hospital cancer case reporting from the SEER program of the National Cancer Institute. Cancer 1995;76: Robles SC, Marrett LD, Clarke EA, Risch HA. An application of capture-recapture methods to the estimation of completeness of cancer registration. J Clin Epidemiol 1988;41: Sobin LH, Wittekind Ch, editors. TNM classification of malignant tumours. 5th ed. New York: John Wiley and Sons, Spitz MR, Batsakis JG. Major salivary gland carcinoma. Arch Otolaryngol 1984;110: Berg JW. Morphologic classification of human cancer. In: Schottenfeld D, Fraumeni JF, editorss. Cancer epidemiology and prevention. Philadelphia: Saunders, 1982: Ederer F, Axtell LM, Cutler LM. End results and mortality trends in cancer. Natl Cancer Inst Monogr 1991;6: Cvitkovic E, Bachouchi M, Armand J-P. Nasopharyngeal carcinoma: biology, natural history and therapeutic implications. Hematol Oncol Clin North Am 1991;5: Statistics Canada. Visible minorities (20% data). Available from URL: statcan. ca/english/census96/ feb17/vmon.htm [accessed June 10, Surveillance, Epidemiolgy, and End Results (SEER). Number of persons by race for SEER participants 1990 census. Available from URL: [accessed May 21, 1999]. 24. Freeman HJ. Cancer in the economically disadvantaged. Cancer 1989;64(Suppl): [SEER] Surveillance, Epidemiolgy, and End Results cancer statistics review Available from URL: www. seer. ims.nci.nih.gov/publications/csr7394 [accessed November 13, Health Canada. Cancer surveillance on-line. Available from URL: accessed November 18, Kosary CL, Ries LAG, Miller BA, Hankey BF Harras A, Edwards BK, editors. SEER Cancer Statistics Review, : tables and Graphs. Bethesda, MD: National Cancer Institute,1995. NIH Pub. No Cella DF, Orav EJ, Kornblith AB, et al. Socioeconomic status and cancer survival. J Clin Oncol 1991;9: Sterling T, Rosenbaum W, Weinkam J. Income, race and mortality. J Natl Med Assoc 1993;85: Marks JE, Phillips JL, Menck HR. The national cancer data base report on the relationship of race and national origin to the histology of nasopharyngeal carcinoma. Cancer 1998;83: Mackillop WJ, Fu H, Quirt CF, Dixon P, Brundage M, Zhou Y. Waiting for radiotherapy in Ontario. Int J Radiat Oncol Biol Phys 1994;30: Mackillop WJ, Zhou Y, Quirt CF. A comparison of delays in the treatment of cancer with radiation in Canada and the US. Int J Radiat Oncol Biol Phys 1995;32: Boyd C, Zhang-Salomons JY, Groome PA, Mackillop WJ. Associations between community income and cancer survival in Ontario and the US. J Clin Oncol 1999;17: Groome PA, Rothwell D, O Sullivan B, Irish J, Bissett RJ, Dixon P, et al. Management and outcome of glottic carcinoma in Canada and the US: results by T-category [abstract no. 486]. Clin Invest Med 1998;(Suppl):S Groome PA, Zhou Y, Skarsgard D, O Sullivan B, Irish J, Dixon P, et al. The management and outcome of supraglottic larynx cancer in Canada and the US: a comparison using population registry data [abstract no. 558]. Clin Invest Med 1996;19(4 suppl):s O Sullivan B, Mackillop WJ, Gilbert R, Gaze M, Lundgren J, Atkinson C, et al. Controversies in the management of laryngeal cancer: results of an international survey of patterns of care. Radiother Oncol 1994;31:23 32.