Fax: (410) 706-8013
Primary malignant neoplasms of the appendix
A population-based study from the surveillance, epidemiology and end-results program, 1973–1998
Article first published online: 12 JUN 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 12, pages 3307–3312, 15 June 2002
How to Cite
McCusker, M. E., Coté, T. R., Clegg, L. X. and Sobin, L. H. (2002), Primary malignant neoplasms of the appendix. Cancer, 94: 3307–3312. doi: 10.1002/cncr.10589
- Issue published online: 12 JUN 2002
- Article first published online: 12 JUN 2002
- Manuscript Accepted: 31 JAN 2002
- Manuscript Received: 30 OCT 2001
- Manuscript Revised: 22 JAN 2001
- National Cancer Institute, Division of Cancer Control and Population Sciences
- carcinoid tumor;
- goblet cell carcinoid;
- mucinous adenocarcinoma;
- signet ring cell carcinoma;
Cancer of the appendix is an uncommon disease that is rarely suspected rarely before surgery. Although several case series of these tumors have been published, little research has been anchored in population-based data on cancer of the appendix.
This analysis included all actively followed cases of appendiceal neoplasms reported to the National Cancer Institute's Surveillance, Epidemiology and End-Results (SEER) program between 1973 and 1998. Tumors were classified as “colonic type” adenocarcinoma, mucinous adenocarcinoma, signet ring cell carcinoma, goblet cell carcinoid, and “malignant carcinoid” (SEER only collects data on carcinoids specifically classified as malignant). We compared incidence, overall survival and survival rates by extent of disease at diagnosis.
Between 1973 and 1998, 2117 appendiceal malignancies were reported to the SEER program, of which 1645 cases were included in the analysis. Age-adjusted incidence of cancer of the appendix was 0.12 cases per 1,000,000 people per year. Demographic characteristics of patients with goblet cell carcinoid tumors were midway between those of patients with malignant carcinoid and all types of adenocarcinomas. After controlling for age and extent of disease at diagnosis, the overall survival rate for patients diagnosed between 1983 and 1997 (n = 1061) was significantly worse for those with signet ring cell carcinoma than for those with any other tumor type (P < 0.01). In addition, overall survival rates were better for patients with malignant carcinoid (P = 0.01).
Demographic characteristics of patients with cancer of the appendix vary by histology. Except for signet ring cell carcinoma and malignant carcinoid, the extent of disease at time of diagnosis is a more important predictor of survival than histology. Cancer 2002;94:3307–12. © 2002 American Cancer Society.
More than 500,000 appendectomies are performed each year in the United States.1 Primary appendiceal cancer is diagnosed in 0.9–1.4% of appendectomy specimens.2, 3 These rare tumors are seldom suspected before surgery and less than one half of the cases are diagnosed intraoperatively.4–7 Several studies describing primary tumors of the appendix have appeared in the literature since 1903, when Elting published a review and case series.8 Only two population-based studies of tumors of the appendix have been published, one a series of seven cases of adenocarcinoma from the Icelandic Cancer Registry and another as a subsection in an overview of gastrointestinal cancers reported to the Surveillance, Epidemiology and End-Results (SEER) program of the National Cancer Institute.9, 10 Most published information on appendiceal malignancies is derived from compilations of case series.
Since Subbuswamy et al.11 coined the term in 1974, goblet cell carcinoid of the appendix has been described by many authors. In 1978, Warkel et al.12 proposed that goblet cell carcinoid is an aggressive subtype of adenocarcinoid tumors, intermediate in prognosis between carcinoids and adenocarcinomas. No study published to date has reported population-based information on these tumors and no direct comparison of survival among the three tumor types exists.
Population-based data sources provide relatively large numbers of cases of rare tumors and are not impacted by selection bias due to institutional referral patterns. Therefore, national cancer registries are an ideal source of information about cancer of the appendix. The purpose of this study was to describe the epidemiology of cancer of the appendix using population data collected by the SEER program. These data provide an excellent opportunity to compare the characteristics and survival of appendiceal tumors by histology.
MATERIALS AND METHODS
The SEER program collects data from all persons diagnosed with cancer residing in specific geographically defined regions. Reporting to cancer registries in these jurisdictions, which currently cover approximately 14% of the U.S. population, is mandated by law.13 Quality control, with a standard case ascertainment set at 98%, is conducted annually in the SEER areas.14 Data from 11 SEER registries were used for this study. Registries in Connecticut, Hawaii, Iowa, New Mexico, Utah, Detroit, and San Francisco/Oakland have participated in the SEER program since 1973; Atlanta and Seattle/Puget Sound joined the program in 1974–1975, and the Los Angeles and San Jose/Monterey registries have participated since 1992.13 The SEER database includes information on demographics, tumor site, and histology.15 Vital status information for living patients is updated annually by the registries.14
Cases of invasive, malignant neoplasms of the appendix reported to the SEER program between 1973 and 1998 were included in the study. For survival analysis, cases diagnosed between 1983 and 1997 were analyzed because only these cases had detailed information on extent of disease spread at the time of diagnosis. They also had at least one possible year of follow-up. Using the International Classification of Diseases for Oncology (ICD-O), second edition, tumors were grouped into five categories: “colonic type” adenocarcinoma (8140, 8141, 8260, 8262, 8440, 8460), mucinous adenocarcinoma (8470, 8471, 8480, 8481), signet ring cell carcinoma (8490), goblet cell carcinoid/adenocarcinoid (8243, 8245), and “malignant carcinoid” (8240, 8241).16 SEER only includes carcinoid tumors designated expressly by pathologists as malignant, which are hereafter referred to as “malignant carcinoid.” Except for incidence calculations, we excluded patients who had a previous diagnosis of cancer, who had tumors diagnosed “in a polyp,” whose tumor was diagnosed after death, cases that were not confirmed microscopically, cases without active follow-up, and cases for which information on ethnicity was unknown. Because tumors diagnosed as goblet cell carcinoid or adenocarcinoid in the SEER database were the same in terms of demographics and survival (data not published), they are combined in this study and are referred to as “goblet cell carcinoid.”
Continuous variables were compared using the Student t test or analysis of variance and categorical variables were compared by Pearson's chi-square. Incidence rates were age adjusted to the 1970 standard million U.S. population and expressed as cases per 1,000,000 persons.13 Observed survival (all-cause) was calculated in 6-month intervals using the life table method. For cases diagnosed between 1983 and 1997, the Cox proportional hazards model was used to determine if tumor histology affected long-term survival rates. Overall hazard of death by histology was adjusted for age and extent of disease spread at time of diagnosis and hazard of death by degree of disease extension was adjusted for age at diagnosis. To avoid the assumption of proportionality, adjustments for age and extent of disease at time of diagnosis were performed by stratifying the baseline hazards of death by these two variables instead of introducing them as covariates in the model. The assumption of proportional hazards was checked by graphing the log of the negative log of the survival function versus the log of follow-up time. Proportionality was inferred if the plotted curves were roughly parallel over time. Because the hazards were not proportional for colonic type adenocarcinoma and goblet cell carcinoid in the subgroup of cases with tumor extension beyond the colon, two models were created so that hazards for the other histologies could be compared. Analysis was performed using SAS statistical software (version 8; SAS Institute, Cary, NC). All P values reflect two-sided tests and significance was set at an alpha of 0.05.
From 1973 to 1998, 2117 primary malignant tumors of the appendix were reported to the SEER program. Of these, 320 cases were excluded because the patient had a previous diagnosis of malignancy, 75 cases were diagnosed as in situ or in a polyp, 12 were missing data on race, 9 were diagnosed at autopsy, 8 had no active follow-up, and 4 did not have histologic confirmation of the diagnosis. Adenocarcinomas were more likely to be excluded due to a previous diagnosis of malignancy (9% vs. 3–4% for the other histologies). No other differences in proportions of cases excluded were identified. There were 1689 cases eligible for inclusion in the study, 1645 of which were analyzed for the five most common histologic tumor types.
The age-adjusted incidence of appendiceal malignancies in the population represented by the SEER program was 0.12 cases per 1,000,000 people per year (crude incidence 0.12 cases per 1,000,000 people per year). No temporal trends in incidence were noted over the 26-year period. Table 1 shows patient characteristics for 1645 tumors in each of the five most common histologic types. Mucinous adenocarcinoma was the most frequent diagnosis, with 613 cases (37% of total).
|Characteristic||Mucinous adenocarcinoma (n = 613)||Colonic type adenocarcinoma (n = 411)||Signet ring cell carcinoma (n = 70)||Malignant Carcinoida (n = 324)||Goblet cell carcinoid (n = 227)|
|Age at diagnosis (yrs)|
|Mean (SD)||60 (15)||62 (16)||58 (14)||38 (17)||52 (16)|
|Male||301 (49)||245 (60)||32 (46)||89 (27)||117 (52)|
|Female||312 (51)||166 (40)||38 (54)||235 (73)||110 (48)|
|White||544 (89)||329 (80)||65 (93)||288 (89)||191 (84)|
|Black||37 (6)||55 (13)||1 (1)||27 (8)||21 (9)|
|Other||32 (5)||27 (7)||4 (6)||9 (3)||15 (7)|
|Tumor extension (%; 1983–1998, n = 1192)b|
|Confined to appendix (no serosa invasion)||78 (17)||88 (29)||5 (7)||44 (39)||74 (35)|
|Through serosa or to mesoappendix||93 (20)||104 (34)||11 (17)||35 (30)||108 (51)|
|Extension to adjacent organs/metastasis||296 (63)||114 (37)||51 (76)||36 (31)||31 (14)|
|Type of surgery (%; 1983–1998, n = 1168)b|
|Hemicolectomy or more||244 (52)||166 (54)||48 (72)||41 (36)||90 (42)|
|Less than hemicolectomy||177 (38)||118 (39)||17 (26)||55 (48)||119 (56)|
|Other surgery||27 (6)||13 (4)||1 (1)||16 (14)||4 (2)|
|No/unknown surgery||19 (4)||9 (3)||1 (1)||3 (2)||0|
|Lymph nodes involved (%; 1983–1998, n = 789)bc|
|Yes||74 (26)||73 (31)||32 (64)||32 (45)||24 (17)|
|No||216 (74)||165 (69)||18 (36)||39 (55)||116 (83)|
The mean age at diagnosis varied significantly among the tumor types. People diagnosed with malignant carcinoid were significantly younger (mean age, 38 years) than people diagnosed with any of the other morphologies (P < 0.01 for all comparisons). The mean age at diagnosis for goblet cell carcinoid (52 years) was intermediate between malignant carcinoid (38 years) and mucinous (60 years) or colonic type (62 years) adenocarcinoma.
A higher proportion of patients with colonic type adenocarcinoma were male compared to patients with mucinous adenocarcinoma (60% vs. 50%; P < 0.01). Roughly equal numbers of men and women developed goblet cell carcinoid, mucinous adenocarcinoma, and signet ring cell carcinoma. The gender ratio for malignant carcinoid was 2.6 women to 1 man, which was significantly different from all other tumor types (P < 0.01).
Tumor extension at the time of diagnosis differed by histology. Patients diagnosed with mucinous adenocarcinoma and signet ring cell carcinoma were significantly more likely to have tumor extension beyond the colon and/or metastases at the time of diagnosis than the other histologies (P < 0.01). Fewer people with goblet cell carcinoid had tumor spread beyond the colon than did those with malignant carcinoid, colonic type adenocarcinoma, or mucinous adenocarcinoma (P < 0.01 for all three comparisons). Patients diagnosed with signet ring cell carcinoma were more likely to undergo hemicolectomy or more extensive surgery than those diagnosed with the other tumor types (P < 0.01). In cases where lymph nodes were examined, 64% of patients with signet ring cell carcinoma had lymph node involvement, a significantly higher proportion than all other histologies (P < 0.05 for all comparisons).
Figure 1 shows the observed survival for all appendix malignancies by histology, unadjusted for patient age or extent of disease at presentation. A total of 1061 cases were included in the survival analysis. After adjustment for age and extent of disease at time of diagnosis, overall survival was worse for patients with signet ring cell carcinoma than for any of the other histologies (P < 0.01). Similarly, overall survival was better for patients with malignant carcinoid, but the difference in survival was not statistically significant between patients with malignant carcinoid and goblet cell carcinoid (P = 0.2).
Table 2 shows hazard ratios (HR) for the histologic types with and without adjustment for extent of disease spread at time of diagnosis. For patients with tumors that extended beyond the serosa at the time of diagnosis, those with malignant carcinoid had significantly better survival than did people with colonic type adenocarcinoma (HR 0.10, 95% confidence interval [CI] 0.01–0.78). Survival for patients with tumor spread beyond the colon and/or metastases was worse in cases of signet ring cell carcinoma compared with colonic type adenocarcinoma (HR 1.82, 95% CI 1.09–3.04) and with goblet cell carcinoid (HR 1.66, 95% CI 1.13–2.44).
|Histology||Hazard ratio, baseline hazards stratified by age groups and tumor extension at time of diagnosis (95% CI)a||Hazard Ratio by tumor extension at time of diagnosis, baseline hazards stratified by age groups (95% CI)ab|
|Confined to appendix||Beyond serosa or to mesoappendix||Extension to adjacent organs/metastasis (excluding colonic type adenocarcinoma)||Extension to adjacent organs/metastasis (excluding goblet cell carcinoid)|
|Colonic type adenocarcinoma (n = 280)||Ref||Ref||Ref||—||Ref|
|Mucinous adenocarcinoma (n = 417)||0.89 (0.71–1.10)||1.35 (0.79–2.29)||1.30 (0.81–2.09)||0.74 (0.57–0.97)c||0.79 (0.51–1.22)|
|Signet ring cell carcinoma (n = 60)||1.74 (1.24–2.43)c||1.07 (0.14–8.19)||1.64 (0.74–3.64)||1.66 (1.13–2.44)c||1.82 (1.09–3.04)c|
|Malignant carcinoid (n = 115)||0.56 (0.36–0.88)c||0.64 (0.20–2.04)||0.10 (0.01–0.78)c||0.63 (0.38–1.04)||0.65 (0.36–1.20)|
|Goblet cell carcinoid (n = 189)||0.78 (0.56–1.07)||0.81 (0.43–1.55)||0.60 (0.34–1.06)||Ref||—|
|P||< 0.01||0.5||< 0.01||< 0.01||< 0.01|
Except for signet ring cell carcinoma and malignant carcinoid, histology did not have a significant impact on survival when age and extent of disease at the time of diagnosis were taken into account. Even though the unadjusted data suggest that a difference in survival does exist among the other tumor types, much of the difference in these data is explained by the degree of disease spread that is present when the tumors are discovered. Survival analysis by extent of disease limited the number of cases that were analyzed because detailed data on tumor spread were not collected before 1983. Therefore, the subsets within each group were small (Table 1). Future analyses of population-based data that include more cases may detect differences in survival between the other histologies. Alternatively, no other differences in survival may be present between the histologies.
The data analyzed for this population-based study confirm many of the findings reported from smaller case series of appendiceal malignancies. As suggested by other authors, the overall biologic behavior of goblet cell carcinoids in our series was intermediate between that of adenocarcinomas and carcinoid tumors in terms of age at diagnosis, extent of disease spread at diagnosis, and number of cases with lymph node involvement.17, 18
The contrast between mucinous adenocarcinoma and the so-called colonic type adenocarcinoma of the appendix has been given much consideration.18, 19 The tumors are separate entities and the survival for patients with mucinous adenocarcinoma is better than that of people with the nonmucinous variety in two studies and worse in one.5, 7, 18 In our study, the hazards of death were equivalent for mucinous adenocarcinoma and colonic type adenocarcinoma. Although no differences in survival were evident, the differences in gender ratio and affected ethnicities support the notion that mucinous adenocarcinoma and colonic type adenocarcinoma are two distinct morphologies.
Primary signet ring cell carcinoma of the appendix is an exceedingly rare entity, and little information on the distinct characteristics of this tumor has been published. Like signet ring cell carcinomas of the stomach and colon, people with signet ring cell carcinoma of the appendix in this study had greater tumor extension at the time of diagnosis. Survival for patients with this type of tumor in the appendix is significantly worse than for patients with other tumor histologies.10 Moreover, the signet ring cell morphology itself did adversely affect survival even after disease extent and age were taken into account. Signet ring cell carcinoma is likewise a distinct tumor type in the appendix that should be considered separately from other carcinomas, especially because of its poor prognosis.
As has been described for carcinoid tumors in general, malignant carcinoids in this analysis were more common in women, had a younger average age at diagnosis, and had a better overall survival than all other histologic types of appendiceal tumors.20, 21 Malignant carcinoid was the only other histology that had an impact on survival independent of age and extent of disease at time of diagnosis.
The data used for this study were collected from many sites around the United States and include diagnoses from hundreds of pathologists. Because of the way in which the data were collected, there are no effective means of establishing the reliability of the tissue diagnoses. Some of the diagnoses submitted to the SEER program may be incorrect, which could impact the reliability of our results. Our findings from this large-scale analysis confirm many of the findings already reported in the literature about appendiceal cancers. Even though our conclusions are limited by the observational nature of this study, these data may serve as a basis for comparison for future studies.
This study is the most extensive comparison of tumors of the appendix by histology published to date. The use of population-based data enhances the generalizability of the findings by including cases reported from a variety of institutions within a defined population. Surveillance data offer a statistically robust yet systematic view of appendiceal cancer, revealing information about the relative importance of histology and extent of disease that can help physicians and their patients to understand the implications of the diagnosis.
- 271,000 human appendix specimens: a final report, summarizing forty years' study. Am J Proctol. 1963; 14: 365–381..
- 13RiesLAG, EisnerMP, KosaryCL, et al., editors. SEER cancer statistics review, 1973–1998, National Cancer Institute. Bethesda, MD [monograph online]. Available from URL: http://seer.cancer.gov/Publications/CSR1973_1998/, 2001 [accessed October 9, 2001].
- 14National Cancer Institute. About SEER. [website online] Last updated April 5, 2001. Available from URL: http://seer.cancer.gov/AboutSEER.html [accessed October 15, 2001].
- 15Surveillance, Epidemiology, and End-Results (SEER) program. Public-use data (1973–1998). Bethesda: National Cancer Institute, DCCPS, Surveillance Research Program. Released April 2001, based on August 2000 submission.
- 16PercyC, Van HoltenV MuirC, editors. International Classification of Diseases for Oncology (ICD-O). 2nd edition. Geneva: World Health Organization, 1990.
- 21A retrospective analysis of 1570 appendiceal carcinoids. Am J Gastroenterol. 1998; 93: 422–428., .Direct Link: