A 35-year retrospective study of carcinoid tumors in Taiwan

Differences in distribution with a high probability of associated second primary malignancies

Abstract

BACKGROUND.

A comprehensive study of carcinoid tumors from United States-based databases indicated that the small intestine, colon, rectum, and bronchopulmonary system are common locations for carcinoid tumors. In addition, certain carcinoid tumors, such as rectal carcinoids, appeared to be overrepresented in nonwhite populations in the United States. High frequencies of associated noncarcinoid malignancies were reported in some articles. The objective of the current study was to address the organ distribution, frequency of metastasis, and survival rates of carcinoid tumors and the associated noncarcinoid tumors in Taiwanese, Asian populations.

METHODS.

Two hundred twenty-eight patients with carcinoid tumors were identified and evaluated from the surgical pathology files and medical records of the Veterans General Hospital, Taipei, Taiwan from January 1970 to December 2005.

RESULTS.

In 228 carcinoid tumors that were analyzed, the rectum (60.5%) was the most common location followed by the lung (20.2%) and the thymus (6.6%). Metastatic lesions were demonstrated in 16.2% of patients. Disease extent was associated with survival. The 5-year survival rates for patients with localized, regional metastatic, and distant metastatic disease were 94.1%, 49.1%, and 0%, respectively (P < .001). Associated noncarcinoid malignancies were noted in 14% of patients with carcinoids, mainly in the gastrointestinal tract (52.9%), lung, and genitourinary system.

CONCLUSIONS.

A different organ distribution of carcinoids was observed in Taiwanese patients, who had with significantly more carcinoids located in the rectum and thymus compared with patients in Western countries. The patients with carcinoids in the current study had a high possibility of developing associated, noncarcinoid neoplasms. Surveillance of the colon, stomach, lung, and genitourinary system for second malignant tumors is recommended. Cancer 2008. © 2007 American Cancer Society.

Carcinoid tumors are slow-growing neoplasms with distinct biologic and clinical characteristics. They are derived from the enterochromaffin/neuroendocrine cell compartments, which usually are scattered throughout the body.1–7 Carcinoid tumors represents approximately 0.49% of all malignancies, and the incidence is estimated from 0.65 to 4.48 per 100,000 population per year.7–12 The gastrointestinal tract and respiratory system usually are the most common sites for carcinoid tumors, reflecting the high density of enterochromaffin/neuroendocrine cells in the mucosa of the digestive system and the respiratory epithelium.2–8, 11–15

It has been suggested that ethnic background and genetic predisposition may play important roles in the incidence and locations of carcinoid tumors.2, 4–6, 16–19 A 5-decade analysis of 13,715 carcinoid tumors in the United States, currently the largest epidemiologic series, demonstrated that the gastrointestinal tract (67.5%) and the bronchopulmonary system (25.3%) are the most commonly affected sites. Within the gastrointestinal tract, carcinoid tumors occurred mostly in the small intestine (41.8%), rectum (27.4%), and stomach (8.7%).7 In a large-scale survey by Modlin et al. of patients with carcinoid tumors who were registered in the Surveillance, Epidemiology, and End Results (SEER) Program at the National Cancer Institute (NCI), a significant difference in the distribution of certain types of carcinoid tumors was noted in white, black, and other races within the United States.5 Compared with whites in the United States, nonwhites in the United States had a higher risk of developing carcinoids of the rectum (with an actual-to-expected ratio of 3.19) and the duodenum (with an actual-to-expected ratio of 2.8) and were less likely to develop carcinoids in the lung (with an actual-to-expected ratio of 0.49). It is noteworthy that individuals in the United States of other ethnic backgrounds (nonwhite and nonblack) were at significantly lower risk of developing carcinoids in the small intestine (with an actual-to-expected ratio of 0.2). Therefore, in this study, we examined the distribution of carcinoids at varying sites in an Asian population.

Carcinoid tumors have a complex disease spectrum. Most have a benign clinical course, but some are aggressively malignant, with invasive growth and metastasis and with significantly shorter survival.5, 7 To date, the criteria for setting the degree of malignancy are not well-defined, and histologic examination has failed to provide accurate prediction for either aggressive or metastatic potential.5–7, 14, 15 Because the prognosis for patients with carcinoid tumors is influenced greatly by the presence of metastasis, the SEER Program at the NCI has classified carcinoid tumors into the following categories: localized (for tumors confined to the organ of origin), regional metastasis (for local invasion or regional lymph node metastases), and distant metastasis (for metastatic dissemination to other organs).3, 5–7, 14 Survival rates are correlated closely with the disease stage, and the 5-year survival rate ranges from 77% in patients with localized disease, to between 63% and 72% in patients with regional metastasis, and to between 27% to 39% in patients with distant metastatic disease.5, 7, 15 These staging categories also were applied in the current study. In carcinoid tumors across all anatomic sites, >20% were associated with noncarcinoid neoplasms. Associated noncarcinoid tumors were frequent in conjunction with carcinoids of the small intestine and, to a lesser degree, with gastric, colonic, and appendiceal carcinoids.7 When the second tumor is a more malignant lesion, the prognosis may be correlated more closely with the noncarcinoid malignant neoplasm.

For the current study, we retrospectively examined patients with carcinoid tumors who were enrolled at the Veterans General Hospital, Taipei, Taiwan from 1970 to 2005. The following issues related to the carcinoid tumors in Taiwanese were addressed: 1) the organ distribution; 2) the frequencies of associated tumors, both benign and malignant; and 3) the frequency of metastasis and survival rates.

MATERIALS AND METHODS

Patients and Clinical Data

Patients with carcinoid tumors were identified and retrieved from the surgical pathology files and medical records of the Veterans General Hospital, Taipei, Taiwan from January 1970 to December 2005. This study was approved by the Institutional Human Subjects Committee. Sections were reviewed according to the World Health Organization Classification of Tumors of the Digestive System and the World Health Organization Classification of Tumors of the Lung, Pleura, Thymus, and Heart.20, 21 Among the 4 groups of endocrine tumors of the gastrointestinal tract, we focused on well-differentiated endocrine tumors (carcinoids, typical carcinoids) and well-differentiated endocrine carcinomas (malignant carcinoids, atypical carcinoids), whereas mixed exocrine-endocrine carcinomas and poorly differentiated endocrine carcinomas (small cell carcinomas) were excluded from this study. Pancreatic carcinoids were not included in this study, because the parameters used to distinguish between pancreatic carcinoids and islet cell neoplasias were not defined clearly.

In the patients with carcinoid tumors who were identified for this study, the following factors were considered: location, staging, metastasis, recurrence, survival, and the presence of secondary neoplasm. The location and staging were based on panendoscopy; complete colonoscopy; barium studies; computed tomography scans of the chest, abdomen, and brain; ultrasonography of the abdomen; whole-body bone scan; and pathology examination. Patients were divided into localized, regional, and distant metastatic groups for each primary carcinoid organ site.

Statistical Analysis

Continuous variables were presented as the mean ± standard deviation, and the differences were compared with the Mann-Whitney test or the Kruskal-Wallis test. Categorical variables were compared using the Fisher exact test or the chi-square test, as appropriate. Logistic regression was used to determine the relative risk of the association with second malignancy. Overall survival was calculated as the time from diagnosis to either death or the last follow-up appointment for surviving patients. Survival probability was calculated for patients' tumor-related factors, including age, sex, site, tumor size, depth of tumor penetration, and staging. Survival was analyzed by using the Kaplan-Meier method, and differences were compared by using the log-rank test. Statistical significance was defined as a P value <.05.

RESULTS

Site

Two hundred twenty-eight patients who were diagnosed with carcinoid tumor at the Veterans General Hospital, Taipei, Taiwan from 1970 to 2005 were recruited for this study. There were 3467,311 patients registered in our hospital during the same period with a ratio of men to women (M/F ratio) of 1.08, and only 4.2% of the registered patients were veterans. Table 1 summarizes data from all 228 patients stratified by the site of the carcinoid tumor. Carcinoids occurred most frequently in the rectum/sigmoid (141 of 228 tumors; 61.8% of total), followed by the lung (46 of 228 tumors; 20.2%), and the thymus (15 of 228 tumors; 6.6%). Elsewhere in the gastrointestinal tract (8.3%), carcinoids were located in the small intestine (11 of 228 tumors; 4.8%), the stomach (7 of 228 tumors; 3.1%), and the appendix (1 of 228 tumors; 0.4%). In the large bowel, carcinoids occurred only in the rectum (138 tumors) and sigmoid (3 tumors). Of the 11 carcinoids of the small intestine, 7 tumors were located in the duodenum, 3 tumors were located in the ileum, and the location was unspecified in 1 tumor. Of the 46 carcinoids located in the respiratory system, 79% (34 of 43 tumors) were located centrally and originated from a major airway, 21% (9 of 43 tumors) were located peripherally, and, in 3 patients, the tumor location was unknown. The thymus was the third most common site of carcinoid tumors (15 tumors). In addition, 7 carcinoid tumors were identified in other locations, including 5 in the head and neck region (2 in the nasal cavity and 3 in the epiglottis or larynx) and 2 in the ovary. The mean age of all patients was 57.5 years (standard deviation, 15.1 years; range, 15–82 years), and no significant difference was evident in the age of patients with carcinoid tumors at different locations. An M/F ratio of 2.4:1 was observed.

Table 1. General Clinical Information and Distribution of 228 Carcinoid Tumors by Site

Site	Total No. of patients (%)	No. of men/women [Ratio]	Mean age & ± SD, y	No. of patients (%)			5-Year survival, %	10-year survival, %
				Metastasis	Second malignancy	Second benign neoplasia
SD indicates standard deviation. * One patient with a thymic carcinoid demonstrated multiple endocrine neoplasia, type 1. The related endocrine neoplasms and hyperplasia were not counted as second noncarcinoid neoplasms.
Rectum/sigmoid	141 (61.8)	100/41 [2.4:1]	57.6 ± 14.9	19 (13.5)	18 (12.8)	31 (22)	86.9	76.7
Lung	46 (20.2)	28/18 [1.6:1]	55.9 ± 15.4	7 (15.2)	7 (15.2)	1 (2.2)	80.2	66.8
Thymus*	15 (6.6)	13/2 [6.5:1]	54.9 ± 14.5	6 (40)	0	0	49.9	39.9
Small intestine	11 (4.8)	9/2 [4.5:1]	62.8 ± 17.3	3 (27.3)	4 (36.4)	2 (18.2)	100	50
Stomach	7 (3.1)	5/2 [2.5:1]	67.5 ± 4.7	0	2 (28.6)	0	100	66.7
Head and neck	5 (2.2)	5/0	47.6 ± 20.8	2 (40)	0	0	0	0
Ovary	2 (0.9)	0/2	53.5 ± 12	0	0	1 (50)	100	100
Appendix	1 (0.4)	0/1	76	0	1 (100)	0	0	0
All sites	228 (100)	160/68 [2.4:1]	57.5 ± 15.1	37 (16.2)	32 (14)	35 (24.8)	82.6	70.4

Associated Neoplasms

Associated noncarcinoid neoplasms were observed in 58 of 228 patients (25.4%) with carcinoid tumors, and 32 of 228 patients (14%) had malignant neoplasms (Table 1). Associated malignancies were identified frequently in conjunction with carcinoids in the small intestine (4 of 11 patients; 36.4%), stomach (2 of 7 patients; 28.6%), lung (7 of 46 patients; 15.2%), and rectum/sigmoid (18 of 141 patients; 12.8%). In the only instance of an appendiceal carcinoid in this study, the patient had an associated pheochromocytoma. No second primary tumors were demonstrated in the 15 patients who had thymic carcinoids or in the 5 patients who had carcinoids located in the head and neck region. Patients with carcinoids of the small intestine had a 3.9-fold higher risk (95% confidence interval, 1.039 ∼ 14.680; P = .044) than patients with carcinoids of rectum/sigmoid of developing a second primary malignancy (SPM). The risks for patients with other carcinoid tumors of having associated noncarcinoid malignancies did not differ significantly from the risk for patients with carcinoids of the rectum/sigmoid. Overall, 92 noncarcinoid neoplasms were identified in 58 patients, including 34 noncarcinoid malignancies in 32 patients and 58 benign noncarcinoid neoplasms in 35 patients. Nine patients with carcinoids had both benign and malignant noncarcinoid neoplasms. The associated noncarcinoid neoplasms are listed in Table 2. The most common benign neoplasm was colon adenoma (93.1%; 54 of 58 neoplasms). The gastrointestinal tract was the system most involved in malignant noncarcinoid neoplasms (52.9%; 18 of 34 neoplasms), with 8 adenocarcinomas and 1 gastrointestinal stromal tumor (GIST) in the colorectal area, 6 adenocarcinomas and 1 GIST in the stomach, 1 esophageal squamous cell carcinoma, and 1 squamous cell carcinoma of the tongue. Malignant neoplasms of the respiratory system (2 squamous cell carcinomas and 2 adenocarcinomas) and the genitourinary system (2 prostatic adenocarcinomas and 1 urothelial cell carcinoma in urinary bladder) also were observed (Table 2).

Table 2. Associated Noncarcinoid Neoplasms in the 228 Carcinoid Patients

Location and diagnosis	No. of neoplasms
GIST indicates gastrointestinal stromal tumors.
Benign
Colorectum, adenoma	54
Lip, neurofibroma	1
Breast, fibroadenoma	1
Cauda equina, schwannoma	1
Ovary, teratoma	1
Malignant
Colon, adenocarcinoma, and GIST	9
Stomach, adenocarcinoma, and GIST	7
Esophagus, squamous cell carcinoma	1
Tongue, squamous cell carcinoma	1
Lung, carcinoma	4
Skin, malignancies	3
Prostate, adenocarcinoma	2
Urinary bladder, urothelial cell carcinoma	1
Cervix, squamous cell carcinoma	2
Breast, carcinoma	1
Thyroid, papillary carcinoma	1
Nasopharyngeal carcinoma	1
Adrenal glands, pheochromocytoma	1

Metastasis Dissemination and Survival Rates

In total, metastatic lesions were demonstrated in 16.2% (37 of 228) of carcinoid tumors. The percentages of metastasized disease ranged from 0% for patients who had carcinoids of the stomach, appendix, and ovary to 40% for patients who had carcinoids of the thymus and the head and neck region. The overall 5-year and 10-year survival rates for all patients with carcinoids were 82.6% and 70.4% (Table 1), respectively, and the estimated mean survival (±standard deviation) was 257.9 ± 22 months. The 5-year survival rates for localized, regional metastatic, and distant metastatic disease for all patients with carcinoid were 94.1%, 49.1%, and 0%, respectively (P < .001). Carcinoids in the rectum/sigmoid, lung, and thymus accounted for 88.6% of the total tumors in the current study. The disease extent and survival of patients with carcinoids in these 3 locations were analyzed further (Table 3, Figs. 1–3). Among these 3 sites, rectal/sigmoid carcinoids were the smallest in size, >87% of them were present as localized lesions when they were diagnosed, and the patients with these tumors had the best 5-year and 10-year survival. Conversely, thymic carcinoids were the largest, >76% were present as nonlocalized lesions when they were diagnosed, and patients with these tumors had the worst survival (Table 1, Fig. 1). In general, survival rates were correlated inversely with the presence of regional and distant metastases (P < .001) (Figs. 2, 3).

Figure 1. The effects of disease site on the survival curve in patients with carcinoid tumors. Patients with thymic carcinoids had significantly poorer survival.

Figure 2. The effects of disease extent on survival curves in patients with colorectal carcinoid tumors. The distant metastatic group (D) had a significantly worse prognosis than the groups with localized (L) and regional (R) colorectal carcinoid tumors.

Figure 3. The effects of disease extent on survival curves in patients with bronchopulmonary carcinoid tumors. Patients with localized bronchopulmonary carcinoid tumors (L) had significantly better survival than patients with regional (R) and distantly metastatic (D) bronchopulmonary carcinoid tumors.

Table 3. Comparison of Age, Survival, Disease Extent, and Size of Carcinoids in Patients With Carcinoids of the Rectum/Sigmoid, Lung, and Thymus

Variable	Rectum/Sigmoid, n = 141	Lung, n = 46	Thymus, n = 15	P
SD indicates standard deviation, —, not applicable because of limited patient numbers. * Kruskal-Wallis test. † Log-rank test. ‡ Chi-square test.
Mean age ±SD, y	57.6 ± 14.9	55.9 ± 15.4	54.9 ± 14.5	.609*
Mean size ±SD, cm	0.9 ± 1	2.6 ± 1.4	8.4 ± 1.8	<.001*
Disease-related survival				<.001†
5-Y survival, %	86.9	80.2	49.9
10-Y survival, %	76.7	66.8	39.9
Estimated mean survival ±SD, mo	266.9 ± 32.6	231.9 ± 23.5	78.9 ± 18.5
Disease extent (%) when first diagnosed, (no./total patients)				<.001‡
Localized	87.9 (124/141)	87 (40/46)	26.7 (4/15)
Regional	6.4 (9/141)	6.5 (3/46)	76.3 (11/15)
Distant	5.7 (8/141)	6.5 (3/46)	0
5-Y survival, %
Localized	94.4	93.4	75	.203†
Regional	74.1	0	39	.045†
Distant	0	0	—	.791†

Carcinoids of the rectum/sigmoid ranged from small lesions that measured <1 cm in greatest dimension (n = 102 tumors), to lesions that measured 1 or 2 cm (n = 25 tumors), to lesions >2 cm (n = 9 tumors). Distant metastatic lesions were noted in 5 patients at diagnosis, and those patients refused further treatment; thus, the size and depth of invasion of the primary lesions in those 5 patients are unknown. Two patients with localized rectal carcinoids developed liver and lymph node metastasis 6 years after treatment. Therefore, metastasis was demonstrated in 19 patients (13.5% of all patients with carcinoids of the rectum/sigmoid). The organs involved in metastasis were lymph nodes (11 patients), liver (10 patients), bone (1 patient), and adrenal gland (1 patient). Tumor size, muscle layer invasion, and disease extent were significant factors that predicted metastasis and survival (P < .001) (Table 4, Fig. 2). Therefore, rectal/sigmoid carcinoids that measured >2 cm, or invaded to muscle layers (T2) or beyond, or that had distant metastasis had significantly more frequent metastases and worse survival, and patients with these tumors should be treated aggressively instead of undergoing excisional biopsy.

Table 4. General Clinical Information and Survival of Patients with Rectal/Sigmoid Carcinoids (N = 141 Patients)

Variable	Metastasis		5-year survival		10-year survival
	No. of patients (%)	P*	%	P†	%	P†
* Fisher exact or chi-square test. † Log-rank test. ‡ The size and depth of invasion were available from 136 of 141 patients. § Lymph node and liver metastasis were demonstrated in 2 patients with localized disease after 6 years of follow-up.
Age, y		.505		.199		.033
≤60, n = 68	10 (14.7)		90.2		84.1
>60, n = 73	8 (11)		84.4		71.1
Size, cm‡		<.001		<.001		<.001
<1, n = 102	1 (1)		94.5		88.4
1–2, n = 25	5 (20)		90.2		80.2
>2, n = 9	8 (89.9)		25.4		0
Depth‡		<.001		<.001		<.001
T1, n = 122	4 (3.3)		95.5		88
T2, n = 7	4 (57.1)		50		0
T3, n = 3	3 (100)		66.7		33.3
T4, n = 4	3 (75)		0		0
Disease extent		<.001		<.001		<.001
Localized, n = 124§	2 (1.6)		94.4		82.6
Regional, n = 9	9 (100)		74.1		74.1
Distant, n = 8	8 (100)		0		0

With 46 bronchopulmonary carcinoid tumors, the respiratory system was the second most common system involved in carcinoid tumors. In the 41 patients with bronchopulmonary carcinoids who had size information available, all 5 patients with metastatic pulmonary carcinoids had primary tumors that measured >2 cm; bronchopulmonary carcinoids that measured <2 cm did not metastasize, and patients with those tumors had significantly better 5-year survival (P = .026) (Table 5). Among the bronchopulmonary carcinoid tumors, 65% were typical carcinoids, and 35% were atypical carcinoids. Typical carcinoids were associated with a significantly lower metastatic rate (P = .04) and a better 10-year survival rate (P = .035) than atypical carcinoids (Table 5). Disease extent appeared to be a strong predictor for survival, because patients who had localized carcinoids had a significantly better prognosis and longer survival than patients who had regional and distant metastatic carcinoids (P < .001) (Table 5, Fig. 3).

Table 5. Survival of Bronchopulmonary Carcinoids Is Related to Size, Pathology Subtype, and Disease Extent

Variable	Metastasis during the whole course of follow-up		5-year survival		10-year survival
	No. of patients (%)	P*	%	P†	%	P†
* Fisher exact or chi-square test. † Log-rank test. ‡ Size was available from 41 of 46 patients. § Location was determined for 43 of 46 patients. ∥ Liver metastasis was demonstrated in 1 patient with localized disease after 2 years of follow-up
Age, y		1.000		.482
≤60, n = 25	4 (16)		78.8		78.8
>60, n = 21	3 (14.3)		81.9		53.6
Size, cm‡		.066		.026		.114
≤2, n = 20	0 (0)		100		78.8
>2, n = 21	5 (23.8)		73.9		64.6
Location§		.589		.445		.057
Central, n = 34	4 (11.8)		85.1		80.6
Peripheral, n = 9	2 (22.2)		75		25
Histology subtypes		.04		.063		.035
Typical carcinoids, n = 30	2 (6.7)		91.1		82
Atypical carcinoids, n = 16	5 (31.3)		64		46.7
Disease extent		<.001		<.001		<.001
Localized, n = 40∥	1 (2.5)		93.4		77.8
Regional, n = 3	3 (100)		0		0
Distant, n = 3	3 (100)		0		0

Thymic carcinoids, which represented 6.6% (15 of 228) of all tumors, were the third most common carcinoid tumors in this study. The estimated mean survival for patients with thymic carcinoids was 78.9 ± 18.5 months, and their 5-year and 10-year survival rates were 49.9% and 39.9%, respectively (Table 3). One patient had multiple endocrine neoplasia, type I, with parathyroid hyperplasia, adrenal cortical hyperplasia, a pancreatic islet cell tumor, and a thymic carcinoid. None of the other 14 patients with thymic carcinoids had associated noncarcinoid neoplasms. Five of 15 thymic carcinoids (33%) were inoperable because of tumor invasion of major vessels, lung, or pericardium. Suitability for operation was the key factor affecting survival (P = .007). The survival rates of the localized and regional groups in the 10 operable patients were not significantly different (Table 6). Patients with thymic carcinoids and with carcinoids of the head and neck region suffered local invasion and local recurrence, they often were inoperable, and they had poorer survival than patients with carcinoid tumors at other locations. Prognostic factors were not studied in carcinoids of the small intestine, appendix, ovary, or head and neck region because of the small numbers of patients.

Table 6. Survival of Thymic Carcinoids Is Related to the Probability of Operation

Variable	5-year survival		10-year survival		Estimated mean survival
	%	P	%	P	Months	P
* The total number of patients with thymic carcinoids was 15. The sizes of 12 of those tumors were available.
Age, y		.249		.345		.345
≤60, n = 9	63.5		47.6		92.9 ± 24.3
>60, n = 6	25		25		47.3 ± 21.6
Size, cm*		.661		.180		.418
<8, n = 6	55.6		55.6		82.1 ± 25.1
>8, n = 6	50		25		66.8 ± 25.7
Operation		.007		.007		.007
Operable, n = 10	66.7		53.3		102.2 ± 21.3
Inoperable, n = 5	20		0		13.4 ± 4
Disease extent, no. of operable patients		.771		.877		.877
Localized, n = 4	75		50		92.5 ± 21.6
Regional, n = 6	60		60		102.8 ± 31.4

DISCUSSION

Distribution of Carcinoids

The current study, which was a large-scale analysis of carcinoid tumors in an Asian population, demonstrated that rectal carcinoids are most common, whereas appendiceal and small intestinal carcinoids are rare. Apparently, the anatomic distribution of carcinoid tumors within the gastrointestinal tract in our cohort varied significantly from the distribution reported in the Western population (Table 7). A similar distribution of carcinoids was reported in a 15-year nationwide registry in Japan: The rates of carcinoids located in the rectum, nonrectal colon, appendix, and ileum were 88.6%, 8.2%, 2.3%, and 0.9% respectively.22 Fifteen small rectal carcinoid tumors that ranged in size from 2 mm to 13 mm were identified in 21,522 healthy teachers who received proctosigmoidoscopy examination, suggesting that there is a high incidence of rectal carcinoids in Japanese populations.23 In a smaller Taiwanese study, the primary sites of carcinoids in the gastrointestinal tract were the rectum (66%), duodenum (14%), stomach (8%), appendix (4%), ileum (4%), colon (2%) and cecum (2%), and the M/F ratio was 2.3.24 Thirty-four carcinoids were identified in 198,950 individuals who were screened for colorectal cancer in Haining County, China.25 Therefore, rectal carcinoids are relatively common in Japan, Taiwan, and China compared with their frequency in Western countries. The incidence of appendiceal carcinoids in Hong Kong was much lower than in Western countries (range, 0.088–0.096% vs 0.3–0.7%).26 The small intestine and the appendix were the most common sites for gastrointestinal tract carcinoids in most studies based on Western populations.2, 4–6, 16–18 Mediastinal/thymic carcinoids were more common in men and accounted for 342 of 8970 (3.8%) gut-pancreatic endocrinomas that were listed of the Japanese Niigata Registry.27 Therefore, thymic carcinoids are relatively common in Japan and have a male predominance, as reported in our patients. In the study by Modlin et al. in the United States, thymic carcinoids, along with carcinoids of other endocrine organs, accounted for only 0.3% of all carcinoids. Therefore, there appears to be a much greater propensity for thymic carcinoids in the Taiwanese and Japanese populations than in the United States population. A male predominance of rectal carcinoids was observed by Matsui et al. (M/F ratio, 2.8), Konishi et al. (M/F ratio, 2.2), and Jetmore et al. (M/F ratio, 1.7).19, 22, 23 The M/F ratio was 2.4:1 for total carcinoids, and it was even higher (6.5:1) for thymic carcinoids in our study. We conclude that there may be a race-related predisposition or protective effect for these sites in carcinoid tumors, although environmental factors or diet effects cannot be ruled out completely.

Table 7. Incidence (%) of Carcinoids at Various Gastrointestinal Locations in Several Studies

Reference	Stomach	Small intestine	Appendix	Colon	Rectum	Others
Expressed as the percentage of total gastrointestinal (GI) carcinoids. * Location of the highest incidence in the GI tract.
Shebani, 199916	4	37*	31	13	12	3
Modlin, 20037 (white)	6.8	44.1*	7.6	22.6	14.8	4.1
Modlin, 20037 (black)	5.7	36.5*	4.1	18.5	28.9	6.3
Modlin, 20037 (total)	6.6	40.9*	6.8	21.5	19.5	4.5
Onaitis, 200017	12	71*	0	4.8	12	0.2
Maggard, 20044	7.3	45.2*	16.9	10.8	19.8	0
Godwin, 19756	2.5	32.2	44.6*	5.1	15.6	0
Van Gompel, 200418	13.8	26.1	26.2*	12.3	10.8	10.8
Jetmore, 199219	5.9	15.3	11.8	5.9	55.3*	5.8
Li, 2008 (current data)	4.4	6.9	0.6	1.9	86.3*	0
Modlin, 20037 (others)	5.8	14.3	3.8	16.4	52.8*	7
Vinik, 19892 (autopsy)	2.8	84.4*	3.9	7.3	1.7	0
Vinik, 19892 (surgery)	3.1	31.6	40.3*	6.7	18.3	0

Associated Neoplasms

The association of second primary neoplasms with carcinoid tumors has been emphasized recently,5–7, 16, 19, 28–31 and the incidence reportedly ranged from 12% to as high as 46%.29, 30 Of the gastrointestinal tract carcinoids, small intestinal carcinoids reportedly had the highest rate of SPM (range, 29%–52%), followed by appendiceal carcinoids (range, 13%–32%), and colorectal carcinoids (range, 5%–32%). The most common site of associated noncarcinoid malignancies was the gastrointestinal tract,5, 6, 16, 19, 28–31 which involved from 32% to 62% of tumors, followed by the genitourinary tract (range, 9%–22%), and the lung/bronchial system (range, 9%–13%). In the current study, SPMs were demonstrated in 14% of all patients, in 12.8% of patients with colorectal carcinoids, in 15.2% of patients with bronchopulmonary carcinoids, in 28.6% of patients with gastric carcinoids, and 36.4% of patients with small intestinal carcinoids. The gastrointestinal tract (52.9%l 18 of 34 SPMs) was the most common location for SPMs followed by the respiratory system (11.8%; 4 of 34 SPMs) and the genitourinary tract (8.8%; 3 of 34 SPMs). The etiology of high risk of SPMs associated with carcinoid tumors remains unclear. The increased incidence of SPMs may arise from growth stimulation induced by neuroendocrine factors secreted by the carcinoid.32 In support of this hypothesis, high levels of neuroendocrine secretory products are observed in small intestinal carcinoids (midgut) compared with colonic/rectal carcinoids (hindgut), in which SPMs are less common. Thus, our data and that from other studies strongly suggest the need for close surveillance of the gastrointestinal tract, respiratory system, and genitourinary tract when a carcinoid tumor is identified. Long-term follow-up is recommended for patients with carcinoid tumors for both delayed metastasis and second malignancies.

Metastases and Survival

The NCI-SEER Program has demonstrated that the 5-year survival rates for patients with carcinoid tumors are correlated closely with disease stage, ranging from 77% in patients with localized disease, to 63% to 72% in patients with regional disease, and to 27% to 39% in patients with distant metastatic disease.5, 7, 15 In our study, the 5-year survival rates for patients with localized, regional metastatic, and distant metastatic were 94.1%, 49.1%, and 0%, respectively (P < .001). In agreement with these rates, our study also demonstrated that survival is correlated inversely with disease extent and tumor size in carcinoids of the rectum/sigmoid, lung, and thymus (P < .001) (Table 3). We also confirmed previous observations that tumor size and muscle layer invasion are important predictive criteria in assessing the malignant nature of rectal carcinoids1, 2, 14–19, 23, 28, 33, 34 and that patients with typical and atypical bronchopulmonary carcinoid tumors display significantly different metastatic possibilities and 10-year postoperative survival rates.35–38 In addition, we demonstrated that thymic carcinoids, also known as well-differentiated neuroendocrine carcinomas, usually bear high risks for recurrence, metastasis, and tumor-associated death, and patients with thymic carcinoids have a worse prognosis compared with bronchial carcinoids.39–44 Surgery is the key factor that influences prognosis in patients with thymic carcinoids (P = .007). When surgical treatment was permissible, the survival rates for patients with regional and localized thymic carcinoids did not differ significantly.

In conclusion, factors that determine the clinical course and outcome of patients with carcinoid tumors are complex and multifaceted. Our data, in combination with data from other studies, suggest that the size and location of primary tumors, the extent of disease (localized, regional, or distant metastatic), the feasibility of complete resection, the MIB-1 proliferating index, p53 expression, and the presence of carcinoid symptoms all are important prognostic factors.2–7, 14–16, 18, 19, 34–36, 39–45