Carcinoid tumors rarely metastasize to the brain. The objectives of the current study were to assess the frequency of brain metastasis from carcinoid tumors, determine correlates of survival, and describe treatment modalities and their outcomes.
Between January 1977 and December 2003, 1633 patients with a carcinoid tumor were registered at The University of Texas M. D. Anderson Cancer Center. Of those, 24 patients (1.5%) had a diagnosis of brain metastasis. The authors collected demographic and clinical data and performed a statistical analysis.
The median age at the time patients were diagnosed with brain metastasis was 60 years. The metastases were treated with whole-brain radiotherapy (WBRT) alone in 7 patients (29%), and 12 patients (50%) underwent surgical resection, 7 of whom (29%) also received WBRT. The median survival time for the entire cohort after diagnosis of the primary tumor was 2.3 years (95% confidence interval [CI], 0.5–4.1 years), and the median survival time after the diagnosis of brain metastasis was 10.0 months (95% CI, 4.0–16.0 months). The longest median survival observed after the diagnosis of brain metastasis (3.2 years) occurred in patients who underwent resection and received WBRT. In the multivariate analysis, the adjusted rate ratio for comparison of all treatments versus combination of neurosurgical intervention and WBRT was 5.7 (95% CI, 1.3–26.1; P = 0.024). A positive effect of surgery followed by WBRT on the duration of survival was detected in patients with a single metastasis (P = 0.084) as well as in those with multiple metastases (P = 0.018).
Carcinoid tumors represent a heterogeneous group of slow-growing neoplasms with malignant potential that arise from widely dispersed enterochromaffin cells. Carcinoid tumors were first described accurately in 1888 by Lubarsch in autopsy findings of two patients with multiple ileal tumors.1 Oberndorfer coined the term “Karzinoide” in 1907 to emphasize the benign nature of many of these tumors.2 The overall incidence of carcinoid tumors in the United States has been estimated to be 1–2 cases per 100,000 individuals.3, 4
Whereas carcinoid tumors often follow an indolent course, numerous reports confirm a propensity of these tumors to metastasize. Metastases to the central nervous system are very rare. Our MedLine search identified 34 reported patients with single or multiple brain metastases from carcinoid tumors.5–30 The only published series located (i.e., Patchell and Posner16) reported a 5% (n = 11 patients) incidence of brain metastasis among 219 patients with carcinoid tumors. The remaining publications consisted of isolated case reports.
Because of the rarity of these metastases, there are no firm guidelines for their treatment. Randomized treatment evaluations are not possible because of the few patients with this tumor histology. We performed a retrospective review of consecutive patients with brain metastasis from carcinoid tumors who were treated at The University of Texas M. D. Anderson Cancer Center (M. D. Anderson; Houston, TX). This included an assessment of the frequency of brain metastasis and correlates of survival as well as a description of the treatment modalities and their outcomes.
MATERIALS AND METHODS
A review of the institutional database at M. D. Anderson identified 1633 patients with carcinoid tumors between January 1, 1977 and December 31, 2003. The characteristics of the carcinoid tumors in these 1633 patients are summarized in Table 1. Of these patients, 24 (1.5%) had a diagnosis of brain metastasis. These patients constituted our study population. The patients did not routinely receive a cranial imaging study, and no standard cranial screening was performed. All lesions were diagnosed because of certain clinical symptoms that prompted a neuroimaging study. This approach was consistent over the whole population of patients, and all patients diagnosed with brain metastases were included in the analyses. The requirement for informed consent was waived by the M. D. Anderson institutional review board. The medical records of the 24 patients were reviewed, and data were collected regarding patient demographic and clinical characteristics and the treatment used for the brain metastasis. The pathology findings of brain metastases from carcinoid tumors were reviewed. The recursive partitioning analysis (RPA) class was assigned retrospectively based on clinical information gathered from the chart review.31 In addition, reports describing the brain tumor imaging characteristics or the actual magnetic resonance imaging (MRI) scans were reviewed, and data were collected regarding the number and localization of lesions. The vital status of the patients at the time of the review and the cause of death, if applicable, were noted. If the cause of death was identifiable, it was categorized as due to either neurologic or systemic complications. Determination of the cause of death was based on an evaluation of several criteria, including the patient's neurologic and systemic status, radiographic imaging findings (neurologic vs. systemic), and the neurosurgeon's and primary oncologist's notes at or around the time of death, whenever available. The exact cause of death often is hard to identify, even in well-controlled prospective studies. The cause of death that we report is the one that is most likely, given the information available on the patient at or near the time of death. Statistical analysis was performed using the SPSS software, Version 10.1 (SPSS Institute, Chicago, IL). Kaplan–Meier estimates of survival were generated,32 and univariate and multivariate analyses were performed using the Cox proportional hazards model.33
Table 1. Characteristics of 1633 Patients with Carcinoid Tumors Registered at The University of Texas M. D. Anderson Cancer Center between January 1, 1977 and December 31, 2003
The clinical characteristics of the patient population with brain metastases from carcinoid tumors are summarized in Table 2. The median age of the 24 patients at the time the carcinoid tumor was diagnosed was 56 years (range, 11–83 years). The median age of the 24 patients at the time of brain metastases from carcinoid tumor was 60 years (range, 18–85 years). The median interval from the diagnosis of the primary tumor to the diagnosis of the brain metastasis was 1.5 years (range, 0.0–16.3 years). Thirteen (54%) patients were men. The pretreatment Karnofsky Performance Scale (KPS) score was ≥ 70 in 21 patients (88%). The distribution of patients by RPA class was determined mainly by pretreatment KPS score. There were 3 patients (12%) in RPA Class III and 21 patients (88%) in RPA Class II. All 24 patients had evidence of extracranial systemic metastasis at the time of diagnosis of the brain tumor. Most patients (23 patients [96%]) experienced symptoms directly related to the brain metastasis. The remaining patient (4%) was asymptomatic, and the tumor was found during a regular metastatic workup. Carcinoid syndrome was present in 12 patients (50%).
Table 2. Characteristics of 24 Patients with Brain Metastases from Carcinoid Tumors
Percentages may not add up to 100% because of rounding.
Patients may have presented with more than one site of systemic metastasis or symptom.
“Eloquent brain” indicates a tumor location near sensitive functional centers (such as sensory or motor cortex, as detailed in Sawaya et al.34). “Noneloquent brain” indicates a tumor location that was distant from sensitive functional centers (also detailed in Sawaya et al.34).
Median age at diagnosis of primary tumor (yrs) (range)
Median age at diagnosis of brain metastasis (yrs) (range)
Median time from diagnosis of primary carcinoid tumor to brain metastasis (yrs)
A single brain lesion was found in 11 patients (46%), 2 lesions were found in 2 patients (8%), and ≥ 3 lesions were found in 11 patients (46%). The tumor was located in the eloquent34 cortex in 7 patients (29%) and in noneloquent34 locations in 17 patients (71%). In 10 patients (42%), ≥ 1 brain tumors had dural attachment.
Brain Tumor Treatment
During the study period, there was no established treatment protocol for patients with metastatic lesions from carcinoid tumors. Seven patients (29%) were treated with whole-brain radiotherapy (WBRT) alone. All of them had multiple brain lesions (Table 3). Twelve patients (50%) underwent surgical resection (a ≥ 95% resection rate was achieved in all 12 patients). Seven of those patients (29%) were treated subsequently with WBRT at the discretion of the treating physician as part of a planned, multimodality approach to therapy. Of those seven patients, four had single lesions. One patient (4%) underwent surgical resection followed by stereotactic radiosurgery (SRS). All patients underwent systemic chemotherapy for treatment of the primary tumor.
Table 3. Treatment for Brain Metastases from Carcinoid Tumors in 24 Patientsa
The median follow-up from the diagnosis of the primary tumor was 2.3 years (range, 0.2–20.3 years). The median follow-up from the diagnosis of the brain metastasis was 11.0 months (range, 3 months–8.3 years).
There were no local disease recurrences during follow-up, but two patients developed distant disease recurrences from their brain tumors. One of the patients had a single brain lesion and the other patient had three metastases. Both of these patients underwent resection followed by WBRT. At the end of the follow-up interval, 1 patient was alive and 23 had died (Table 4). The cause of death was neurologic in 8 patients (35%) and systemic in the remaining 15 (65%).
Table 4. Outcome of 24 Patients with Brain Metastases from Carcinoid Tumors
All patients (n = 24)
Patients with a single metastasis (n = 11)
Patients with multiple metastases (n = 13)
No. of patients alive at the end of follow-up
Cause of death
The median survival time of the entire group of patients after diagnosis of the primary tumor was 2.3 years (95% confidence interval [CI], 6.1 months–4.1 years), with a Kaplan–Meier estimated 1-year survival rate of 67% (82% for patients with a single brain lesion and 54% for patients with multiple lesions; P = 0.95 for the comparison of the 2 survival curves).
The median survival time of the entire group from the time that the brain metastasis was diagnosed was 10.0 months (95% CI, 4.0–16.0 months; Fig. 1), with a Kaplan–Meier estimated 1-year survival rate of 45% for patients with single lesions and 39% for patients with multiple lesions (P = 0.79 for the comparison of the 2 survival curves; Fig. 2).
The median survival time of patients from the time of diagnosis of brain metastasis was 1.0 year (95% CI, 0.0–2.1 years) for patients in RPA Class II and 6.0 months (95% CI, 0.0–13.9 months) for patients in RPA Class III (P = 0.15 for the comparison of the 2 survival curves). The median survival time of patients with brain metastasis who had ≥ 1 lesion attached to the dura was 1.3 years (95% CI, 0.0–2.9 years) versus 7.1 months (95% CI, 0.0–14.4 months) for patients without this imaging characteristic (P = 0.97 for the comparison of the 2 survival curves). The Kaplan–Meier estimate of survival from the time of diagnosis of brain metastasis for patients with symptoms of carcinoid syndrome was 7.0 months (95% CI, 1.9–12.0 months) and 1.8 years (95% CI, 2.6 months–3.3 years) for patients without symptoms of carcinoid syndrome (P = 0.02 for the comparison of the 2 survival curves; Fig. 3).
Kaplan–Meier estimates of survival for each treatment modality were calculated (Fig. 4A). The overall median survival time from the time of the brain treatment was 10.1 months (95% CI, 4.0–16.2 months). The median survival time of patients treated with brain surgery alone was 4.8 months (95% CI, 0.1–22.8 months) and 6.0 months (95% CI, 0.1–20.4 months) for patients treated with WBRT alone. The longest median survival time of 3.2 years was seen in patients who underwent resection followed by WBRT (95% CI, 0.8–5.5 years; P < 0.01 compared with all other treatments; Fig. 4B). The median survival time was 5.1 months for patients with a single brain lesion who underwent surgery alone. The Kaplan–Meier estimate of survival time for patients with a single metastasis treated with resection and WBRT was 1.3 years (95% CI, 0.0–5.4 years) and 5.1 months (95% CI, 2.5–7.6 months) for patients with a single metastasis who underwent any other treatment (P = 0.08). The Kaplan–Meier estimate of survival time for patients with multiple metastases treated with resection and WBRT was 3.2 years (95% CI, 1.7–4.7 years) and 7.1 months (95% CI, 0.0–19.3 months) for patients with multiple metastasis who underwent any other treatment (P = 0.02).
The Cox proportional hazards model was used to determine predictors of survival (Table 5). The statistically significant variables that predicted worse survival in the univariate analysis included the following: age ≥ 65 years (univariate rate ratio 9.5; 95% CI, 2.7–33.9; P = 0.001); a < 1-year interval from diagnosis of primary tumor to brain metastasis (univariate rate ratio 3.6; 95% CI, 1.2–10.2; P=0.02); and WBRT as the only therapy (univariate rate ratio 3.03; 95% CI, 1.1–8.7; P = 0.04). There was a trend toward poorer survival for patients who underwent resection as the only therapeutic intervention (univariate rate ratio 4.4; 95% CI, 0.7–26.8; P = 0.1). The absence of symptoms of carcinoid syndrome predicted improved survival in the univariate analysis (univariate rate ratio 0.33; 95% CI, 0.1–0.9; P = 0.03). In the Cox multivariate analysis, the adjusted rate ratio for comparison of all treatments versus surgery followed by WBRT was 5.7 (95% CI, 1.3–26.1; P = 0.024), and the adjusted rate ratio for comparison of age at diagnosis of brain metastasis ≥ 65 years versus < 65 years was 7.3 (95% CI, 2.1–25.8; P = 0.002).
Table 5. Univariate Predictors and Multivariate Analyses of Survival in 24 Patients with Brain Metastasis from Carcinoid Tumors
Using multivariate analysis, the adjusted rate ratio for all treatments versus surgery and whole-brain radiotherapy was 5.7 (95% CI, 1.3–26.1; P = 0.024) and the adjusted rate ratio for patient age ≥ 65 versus age < 65 years was 7.3 (95% CI, 2.1–25.8; P = 0.002).
Indicates referent group (i.e., the group with which other groups were compared). A rate ratio > 1.0 indicates a shorter survival, a rate ratio < 1.0 indicates a longer survival, and a rate ratio of 1.0 indicates a survival that was similar to the referent group.
“Eloquent brain” indicates a tumor location near sensitive functional centers (such as the sensory or motor cortex, as detailed in Sawaya et al.34). “Noneloquent brain” indicates a tumor location that was distant from sensitive functional centers (as detailed in Sawaya et al.34).
The overall incidence of carcinoid tumors in the United States has been estimated to be 1–2 cases per 100,000 individuals.3, 4 An analysis of 2837 cases in the United States using data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program (1950–1969) and the Third National Cancer Survey (1969–1971) found that the appendix was the most common site of carcinoid tumors, followed by the rectum, ileum, lungs, and stomach.3 An analysis of 5486 cases identified by the SEER program between 1973 and 1991 found a relative increase in the proportion of carcinoids of the lung and stomach and a decrease in the proportion of appendiceal carcinoids.4 These changes in relative incidence, however, are probably due to variations in the detection and reporting of carcinoid tumors. In the current study, appendiceal carcinoids represented 13% of carcinoid tumors registered at M. D. Anderson between January 1, 1977 and December 31, 2003. Overall, the most common site of carcinoid tumors in our study was the small intestine (29%). Carcinoids of lung origin were detected in 20% of our patients, but they contributed 71% of cases of brain metastases. The characteristics of the whole population of 1633 patients as presented in Table 1 are very similar to characteristics presented in previously published studies.2, 3
The frequency of brain metastases from carcinoid tumors in our series was 1.5%. This probably is an underestimate of the true frequency. Patients in the current series did not undergo routine brain imaging as part of the metastatic workup. Consequently, asymptomatic brain lesions may have been missed. Patients with a diagnosis of brain metastasis after their last follow-up at M. D. Anderson would also have been missed. In the one previously published series (Patchell and Posner16), the frequency of brain metastasis in patients with carcinoid tumors was 5% (11 patients). The remaining reports in the literature concerning brain metastases from carcinoid tumors originate from 23 patients described in isolated case reports.5–30 To our knowledge, there are no autopsy series available in the literature that describe the incidence of brain metastatic lesions in patients with carcinoids.
Twenty-six (76%) of the 34 patients with carcinoid tumors described in the literature had single brain metastases. In our study, single brain metastases were found in 11 patients (46%) and multiple metastases were found in 13 patients (54%). The majority of the lesions reported were purely intraparenchymal, with only six showing meningeal involvement. Smith et al.14 published a case report of a carcinoid tumor metastatic to a meningioma. Huang and Oz29 published a case report of a patient with carcinoid metastasis mimicking a meningioma. We recently performed surgical resction on a patient with multiple dural-based carcinoid metastases. The larger parafalcine lesion simulated a meningioma not only in MRI appearance but also during surgical intervention (Fig. 5). In our series, the metastatic lesions were dural based in 10 patients (42%). Although we did not find this feature to be predictive of survival (P = 0.97), it is necessary to be aware of it, mainly in patients with a meningioma-like lesion and a history of carcinoid tumors.
Carcinoid metastasis to the brain occurs after a variable time course. Patchell and Posner16 reported a median interval of 16 months (range, 0–76 months) between diagnosis of the primary site and development of a brain metastasis. Brain metastases may occur very late. Bouldin et al.30 reported a patient with carcinoid tumor metastatic to the brain 10 years after the resection of a rectal carcinoid tumor. In the current study, the median interval from the diagnosis of the primary tumor to brain metastasis was 1.5 years (range, 0.0–16.3 years).
Neurologic symptoms from brain metastases as the first manifestation of carcinoid tumors are exceptionally rare. Only two such patients have been reported.6, 8 All our patients were diagnosed with carcinoid tumors before the diagnosis of a brain metastasis was made. In 96% of patients, the diagnosis of a brain metastasis was made or the diagnostic process was initiated because of symptoms evoked by the cerebral lesion. In only one patient was the brain metastasis detected during the staging of the patient's disease.
Because of their neuroendocrine nature, carcinoid tumors produce and store a variety of biogenic amines, neuropeptides, and prostaglandins.35 The carcinoid syndrome, which consists of flushing, intermittent diarrhea, and paroxysmal dyspnea or asthma, is caused by the oversecretion of serotonin and other products including bradykinin, prostaglandins, and histamine. The syndrome occurs in approximately 10% of patients with carcinoid tumors.16, 35 It usually occurs in the setting of widely metastatic disease with liver involvement, although Ozawa et al.20 demonstrated the presence of immunoreactive serotonin within an operative specimen of brain metastatic carcinoid despite normal serum serotonin and urine 5-HIAA concentrations. In the current study, symptoms of the carcinoid syndrome were present in 12 patients (50%). In all these patients, metastatic disease was present in the liver. In our patients with symptoms of carcinoid syndrome, the median survival time from the time of diagnosis of brain metastasis was 7.0 months (95% CI, 1.9–12.0 months), but in patients without any symptoms of carcinoid syndrome, the median survival time was 1.8 years (95% CI, 2.6 months–3.3 years; P = 0.02).
Extracranial systemic metastases were present in all patients in our study. Based on the literature, in addition to the lymph nodes (89.8%), the most frequent sites of metastatic carcinoids were the liver (44.1%), the lungs (13.6%), the peritoneum (13.6%), and the pancreas (6.8%).3 In their series, Olney et al.36 reported that ileal (40%), cecal (66.7%), and colonic lesions (44.7%) metastasized most frequently. In our series, metastatic involvement of lymph nodes was detected in 79% of patients, and liver metastases occurred in 50% of patients. In general, most patients with brain metastasis have either primary lung tumors or metastases to the lung at the time of brain metastasis. The usual mechanism of metastasis to the brain involves hematogenous dissemination from the lung. However, another mechanism is possible, especially in patients with pelvic or abdominal tumors. Batson's venous plexus connects pelvic structures with the brain through a valveless set of veins that run in the spinal epidural space.37 Given the percentage of primary carcinoid tumors of the lungs in our series (71%) and the presence of liver metastases in all patients with the other primary sites of carcinoid tumor, our study seems to confirm both of these mechanisms.
The overall 5-year survival rate for patients having any carcinoid tumor regardless of primary site is 50.4% ± 6.4%.3 There are limited data available on the survival in patients with carcinoid metastases to the brain. Survival time presented in case reports varies from several weeks to several months.5–15, 17–30 Patchell and Posner16 reported a median survival time of 7 months, with 33 % of these patients surviving for > 1 year. In our patients, the overall median survival time after the diagnosis of brain metastasis was 10 months (95% CI, 4–16 months), with 42 % of patients surviving for ≥ 1 year. We did not find any difference in survival time after the diagnosis of brain metastasis between patients with single and multiple metastases (P = 0.79). The longest observed survival intervals from the time of diagnosis of brain metastasis were 8.3 years in a patient with a single metastasis and 4.2 years in a patient with multiple metastases.
There are no established treatment guidelines for patients with brain metastases from carcinoid tumors, based on the published literature. Patchell and Posner16 observed initial improvement in all patients after WBRT. Treatments implemented in reported cases cover a spectrum of possibilities from chemotherapy as the only treatment, to WBRT, SRS, surgery, or to surgical intervention followed by WBRT.5–30 The therapies most often used in our study were WBRT alone (in 29% of patients) and surgery followed by WBRT (in 29% of patients as well). Systemic chemotherapy was used in all patients but as the only treatment in 5 patients (21%). Application of the multivariate Cox hazards model to our results revealed an adjusted rate ratio of 5.7 (95% CI, 1.3–26.1; P = 0.024) for patients treated with surgical intervention followed by WBRT versus all other treatment protocols. The positive effect of surgery followed by WBRT on the duration of survival was detected in patients with a single metastasis (P = 0.08) as well as in those with multiple metastases (P = 0.02). Given the improved survival time in patients who underwent surgery and received WBRT, this combination of therapeutic interventions seems to be indicated for all patients with brain metastases from carcinoid tumors whenever their clinical condition permits it.
The authors thank David M. Wildrick, Ph.D., for editorial assistance with the article.