Ann Thorac Surg 2006;82:2042-2049
© 2006 The Society of Thoracic Surgeons
Original Articles: General Thoracic
Brain Metastases From Esophageal Cancer: A Phenomenon of Adjuvant Therapy?
Thomas W. Rice, MDa,b,*,
Deepak Khuntia, MDa,c,
Lisa A. Rybicki, MSa,d,
David J. Adelstein, MDa,e,
Michael A. Vogelbaum, MD, PhDa,f,
David P. Mason, MDa,b,
Sudish C. Murthy, MD, PhDa,b,
Eugene H. Blackstone, MDa,b,d
a The Center for Swallowing and Esophageal Disorders, Cleveland Clinic, Cleveland, Ohio
b Department of Thoracic and Cardiovascular Surgery, Cleveland Clinic, Cleveland, Ohio
c Department of Radiation Oncology, Cleveland Clinic, Cleveland, Ohio
d Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, Ohio
e Department of Hematology and Medical Oncology, Cleveland Clinic, Cleveland, Ohio
f Cleveland Clinic Brain Tumor Institute, Cleveland Clinic, Cleveland, Ohio
Accepted for publication June 30, 2006.
* Address correspondence to Dr Rice, Cleveland Clinic, 9500 Euclid Avenue/Desk F24, Cleveland, OH 44195 (Email: ricet{at}ccf.org).
 |
Abstract
|
|---|
BACKGROUND: Brain metastases from esophageal cancers are uncommon, yet our impression was that they occurred more frequently than expected after esophagectomy plus adjuvant therapy. Therefore, we determined (1) incidence and prevalence of, risk factors for, and survival after development of brain metastases following esophagectomy for esophageal cancer, and (2) their association with adjuvant therapy.
METHODS: From 1985 to 2002, 403 patients (52%) underwent esophagectomy alone and 369 esophagectomy plus adjuvant therapy (118 [15%] preoperative only, 124 [16%] postoperative only, and 127 [16%] both). Hazard-function methodology was used to characterize time-related occurrence of brain metastases and risk factors. Inferences were confirmed by propensity analysis.
RESULTS: Twenty-nine patients developed brain metastases, 20 within 1 year; 6 had undergone surgery alone, and 23 had adjuvant therapy. Prevalence was 2.5% 5 years after surgery alone, but 8.4%, 7.0%, and 18.4% after preoperative adjuvant therapy only, postoperative adjuvant therapy only, and both, respectively (p < 0.0001). Greater number of locoregional lymph node metastases was associated with brain metastases after surgery alone (p = 0.04). Distant metastases (p = 0.03) and both preoperative and postoperative adjuvant therapy (p = 0.004) were risk factors. Median survival after diagnosis of brain metastases was 3.5 months. Postesophagectomy propensity-matched survival was shorter after adjuvant therapy than after surgery alone; thus, time available for developing brain metastases after surgery alone was slightly lower.
CONCLUSIONS: A dose-related increased incidence of brain metastases after adjuvant therapy for esophageal cancer cannot be explained by increased longevity. Adjuvant therapy itself, not just advanced disease, appears to create conditions conducive to developing these rapidly fatal metastases.
|
Introduction
|
|---|
Brain metastases from primary esophageal cancers are uncommon [1–14]. Yet, it was our clinical impression that patients receiving surgery plus adjuvant therapy were developing more brain metastases than those undergoing surgery alone. We presumed that the survival advantage of adjuvant therapy allowed time for uncommon distant metastases to occur. To our surprise, however, brain metastases developed soon after esophagectomy, challenging this hypothesis.
Therefore, purposes of this study were to determine (1) incidence and prevalence of, risk factors for, and survival after development of brain metastases following esophagectomy for esophageal cancer, and (2) association of adjuvant therapy and its intensity with their development.
|
Patients and Methods
|
|---|
Patients
From 1985 to 2002, 772 patients underwent esophagectomy for primary esophageal cancer at Cleveland Clinic. They were identified from the Thoracic Surgery Esophagectomy Database, which the Institutional Review Board (IRB) has approved for use in research, with patient consent waived.
Cross-sectional follow-up was performed annually for vital and disease status using an IRB-approved process that included patient consent at each follow-up. Median follow-up of all patients was 17 months (range, 1 week to 206 months); among 293 patients alive at last follow-up, median follow-up was 39 months (range, 1 week to 206 months). At the time of analysis, 29 patients were recognized as having developed brain metastases.
Adjuvant Therapy for Esophageal Cancer
Four hundred three patients underwent surgery alone, and 369 received adjuvant therapy, 118 (15%) preoperative only, 124 (16%) postoperative only, and 127 (16%) both preoperative and postoperative. The majority of patients receiving preoperative adjuvant therapy (n = 231 of 245, 94%) had one to two courses of chemotherapy, either cisplatin and 5-fluorouracil or cisplatin and paclitaxel. Accelerated fractionated radiation therapy (1.5 Gy twice daily to a total dose of 30 to 45 Gy) was administered concurrently with chemotherapy in a split fashion over 4.5 weeks. Preoperative adjuvant chemotherapy alone was given to 13 patients (5.3%) early in the series, including two planned courses of cisplatin, etoposide, and doxorubicin. One patient received preoperative radiation only. Esophagectomy was performed 4 to 6 weeks later [1, 2]. The majority of patients receiving postoperative adjuvant therapy (n = 185 of 251, 74%) had chemoradiotherapy similar to the preoperative regime. This was begun 3 to 10 weeks after esophagectomy [3]. Postoperative adjuvant chemotherapy alone was given to 14 patients (5.6%), and 52 patients (21%) received radiation only.
Patients were offered preoperative adjuvant therapy for clinically staged cT3, cT4, cN1, and cM1a cancers, but some patients with less advanced cancer received it at the discretion of the treating oncologist. Postoperative adjuvant therapy was planned for all patients receiving induction therapy, but was administered in only about half; it was also offered to patients who had surgery first and were found to have pT3, pT4, or pN1 cancers. However, some patients with less advanced cancer received it at the discretion of the treating oncologist. When adjuvant therapy protocols were unavailable, as they often were early in the experience, esophagectomy alone was performed; patients with cT2N0M0 or less-advanced cancers generally had esophagectomy alone, as did those with more advanced cancers who refused adjuvant therapy. Serendipitously, this provided important treatment heterogeneity of all cancer stages, facilitating comparison of outcomes between adjuvant therapy groups and surgery alone.
Esophagectomy was performed by thoracotomy with two-field lymphadenectomy in 582 patients (75%), transhiatally with lymph node sampling in 181 (23%), and by laparotomy in 9 (1.2%; Table 1). Of the 772 patients, 755 (98%) had locoregional lymph nodes resected—13 ± 9 per patient (median 11). The number of nodes positive for metastases ranged from 0 to 27.
Brain Metastases and Their Treatment
At diagnosis of brain metastases, the lesion was solitary in 48% and unilateral in 60% (Table 2). One patient had leptomeningeal metastases. In all cases, the esophageal tumor was adenocarcinoma. In most cases, the esophageal tumor was considered to be locally controlled (Table 2), but in 44% there was evidence of other sites of metastasis. Whole-brain radiation therapy, either alone or in combination with local therapy, was the most common treatment for brain metastases (see Table 2) [4].
View this table:
[in this window]
[in a new window]
|
Table 2. Characteristics of 29 Patients at Time Brain Metastases Were Diagnosed, Their Tumor Characteristics, and Their Management
|
|
Data Analysis
Incidence and prevalence of brain metastases
The hazard function (instantaneous risk) for brain metastases was estimated separately for the adjuvant therapy and surgery-alone groups by a parametric hazard decomposition model [15]. (For additional details, see http://www.clevelandclinic.org/heartcenter/hazard.) This was done because the shape of the hazard function was unique to adjuvant therapy groups compared with surgery alone, precluding their combination.
Risk factors for brain metastases
Risk factors for brain metastases were identified separately for the adjuvant therapy and surgery-alone groups. Potential risk factors examined were those in Table 1, to which were added number of locoregional lymph nodes sampled, number of these positive for cancer, presence or absence of cancer-positive resection margins, and (for adjuvant therapy groups) timing of adjuvant therapy (preoperative, postoperative, both). Bootstrap aggregation (bagging) was used for variable selection [16, 17]. Briefly, automated stepwise variable selection was performed on 1,000 bootstrap samples and the results aggregated both by individual factors and by clusters of related factors (such as candidate transformations of scale of continuous and ordered variables). Those appearing in 50% or more of the analyses were considered reliably significant at the p 
0.1 level.
Association of adjuvant therapy with development of brain metastases
Because adjuvant therapy was preferentially administered for advanced disease, it may simply be a surrogate for advanced disease. To account for disease extent, a propensity model was developed to identify well-matched groups of patients receiving or not receiving adjuvant therapy [18, 19]. This model was based on clinical (c)TNM classification. The probability of receiving adjuvant therapy versus surgery alone was estimated from logistic regression models containing age, gender, race, cTNM classifications, tumor histology, and date of operation. Adjuvant therapy and surgery-alone patients were then matched using these propensity scores. The number of matched pairs obtained using clinical stage was 131. Development of brain metastases was compared using the log-rank test (Appendix,*
Table 1). We also developed propensity models based on pathologic staging, with similar results (not shown; Appendix,*
Table 2).
View this table:
[in this window]
[in a new window]
|
Appendix, Table 1. Variables Stratified by Adjuvant Therapy and Matching Using Only Patient and Tumor Characteristics
a
b
|
|
View this table:
[in this window]
[in a new window]
|
Appendix, Table 2. Variables Stratified by Adjuvant Therapy and Matching Using Patient, Pathologic Tumor, and Operative Characteristics
a
b
|
|
Survival
Survival after esophagectomy and survival after diagnosis of brain metastases were estimated nonparametrically by the Kaplan-Meier method [20] and parametrically by hazard decomposition [15]. Comparison of survival after surgery alone versus after adjuvant therapy used propensity-matched pairs as described above.
Presentation
Analyses were performed using SAS statistical software (versions 8 and 9; SAS Institute, Inc, Cary, NC). Data are presented as frequencies, medians with ranges, or means ± standard deviations, as appropriate. Regression coefficients are accompanied by their standard error. Proportions and other point estimates are accompanied by asymmetric confidence limits (CL) equivalent to 1 standard error (68%).
|
Results
|
|---|
Incidence and Prevalence of Brain Metastases
During follow-up, 29 patients developed brain metastases, 20 within the first year after esophagectomy. Of these 29 patients, 23 had received adjuvant therapy. Hazard functions for adjuvant therapy groups and surgery alone were different (Fig 1A). For patients who received adjuvant therapy, incidence peaked within the first year before stabilizing at a constant level, but for patients undergoing surgery alone, risk was low and fell steadily toward zero. Prevalence of brain metastases at 1 and 5 years was 6.1% and 10.9% after use of adjuvant therapy, in contrast to 1.1% and 2.5% after surgery alone (Fig 1B; p < 0.0001). Among adjuvant therapy groups, peak incidence of brain metastases after only postoperative adjuvant therapy was broader than the sharp peak after only preoperative therapy; however, when both were used, the peak was broad with a long right tail (Fig 1C). Prevalence of brain metastases at 1 and 5 years was 6.2% and 8.4% for preoperative only, 3.0% and 7.0% for postoperative only, and 9.3% and 18.4% for both preoperative and postoperative therapy (Fig 1D; p = 0.01).
View larger version (16K):
[in this window]
[in a new window]
|
Fig 1. Brain metastases after esophagectomy for esophageal cancer. (A) Hazard function (instantaneous risk) for brain metastases after surgery alone or adjuvant therapy. Solid lines are parametric estimate, and dashed lines are 68% confidence limits. (B) Prevalence of brain metastases after surgery alone or adjuvant therapy. Each step represents an occurrence, and short vertical lines are patients remaining alive (censored); vertical bars represent ±1 standard error. (C) Hazard functions for brain metastases as in (A), but with adjuvant therapy group broken down according to adjuvant therapy administered preoperatively only, postoperatively only, or both. (D) Prevalence of brain metastases in adjuvant therapy groups. Format is as in (B).
|
|
Risk Factors for Brain Metastases
In surgery-alone patients, those operated on early in the series were more likely to have developed brain metastases (p = 0.0004). These early patients were more likely to have advanced disease because adjuvant therapy was infrequently used. Brain metastases were also more likely to occur in surgery-alone patients with a greater number of locoregional lymph nodes positive for metastatic disease (p = 0.04; Fig 2). In adjuvant therapy groups, patients having other distant metastases at esophagectomy were also more likely to develop brain metastases (p = 0.03), as were those who received both preoperative and postoperative adjuvant therapy (p = 0.004).
View larger version (11K):
[in this window]
[in a new window]
|
Fig 2. Occurrence of brain metastases after surgery alone according to number of locoregional lymph nodes positive for cancer. Each step represents an occurrence, and short vertical lines are patients remaining alive (censored); vertical bars represent ±1 standard error.
|
|
Association of Adjuvant Therapy With Development of Brain Metastases
Because patients with locoregional lymph node metastases were more likely to receive adjuvant therapy (see Table 1), we initially reasoned that it was advanced disease, not adjuvant therapy per se, that made it more likely to observe brain metastases. However, in propensity-matched pairs analysis (Fig 3), patients receiving adjuvant therapy over all tumor stages (clinical, pathologic, or both) developed more brain metastases than patients undergoing surgery alone. That this difference persisted and was similar in unmatched patients points to a direct association of brain metastases with adjuvant therapy.
View larger version (15K):
[in this window]
[in a new window]
|
Fig 3. Occurrence of brain metastases in propensity-matched groups (in this illustration, matched on patient, pathologic tumor, and surgical factors). For comparison, unmatched patients are also shown. Each step represents an occurrence, and short vertical lines are patients remaining alive (censored); vertical bars represent ±1 standard error.
|
|
Survival After Occurrence of Brain Metastases
Patients died rapidly after diagnosis of brain metastases. Median survival was 3.5 months (CL 3.3 to 4.1), irrespective of whether adjuvant therapy was given (Fig 4).
View larger version (8K):
[in this window]
[in a new window]
|
Fig 4. Survival after occurrence of brain metastases following surgery alone or adjuvant therapy (p = 0.1). Each step represents an occurrence, and short vertical lines are patients remaining alive (censored); vertical bars represent ±1 standard error.
|
|
Adjuvant Therapy
Despite this important difference in pattern of incidence of brain metastasis among patients receiving adjuvant therapy versus those undergoing surgery alone, it may still be postulated that higher occurrence of brain metastases in the adjuvant therapy group was a result of its beneficial effect in prolonging survival. However, survival after esophagectomy was shorter in the adjuvant therapy group (Fig 5A), and even somewhat shorter in propensity-matched patients (Fig 5B). Thus, patients undergoing surgery alone had more, not less, time to develop brain metastases stage-per-stage, contrary to our original hypothesis. The idea, therefore, that prolongation of life by adjuvant therapy allows more brain metastases to become manifest (unmasking) is not the explanation for a high prevalence of these metastases in this group.
View larger version (12K):
[in this window]
[in a new window]
|
Fig 5. Survival after either surgery alone or adjuvant therapy (p < 0.001). Each step represents an occurrence, and short vertical lines are patients remaining alive (censored); vertical bars represent ±1 standard error. (A) Unadjusted survival. (B) Propensity-matched survival comparison (p = 0.04), based on clinical TNM.
|
|
Additional evidence that adjuvant therapy fosters brain metastases is depicted in Figure 6. Unlike surgery alone, where regional lymph nodal status predicts brain metastases, occurrence of brain metastases in patients receiving adjuvant therapy was independent of pathologic nodal status.
View larger version (14K):
[in this window]
[in a new window]
|
Fig 6. Occurrence of brain metastases in patients undergoing surgery alone versus adjuvant therapy, stratified by pathologic nodal status. Each step represents an occurrence, and short vertical lines are patients remaining alive (censored); vertical bars represent ± 1 standard error.
|
|
|
Comment
|
|---|
Prevalence and Incidence of Brain Metastases
Unlike many cancers, such as lung and breast, metastases to the brain from esophageal cancers are exceedingly rare. Prevalence of brain metastases in autopsied patients who died from esophageal cancer is reported to be 0% to 1.8% [5, 6], except for a single report of 5% to the "nervous system" [7–9]. In one study [10], after curative esophagectomy, autopsy revealed no brain metastases, despite recurrent or residual cancer in 63% of patients. In another study, Mafune and colleagues [11] observed no brain metastases despite recurrent or residual cancer at autopsy in 72% of resected patients and 94% of nonresected patients [11]. In our series, patients who did not receive adjuvant therapy had a similarly low prevalence of brain metastases. In marked contrast, prevalence of brain metastases was already sixfold greater by one year after esophagectomy among patients receiving adjuvant therapy, and ninefold greater if both preoperative and postoperative adjuvant therapy was received.
Incidence of brain metastases in 722 patients with esophageal cancer seen at the University of Michigan during a 9-year period was approximately 2.1% per year [12]. Brain metastases occurred in 12 of 334 patients undergoing esophagectomy (approximate incidence 3.6% per year), in 2 of 293 patients not resected (approximate incidence 7% per year), and in 1 of 95 registry patients (approximate incidence 1.1% per year). From the M. D. Anderson Cancer Center Familial Brain Tumor Registry of 1,588 patients followed for a median of 12 months, 27 were diagnosed with brain metastases, an incidence of 1.7% per year [13]. These figures are consistent with incidence of brain metastases in our surgery-alone patients. Shortly after esophagectomy, however, patients who had received adjuvant therapy exhibited a remarkably high peaking incidence of brain metastases before it fell to a level consistent with reported incidence. This led us to hypothesize that adjuvant therapy, by an unknown mechanism, fosters early dissemination of locally advanced cancer to the brain, an otherwise protected site.
Of interest, risk of brain metastases in the postoperative adjuvant therapy group was displaced forward in time from that of patients receiving preoperative therapy plus surgery, about equivalent to the temporal difference between these two strategies (see Fig 1C). This suggests that the important event was not surgery, but adjuvant therapy. Further, the higher peak and prolonged tail of risk after both preoperative and postoperative therapy suggest an additive, dose-related effect (albeit a coarse categorization of "dose").
Risk Factors for Brain Metastases
The only reported preoperative risk factor for developing esophageal brain metastases has been longer length of primary tumor [12]. This may be a surrogate for advanced stage. In our study, regional lymph node metastases (pN1) were associated with increased risk of developing esophageal brain metastases after surgery alone. At Cleveland Clinic, patients with esophageal cancer who are candidates for resection undergo preoperative clinical staging with endoscopic esophageal ultrasound and, when possible, endoscopic ultrasound-directed fine-needle aspiration of regional and nonregional lymph nodes. Patients found to have locally advanced disease, defined as invasion beyond the esophageal wall (cT3 or cT4), or lymph node metastases (cN1 or cM1a) are offered induction chemoradiotherapy. Despite this clear linkage of advanced disease with adjuvant therapy, the analysis identified it as a strong predictor of esophageal brain metastases in stage-matched patients. Patients at highest risk of developing esophageal brain metastases are those with locally advanced disease and those receiving both preoperative and postoperative adjuvant therapy.
Survival After Development of Brain Metastases
Weinburg and colleagues [13] and Ogawa and colleagues [14] reported that survival after diagnosis of brain metastases in esophageal cancer patients was 3.8 months (95% CL 1.1 to 6.5 months) and 3.9 months, respectively, consistent with our series [13, 14]. Although most patients are dead within months, there has been one report of a long-term survivor (>9 years) who underwent aggressive trimodality treatment of the primary esophageal cancer, cerebral metastasectomy with whole-brain radiation therapy, and renal metastasectomy [21].
Our data do not indicate the mechanism of death in these patients (ie, progression of primary or non–central nervous system metastatic disease versus neurologic decline due to the brain metastases). Of note, only about a fifth of patients received aggressive treatment of their brain metastases (whole-brain radiation therapy plus surgery or radiosurgery), and another fifth received no treatment at all. Poorly controlled systemic disease has been identified as one of the most important risk factors for death in patients with brain metastases [22]. We cannot determine from this study whether the overall poor survival was due to aggressiveness of the systemic disease or resistance to therapy of the brain metastases.
Adjuvant Therapy Phenomenon
Although the mechanism by which adjuvant therapy facilitates development of brain metastases is unknown, we offer the following speculation. The process of metastasis requires a series of steps: a "metastatic cascade," including intravasation of cancer cells at the primary site, transportation of these cells, extravasation at the secondary site, survival, and eventual proliferation [23]. The brain is a protected metastatic site due to many factors, including the blood–brain barrier, absence of cerebral lymphatic vessels, and a unique environment high in chloride. For unknown reasons, esophageal cancer cells are unlikely to metastasize to the brain, whereas it is a favored site of metastasis for neuroepithelial cells such as small-cell lung cancer and melanoma. This homing influence is believed to result from embryonic similarity of cancer and brain cells [24].
The mechanism by which adjuvant therapy modulates some step or steps in the metastatic cascade, overcomes protective barriers to metastases, or provides a homing influence to increase occurrence of brain metastases is unknown, and no insight can be gained from this study. The mechanism may be mediated by a direct effect on the primary cancer cells, assisted intravasation, enhanced survival, or facilitated proliferation. Whether the influence is on "seed," "soil," or both is speculative.
This study illustrates the therapeutic dilemma. Propensity-matched patients receiving surgery alone had better survival than those receiving adjuvant therapy. This comparison is most fair for clinically staged patients. The data demonstrate the need to identify patients who are likely to benefit from adjuvant therapy, so that the toxicity of adjuvant therapy is avoided in the majority of patients whose tumors are unlikely to respond. Because adjuvant therapy is toxic, and this study demonstrates one of these (brain metastases), and because we are unable to identify patients whose tumors will downstage from adjuvant therapy (responders), we believe surgery should be the primary modality for curative treatment. Postoperative adjuvant therapy should be reserved for pathologically advanced tumors and preoperative adjuvant therapy for the patient who requires downstaging to permit resection.
Strengths and Limitations
This is a single-institution prospective experience. Brain metastases were diagnosed clinically and by symptom-directed investigations. Thus, number of brain metastases diagnosed represents a minimum number; subclinical metastases would not be detected. Use or not of adjuvant therapy was not randomized and, indeed, would not be recommended for early-stage cancers. To minimize this selection bias, propensity analysis was used. Because no patient with squamous cell carcinoma developed brain metastases, the analysis was not able to establish that the adjuvant therapy phenomenon is mainly associated with adenocarcinoma, despite the fact that patients with squamous cell carcinoma were twice as likely to have received adjuvant therapy. This illustrates another limitation of the study, which is the small number of events that precludes in-depth analysis of some details of the phenomenon.
Conclusions
Brain metastases after esophagectomy are uncommon and rapidly lethal. However, they are most likely to occur in patients with regional nodal metastases and, particularly, those patients receiving adjuvant therapy. This appears to be a dose-related effect.
|
Footnotes
|
|---|
The Appendix, which consists of two tables, is available only online. To access it, please visit: http://ats.ctsnetjournals.org and search for the article by Rice, Vol. 82, pages 2042–9.e1–2.
* The Appendix, which consists of two tables, is available only online. To access it, please visit: http://ats.ctsnetjournals.org and search for the article by Rice, Vol. 82, pages 2042–9.e1–2. 
|
References
|
|---|
- Adelstein DJ, Rice TW, Becker M, et al. Use of concurrent chemotherapy, accelerated fractionation radiation, and surgery for patients with esophageal carcinoma Cancer 1997;80:1011-1020.[Medline]
- Adelstein DJ, Rice TW, Rybicki LA, et al. Does paclitaxel improve the chemoradiotherapy of locoregionally advanced esophageal cancer?A nonrandomized comparison with fluorouracil-based therapy. J Clin Oncol 2000;18:2032-2039.[Abstract/Free Full Text]
- Rice TW, Adelstein DJ, Chidel MA, et al. Benefit of postoperative adjuvant chemoradiotherapy in locoregionally advanced esophageal carcinoma J Thorac Cardiovasc Surg 2003;126:1590-1596.[Abstract/Free Full Text]
- Khuntia D, Sajja R, Chidel MA, et al. Factors associated with improved survival in patients with brain metastases from esophageal cancer: a retrospective review Technol Cancer Res Treat 2003;2:267-272.[Medline]
- Attah EB, Hajdu SI. Benign and malignant tumors of the esophagus at autopsy J Thorac Cardiovasc Surg 1968;55:396-404.[Medline]
- Anderson LL, Lad TE. Autopsy findings in squamous-cell carcinoma of the esophagus Cancer 1982;50:1587-1590.[Medline]
- Bosch A, Frias Z, Caldwell WL, Jaeschke WH. Autopsy findings in carcinoma of the esophagus Acta Radiol Oncol Radiat Phys Biol 1979;18:103-112.[Medline]
- Sons HU, Borchard F. Esophageal cancerAutopsy findings in 171 cases. Arch Pathol Lab Med 1984;108:983-988.[Medline]
- Chan KW, Chan EY, Chan CW. Carcinoma of the esophagusAn autopsy study of 231 cases. Pathology 1986;18:400-405.[Medline]
- Katayama A, Mafune K, Tanaka Y, Takubo K, Makuuchi M, Kaminishi M. Autopsy findings in patients after curative esophagectomy for esophageal carcinoma J Am Coll Surg 2003;196:866-873.[Medline]
- Mafune KI, Tanaka Y, Takubo K. Autopsy findings in patients with esophageal carcinoma: comparison between resection and nonresection groups J Surg Oncol 2000;74:196-200.[Medline]
- Gabrielsen TO, Eldevik OP, Orringer MB, Marshall BL. Esophageal carcinoma metastatic to the brain: clinical value and cost-effectiveness of routine enhanced head CT before esophagectomy AJNR Am J Neuroradiol 1995;16:1915-1921.[Abstract]
- Weinberg JS, Suki D, Hanbali F, Cohen ZR, Lenzi R, Sawaya R. Metastasis of esophageal carcinoma to the brain Cancer 2003;98:1925-1933.[Medline]
- Ogawa K, Toita T, Sueyama H, et al. Brain metastases from esophageal carcinoma Cancer 2002;94:759-764.[Medline]
- Blackstone EH, Naftel DC, Turner Jr ME. The decomposition of time-varying hazard into phases, each incorporating a separate stream of concomitant information J Am Stat Assoc 1986;81:615-624.
- Breiman L. Bagging predictors Machine Learning 1996;24:123-140.
- Blackstone EH. Breaking down barriers: helpful breakthrough statistical methods you need to understand better J Thorac Cardiovasc Surg 2001;122:430-439.[Free Full Text]
- Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects Biometrika 1983;70:41-55.[Abstract/Free Full Text]
- Blackstone EH. Comparing apples and oranges J Thorac Cardiovasc Surg 2002;123:8-15.[Free Full Text]
- Kaplan EL, Meier P. Nonparametric estimation from incomplete observations J Am Stat Assoc 1958;53:457-481.
- Mao YS, Suntharalingam M, Krasna MJ. Management of late distant metastases after trimodality therapy for esophageal cancer Ann Thorac Surg 2003;76:1742-1743.[Abstract/Free Full Text]
- Murray KJ, Scott C, Greenberg HM, et al. A randomized phase III study of accelerated hyperfractionation versus standard in patients with unresected brain metastases: a report of the Radiation Therapy Oncology Group (RTOG) 9104 Int J Radiat Oncol Biol Phys 1997;39:571-574.[Medline]
- Nathoo N, Chahlavi A, Barnett GH, Toms SA. Pathobiology of brain metastases J Clin Pathol 2005;58:237-242.[Abstract/Free Full Text]
- Marchetti D, Denkins Y, Reiland J, et al. Brain-metastatic melanoma: a neurotrophic perspective Pathol Oncol Res 2003;9:147-158.[Medline]
This article has been cited by other articles:
|
|
|
|
 
P. C. Austin
Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for improvement.
J. Thorac. Cardiovasc. Surg.,
November 1, 2007;
134(5):
1128 - 1135.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
