SEARCH

SEARCH BY CITATION

Keywords:

  • esophageal cancer;
  • outcomes;
  • Surveillance, Epidemiology, End Results;
  • signet ring;
  • prognosis

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURE
  9. REFERENCES

BACKGROUND

Signet ring cell esophageal adenocarcinoma histology has been difficult to study in single institution series because of its relative rarity, yet has an anecdotal reputation for poor prognosis. The Surveillance, Epidemiology, and End Results (SEER) database was examined to assess the prognostic implications of this esophageal adenocarcinoma subtype.

METHODS

All patients with esophageal adenocarcinoma in the SEER database from 2004 to 2009 were included. Univariate and multivariate analyses examining the relationship of signet ring cell histology with overall survival were performed in all patients, as well as those undergoing surgical resection.

RESULTS

A total of 596 of 11,825 (5%) study patients had signet ring cell histology. Patients with signet ring cell histology were similar in age, race, and sex distribution, but had a higher grade (P <  .001) and higher stage (P <  .001) at diagnosis. In both the all-patient group as well as those undergoing surgical resection, univariate analyses showed a worse survival in patients with signet ring cell esophageal cancer (hazard ratio [HR] = 1.24; 95% confidence interval [CI] = 1.13-1.36 and HR = 1.57; 95% CI = 1.29-1.93, respectively). In multivariate analyses adjusting for covariates, patients with signet ring cell cancer had a worse prognosis than those without (HR = 1.18; 95% CI = 1.07-1.30). In surgically resected patients, this remained a trend, but did not reach statistical significance (HR = 1.16; 95% CI = 0.94-1.42).

CONCLUSIONS

This large study of esophageal adenocarcinoma confirms the clinical impression that signet ring cell variant of adenocarcinoma is associated with an advanced stage at presentation and a worse prognosis independent of stage of presentation. Cancer 2013;119:3156–3161. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURE
  9. REFERENCES

The last few decades have witnessed an increase in the incidence of esophageal cancer in the United States. This is constituted mainly by an increase in the incidence of esophageal adenocarcinoma, a disease that is marked with a dismal prognosis.[1] Just as in many other organs, adenocarcinoma of the esophagus has several morphological subtypes. Signet ring cell adenocarcinoma is a variant of esophageal adenocarcinoma characterized by a significant proportion of tumor cells (usually > 10%-50%) showing an eccentric nucleus pushed to the cellular periphery by cytoplasmic mucin. Signet ring cells in other organs have been associated with a poor prognosis,[2, 3] but the clinical relevance of this histology in esophageal cancer is largely unknown. Current evidence largely stems from case reports[4-6] and single-institution reports.[7, 8] The main problem with studying this tumor type is its relative rarity, ranging from 0.6% to 9% of esophageal adenocarcinomas,[5, 8] thus limiting the statistical power of such studies.

The Surveillance, Epidemiology, and End Results (SEER; seer.cancer.gov) database collects information on the incidence, prevalence, and survival of cancer from specific geographic areas representing 28% of the US population. Therefore, the SEER database provides an ideal resource for the study of relatively rare tumors and understanding their association with survival. In this study, we sought to examine the demographic and clinical characteristics of patients in the SEER database with signet ring cell adenocarcinomas of the esophagus and to compare them with those of adenocarcinomas without documented signet ring cell features.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURE
  9. REFERENCES

The latest version of the SEER database (limited dataset) containing data until 2009 was queried to identify all patients with esophageal cancer. To keep the conclusions drawn from analyses relevant to current practice, only patients diagnosed from 2004 onward were included. Among these patients, only those with adenocarcinoma (not otherwise specified [NOS]) and those with signet ring cell adenocarcinoma (histology codes 8140 and 8490, respectively) were selected for further analysis. Patients with diagnosis made at autopsy and those with unknown surgery or radiation therapy, or with unknown radiation–surgery sequence, were excluded. The location of the tumor was categorized into 3 groups: upper third, middle third, and lower third. Cervical tumors were included in the upper third group, thoracic tumors were included in the middle third group, and abdominal tumors were included in the lower third group. The distributions of demographic and disease-related variables among the signet ring cell and non–signet ring cell populations were compared using the Wilcoxon rank sum test (or Kruskal-Wallis test) in the case of ordinal variables and the Pearson chi-square test for categorical variables. For all survival analyses, follow-up was truncated at 72 months. This was done so that overall survival (OS) better reflects disease status. Univariate survival analyses using the Kaplan-Meier method were performed to examine the association of variables with OS. Multivariate analyses were then performed to examine the association of histology with OS after adjusting for other covariates, using Cox regression. These covariates included age, sex, race, grade, stage (derived from American Joint Commission on Cancer stage, 6th edition), esophagectomy status, tumor location, and radiation therapy. All analyses were performed in the whole population as well as the subset of patients undergoing esophagectomy. Results were deemed statistically significant if P < .05. All analyses were performed with SAS, version 9.3.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURE
  9. REFERENCES

The SEER database has data on 64,799 patients with esophageal cancer. After the inclusion criteria and exclusion criteria were applied, a total of 11,825 patients were deemed eligible for analysis. The effect of each selection criterion on the study patient number is detailed in Table 1. Of the 11,825 patients selected for final analyses, 5% had signet ring cell histology (Table 2). There were no statistically significant differences between the 2 comparison populations in demographic characteristics such as age, sex, and race. Interestingly, there was no difference in the location of the tumor between both groups, with most patients having a tumor in the lower third of the esophagus (79.1%). However, signet ring cell histology was associated with a higher grade with 81% of patients having grade III or IV disease (versus 43% in non–signet ring cell adenocarcinoma). Consistent with this finding, patients with signet ring cell were diagnosed at a higher stage compared with their non–signet ring cell counterparts and were more likely to get neoadjuvant radiation therapy. For comparison, staging using both historic staging and conventional AJCC-based staging are presented, confirming a higher stage at diagnosis in patients with signet ring cell histology (P < .001). Despite this higher stage, there is no statistically significant difference in the proportion of patients undergoing esophagectomy in either group.

Table 1. Steps in Sample Selection and Effect on Sample Size
StepSelection criteriaCount
1Select Respiratory SEER Data with Schema–esophageal64,799
2Year of diagnosis > 200322,886
3Only patients with esophageal adenocarcinoma and signet ring cell adenocarcinoma12,116
4Exclude patients diagnosed at autopsy, patients with unknown surgery and radiation therapy11,825
Table 2. Clinical Characteristics of Patients in Each Histologic Group
CharacteristicAdenocarcinomaSignet Ring Cell HistologyP
  1. Abbreviations: AJCC, American Joint Commission on Cancer; NOS, not otherwise specified; SEER, Surveillance, Epidemiology, and End Results.

Count11,229 (95%)596 (5%) 
Age, y, mean (standard deviation)67.1 (12.2)66.9 (12.5).94
Sex (% female)14.6%14.9%.82
Race (%)   
White9595.5.82
Black2.42.3 
Others2.62.2 
Esophagectomy (%)28.2%29.2%.64
Grade (%)  <.001
I4.90.3 
II33.72.5 
III41.778.2 
IV1.32.8 
Unknown18.416.1 
SEER historic stage A (%)  <.001
In situ1.60 
Localized2419.6 
Regional28.037.9 
Distant37.333.6 
Unstaged9.08.9 
Radiation therapy (% yes)49.9%56.6%.001
No radiation9265 (82.5%)466 (78.2%).024
Presurgical radiation1394 (12.4%)90 (15.1%) 
Postsurgical radiation570 (5.1%)40 (6.7%) 
Location  .718
Upper146 (1.3%)7 (1.2%) 
Middle848 (7.6%)43 (7.2%) 
Lower8870 (79.0%)482 (80.9%) 
Other1365 (12.2%)64 (10.7%) 
Derived AJCC stage, 6th edition (%)<.001
Stage 01.60 
Stage I18.213.8 
Stage IIA9.69.4 
Stage IIB6.77 
Stage III16.324.8 
Stage IV NOS2.73.5 
Stage IVA3.74 
Stage IVB2822 
Stage unknown13.115.4 

Examination of the association of individual variables with OS showed that increasing age, female sex, and black race (compared with white race) portended worse prognoses (Table 3). Consistent with expectations, increasing grade and stage corresponded with worse OS. Patients undergoing esophagectomy and those given radiation therapy either before or after surgery had significantly better prognoses than those who did not (Table 4). Patients with signet ring cell histology fared worse than the non–signet ring cases (HR = 1.24; Fig. 1A). Multivariate analysis demonstrated that this association persists despite adjusting for covariates (HR = 1.18). Other factors that were associated with survival on this analysis are shown in Table 3.

Table 3. Variables Associated With Overall Survival in Univariate and Multivariate Analyses for the Entire Study Population
VariableReference LevelUnivariate Analysis HR (95% CI)Multivariate Analysis HR (95% CI)
  1. Abbreviations: CI, confidence interval; HR, hazard ratio; NOS, not otherwise specified.

Age 1.02 (1.02-1.02)1.02 (1.02-1.02)
Female sexMale1.18 (1.11-1.26)1.05 (0.99-1.12)
Race   
BlackWhite1.17 (1.02-1.35)1.19 (1.03-1.37)
Others 1.05 (0.91-1.21)0.96 (0.84-1.11)
EsophagectomyNo resection0.29 (0.27-0.30)0.46 (0.42-0.50)
Presurgical radiationNo radiation0.39 (0.36-0.43)0.87 (0.78-0.97)
Postsurgical radiationNo radiation0.57 (0.51-0.64)0.82 (0.73-0.92)
Signet ring cellAdenocarcinoma1.24 (1.13-1.36)1.18 (1.07-1.30)
Location   
LowerUpper1.04 (0.85-1.26)1.02 (0.84-1.25)
Middle 1.22 (0.99-1.51)1.16 (0.94-1.44)
Other 1.34 (1.09-1.64)1.18 (0.96-1.45)
Grade   
IIGrade I1.33 (1.18-1.50)1.11 (0.98-1.26)
III 1.88 (1.67-2.12)1.47 (1.31-1.66)
IV 1.54 (1.24-1.91)1.31 (1.05-1.62)
Unknown 1.22 (1.07-1.39)1.00 (0.88-1.13)
Stage   
Stage IStage 02.24 (1.66-3.02)2.07 (1.53-2.80)
Stage IIA 2.66 (1.97-3.61)2.67 (1.96-3.64)
Stage IIB 3.31 (2.43-4.49)3.37 (2.47-4.61)
Stage III 4.18 (3.11-5.64)4.39 (3.24-5.94)
Stage IV NOS 5.87 (4.26-8.08)4.74 (3.42-6.57)
Stage IVA 4.98 (3.64-6.80)4.59 (3.34-6.32)
Stage IVB 11.49 (8.55-15.44)8.67 (6.42-11.71)
Stage unknown 7.02 (5.21-9.46)4.73 (3.50-6.40)
Table 4. Clinical Characteristics of Patients Who Had an Esophagectomy in Each Histologic Group
CharacteristicAdenocarcinomaSignet Ring Cell HistologyP
  1. Abbreviations: AJCC, American Joint Commission on Cancer; NOS, not otherwise specified; SEER, Surveillance, Epidemiology, and End Results.

Count3168 (94.8%)174 (5.2%) 
Age, mean (standard deviation)63.0 (10.2)62.2 (11.1).43
Sex (% female)10.0%13.2%.18
Race (%)   
White96.296.6.97
Black2.01.7 
Others1.91.7 
Radiation therapy (% yes)53.467.2<.001
No radiation1479 (46.7%)57 (32.8%)<.001
Presurgical radiation1348 (42.6%)88 (50.6%) 
Postsurgical radiation341 (10.8%)29 (16.7%).363
Location   
Upper25 (0.8%)3 (1.7%) 
Middle171 (5.4%)13 (7.5%) 
Lower2756 (87.0%)147 (84.5%) 
Other216 (6.8%)11 (6.3%) 
Grade (%)  <.001
I6.90 
II38.02.9 
III40.581 
IV1.63.4 
Unknown13.012.6 
SEER historic stage A (%)  <.001
In situ2.00 
Localized35.218.4 
Regional51.165.5 
Distant10.214.9 
Unstaged1.61.1 
Derived AJCC stage 6th edition (%)  <.001
Stage 02.00 
Stage I27.212.6 
Stage IIA18.615.5 
Stage IIB12.49.2 
Stage III28.946.6 
Stage IV NOS1.31.7 
Stage IVA3.84.6 
Stage IVB35.7 
Stage unknown2.74 
image

Figure 1. Univariate analysis demonstrates poorer overall survival of esophageal adenocarcinoma patients with signet ring histology in (A) the entire study population and in (B) those undergoing esophagectomy. Abbreviation: CI, confidence interval.

Download figure to PowerPoint

Next, similar analyses were performed in the 3342 patients that had an esophagectomy. Again, 5.2% of these patients had signet ring cell histology. There were no differences between populations with respect to age, sex, race, and tumor location, but patients with signet ring cell histology had a higher grade, more advanced stage at diagnosis and were more likely to receive treatment with radiotherapy. Univariate survival analyses showed that increasing age, grade, stage, and signet ring cell histology were associated with poorer OS, but no relationship between survival and sex or race was evident (Table 5). Patients receiving radiation therapy had a better outcome than those who did not. In fact, the strength of association between signet ring cell histology and OS was stronger in this subset than in the overall population (HR = 1.57; Fig. 1B). However, on multivariate analysis, patients with signet ring cell histology had a trend toward worse survival (HR = 1.16) that did not reach statistical significance.

Table 5. Variables Associated With Overall Survival in Univariate and Multivariate Analyses in Patients Who Had an Esophagectomy
VariableReference LevelUnivariate AnalysisMultivariate Analysis
  HR (95% CI)HR (95% CI)
  1. Abbreviations: CI, confidence interval; HR, hazard ratio; NOS, not otherwise specified.

Age 1.02 (1.02-1.03)1.03 (1.02-1.03)
Female sexMale0.91 (0.76-1.08)0.86 (0.72-1.03)
Race   
BlackWhite0.93 (0.63-1.38)0.98 (0.65-1.46)
Others 0.87 (0.57-1.31)1.00 (0.66-1.51)
Signet ring cellAdenocarcinoma1.57 (1.29-1.93)1.16 (0.94-1.42)
Grade   
IIGrade I1.27 (0.97-1.65)0.97 (0.75-1.27)
III 2.14 (1.65-2.76)1.38 (1.06-1.80)
IV 1.61 (1.01-2.56)1.21 (0.76-1.93)
Unknown 0.73 (0.53-1.00)0.72 (0.52-0.99)
Presurgical radiationNo radiation1.28 (1.14-1.43)0.81 (0.71-0.91)
Postsurgical radiationNo radiation1.40 (1.18-1.65)0.77 (0.65-0.92)
Location   
LowerUpper1.96 (0.93-4.11)1.32 (0.63-2.78)
Middle 2.02 (0.93-4.36)1.48 (0.68-3.20)
Other 1.67 (0.77-3.60)1.30 (0.60-2.81)
Stage   
Stage IStage 01.67 (0.91-3.08)1.18 (0.63-2.22)
Stage IIA 3.55 (1.94-6.51)2.52 (1.34-4.77)
Stage IIB 4.46 (2.42-8.20)3.18 (1.67-6.03)
Stage III 6.68 (3.68-12.14)5.14 (2.73-9.68)
Stage IV NOS 8.34 (4.12-16.89)5.85 (2.80-12.21)
Stage IVA 6.83 (3.63-12.88)5.82 (2.99-11.34)
Stage IVB 10.68 (5.68-20.09)7.55 (3.89-14.67)
Stage unknown 5.04 (2.61-9.72)3.53 (1.78-6.99)

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURE
  9. REFERENCES

Histological subtyping remains an important factor in the prognostication of several solid tumors, including esophageal carcinoma. Although esophageal adenocarcinoma is usually lumped together clinically as a single entity for clinical purposes, different histological subtypes may be associated with distinct clinical behaviors. The results of the analyses presented above are consistent with this expectation. The primary inference of our work is that signet ring cell histology portends a poorer survival among esophageal adenocarcinoma patients after adjusting for oncologically relevant covariates.

The significance of signet ring cell histology has been extensively studied in gastric cancer. Based on studies from Asia, the signet ring cell carcinoma variant of gastric cancer has been associated with variable prognostic meaning, depending on stage of the disease. A study by Chiu et al[9] evaluated 2439 patients with gastric cancer and the signet ring cell histotype was found in 20.7%. On comparing these patients with those showing other histologic patterns, they found that patients with signet ring cells tended to be younger, female patients with a lower risk for lymph node metastasis, and better survival. Hyung et al[10] reported similar results, supporting signet ring cell histology as a good prognostic factor in early-stage gastric cancer. In advanced-stage disease, however, signet ring cell histology has been associated with aggressive tumor biology and worse clinical outcome.[11, 12] Additional studies have reported a lack of association with prognosis after adjusting for stage of disease.[13] Interestingly, a recent analysis of gastric signet ring cell carcinoma using the SEER database showed no difference in OS in this subtype.[14]

Although a number of studies have examined the influence of signet ring cell histology in gastric cancer, relatively little data exists in the field of esophageal cancer. Most information comes in the context of medium to large case series.[7, 8] In the study by Yoon et al, 796 patients with resected esophageal adenocarcinoma were examined, and signet ring cell histology was statistically significantly associated with reduced OS in univariate analysis. This association did not hold true on multivariate analysis, presumably due to its strong correlation with tumor grade. This is similar to our observation in surgically resected adenocarcinoma patients. This apparent lack of association may also be due to the smaller number of resected patients in our study compared to the entire population.

Our data demonstrate that the multivariate prognostic effect of signet ring cell histology is reduced in cancer patients who are treated with an esophagectomy. In addition, patients treated with esophagectomy and radiotherapy fared better than patients treated with esophagectomy alone. This observation suggests that patients with signet ring cell esophageal adenocarcinoma should be treated with radiotherapy combined with esophagectomy. Notably, there was no data on chemoradiotherapy for treatment; however, concurrent chemoradiotherapy represents a commonly employed standard of care for advanced esophageal cancer patients as compared with radiotherapy alone in the treatment period for this study (2004-2009). It might be inferred that radiotherapy may be indicative of chemoradiotherapy, but the data are not available to confirm what proportion of “radiotherapy” patients received chemoradiotherapy from this group. One way to overcome this lack of chemotherapy data is to use the SEER-Medicare database. However, use of the SEER-Medicare database for this purpose decreases the sample size significantly, because 42.3% of our study sample is under the age of 65 years. Apart from a significant decrease in size that would ensue from such a strategy, the results obtained from such an analysis may not be applicable to patients under 65, limiting the generalizability of the study.

Inferences drawn from a study such as ours must be tempered by the realization that the SEER database does not have other relevant data such as, the use, type and duration of chemotherapy, and validation of histologic classification by expert pathologists. Potentially the data may be subject to misclassification and no independent audit of signet ring cell histology in the SEER data has been reported, but signet ring morphology is a widely and easily recognized cellular phenotype. Given the rarity of these tumors, the proportion of signet ring morphology required for classification as signet ring adenocarcinoma is admittedly not standardized in the esophagus. Furthermore, it remains possible that some tumors grouped in the SEER database as esophageal primaries could be primary gastric carcinoma with dominant spread upward into the esophagus. Despite the limitations mentioned above, this large study of esophageal cancer demonstrates that signet ring cell histology portends a poorer prognosis in patients with esophageal adenocarcinoma.

So what are the clinical implication of these results? A hazard ratio of 1.18 is small when compared to those associated with stage or multimodality treatment. We do not suggest that this finding should be used to refine staging systems. However, the hazard ratio is significant enough for a clinician to take into consideration while choosing adjuvant treatment options. In addition, one may expect that in the future all known clinical, pathologic and genetic factors could be combined to develop a probability score to guide treatment decisions. These results can be viewed as a small step in that direction.

In summary, this analysis of the SEER database demonstrates that esophageal adenocarcinoma with signet ring cell histology has a worse prognosis than adenocarcinoma without signet ring cell histology. This finding deserves further study and validation in other large prospective databases and has potential clinical import.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. CONFLICT OF INTEREST DISCLOSURE
  9. REFERENCES