Effect of the number of lymph nodes sampled on postoperative survival of lymph node-negative esophageal cancer

Authors

  • Alexander J. Greenstein MD,

    1. Department of General Surgery, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Virginia R. Litle MD,

    1. Department of Cardiothoracic Surgery, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Scott J. Swanson MD,

    1. Department of Cardiothoracic Surgery, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Celia M. Divino MD,

    1. Department of General Surgery, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Stuart Packer MD,

    1. Division of Hematology-Oncology, Department of Medicine, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Juan P. Wisnivesky MD, MPH

    Corresponding author
    1. Division of General Internal Medicine, Department of Medicine, Mount Sinai School of Medicine, New York, New York
    2. Division of Pulmonary, Critical Care, and Sleep Medicine, Department of Medicine, Mount Sinai School of Medicine, New York, New York
    • Department of Medicine, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1087, New York, NY 10029
    Search for more papers by this author
    • Fax: (212) 831-8116


Abstract

BACKGROUND

The presence of lymph node (LN) metastases in esophageal cancer has important prognostic and treatment implications. However, the optimal number of LNs that should be examined for accurate staging is controversial. In the current study, the association between survival and the number of LNs evaluated was examined in patients who underwent resection of lymph node-negative (American Joint Committee on Cancer [AJCC] TNM stage I-IIA) esophageal cancer.

METHODS

All patients were identified who underwent surgery for lymph node-negative esophageal cancer between 1988 and 2003 from the Surveillance, Epidemiology, and End Results cancer registry. Patients were classified into 3 groups by the number of negative LNs sampled during surgery (≤10 LNs, 11-17 LNs, and ≥18 LNs). Esophageal cancer-specific survival was compared among these LN groups using Kaplan-Meier curves. Stratified and Cox regression analyses were used to evaluate the association between survival and the number of negative LNs after adjusting for potential confounders.

RESULTS

A total of 972 patients were included in the study. Disease-specific survival rates increased with a higher number of negative LNs. The 5-year disease-specific survival rate was 55% among patients with ≤10 negative LNs, compared with 66% and 75%, respectively, for those with 11 to 17 negative LNs and ≥18 negative LNs. The number of negative LNs was found to be significantly associated with survival in analyses stratified by tumor status. On multivariate regression controlling for age, race/ethnicity, sex, histology, tumor status, and postoperative radiotherapy, a higher number of negative LNs was found to be independently associated with higher disease-specific survival.

CONCLUSIONS

The presence of LN metastases in patients with esophageal cancer appears to have important prognostic and treatment implications. Data from the current study suggest that patients undergoing surgical resection for esophageal cancer should have at least 18 LNs removed. Cancer 2008. © 2008 American Cancer Society.

Approximately 90% of the 14,550 patients diagnosed annually with esophageal cancer in the U.S. will die of their disease.1 Esophageal resection historically has been associated with morbidity and mortality rates of approximately 40% and 10%, respectively.2 However, surgical resection remains the standard of care in trying to cure this deadly disease and improvements in techniques and care have resulted in better surgical outcomes.3

The presence of lymph node (LN) metastases in patients with esophageal cancer has important prognostic implications. The current American Joint Committee on Cancer (AJCC) system classifies patients into 2 groups according to the LN (N) status: N0 disease for patients without involvement of regional LNs versus N1 disease for those with metastases to regional LNs. As with other cancer types, LN status remains one of the most important prognostic factors for resected cases.4, 5 Fewer than 10% of patients with LN involvement survive 5 years from diagnosis compared with >50% of patients with LN-negative disease.4, 6, 7 Adjuvant treatment regimens also depend on the LN status, particularly in patients with stage I or IIA disease. Although it is currently recommended to observe most N0 cases, adjuvant chemotherapy is, in general, recommended for patients with N1 disease.8

Despite the prognostic significance of identifying LN metastases in patients with esophageal cancer, the minimum number of LNs that need to be removed during surgery is controversial. Some recent studies have addressed staging issues in patients with esophageal carcinoma, including the prognostic significance of the extent of LN involvement,9–11 the number of LNs examined, and the proportion of positive LNs.12, 13 However, these studies did not focus on assessing the significance of the number of LNs sampled on survival among a homogenous cohort of patients with N0 esophageal cancer.

We used the Surveillance, Epidemiology, and End Results (SEER) cancer registry, a nationally representative, population-based cancer data source, to evaluate the association between the number of negative LNs and survival of patients with stage I and IIA LN-negative esophageal carcinoma. We hypothesized that a greater number of negative LNs removed during surgery is associated with more accurate staging and therefore improved disease-specific survival.

MATERIALS AND METHODS

Data Source and Selection of Study Population

Cases were selected from the SEER program, a national database that collects information on all incident cancer cases in selected areas of the U.S. and covers nearly 26% of the U.S. population.14, 15 From the SEER database, we identified all cases of primary esophageal tumors (tumor site codes 15.0-15.5, 15.8, and 15.9 and International Classification of Diseases-Oncology-second revision morphology codes 8000-8030, 8033, 8041, 8051, 8052, 8070-8075, 8123, 8140-8145, 8200, 8210, 8211, 8260-1, 8430, 8480-1, 8560, and 8980) diagnosed between 1988 and 2003. Among these patients, we narrowed the focus to cases not diagnosed at autopsy or from death certificate data. Using SEER information on the extent of tumor involvement, we identified LN-negative stage I and IIA cancers based on the American Joint Committee on Cancer (AJCC) TNM classification.16 We limited our analyses to patients treated surgically from whom there was pathologic stage data available in SEER. The objective of the current study was to determine whether the number of negative LNs removed was associated with improved patient survival. Thus, to remove the effect of tumor or LN downstaging by neoadjuvant radiation all patients who received preoperative radiation were excluded. The final study cohort consisted of the combined 972 cases of stage I (T1N0M0) and stage IIA (T2-3N0M0) esophageal cancer.

On the basis of the distribution of the number of LNs sampled in SEER and on data from prior studies,12, 13 patients were classified into 3 groups according to the number of negative LNs removed during surgery: ≤10, 11 to 17, and ≥18 negative LNs. To create these categories, we also evaluated whether further numbers of groups were necessary by visually inspecting the Kaplan-Meier plots and by fitting Cox regression models. Based on these analyses, we decided to merge groups demonstrating similar survival rates, leading to the final grouping for the study.

Information regarding age at diagnosis, sex, race/ethnicity, tumor histology, and cancer treatment was obtained from SEER. Histology subtypes were classified into categories of adenocarcinoma, squamous cell carcinoma, and other histologic types. Cases were classified as resected if the SEER site-specific variable indicated that a surgical procedure had been performed within 4 months of diagnosis (site-specific surgery codes 30, 40, 50-55, 60-65, and 70). Patients were considered as having received radiotherapy if the SEER radiation code showed that the patient had undergone beam radiation.

Statistical Analysis

Differences in the distribution of sex, age category (<60 years, 60-70 years, and >70 years), race/ethnicity, T classification, histology, and postoperative radiotherapy use among patients in the 3 LN categories were evaluated using the chi-square test.

The Kaplan-Meier method was used to estimate esophageal cancer-specific survival curves separately for patients with ≤10, 11 to 17, and ≥18 negative LNs.17 Differences in survival among groups were examined using the log-rank test. Because the purpose of the study was to evaluate the association between the number of negative LNs and esophageal cancer prognosis, we used disease-specific mortality for all comparisons because it allows for controlling for unrelated causes of death. To estimate cancer-specific survival, deaths attributed to causes other than esophageal cancer were censored at the date of death. Data regarding the cause of death in the SEER registry is abstracted from the National Center for Health Statistics and obtained from state death certificates. Survival was determined as the interval from the date of cancer diagnosis to the SEER date of death. To avoid calculating survival based on a small number of observations, survival curves were not estimated beyond 10 years of esophageal cancer diagnosis. For these analyses, individuals surviving past 10 years were classified as censored at this timepoint. All other surviving patients were censored at the last follow-up time available in SEER (December 31, 2003).

We also performed stratified analyses by sex, age, race/ethnicity, T classification, histology, and postoperative radiation use comparing survival among patients within the different LN groups. In particular, we focused on the analyses stratified by T classification because the extent of local tumor involvement is a well-established independent prognostic factor for esophageal cancer patients. Cox regression models were used to evaluate the association between the number of negative LNs and esophageal cancer-specific survival after adjusting for potential confounders such as sex, age, race/ethnicity, T classification, histology, and postoperative radiotherapy use. Analyses were performed using SPSS statistical package (version 11.5; SPSS Inc, Chicago, Ill) using 2-sided tests.

RESULTS

A total of 972 patients who underwent resection for stage I to IIA primary esophageal cancer were identified from the SEER registry. The median follow-up time for the entire cohort was 27 months and esophageal cancer mortality comprised 66% of all deaths. Patients with ≤10, 11 to 17, and ≥18 LNs sampled during surgery represented 70%, 19%, and 11% of the study cohort, respectively. The baseline patient characteristics according to the number of negative LNs are summarized in Table 1. Overall, there were no significant differences with regard to the distribution of sex, age, race/ethnicity, histology, T classification, or postoperative radiation use between patients in the 3 different categories according to the number of negative LNs.

Table 1. Demographics and Tumor Characteristics for Patients With Stages I-IIA Esophageal Cancer
CharacteristicAll patients (n = 972)No. of LNs sampledP
1-10 LNs (n = 685)11-17 LNs (n = 174)≥18 LNs (n = 113)
No.%%%%
  1. LNs indicates lymph nodes.

Sex      
 Male77079798079.91
 Female20221212021 
Age, y      
 <6030331313429.63
 60–7034435353735 
 >7032534342936 
Race/ethnicity      
 White78581818282.63
 Black8088710 
 Hispanic535634 
 Other546584 
Histology      
 Adenocarcinoma61263645963.36
 Squamous cell  carcinoma31332313635 
 Other cell type475552 
T classification      
 T144746474445.35
 T224625243123 
 T327929292532 
Postoperative radiotherapy    
 No86088888891.64
 Yes1121212129 

Unadjusted and Stratified Survival Analyses

Unadjusted analysis for the entire cohort demonstrated that esophageal cancer-specific survival was significantly higher with an increasing number of negative LNs (P < .001) (Fig. 1). The 5-year esophageal cancer-specific survival rate was 55% (95% confidence interval [95% CI], 51-60%) for patients with ≤10 negative LNs compared with 66% (95% CI, 58-74%) and 75% (95% CI, 66-84%) for those with 11 to 17 and ≥18 LNs sampled, respectively (P < .001). Similarly, patients in the categories with the higher number of negative LNs had better disease specific-survival in analyses stratified by T classification, an established esophageal cancer prognostic factor. However, these analyses demonstrated that the minimal number of negative LNs associated with improved survival depends on the T classification of the tumor. For patients with T1 tumors, esophageal cancer-specific survival curves suggest that patients can be grouped into 2 categories with respect to prognosis (<18 vs ≥18 negative LNs) (Fig. 2A). Conversely, among T2 or T3 cases, patients with >10 negative LNs had the best disease-specific survival rates (Figs. 2B and 2C).

Figure 1.

Esophageal cancer-specific survival according to the number of negative lymph nodes (LNs). Disease-specific survival was found to be significantly associated with the number of negative LNs sampled during surgery (P < .01).

Figure 2.

(A) Esophageal cancer-specific survival among patients with T1 disease according to the number of negative lymph nodes (LNs). Esophageal cancer-specific survival was found to be significantly higher among patients with ≥18 negative LNs sampled during surgery compared with those with <18 LNs sampled (P < .01). (B) Esophageal cancer-specific survival among patients with T2 disease according to the number of negative LNs. Esophageal-cancer specific survival was found to be significantly greater among patients with >10 negative LNs sampled during surgery compared with those with ≤10 LNs sampled (P = .05). (C) Esophageal cancer-specific survival among patients with T3 disease according to the number of negative LNs. Esophageal cancer-specific survival was found to be significantly higher among patients with >10 negative LNs sampled during surgery compared with those with ≤10 LNs sampled (P < .01).

Analyses stratified by age, race/ethnicity, and postoperative use of radiotherapy also demonstrated significant differences in esophageal-specific survival according to the number of negative LNs (Table 2). However, the effect of the number of negative LNs on survival was not present when analyses were limited to females (P = .23), patients aged >70 years (P = .41), or those with squamous cell carcinoma (P = .51).

Table 2. Stratified Analyses of the Risk of Esophageal Cancer Deaths According to the Number of Negative Lymph Nodes
CharacteristicNo. of negative LNsP
11-17 LNs≥18 LNs
HR*(95% CI)HR*(95% CI)
  • LNs indicates lymph nodes; HR, hazards ratio; 95% CI, 95% confidence interval.

  • *

    The HR indicates the risk of disease-specific death among patients in the corresponding LN category compared with patients with ≤10 LNs sampled.

  • Given the small number of patients in this subgroup, no large sample estimate of the HR could be obtained.

Sex     
 Male0.550.39–0.790.520.33–0.82<.001
 Female1.060.60–1.900.500.20–1.24.23
Age, y     
 <600.560.31–1.010.280.11–0.79<.001
 60–700.630.38–1.030.460.22–0.95.02
 >700.790.48–1.310.730.41–1.31.41
Race/ethnicity     
 White0.570.40–0.810.450.28–0.72<.001
 Black0.380.12–1.270.280.07–1.16.04
 Hispanic2.960.94–9.33.06
 Other1.000.39–2.542.981.07–8.30.16
Histology     
 Adenocarcinoma0.570.37–0.880.300.15–0.59<.001
 Squamous cell carcinoma0.800.52–1.230.840.49–1.42.51
 Other0.030.00–9.23.02
T classification     
 T10.720.41–1.250.080.01–0.60<.001
 T20.660.41–1.070.710.38–1.32.15
 T30.560.32–0.970.560.31–0.99.02
Postoperative radiotherapy   
 No0.630.45–0.880.580.38–0.88.01
 Yes0.750.35–1.580.150.02–1.10.03

Adjusted Survival Analysis

We performed Cox regression analysis to evaluate whether the number of negative LNs was associated with survival after adjusting for potential confounders. As shown in Table 3, in comparison with the baseline group (≤10 negative LNs), patients with 11 to 17 and ≥18 negative LNs had significantly better esophageal cancer-specific survival after adjusting for age, sex, race/ethnicity, T classification, histology, and postoperative radiotherapy use (P < .001). As expected, older age (P < .001) and a more advanced T classification (P < .001) at diagnosis were associated with an increased risk of death from esophageal cancer. In addition, we found a statistically significant interaction between the number of LNs sampled and tumor histology, indicating that the survival advantage among patients with a higher number of negative LNs was limited to patients with esophageal adenocarcinoma. No significant interaction between LN status and sex or age was observed after adjusting for the effect of tumor histology.

Table 3. Multivariate Analysis of Factors Predicting Survival Among Patients With LN-Negative Esophageal Cancer
VariableHR95% CIP
  1. LN indicates lymph node; HR, hazards ratio; 95% CI, 95% confidence interval.

No. of negative LNs  
 1–10Reference<.001
 11–170.560.37–0.87 
 ≥180.270.14–0.54 
No. of negative LNs by squamous histology interaction 
 11–17 LNs by squamous cell1.320.71–2.42.02
 ≥18 LNs by squamous cell3.161.34–7.43 
Sex   
 MaleReference.19
 Female0.840.64–1.10 
Age, y   
 <60Reference<.001
 60–701.300.98–1.71 
 >701.711.30–2.25 
Race/ethnicity   
 WhiteReference.40
 Black1.040.69–1.56 
 Hispanic0.790.47–1.34 
 Other1.320.88–1.98 
Histology   
 AdenocarcinomaReference.30
 Squamous cell carcinoma1.130.85–1.51 
 Other0.730.39–1.35 
T classification   
 T1Reference<.001
 T22.732.06–3.62 
 T32.962.22–3.94 
Postoperative radiotherapy  
 NoReference.49
 Yes1.120.82–1.52 

DISCUSSION

LN status is a critical determinant of the prognosis and management of esophageal carcinoma, although to our knowledge current surgical guidelines have not yet established an optimal surgical strategy with regard to the number of LNs that should be sampled. Using nationally representative data from 972 cases of surgically resected stage I to stage IIA esophageal carcinoma, we found that an increased number of negative LNs is associated with a significant increase in disease-specific survival, which peaks at 11 to 18 negative LNs depending on the T classification. Because T classification cannot be determined in the majority of cases before pathologic staging, these data suggest that at least 18 LNs should be removed during surgery.

Adequate LN staging is very important for both patients and clinicians because the level of LN involvement is the most significant prognostic factors for patients with resected esophageal cancer. The current standard of care for esophageal cancer consists of postoperative chemotherapy for lymph node-positive adenocarcinoma patients.8, 18–21 Although current guidelines do not consistently recommend postoperative chemotherapy for patients with squamous cell esophageal carcinoma, regardless of LN status, adjuvant treatments are commonly used in clinical practice. Thus, the status of regional LNs is critical when selecting the most appropriate postoperative management of these patients.

The most likely explanation for the observed association between an increasing number of negative LNs and improved survival is incorrect staging of patients in the current study cohort. Most likely, a proportion of patients classified as having lymph node-negative disease may in fact have had cancer disseminated to regional LNs. As the number of LNs removed during surgery increases, the probability of missing a positive LN decreases and so does the proportion of patients with higher-stage disease who are misclassified as having stage I or stage IIA cancer. This resulting stage migration phenomenon may result in poor survival associated with a low LN harvest.10, 22 An alternative explanation for the current findings is that the resection of more LNs leads in and of itself to higher cure rates. Although this is possible, all patients in the current study had pathologically negative LNs, thus suggesting that stage migration is a more likely mechanism of our findings.

Another important factor to consider when interpreting the results of the current study is the potential association between the number of LNs sampled during surgery and institutional and/or surgeon practice patterns. As with other complex surgical procedures, increased surgeon and hospital volume have been correlated with lower perioperative mortality and improved clinical outcomes among patients undergoing esophagectomy.23–25 It is possible that high-volume centers with better perioperative management of esophagectomy patients and, hence, better outcome, have surgeons using approaches that remove a greater number of LNs. This, in turn, would render the extent of LN dissection a surrogate for surgical expertise and thus better survival. Given the lack of detailed provider information in SEER, we were not able to link data regarding surgical or hospital volume with the number of LNs sampled to assess whether it confounded the results.

Based on the findings of the current study, we concluded that patients undergoing surgery for esophageal cancer should have at least 18 LNs sampled. Sampling a higher number of LNs did not appear to improve outcomes of these esophageal cancer patients. Although 11 to 17 LNs was associated with improved survival among patients with T2 or T3 disease, in general surgeons do not know the true T classification until postoperative pathologic staging is completed. Similarly, Rizk et al.13 found that 18 LNs should be removed for adequate staging purposes based on data from a single institution study of 336 cases of stage I to stage IV esophageal carcinoma. In a study of 198 cases (also including all tumor and LN stages), Bollschweiler et al.12 suggested that 15 LNs be sampled to minimize misclassification of LN-positive patients. Thus, our results are consistent with prior recommendations regarding the surgical management of these patients. However, the current study is based on a larger population-based sample of homogenous patients (N0 cases), which allowed us to perform stratified and adjusted analyses to control for potential confounders and to obtain more precise estimates of the effect of LN status on survival. In addition, the size of the cohort allowed us to explore how the influence of the number of sampled LNs varied by T classification and histology. The current results are also consistent with the findings of recent studies evaluating the correlation between the number of negative LNs removed during surgery and survival for other cancer types, such as colon cancer26, 27 and nonsmall cell lung cancer.28

The results of the stratified and multivariate analyses suggest that the effect of the number of negative LNs varies according to the tumor's cell type. In particular, the number of LNs sampled had limited effect on survival in the squamous cell group. Prior studies have shown that, in comparison with patients with adenocarcinoma, those with esophageal squamous cell carcinoma have a worse overall prognosis (in part related to a higher prevalence of comorbidities), a different pattern of lymphatic involvement, and a greater tendency for the disease to spread locally rather than systemically.29–32 These factors may explain the lack of association between LN status and survival of squamous cell carcinoma patients noted in the current study. In addition, our findings did not hold within the female subgroup. However, women represented only 20% of the study population and were more likely to have squamous cell carcinomas. Furthermore, the interaction between sex and survival was not found to be statistically significant on the multivariate analysis, suggesting that our null findings were because of negative confounding by histology in combination with limited power secondary to a low number of females in our cohort.

When interpreting the results of the current analysis, it is important to consider the strengths and limitations of SEER data. We used disease-specific survival in our analyses to control for unrelated causes of death because we were interested in assessing the effect of the number of negative LNs on esophageal cancer prognosis. Information regarding the cause of death in SEER is abstracted from death certificates. Although the death certificate is an important source of data concerning disease incidence, prevalence, and mortality, inaccuracies in the reported cause of death in this document have been previously observed.33, 34 In addition, the SEER database does not include detailed clinical information regarding the type of surgical approach used, or surgeon training and experience. Similarly, there is no information in SEER regarding chemotherapy use. Thus, it is possible that some patients in the study may have had pathologically negative LNs after receiving neoadjuvant chemotherapy. However, chemotherapy is not the standard therapy for patients with N0 disease, and it is unlikely that treatment was administered on the basis of the number of negative LNs. Moreover, we excluded patients who received preoperative radiotherapy. Because chemotherapy and radiotherapy are not infrequently given together, the exclusion of these patients should have effectively reduced the number of patients in the study who received neoadjuvant chemotherapy.

The large sample size of the SEER database enabled us to have sufficient power for detecting relatively moderate associations and permitted the multivariate analyses. In addition, the SEER registry contains population-based data and therefore is less affected by referral patterns and other sources of bias that might be associated with hospital-based case series. The rigorous data extraction and coding procedures allow for the high level of ascertainment (>97% of all cancer cases) and guarantees the quality of cancer data in the SEER registry.

In summary, the results of the current study indicate that the removal of ≥18 LNs during an esophagectomy with curative intent results in improved survival in esophageal cancer patients, particularly those with adenocarcinoma. Given that >80% of patients in this national sample of esophageal cancer cases had <18 LNs removed during surgery, the results of the current study may have a considerable impact on current surgical practices and potentially lead to improvements in the staging and postoperative management of these patients. Additional research will be needed to reveal the potential underlying mechanisms of the current study findings. If validated in future studies, these findings should be incorporated into future surgical guidelines.

Ancillary