Annals of Surgical Oncology Cite Track
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
10.1245/s10434-006-9016-x
Annals of Surgical Oncology 13:1189-1200 (2006)
© 2006 Society of Surgical Oncology
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Schwarz, R. E.
Right arrow Articles by Smith, D. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Schwarz, R. E.
Right arrow Articles by Smith, D. D.

Original Article

Extent of Lymph Node Retrieval and Pancreatic Cancer Survival: Information from a Large US Population Database

Roderich E. Schwarz, MD1 and David D. Smith, PhD2

1 Division of Surgical Oncology, The Cancer Institute of New Jersey, 195 Little Albany Street, New Brunswick, New Jersey 08903, USA
2 Division of Biostatistics, City of Hope Cancer Center, 1500 E. Duarte Road, Duarte, California 91010, USA

Correspondence: Address correspondence and reprint requests to: Roderich E. Schwarz, MD; E-mail: r.schwarz{at}umdnj.edu


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Background: Operative therapy of pancreatic cancer is associated with poor survival because of high recurrence rates after pancreatectomy. The effect of lymph node (LN) dissection on survival continues to be debated.

Methods: A pancreatic cancer data set was created through structured queries to the Surveillance, Epidemiology, and End Results 1973 to 2000 database. Stage information was created according to 6th edition American Joint Committee on Cancer tumor-node-metastasis criteria, and the effect of LN number on survival was analyzed.

Results: Out of a cohort of 20,631 patients with carcinomas of the exocrine pancreas, surgical details were available for 2,787 patients. Procedures included pancreatoduodenectomies (n = 1848; 66%), radical regional pancreatectomies (n = 516; 19%), other partial resections (n = 316; 11%), and total pancreatectomies (n = 107; 4%). For 1666 of these patients with complete clinicopathologic information, the median age was 66 years (range, 22–96 years), with an equal sex ratio. The median number of total LNs examined was 7 (range, 1–52), of positive LNs was 1 (range, 0–34), and of negative LNs was 6 (range, 0–30). Multivariate survival analysis yielded these prognostic variables: number of LNs examined, number of positive LNs, tumor size, extrapancreatic extension, radiotherapy (all P < .0001), and age (P = .0009). The greatest survival differences were observed for negative LN counts of 10 to 15.

Conclusions: Stage-based survival prediction of pancreatic cancer is strongly influenced by total LN counts and numbers of negative LNs obtained. Although the mechanism remains unclear and could reflect confounding factors (margin status and institutional volume), an attempt to resect and examine at least 15 LNs to yield preferably between 10 and 15 negative LNs seems sensible for curative-intent pancreatectomy.

Key Words: Pancreatic cancer • Postoperative survival • Lymph node counts • Lymphadenectomy • SEER population data


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Operative therapy of pancreatic cancer remains the only potentially curative treatment modality, but it is associated with high recurrence rates after pancreatectomy. Postoperatively, both distant sites of relapse and locoregional recurrences are encountered in most patients.1 Adjuvant treatment strategies have failed to clearly improve overall cure rates to date, and even positive studies have demonstrated only a modest gain in median survival.2,3 For this reason, several attempts have been made in the past decades to enhance surgical radicality. Retrospective analyses of mesenteric vascular resections during pancreatoduo-denectomy for advanced primary tumors4,5 or of regional pancreatectomy with mesenteric vascular resection and reconstruction6 have resulted in postoperative survival comparable but not superior to that achieved after standard pancreatic R0 resection. The concept of extended lymphadenectomy has been used for other gastrointestinal malignancies, such as gastric or rectal cancer, to achieve better regional disease control.7,8 Radical pancreatectomy with extended lymph node dissection (ELND) has been prospectively evaluated in pancreatic adenocarcinoma in only two randomized trials.9,10 Overall, neither of these efforts demonstrated any obvious survival benefit after ELND. Of note, the larger prospective trial evaluating radical pancreatoduo-denectomy included other periampullary cancers of nonpancreatic origin as well10; the smaller study apparently is not powered to detect a possible small survival difference. In this series, a survival difference could be found for patients with nodal positive disease in a post-hoc subgroup analysis.9 A well-documented, problematic finding has been the increase in postoperative morbidity after ELND during pancre-atoduodenectomy,10 which nevertheless did not translate into any long-term quality-of-life reduction.11

Reasons for an apparent failure to improve pancreatic cancer survival with ELND at the time of an R0 pancreatic resection have been proposed.12 If one assumes a curability potential after ELND only for those patients with isolated nodal involvement of second-echelon lymph nodes (LNs) in the absence of more distant micrometastases, the group benefiting from ELND would be rather small, and extraordinarily large cohorts would be required to demonstrate such benefit.12 However, ELND may still result in better regional disease clearance, thus improving locoregional recurrence and potentially affecting survival beneficially without altering the overall long-term cure rates. To identify such potential effects would also require patient numbers larger than feasible from any single-center surgical series. We thus investigated the relationship between LN numbers examined and survival after pancreatectomy for pancreatic cancer, using information from a large US population–based cancer database.


   PATIENTS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
A pancreatic cancer data set was created through structured queries to the publicly available version of the Surveillance, Epidemiology, and End Results (SEER) database, covering the years 1973 to 2000.13 The SEER program collects clinical information from 14 cancer registries across the United States. Stage information for exocrine carcinomas was created according to the American Joint Committee on Cancer tumor-node-metastasis criteria, 6th edition.14 From a cohort of 20,631 patients with a diagnosis of pancreatic malignancy, individuals were selected on the basis of completeness of clinicopathologic information, presence of an exocrine pancreatic cancer, therapy through a surgical resection, M0 status, and absence of documented arterial invasion (less than T4 status), with at least 1 LN examined. Patients who received radiation treatment or chemotherapy as surgical adjuvants were kept within the analysis. Patients with incomplete resection information, such as "surgery, not otherwise specified," were kept in the analysis as long as sufficient information was available to document that resection of the primary tumor had taken place, usually through details in the pathologic findings. If a pancreatic resection could not be corroborated in this fashion, individuals were excluded. Patients with a positive margin status (R1 and R2), whenever stated, were excluded as well.

The stepwise process of data extraction is depicted in Fig. 1Go. Relationships between the number of LNs examined and survival outcomes were analyzed for the entire cohort, for node-negative or node-positive groups separately, or for four stage subcategories of interest. These included the node-negative groups T1/ 2N0 and T3N0 and the node-positive groups T1/2N1 and T3N1. Differences in LN count between patient groups were calculated via Wilcoxon/Mann-Whitney test. The primary outcome parameter of interest was overall survival. Survival time, as tabulated by SEER in monthly increments, was the time from diagnosis until last contact, the date of death, or the date used as a cutoff for the SEER database. Univariate, multivariate, cutpoint, and model projection survival analyses were performed, analogous to a previously reported analysis of gastric cancer patients.15 Actuarial survival was calculated with the Kaplan-Meier method,16 and univariate comparison between groups was performed by using the log-rank test.17 A regression model to correlate LN counts with survival was fit according to Kaplan-Meier 5-year survival estimates for each LN count interval. The independent variable was constructed by using the LN count interval midpoints. Cox regression served as a multivariate technique, and a backward-elimination model was used for all covariates.18 The threshold for keeping a variable in the Cox model under backward elimination was P = .05. Independent variables entered into the Cox multivariate model were T-stage category, including the elements of tumor size (as a continuous or categorical variable) and extrapancreatic tumor extension; intrapancreatic location (or type of resection performed); grade; number of LNs examined; number of positive LNs; age at diagnosis; sex; ethnicity; year of diagnosis; and radiation or chemotherapy administered. To link survival outcomes with LN count categories, we used Monte Carlo simulation for exact P values of the Kruskal-Wallis test. All calculations were performed by using the SAS 8.2 statistical software package (SAS, Cary, NC). Significance of differences was assumed at P values of <.05.


Figure 1
View larger version (18K):
[in this window]
[in a new window]
  		FIG. 1. Surveillance, Epidemiology, and End Results (SEER) database data extraction process. LN, lymph node.

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Patient Selection Based on Operative and Pathologic Staging Information
Out of a cohort of 20,631 patients with carcinomas of the exocrine pancreas, 5,314 individuals were stated as having undergone "surgery." This term included patients with mere biopsy procedures or nonresective bypass operations. Specific information for resections regarding the pancreatectomy extent was available for only 2787 patients. These were classified as pancreatoduodenectomies (n = 1848; 66%), radical regional pancreatectomies (n = 516; 19%), other partial resections (n = 316; 11%), and total pancreatectomies (n = 107; 4%). Of these 2787 patients with sufficient operative information, 2406 individuals had M0 pancreatic adenocarcinoma (Fig. 1Go). Margin assessment of the operative specimen has not been included into the SEER data collection effort until recently. To optimize the possibility of analyzing patients with at least gross margin-negative (e.g., R0 and R1) resections, those with a T4 tumor extent were excluded as well. This resulted in 1877 patients. Finally, sufficient histopathologic information regarding nodal status, i.e., at least one LN examined, with documented information on nodal positivity versus negativity, was required for this analysis. This extraction step yielded 1666 patients, who represent the cohort for further analyses. Among these, the median age was 66 years (range, 22–96 years), and the male:female ratio was .97. The median number of total LNs examined was 7 (range, 1–52), of positive LNs was 1 (range, 0–34), and of negative LNs was 6 (range, 0–30). The median tumor size measured 30 mm (range, 1–320 mm). Tumors >100 mm were reported in 30 patients (1.8%), and those >50 mm were reported in 294 individuals (17.6%). Most of these large tumors were unspecified adenocarcinomas or mucinous cystic adenocarcinomas. Histopathologic diagnoses included unspecified adenocarcinomas (n = 1088; 65.3%), infiltrating ductal carcinomas (n = 337; 20.2%), mucinous or mucin-producing adenocarcinomas (n = 112; 6.7%), mucinous cystadenocarcinomas (n = 32; 1.9%), and others, including squamous cell cancers (n = 97; 5.8%). Seven hundred seventy individuals (46%) had N0 disease, of which 279 were staged as T1/2N0 and 491 as T3N0. Eight hundred ninety-six individuals (54%) had N1 disease, of which 174 were staged as T1/2N1 and 722 as T3N1. Of all patients, 683 (41%) had received adjuvant radiotherapy.

Number of LNs Examined by Stage Subgroup
Frequencies of total LN counts within the surgical specimen, separated by stage group, are depicted in Fig. 2Go. For patients with N0 cancers, a median of 6 (range, 1–52) LNs had been examined. In cases of N1 disease, a median total LN count of 9 (range, 1–52) was obtained. In both N0 stage subgroups, 39% of patients had fewer than 5 LNs examined, whereas 65% (T1/2N0) or 70% (T3N0) had fewer than 10 LNs identified. In the N1 subgroups, the corresponding rates for fewer than 5 LNs were 17% (T1/2N1) and 23% (T3N1) and for fewer than 10 LNs were 54% (T1/2N1) and 55% (T3N1). No differences in mean LN counts were identified between patients with different T categories. However, patients with N1 disease had a larger mean number of LNs resected than those classified as N0, with means of 10.2 (N1) and 7.8 (N0), respectively (P < .0001; Wilcoxon/ Mann-Whitney test).


Figure 2
View larger version (28K):
[in this window]
[in a new window]
  		FIG. 2. Frequency of the categorized number of lymph nodes examined for each stage subgroup.

 
Multivariate Survival Analysis
A backward-elimination multivariate survival analysis yielded these prognostic variables: number of LNs examined, number of positive LNs, tumor size, extrapancreatic extension, radiotherapy (all P < .0001), and age (P = .0009). Risk ratios and 95% confidence intervals are listed in Table 1Go. The number of negative LNs obtained and the total number of LNs examined behaved interchangeably and maintained a similar significance level when substituted for one another. Grade, sex, ethnicity, and year of diagnosis failed to retain significance in this model.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Multivariate survival analysis
 
Univariate Survival Analysis
On the basis of a median follow-up of 14 months (range, 0–151 months: for survivors, 68 months; range, 1–151 months), overall survival was calculated for the entire patient cohort and for separate stage subgroups. Univariate comparisons, based on total LN count cutoff levels of 10, 15, and 25, are depicted in Figs. 3Go to 5GoGo. For the entire cohort and for the N0 or N1 subgroups, survival was significantly different at LN count cutoff levels of 10 or 15, always in favor of the group with more LNs retrieved (Figs. 3Go and 4Go). These differences were more pronounced for patients with N0 disease. Survival differences failed to reach statistical significance at a cutoff level of 20 or 25 LNs examined for the entire cohort and also both subgroups (Fig. 5Go). This may in part be due to small patient numbers in the group with ≥25 LNs examined. When 3-year overall survival was tabulated and compared by increasing the total LN count categories, higher LN counts were generally associated with better survival (Table 2Go). The best survival results in N0 patients were observed when >30 total LNs had been identified on pathologic examination. For N1 patients, the best survival was encountered with 20 to 29 total LNs examined. For the entire cohort, the best survival result was achieved in the 10 to 19 total LN category, and this survival outcome remained similar with higher LN counts.


Figure 3
View larger version (23K):
[in this window]
[in a new window]
  		FIG. 3. Actuarial overall survival curves, by stage group, comparing patients with fewer than 10 lymph nodes examined and those with ≥10 lymph nodes examined. (A) Entire cohort; (B) T1–3N0; (C) T1–3N1.

 

Figure 4
View larger version (22K):
[in this window]
[in a new window]
  		FIG. 4. Actuarial overall survival curves, by stage group, comparing patients with fewer than 15 lymph nodes examined and those with ≥15 lymph nodes examined. (A) Entire cohort; (B) T1–3N0; (C) T1–3N1.

 

Figure 5
View larger version (20K):
[in this window]
[in a new window]
  		FIG. 5. Actuarial overall survival curves, by stage group, comparing patients with fewer than 25 lymph nodes examined and those with ≥25 lymph nodes examined. (A) Entire cohort; (B) T1–3N0; (C) T1–3N1.

 

View this table:
[in this window]
[in a new window]
  		TABLE 2. Three-year actuarial overall survival by stage subgroup and number of total lymph nodes examined
 
Cutpoint Survival Analysis
In an attempt to identify the optimal LN count cutoff, survival comparisons were created for all stage groups at increasing total LN counts between 3 and 26. Generally, significant survival differences at each cutoff value always favored the higher LN count group, with a resulting superior survival (Table 3Go). Ranges of significant cutoff points varied between stage groups. For instance, the greatest survival difference for the entire cohort was observed at a cutoff point of 11 total LNs, but the best survival was reached at a cutoff of 16 LNs. These values were 11 and 15 in N0 patients and 14 and 19 in N1 patients, respectively. It is interesting to note that the T category had a great effect on this cutpoint analysis. Only a narrow range of significant cutpoint levels were observed for T1/2N0 lesions, and no differences were found for T1/2N1 disease comparisons. Wide ranges sof cutpoint numbers with significant survival differences were observed within T3 category subgroups. For this T category, the best survival results were encountered at cutpoints of 21 (T3N0) or 20 (T3N1) total LNs examined.


View this table:
[in this window]
[in a new window]
  		TABLE 3. Results of univariate overall survival comparison of various total lymph node count cutpoints, by stage subgroup
 
Survival Effect of Negative LN Counts
Three-year survival results, based on categories of increasing counts of negative LNs obtained, showed considerable variations, again with an obvious trend toward better survival results for higher negative LN counts (Table 4Go). The smallest survival difference between negative LN numeric categories observed was 8% (T1/2N1), and the largest was 16% (T3N0). Generally, negative LN counts of ≥20 were linked with the best survival results for the entire patient cohort and for all stage subgroups containing T3 disease. For the T1/2N0 and T1/2N1 subgroups, the best survival results were observed in the category with negative LN counts of 10 to 19. A cutpoint analysis yielded the ability to detect the greatest survival differences in the entire patient cohort for negative LN counts of between 10 and 15 (Table 5Go). Although the differences at these two cutpoints are statistically significant, gains in 3-year survival (3%) and median survival (2–3 months) were modest. Nevertheless, these gains did extend to the 5-year survival mark, where they represent a 3% (cutoff level: 10 negative LNs) or 6% (cutoff level: 15 negative LNs) survival increase.


View this table:
[in this window]
[in a new window]
  		TABLE 4. Three-year actuarial overall survival by stage subgroup and number of negative lymph nodes obtained
 

View this table:
[in this window]
[in a new window]
  		TABLE 5. Overall survival by number of negative lymph nodes obtained (all stage groups)
 
Projected Numeric LN Effect on Overall Survival
On the basis of the statistically assumed linearity as best fit, the total LN number effect on overall survival was calculated. The baseline model-predicted 5-year survival with only one LN examined was 17% (T1–3N0), 10% (T1–3N1), or 14 % (entire cohort). For every single extra LN dissected, this calculated base line overall survival improved by .6% in the T1–3N0 subgroup (P = .022). In patients with T1–3N1 disease or for the entire cohort, a significant survival advantage by increasing LN counts could not be substantiated in this model. Similarly, when 3-year survival was analyzed in this fashion, a survival advantage with increasing total LN counts was observed only for patients with T1–3N0 cancers (P = .034), but not for those with N1 disease or the entire cohort, as displayed in Fig. 6Go. The corresponding statistical implications are listed in Table 6Go. However, the number of negative LNs obtained correlated with 3-year survival in patients with T1–3N1 cancers (baseline, 11%; increase of .4% for each negative LN obtained; P = .04). As indicated by the relatively flattened slope of the curve in Fig. 7Go, negative LN numbers added only a minimal detectable survival advantage in patients with N1 disease or in the entire cohort. Although the trend is visible, the effects in two of these patient groups failed to reach statistical significance (Table 7Go).


Figure 6
View larger version (22K):
[in this window]
[in a new window]
  		FIG. 6. Plots of Kaplan-Meier overall survival at 3 years versus the total number of lymph nodes (LNs) dissected, by stage group. (A) Entire cohort; (B) T1–3N0; (C) T1–3N1. The shaded area represents the 95% confidence intervals.

 

View this table:
[in this window]
[in a new window]
  		TABLE 6. Projected numerical total lymph node effect on 3-year overall survival
 

Figure 7
View larger version (20K):
[in this window]
[in a new window]
  		FIG. 7. Plots of Kaplan-Meier overall survival at 3 years versus the number of negative lymph nodes dissected, by stage group. (A) Entire cohort; (B) T1–3N0; (C) T1–3N1. The shaded area represents the 95% confidence intervals.

 

View this table:
[in this window]
[in a new window]
  		TABLE 7. Projected numerical negative lymph node effect on 3-year overall survival
 
Early Postoperative Deaths Based on LN Numbers
Among the entire cohort, there were 128 deaths (8%) within 1 month of diagnosis and 208 deaths (12%) within 3 months of diagnosis. T3 patients were more likely to die within 1 or 3 months than T1/2 patients (P = .0364 and P = .0025, respectively). There were no statistically significant differences in 1-or 3-month mortality rates when we compared N0 and N1 patients. We analyzed early mortality by the number of LNs resected, choosing total LN categories as used in Figs. 6Go and 7Go. For T1/2 patients, no differences between LN count categories were identified at 1 month (P = .1244) or at 3 months (P = .4740). For T3 patients, similar analyses yielded P = .2488 at 1 month and P = .0100 at 3 months. Mortality at 3 months in the T3 group was skewed toward those with fewer LNs resected. For those T3 patients who had between one and four LNs resected, 18% (64 of 359) were dead at 3 months. Three-month mortality rates for T3 patients in the other LN count categories were as follows: 5 to 9 LNs, 47 (12%) of 383; 10 to 19 LNs, 54 (15%) of 364; 20 to 29 LNs, 2 (2%) of 83; and ≥30 LNs, 2 (8%) of 24.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Pancreatic cancer remains the most lethal of all human malignancies, with a 5-year survival rate of merely 4%.19 Although complete resection of localized disease is still the sole curative modality to date, most postresection patients experience treatment failure due to locoregional disease recurrence or distant relapse. Attempts to improve outcomes through wider regional resections have been limited to various reports of predominantly single-institution series, in all of which ELND at the time of pancreatic resection failed to translate into an obvious survival advantage.9,10,20–23 Improved survival after ELND has been reported in some subset analyses of these studies, either for node-negative disease22 or for node-positive patients.9 In addition, a trend toward superior survival after ELND has been observed in the first 2 years after pancreatectomy, but not for the long-term.23 One would conclude from these efforts that if there is a survival benefit to ELND, it apparently is small, and sufficient power to detect such a benefit cannot be assumed in single-institutional trials. Indeed, a review of a large multi-institutional Japanese database with 1001 patients showed a survival benefit to ELND for patients with N1 disease; in addition, survival comparable to that for N1 patients was accomplished for N2 disease (by Japanese staging criteria) through an ELND approach.24 Although an analysis of SEER data on the surgical outcomes of 396 patients with pancreatic cancer was performed earlier, that study did not investigate the numerical LN status and its effect on survival.25 To date, only one single-institution review attempted to examine the effect of LN counts on postoperative survival, but it found no correlation; the metastatic LN ratio was found to be predictive instead, but a ratio of 0 was part of this analysis.26 The results of our analysis of SEER data now show that stage-based postoperative survival prediction of pancreatic cancer is significantly influenced by total LN counts and the number of negative LNs obtained. The effect is most obvious in patients with N0 disease, but it remains detectable in node-positive stage groups as well. It seemed to prevail even in a multimodality therapy setting where a substantial number of patients had undergone adjuvant radiotherapy. Although the resulting incremental survival benefit is small, the potential mechanisms deserve attention by surgeons dedicated to pancreatic cancer management.

The stronger effect of LN counts on survival in the N0 stage groups than in nodal positivity suggests that stage migration is a significant mechanism here. N0 stage assignment correlated with fewer LNs examined, and the likelihood of having missed involved LNs is higher with fewer nodes analyzed. The data do not enable us to separate between stage migration and the therapeutic survival effect of numerical LN evaluation, but a certain therapeutic effect can be assumed due to the observed survival extension in patients with N1 disease. In contrast to gastric cancer, node-positive patients with pancreatic cancer cannot be assigned to any higher-stage subgroup because of more extensive LN retrieval for lack of higher N categories or any M1 assignment should numerous or distant involved nodes be identified. Traditionally, positive LNs in gastrointestinal cancers, especially pancreatic cancer, are primarily thought to carry prognostic importance, but not therapeutic relevance.27 For instance, node-positive patients may have a 5-year survival of 8% after pancreatoduodenectomy, compared with 40% for those who are node negative.28 However, a curative role to ELND after R0 resection of primary tumors could be proposed if any obvious or occult metastatic cancer is truly limited to those LNs completely removed through an ELND. In this context, more negative LNs would indicate a lesser risk for residual regional disease. Could this possibly be the case in pancreatic cancer?

In node-positive adenocarcinomas of the pancreatic head, posterior pancreaticoduodenal LNs seem to be the most common site for nodal involvement (approximately 50%), followed by superior mesenteric artery LNs (38%) and para-aortic LNs (18%).29 Positive perigastric LNs have been found in only 14% of specimens.30 Para-aortic LN involvement has rarely been observed if peripancreatic LNs are negative31; in addition, para-aortic LN involvement is rarely diffuse and is most commonly detected up to the dorsal aspect of the renal arteries.32 In contrast, a more diffuse pattern of lymphatic metastasis and neural plexus involvement has been observed for cancers arising in the pancreatic tail.33 No patient with proven para-aortic LN involvement has survived beyond 5 years.34 On the basis of these findings, the potential to benefit from ELND can be assumed for a defined, albeit small, patient subset with cancers at risk for more limited lymphatic spread, sparing para-aortic or distant sites, and carrying no hematogenous progression risk. Because attempts to optimize loco-regional therapy through radical resection and intraoperative radiation lead to a predominantly distant relapse pattern,35 this subset has to be limited in size.

We have to acknowledge several shortcomings in our analysis. SEER data have obvious limitations regarding necessary details of operative therapy or incomplete information, and, to allow for this analysis, individuals had to be highly selected. The margin status of surgical specimens has not been collected appropriately. In this context, the number of LNs examined may reflect not only the extent of regional dissection by the surgeon, but also the effort and diligence of the pathologic analysis. We have discussed these implications in detail before:15 is the superior survival with greater LN counts a reflection of higher standards of clinical practice, high-volume clinical settings, or a more rigorous selection of healthier patients? The significantly lower 3-month mortality after pancreatectomy with ≥20 LNs examined would support such an assumption. Is the LN number possibly a surrogate for other clinical parameters, such as weight loss or pain, which in a nomogram based on clinicopathologic characteristics carry a greater prognostic impact than LN counts?36 Although these questions remain unanswerable, the primary value of our SEER data is that they reflect widespread, national patterns of care better than series from single high-volume institutions. Median LN counts in this series (6 for LN-negative and 9 for LN-positive patients) fall short of those obtained in the 2 randomized trials even after a limited dissection extent—namely, 139 or 16.10 Although ELND with an increase in LN counts to 20 or 26, respectively, did not lead to any obvious survival advantage in these trials, the increase from 7 (SEER median) to 16 (Johns Hopkins, "limited dissection" group median) may well carry this advantage, and this may reflect the difference in the standard of care linked to increased LN counts leading to improved outcomes. In this sense, our results cannot call for performing "extended lymph node dissections" per se, but they do urge surgeons to consider obtaining a certain optimal number of LNs through diligent dissection and examination.

We conclude that our results obtained from a US population data source demonstrate that removal and pathologic examination of increased LN numbers can influence staging quality and overall survival after pancreatectomy for cancer. We do not anticipate major survival effects due to ELND in pancreatic cancer, nor do we expect that any need for additional therapies, especially radiation to regional sites, can be obviated through ELND alone. However, according to our results, the attempt to resect and examine enough LNs to yield at least 15 total LNs, or approximately 10 negative LNs for curative-intent pancreatectomy, seems advisable to optimize operative benefits, as long as this is not associated with increased morbidity. As multimodal therapy concepts indicate the future direction of surgical care for pancreatic cancer, potentially beneficial implications of multinodal resections should be kept in mind.

Received for publication October 19, 2005. Accepted for publication March 11, 2006.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 PATIENTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Sperti C, Pasquali C, Piccoli A, Pedrazzoli S. Recurrence after resection for ductal adenocarcinoma of the pancreas. World J Surg 1997; 21:195–200.[CrossRef][Medline]
  2. Pisters PW, Wolff RA, Crane CH, Evans DB. Combined-modality treatment for operable pancreatic adenocarcinoma. Oncology (Williston Park) 2005; 19:393–404; 409; discussion 409–10, 412–6.[Medline]
  3. Stocken DD, Buchler MW, Dervenis C, et al. Meta-analysis of randomised adjuvant therapy trials for pancreatic cancer. Br J Cancer 2005; 92:1372–81.[CrossRef][Medline]
  4. Harrison LE, Klimstra DS, Brennan MF. Isolated portal vein involvement in pancreatic adenocarcinoma. A contraindication for resection? Ann Surg 1996; 224:342–7; discussion 347–9.[CrossRef][Medline]
  5. Tseng JF, Raut CP, Lee JE, et al. Pancreaticoduodenectomy with vascular resection: margin status and survival duration. J Gastrointest Surg 2004; 8:935–49; discussion 949–50.[CrossRef][Medline]
  6. Fortner JG, Klimstra DS, Senie RT, Maclean BJ. Tumor size is the primary prognosticator for pancreatic cancer after regional pancreatectomy. Ann Surg 1996; 223:147–53.[CrossRef][Medline]
  7. Kodera Y, Schwarz RE, Nakao A. Extended lymph node dissection in gastric carcinoma: where do we stand after the Dutch and British randomized trials? J Am Coll Surg 2002; 195:855–64.[CrossRef][Medline]
  8. Havenga K, Enker WE, Norstein J, et al. Improved survival and local control after total mesorectal excision or D3 lymphadenectomy in the treatment of primary rectal cancer: an international analysis of 1411 patients. Eur J Surg Oncol 1999; 25:368–74.[CrossRef][Medline]
  9. Pedrazzoli S, DiCarlo V, Dionigi R, et al. Standard versus extended lymphadenectomy associated with pancreatoduo-denectomy in the surgical treatment of adenocarcinoma of the head of the pancreas: a multicenter, prospective, randomized study. Lymphadenectomy Study Group. Ann Surg 1998; 228:508–17.[CrossRef][Medline]
  10. Yeo CJ, Cameron JL, Lillemoe KD, et al. Pancreaticoduo-denectomy with or without distal gastrectomy and extended retroperitoneal lymphadenectomy for periampullary adeno-carcinoma, part 2: randomized controlled trial evaluating survival, morbidity, and mortality. Ann Surg 2002; 236:355–66; discussion 366–8.[CrossRef][Medline]
  11. Nguyen TC, Sohn TA, Cameron JL, et al. Standard vs. radical pancreaticoduodenectomy for periampullary adenocarcinoma: a prospective, randomized trial evaluating quality of life in pancreaticoduodenectomy survivors. J Gastrointest Surg 2003; 7:1–9; discussion 11.[CrossRef][Medline]
  12. Pawlik TM, Abdalla EK, Barnett CC, et al. Feasibility of a randomized trial of extended lymphadenectomy for pancreatic cancer. Arch Surg 2005; 140:584–9; discussion 589–91.[Abstract/Free Full Text]
  13. Surveillance, Epidemiology, and End Results (SEER) Program: public use data (1973–2002). 2005. Available at: http://www.seer.cancer.gov/publicdata/access.html. Accessed April 8, 2005.
  14. Greene FL, Page DL, Fleming ID, et al. (2002) American Joint Committee on Cancer Staging Manual6 Springer Verlag, New York, Greene FL, Page DL, Fleming ID, et al. American Joint Committee on Cancer Staging Manual. 6th ed. New York: Springer Verlag, 2002.
  15. Smith DD, Schwarz RR, Schwarz RE. Impact of total lymph node counts on staging and survival after gastrectomy for gastric cancer: data from a large U.S. population database. J Clin Oncol 2005; 23:7114–24.[Abstract/Free Full Text]
  16. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. Am Stat Assoc J 1958; 53:457–81.[CrossRef]
  17. Peto R, Pike MC. Conservatism in the approximation E(0–E)2/E in the log rank test for survival data or tumor incidence data. Biometrics 1973; 29:579–84.[CrossRef][Medline]
  18. Cox DR. Regression models and life tables. J R Stat Assoc B 1972; 29:187–220.
  19. Jemal A, Murray T, Ward E, et al. Cancer statistics, 2005. CA Cancer J Clin 2005; 55:10–30.[Abstract/Free Full Text]
  20. Henne-Bruns D, Vogel I, Luttges J, Kloppel G, Kremer B. Surgery for ductal adenocarcinoma of the pancreatic head: staging, complications, and survival after regional versus extended lymphadenectomy. World J Surg 2000; 24:595–601; discussion 602.[CrossRef][Medline]
  21. Gazzaniga GM, Cappato S, Papadia F, Mori L, Filauro M. D1 versus D2 pancreatoduodenectomy in surgical therapy of pancreatic head cancer. Hepatogastroenterology 2001; 48:1471–8.[Medline]
  22. Popiela T, Kedra B, Sierzega M. Does extended lymphadenectomy improve survival of pancreatic cancer patients? Acta Chir Belg 2002; 102:78–82.[Medline]
  23. Capussotti L, Massucco P, Ribero D, Vigano L, Muratore A, Calgaro M. Extended lymphadenectomy and vein resection for pancreatic head cancer: outcomes and implications for therapy. Arch Surg 2003; 138:1316–22.[Abstract/Free Full Text]
  24. Hirata K, Sato T, Mukaiya M, et al. Results of 1001 pancreatic resections for invasive ductal adenocarcinoma of the pancreas. Arch Surg 1997; 132:771–6; discussion 777.[Abstract/Free Full Text]
  25. Lim JE, Chien MW, Earle CC. Prognostic factors following curative resection for pancreatic adenocarcinoma: a population-based, linked database analysis of 396 patients. Ann Surg 2003; 237:74–85.[CrossRef][Medline]
  26. Berger AC, Watson JC, Ross EA, Hoffman JP. The metastatic/ examined lymph node ratio is an important prognostic factor after pancreaticoduodenectomy for pancreatic adenocarcinoma. Am Surg 2004; 70:235–40; discussion 240.[Medline]
  27. Gervasoni JE Jr, Taneja C, Chung MA, Cady B. Biologic and clinical significance of lymphadenectomy. Surg Clin North Am 2000; 80:1631–73.[CrossRef][Medline]
  28. Benassai G, Mastrorilli M, Mosella F, Mosella G. Significance of lymph node metastases in the surgical management of pancreatic head carcinoma. J Exp Clin Cancer Res 1999; 18:23–8.[Medline]
  29. Kayahara M, Nagakawa T, Ueno K, Ohta T, Tsukioka Y, Miyazaki I. Surgical strategy for carcinoma of the pancreas head area based on clinicopathologic analysis of nodal involvement and plexus invasion. Surgery 1995; 117:616–23.[CrossRef][Medline]
  30. Nakao A, Harada A, Nonami T, et al. Lymph node metastases in carcinoma of the head of the pancreas region. Br J Surg 1995; 82:399–402.[Medline]
  31. Sakai M, Nakao A, Kaneko T, et al. Para-aortic lymph node metastasis in carcinoma of the head of the pancreas. Surgery 2005; 137:606–11.[CrossRef][Medline]
  32. Nagakawa T, Kobayashi H, Ueno K, Ohta T, Kayahara M, Miyazaki I. Clinical study of lymphatic flow to the paraaortic lymph nodes in carcinoma of the head of the pancreas. Cancer 1994; 73:1155–62.[CrossRef][Medline]
  33. Kayahara M, Nagakawa T, Futagami F, Kitagawa H, Ohta T, Miyazaki I. Lymphatic flow and neural plexus invasion associated with carcinoma of the body and tail of the pancreas. Cancer 1996; 78:2485–91.[CrossRef][Medline]
  34. Yoshida T, Matsumoto T, Sasaki A, Shibata K, Aramaki M, Kitano S. Outcome of paraaortic node-positive pancreatic head and bile duct adenocarcinoma. Am J Surg 2004; 187:736–40.[CrossRef][Medline]
  35. Schwarz RE, Smith DD, Keny H, et al. Impact of intraoperative radiation on postoperative and disease-specific outcome after pancreatoduodenectomy for adenocarcinoma: a propensity score analysis. Am J Clin Oncol 2003; 26:16–21.[CrossRef][Medline]
  36. Brennan MF, Kattan MW, Klimstra D, Conlon K. Prognostic nomogram for patients undergoing resection for adenocarcinoma of the pancreas. Ann Surg 2004;240:293–8.[Medline]



This article has been cited by other articles:


Home page
 Proc. Natl. Acad. Sci. USAHome page
From the Cover: Nanoparticle-mediated drug delivery to tumor vasculature suppresses metastasis
PNAS, July 8, 2008; 105(27): 9343 - 9348.


Home page
 Arch SurgHome page
K. Y. Bilimoria, M. S. Talamonti, J. D. Wayne, J. S. Tomlinson, A. K. Stewart, D. P. Winchester, C. Y. Ko, and D. J. Bentrem
Effect of Hospital Type and Volume on Lymph Node Evaluation for Gastric and Pancreatic Cancer
Arch Surg, July 1, 2008; 143(7): 671 - 678.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
K. Y. Bilimoria, C. M. Balch, D. J. Bentrem, M. S. Talamonti, C. Y. Ko, J. R. Lange, D. P. Winchester, and J. D. Wayne
Complete Lymph Node Dissection for Sentinel Node-Positive Melanoma: Assessment of Practice Patterns in the United States
Ann. Surg. Oncol., June 1, 2008; 15(6): 1566 - 1576.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
M. B. Slidell, D. C. Chang, J. L. Cameron, C. Wolfgang, J. M. Herman, R. D. Schulick, M. A. Choti, and T. M. Pawlik
Impact of Total Lymph Node Count and Lymph Node Ratio on Staging and Survival after Pancreatectomy for Pancreatic Adenocarcinoma: A Large, Population-Based Analysis
Ann. Surg. Oncol., January 1, 2008; 15(1): 165 - 174.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
A. Govindarajan, J. C. C. Tan, N. N. Baxter, N. G. Coburn, and C. H. L. Law
Variations in Surgical Treatment and Outcomes of Patients With Pancreatic Cancer: A Population-Based Study
Ann. Surg. Oncol., January 1, 2008; 15(1): 175 - 185.
[Abstract] [Full Text] [PDF]

Home page
 JNCI J Natl Cancer InstHome page
R. Ricciardi and N. N. Baxter
Association Versus Causation Versus Quality Improvement: Setting Benchmarks for Lymph Node Evaluation in Colon Cancer
J Natl Cancer Inst, March 21, 2007; 99(6): 414 - 415.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Schwarz, R. E.
Right arrow Articles by Smith, D. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Schwarz, R. E.
Right arrow Articles by Smith, D. D.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS