| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
1 Department of Surgery, Beth Israel Deaconess Medical Center, Harvard Medical School, Stoneman 9, 330 Brookline Avenue, Boston, Massachusetts 02215, USA
2 Department of Surgery, University of Toronto, Toronto, Ontario, Canada
3 Department of Medical Imaging, University Health Network and Mount Sinai Hospitals, University of Toronto, Toronto, Ontario, Canada
4 Department of Pathology, University of Toronto, Toronto, Ontario, Canada
5 Department of Surgery, University of Calgary, Calgary, Alberta, Canada
Correspondence: Address correspondence and reprint requests to: Charles M. Vollmer Jr., MD; E-mail: cvollmer{at}bidmc.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: Ninety-four consecutive patients underwent attempted curative resection of periampullary tumors between September 2001 and June 2003. A single radiologist recorded in a retrospective, blinded fashion the short- and long-axis measurements of the CBD and GDA nodes.
Results: Sixty-one percent (n = 57) of tumors were resectable by pancreaticoduodenectomy. Overall, actual 6-, 12-, and 18-month survival was 87%, 68%, and 63%, respectively. Enlarged radiographical nodal size by either axis was not associated with the presence of metastasis to these lymph nodes or with reduced overall patient survival. Only a CBD node short-axis size >10 mm predicted unresectability (odds ratio, 3.2; P = .036). Liver metastasis and/or carcinomatosis were present in 43% of unresectable patients, and this was associated with decreased survival at both 1 year (25% vs. 77%; P < .001) and 18 months (19% vs. 72%; P <.001). A pathologic diagnosis of metastasis to the GDA node, but not the CBD node, was associated with a similarly decreased survival (1 year: 33% vs. 78%, P = .028; 18 months: 22% vs. 70%, P = .023).
Conclusions: For presumed periampullary malignancy, a CBD node short-axis size >10 mm predicts tumor unresectability. Metastatic disease to the GDA node, particularly for pancreatic adenocarcinoma, portends a poor prognosis equivalent to that of hepatic or peritoneal spread. Given these findings, radiographical CBD lymph node measurements may guide selection for performing laparoscopic staging with or without ultrasonography in conjunction with GDA nodal biopsy in patients with periampullary malignancy.
Key Words: Periampullary tumors • Staging laparoscopy • Lymph nodes • Resectability
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Unfortunately, most patients initially present with unresectable disease, which has traditionally been described as lymph node metastases beyond the resection margins, vascular invasion that precludes resection, and metastatic spread to the peritoneal cavity, liver, or both. Furthermore, of those patients considered to be resectable after a state-of-the-art preoperative radiographical and clinical assessment, approximately 15% to 40% will still undergo a non-curative laparotomy as a result of these findings.
Multiple imaging modalities, including multiphasic enhanced computed tomography (CT), transabdominal ultrasonography, magnetic resonance imaging, and endoscopic ultrasonography (EUS), are currently used to detect unresectable disease before surgery, to minimize the number of patients who undergo non-curative laparotomy. Although some studies have reported EUS to be more accurate in the local assessment of periampullary tumors,4,5 specifically for tumor size and lymph node involvement, the noninvasive nature of CT offers a distinct advantage. Furthermore, the enhanced resolution of thinly sliced CT scan images has improved its diagnostic and predictive application by better identifying regional and lymphatic spread, vascular invasion, and liver metastases.6–9
Staging laparoscopy is another tool that enhances the diagnosis of advanced locoregional disease and minimizes the incidence of noncurative laparotomy for periampullary tumors.10,11 However, its value in these patients is not universally accepted. Opinions range from recommending its routine use for all patients before laparotomy to not performing laparoscopy in any circumstance.12,13 The true benefit of staging laparoscopy is most likely dependent on careful patient selection.14
The current criterion for suspecting neoplastic involvement of lymph nodes by CT assessment is a short-axis diameter (anterior-posterior dimension) >10 mm, although this finding should not preclude attempted curative resection in a patient who otherwise seems to have resectable disease.15 For periampullary tumors, the prognostic value of using this short-axis radiographical criterion for lymph node evaluation is unclear, and no standard currently exists for long-axis measurements (lateral dimension). Two distinct lymph nodes reproducibly assessed by CT in the evaluation of periampullary tumors are (1) the common bile duct (CBD) node, located superoposterolateral to and behind the insertion of the CBD into the pancreatic head, and (2) the gastroduodenal artery (GDA) node, located anteromedial to the origin of the GDA from the hepatic artery.
Our aim was to determine whether there was added value in a focused and detailed CT analysis of these two regional lymph nodes that might improve our ability to predict nodal metastasis, tumor resectability, and overall survival for patients with periampullary tumors.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
All CT scans were performed by using four or eight multidetector scanners (LightSpeed QX/i and Light-Speed Ultra; GE Medical Systems, Waukesha, WI). Images were obtained at pancreatic parenchymal and venous phases at 50 and 70 seconds after injection of intravenous contrast material, respectively. Images were reconstructed at a slice thickness of 2.5 mm every 1.25 mm for the parenchymal and 5 mm every 2.5 mm for the venous phases. All images were reviewed on softcopy. A single radiologist (K.K.) recorded, in a retrospective, blinded fashion, short-and long-axis measurements of the CBD and GDA nodes for all 94 patients. The anatomical location of these specific lymph nodes is demonstrated in Fig. 1
. Short-axis measurements were categorized into two groups by using the standard criterion of 10 mm as the cutoff value (
10 mm or >10 mm). Because there is no defined standard for predicting malignant involvement by long- axis diameter, the mean value of the long-axis measurements was chosen as the cutoff value to categorize them into two dichotomous groups.
|
Univariate logistic regression analysis was performed to determine which lymph node dimensions (as measured on CT) were predictors of tumor resectability, nodal metastasis, and overall long-term survival. Lymph node measurements were analyzed both as continuous variables and after being divided into two categories based on the above-stated criteria. Statistically significant variables (P < .05) on univariate analysis were subsequently added progressively to develop a multivariate logistic regression model that reliably predicted any of the three main outcomes. Model performance was assessed with formal goodness-of-fit (Hosmer-Lemeshow) tests, in which a statistically nonsignificant goodness-of-fit statistic implies that the model has outcome-predictive value.16 Lymph node measurements and clinical outcomes were also analyzed by using 
2 testing to assess for significant associations. P < .05 was defined as significant. Data were analyzed by using SPSS 12.0 for Windows (SPSS, Inc., Chicago, IL).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
details the spectrum of the final pathologic diagnoses.
|
Univariate regression analysis showed that enlarged CBD node short-axis measurements, both as a continuous and categorical variable (divided at 10 mm), and GDA node long-axis measurements as a continuous variable predicted tumor unresectability (CBD short-axis continuous variable: odds ratio [OR], 3.74; 95% confidence interval [CI], 1.06–13.13; P = .04; CBD short-axis categorical variable: OR, 3.86; 95% CI, 1.37–10.99; P = .011; GDA node long-axis continuous variable: OR, 2.82; 95% CI, 1.09–7.25; P = .032). None of the other lymph node measurements had any significant association with tumor resectability.
However, when the above three variables were used for multivariate regression analysis, only an enlarged CBD node short-axis measurement >10 mm (categorical variable) predicted tumor unresectability (OR, 3.16; 95% CI, 1.08–9.26; P = .036). Similarly, 2 analysis showed that only a CBD node short-axis measurement >10 mm was significantly associated with tumor unresectability (P = .017). These findings are reviewed in Table 2. Thirty-five percent (13 of 37) of patients with unresectable disease had a CBD node short-axis measurement >10 mm, whereas only 12% (7 of 57) of patients with resectable disease met this radiographical criterion. Of these 13 unresectable patients with a CBD node >10 mm, 12 patients (92%) had unresectable disease secondary to vascular invasion (further detailed in Table 3). Furthermore, there was no significant relationship between CBD node short-axis size and the presence of a biliary stent.
|
|
Overall Survival
Using the Canadian Registrar General’s database, accurate survival data for all patients were obtained. Overall 6-month, 1-year, and 18-month actual survival for all 94 patients was 87%, 68%, and 63%, respectively. As expected, patients with unresectable tumors had reduced survival compared with resected patients (6 months, 73% vs. 97%; 1 year, 46% vs.83%; 18 months, 41% vs. 77%; P < .003 for all three). Enlarged radiographical size of either the CBD or GDA lymph node in either axis did not predict mortality at any of the three time points.
As would be expected, the presence of liver metastasis and/or carcinomatosis was significantly associated with reduced survival (compared with all other patients without these findings) at each time point (6 months, 63% vs. 88%; 1 year, 25% vs. 77%; 18 months, 19% vs. 72%; P < .004 for all three). A pathologic diagnosis of tumor metastasis to the GDA node was also associated with a similarly decreased survival at 1 year and 18 months after surgery (1 year, 33% vs. 78%; 18 months, 22% vs. 70%; P < .03 for both), thus mirroring the survival rates for those patients with carcinomatosis, liver metastasis, or both. Conversely, this finding did not hold true for metastasis to the CBD node (1 year, 60% vs. 81%; 18 months, 60% vs. 73%; not significant for both).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Numerous studies have reported on the efficacy of the various available imaging modalities to identify the extent of tumor burden before surgery in an effort to minimize the incidence of nontherapeutic laparotomies in these patients. Helical CT is a valuable staging tool for multiple reasons, including its widespread availability and its ability to accurately discriminate resectable disease. Unlike EUS, it is also noninvasive, operator independent, and more universally available. In a prospective comparison of helical CT, magnetic resonance imaging, angiography, and EUS, Soriano et al.17 reported the superiority of helical CT in accurately assessing the extent of the primary tumor (73%), locoregional extension (74%), vascular invasion (83%), distant metastases (88%), and tumor resectability (83%). Helical CT is also superior in demonstrating involvement of peripancreatic vessels and vascular invasion when compared with CT angiography, because the latter detected only 86% of the cases of vascular invasion detected by helical CT.9 Legmann et al.18 reported identical overall accuracies for both dual-phase helical CT and EUS in staging and predicting resectability of 93% and 90%, respectively, for 30 patients with pancreatic tumors. Similarly, Howard et al.19 report a sensitivity of 63%, a specificity of 77%, and an overall accuracy of 86% for helical CT in determining resectability for 21 patients with periampullary tumors. Even with such a powerful preoperative staging tool, the incidence of noncurative laparotomy for periampullary tumors remains unacceptably high. This is underscored by another study that evaluated 76 patients with suspected pancreatic cancer and found that the main limitation of helical CT for accurately predicting resectability was its inability to reveal small hepatic metastases.20
A relative weakness of CT lies in its poor ability to detect malignancy in regional lymph nodes.21 Although not rigorously tested, standard radiographical criteria for defining a lymph node to be suspicious for harboring malignancy is a short-axis measurement >10 mm.15 We sought to determine whether strict application of this criterion to two lymph nodes reproducibly assessed by CT for periampullary tumors would further enhance our ability to detect unresectable tumors among patients otherwise thought to have resectable disease. Furthermore, we assessed whether enlarged radiographical nodal measurements had any predictive value regarding the presence of metastatic tumor in these lymph nodes and/or a patient’s long-term survival.
We found that an enlarged CBD node short-axis measurement >10 mm was significantly associated with unresectability, because multivariate regression analysis showed that this radiographical finding conferred a 3.2 times increased risk of the primary tumor being unresectable. Neither the CBD node long-axis measurement nor GDA node measurements in either axis had any predictive value for tumor resectability. Using routine histopathologic techniques to assess for lymph node malignancy, we found no significant associations between any of the CT lymph node measurements and the presence of nodal malignant disease. This finding may be due to the fact that, although all patients underwent pre-operative CT evaluation, most (77%) lymph nodes retrieved for histopathologic analysis were obtained from patients with resectable tumors. Similarly, there was no association between lymph node measurements and long-term survival. Although a CBD node short-axis measurement >10 mm was associated with unresectability, which in turn was associated with reduced survival, we found no direct association between CBD node measurements and survival. This apparent discrepancy is likely a function of the number of patients included in our study, or perhaps a longer follow-up period would reveal such an association. Furthermore, the universal poor prognosis of pancreatic adenocarcinoma, a diagnosis that comprised half of our patient population, likely also contributed to our inability to detect an association between CBD node measurements and long-term survival.
A previous study by Schwarz et al.22 reported that biliary stenting reduced the accuracy of CT in diagnosing periampullary malignancy from 88% to 73%, thus suggesting that the presence of a stent may interfere with a complete radiographical assessment. However, in our current study, we did not find any association between an enlarged CBD lymph node short-axis measurement and the presence of a biliary stent. Thus, it does not seem that any reactive inflammation that may have been present secondary to the placement of a biliary stent significantly influenced the CBD lymph node size as measured by CT. One limitation of this finding is that the time interval between stent placement and CT acquisition is not known; thus, our negative finding may be a function of a critical time interval that allowed for resolution of inflammation. Nevertheless, and more importantly, it seems that the value of CBD lymph node size in predicting tumor resectability is independent of biliary stent placement.
Laparoscopy in patients with periampullary tumors is an effective staging adjunct, yet its value continues to be debated. Laparoscopy is an excellent tool for visualizing liver metastases and carcinomatosis; its main limitation is determining critical vascular invasion and lymph node involvement.23 In a large series from Johns Hopkins Hospital, 50% of unresectable periampullary tumors were secondary to metastatic disease.24 This evidence of metastasis would presumably be visualized during laparoscopy, thus preventing a noncurative laparotomy.10,11 Reddy et al.12 reported that 29 (29%) of their 99 patients avoided a noncurative laparotomy after laparoscopic identification of metastases, thus leading them to advocate performing staging laparoscopy in all cases. Conversely, Friess et al.13 found a benefit in only 10% of their patients with periampullary tumors. Routine staging laparoscopy is impractical and cost-inefficient and therefore cannot be recommended for all patients with periampullary tumors. Instead, as advocated by the groups from Washington University and Memorial Sloan-Kettering Cancer Center, the true benefit of staging laparoscopy is likely a function of careful patient selection.14,25
In our study, it is important to note that all of the patients were preoperatively thought to have resectable disease by traditional CT criteria and that staging laparoscopy was not performed. Accordingly, 39% (n = 37) of these patients were found to have unresectable tumors, of which 43% (n = 16) had visible evidence of liver metastases and/or peritoneal implants on operative exploration. In-depth investigation of radiographical lymph node appearance revealed that a CBD lymph node short-axis measurement >10 mm predicts unresectability more than one third of the time. Thirty-five percent of patients with unresectable disease, but only 12% of resectable patients, met this radiographical criterion. If 43% of unresectable patients have visible evidence of meta-static disease on initial exploration, then it is possible that using this radiographical finding of a CBD node short-axis measurement >10 mm as a selection criterion for performing staging laparoscopy in patients with periampullary tumors (otherwise thought to have resectable tumors) may have prevented 16% of all noncurative laparotomies. This percentage breakdown is graphically depicted in Fig. 2. Of the 20 patients in our study who would have been selected for staging laparoscopy by using this CT radiographical criterion, 30% would have likely had their operative course altered or abandoned before laparotomy, on the basis of findings during laparoscopy (Fig. 2). Furthermore, this strategy of using CT criteria to guide selective laparoscopy will contain costs by minimizing valuable resources spent toward performing unnecessary staging procedures. In this study, an overwhelming majority (50 of 57; 88%) of patients with resectable disease would not have been selected to undergo staging laparoscopy on the basis of their CBD node short-axis measurements.
|
). Thus, it is probable that LUS would have further augmented the detection of unresectable disease before full laparotomy over that conferred by performing simple staging laparoscopy. This detection of inoperable disease by LUS in radiographically selected patients will prevent noncurative laparotomies while also minimizing the number of negative LUS procedures performed. Lymph node metastasis portends a poor prognosis for patients with periampullary tumors. Previous reports indicate that patients who underwent tumor resection but were found to have lymph node metastases in the resected specimen have significantly decreased long-term survival compared with patients without nodal disease.26–28 In a series of 123 patients with pancreatic adenocarcinoma from our institution (University of Toronto), only 4 of the 18 patients who survived 5 years had evidence of lymph node metastasis at the time of resection.3 In particular for pancreatic and bile duct adenocarcinoma, involvement of para-aortic lymph nodes (outside the common margins of resection) seems to be a key poor prognostic factor, and an intraoperative, histologically confirmed biopsy positive for tumor of these lymph nodes should arguably terminate the procedure.29,30 This is underscored by recent data indicating no survival advantage in performing an extended lymphadenectomy with pancreaticoduodenectomy for periampullary adenocarcinoma.31
Eighteen percent of the GDA lymph nodes excised in this study were positive for tumor on routine histopathologic analysis. This finding was associated with a decreased survival similar to that with the presence of liver metastasis and/or carcinomatosis (1 year, 33% vs. 25%; 18 months, 22% vs. 19%) and mirrors the general frequency of locoregional tumor involvement. Furthermore, in a recent report from Bogoevski et al.,32 the presence of microinvolvement of pancreatic adenocarcinoma in the superior-anterior nodal compartment (corresponding to where the GDA node is located) confers an independent poor prognostic effect on overall survival beyond that seen from involved nodes in other drainage basins. Thus, intraoperative analysis of this reproducible GDA node (either with staging laparoscopy or during open exploration) should be considered in patients, particularly those with either pancreatic or bile duct adenocarcinoma, who initially meet our radiographical criteria to undergo laparoscopic evaluation but have negative findings during staging laparoscopy and LUS. Although we did not find an association between radiographical GDA node enlargement and the presence of metastasis using routine histopathology, it is possible that further evaluation with more refined histological techniques similar to those used by Bogoevski et al. (immunochemical analysis with antiepithelial monoclonal antibodies) would yield such an association. Nevertheless, a histopathologic positive finding of tumor involvement for pancreatic and/or bile duct primary tumors in this specific GDA lymph node should arguably terminate the procedure, given the dismal survival demonstrated with such involvement.
Given these findings, we propose that one should consider performing staging laparoscopy in those patients with periampullary tumors who have a CBD lymph node that measures >10 mm in short-axis diameter. At the time of laparoscopy, one can also use laparoscopic ultrasonography to aid in the diagnosis of local vascular invasion and/or lymph node disease that would preclude curative resection. For adenocarcinoma of the pancreatic head and distal bile duct in particular, histological evidence of metastasis to the GDA lymph node, which is located in the celiac lymph node basin, portends a dismal prognosis that may negate the benefits of resection.
|
FOOTNOTES |
|---|
Received for publication October 26, 2005. Accepted for publication April 5, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. Falconi, S. Crippa, I. Dominguez, G. Barugola, P. Capelli, S. Marcucci, S. Beghelli, A. Scarpa, C. Bassi, and P. Pederzoli Prognostic Relevance of Lymph Node Ratio and Number of Resected Nodes after Curative Resection of Ampulla of Vater Carcinoma Ann. Surg. Oncol., November 1, 2008; 15(11): 3178 - 3186. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. A. Zamboni, J. B. Kruskal, C. M. Vollmer, J. Baptista, M. P. Callery, and V. D. Raptopoulos Pancreatic Adenocarcinoma: Value of Multidetector CT Angiography in Preoperative Evaluation Radiology, December 1, 2007; 245(3): 770 - 778. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. D. D. Brennan, G. A. Zamboni, V. D. Raptopoulos, and J. B. Kruskal Comprehensive Preoperative Assessment of Pancreatic Adenocarcinoma with 64-Section Volumetric CT RadioGraphics, November 1, 2007; 27(6): 1653 - 1666. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Cordera, C. A. Arciero, T. Li, J. C. Watson, and J. P. Hoffman Significance of Common Hepatic Artery Lymph Node Metastases During Pancreaticoduodenectomy for Pancreatic Head Adenocarcinoma Ann. Surg. Oncol., August 1, 2007; 14(8): 2330 - 2336. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. D. Brennan, G. Zamboni, J. Sosna, M. P. Callery, C. M.V. Vollmer, V. D. Raptopoulos, and J. B. Kruskal Virtual Whipple: Preoperative Surgical Planning with Volume-Rendered MDCT Images to Identify Arterial Variants Relevant to the Whipple Procedure Am. J. Roentgenol., May 1, 2007; 188(5): W451 - W455. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
