| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Mayo Medical School, Mayo Clinic, Rochester, MN.
2 Department of Radiology, Mayo Clinic Rochester, 200 First St. SW, Rochester,
MN 55905.
3 Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester,
MN.
4 Division of Biostatistics, Mayo Clinic, Rochester, MN.
5 Department of Internal Medicine, Division of Gastroenterology, Mayo Clinic,
Rochester, MN.
Received February 11, 2004;
accepted after revision April 28, 2004.
Address correspondence to J. G. Fletcher.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. A single unblinded reviewer characterized the presence and appearance of recurrent intraductal papillary mucinous neoplasms on 66 CT scans of 17 patients with proven recurrence, noting location and appearance of recurrent neoplasm. These results, described in this article, were summarized in tabular format and shown to three blinded observer. The observers then evaluated one postoperative CT examination from every patient at our institution who underwent surgical removal of intraductal papillary mucinous neoplasms (n = 45) for the presence or absence of local or distant recurrence.
RESULTS. The unblinded reviewer found 11 cases of local recurrence.
Extrapancreatic local recurrences tend to have solid components (5/6), tend to
be located adjacent to the resection margin (5/6), and may exhibit vascular
invasion (2/6). Intrapancreatic neoplasms are usually cystic (4/5). Nine
patients had distant metastases. Prospective sensitivity for recurrent tumor
ranged from 76% (13/17) to 94% (16/17). Sensitivity for local recurrence
ranged from 55% (6/11) to 82% (9/11). Specificity ranged from 79% (22/28) to
96% (27/28). Interobserver agreement for predicting recurrence was moderate to
substantial (
= 0.51-0.65).
CONCLUSION. Locally recurrent intraductal papillary mucinous neoplasms of the pancreas tend to be either extrapancreatic and solid at the resection margin or intrapancreatic and cystic. CT can detect most recurrent intraductal papillary mucinous neoplasms of the pancreas with moderate to substantial interobserver agreement.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Intraductal papillary mucinous neoplasm is predominantly a disease of elderly individuals, as seen in our patients who had a mean age at diagnosis of 65 years. There is a male preponderance with a male-to-female ratio of 2:1. The presenting symptoms are often secondary to duct obstruction by the tumor and mucin hypersecretion. The typical symptoms, which may be present for many years before the diagnosis is made, include epigastric pain, recurrent pancreatitis, and hyper-amylasemia. Obstruction of the duct leads to exocrine pancreatic insufficiency that manifests as steatorrhea, with atrophy and destruction of the gland leading to diabetes. Tumors involving the head can cause biliary duct obstruction and jaundice.
Approximately 40-80% of intraductal papillary mucinous neoplasms harbor an invasive carcinoma at diagnosis, but there is a lack of consistent imaging criteria on which to base a presurgical prediction of malignant potential with a high degree of accuracy [1, 6, 7]. Thus, surgical resection of the tumor and involved pancreas with wide margins is the treatment of choice [8, 9]. Although most series report excellent prognosis for patients with noninvasive intraductal papillary mucinous neoplasms, the prognosis of patients with invasive intraductal papillary mucinous neoplasms is much worse but is still better than that of patients with ductal adenocarcinoma [8, 10]. Nevertheless, the long-term survival of patients with metastatic intraductal papillary mucinous neoplasm is poor and comparable to those with ductal adenocarcinoma.
Intraductal papillary mucinous neoplasm, both invasive and noninvasive, has been reported to recur after resection. Diagnosis of intraductal papillary mucinous neoplasm accompanied by subsequent surgical resection is increasing at our institution much as it is worldwide. Consequently, surveillance for tumor recurrence after resection using contrast-enhanced CT has become an important aspect of patient follow-up. Primary intraductal papillary mucinous neoplasm has a distinct spectrum of radiologic appearances on cross-sectional imaging that have been well documented in several studies [6, 7, 11-14]. However, to our knowledge, no reports examining the CT appearance of recurrent intraductal papillary mucinous neoplasms of the pancreas or estimating the ability of CT to detect such recurrences have yet been published.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We initially performed an unblinded review of all cases of known recurrence to characterize the CT findings of recurrent intraductal papillary mucinous neoplasm. This was followed by a blinded multiobserver study to estimate the ability of CT to detect recurrence. For the unblinded review, all CT scans from patients with confirmed intraductal papillary mucinous neoplasm recurrence were reviewed for signs of recurrence by a gastrointestinal radiologist who correlated CT findings with clinical history and surgical or pathology reports. Each CT scan was reviewed for scanning technique, time interval from initial surgical resection to scanning, and visible evidence of tumor recurrence. For CT scans depicting intraductal papillary mucinous neoplasm recurrence, the findings were further classified by the location of recurrence (intrapancreatic vs extrapancreatic and adjacent vs remote from prior resection margin). Intrapancreatic lesions were defined as those predominantly surrounded by pancreatic parenchyma. Extrapancreatic recurrences were similarly defined as being predominantly surrounded by nonpancreatic retroperitoneal structures or fat. The internal nature of a recurrent mass (solid vs cystic vs mixed solid and cystic) was characterized, and the presence of local vascular invasion and distant metastases was noted. "Cystic" lesions were of fluid attenuation and could represent focal cystic lesions or focal dilatation of the pancreatic ducts. The time interval between surgical resection and the date imaging findings of recurrence were first visible on CT was recorded. All results were tabulated and reviewed. A list summarizing these CT findings associated with locally recurrent intraductal papillary mucinous neoplasms of the pancreas was generated.
To assess the ability of CT to detect recurrent intraductal papillary mucinous neoplasms after surgical resection using a multiobserver design, we chose all patients who had undergone surgical resection for intraductal papillary mucinous neoplasm and follow-up CT at our institution. For this analysis, from the group of 17 patients with proven tumor recurrence, one CT scan per patient was randomly selected by a statistical collaborator to reduce bias from choosing obvious cases of recurrence. Each of these CT examinations was then matched to postoperative CT examinations from patients with resected intraductal papillary mucinous neoplasm but without documented tumor recurrence, matching time interval from surgical resection for primary intraductal papillary mucinous neoplasm to follow-up CT (within 3 months). This process was repeated until all CT scans in patients without known recurrence were matched to CT scans in patients with tumor recurrence. The postoperative CT examinations of patients with disease and those without disease were then placed in random order.
Three radiologists were given the list summarizing the CT findings associated with locally recurrent intraductal papillary mucinous neoplasms of the pancreas that was generated in the first phase of our study. These observers then analyzed the CT examinations performed after intraductal papillary mucinous neoplasm resection in all patients. Observers were blinded to previous radiologic examinations and to all clinical, pathologic, and surgical follow-up findings. They interpreted these scans for the presence of recurrence (local or distant), location of local recurrence (extrapancreatic vs intrapancreatic and adjacent vs remote from prior resection margin), nature of recurrence (solid vs cystic vs mixed), and presence of vascular invasion and distant metastases.
Agreement between each pair of observers for diagnosis of recurrent intraductal papillary mucinous neoplasm and local recurrence was estimated using the kappa statistic along with a 95% confidence interval. Similarly, interobserver agreement was also estimated regarding the interpretation of characteristics of each tumor (location of local recurrence, solid vs cystic, vascular invasion). As described by Landis and Koch [15], the kappa statistic may be interpreted as follows: 0.8-1.0, almost perfect agreement; 0.6-0.79, substantial agreement; 0.40-0.59, moderate agreement; 0.2-0.39, fair agreement; 0-0.19 slight agreement; and 0 to -1.0, poor agreement. Sensitivity and specificity for the diagnosis of recurrence of intraductal papillary mucinous neoplasm and local recurrence of intraductal papillary mucinous neoplasm were estimated for each observer along with the 95% confidence intervals. The specificity for the absence of intraductal papillary mucinous neoplasm recurrence along with the 95% confidence intervals was also estimated.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A total of 66 CT scans of 17 patients were interpreted by the unblinded observer for signs of recurrent neoplasm. Eleven (65%) of the 17 patients with recurrence had local tumor recurrence. Of these 11 patients with local recurrence, six (55%) had tumors that recurred in the extrapancreatic tissues and five (45%) had recurrences in the remaining pancreas. In the six patients with local extrapancreatic recurrences, five (83%) had recurrences that were solid (four entirely solid and one solid and cystic). Five (83%) also had recurrences adjacent to the surgical resection margin (Figs. 1A, 1B, 1C, 1D). In two patients (33%), vascular invasion was seen (Figs. 2A, 2B). Four (67%) of the six patients with local recurrence in extrapancreatic tissues initially had invasive tumors, and two initially had noninvasive tumors.
|
|
|
|
|
|
Five patients had local intrapancreatic recurrences. Four (80%) of these five patients had intrapancreatic neoplasms that were cystic (Figs. 3A, 3B and 4A, 4B, 4C, 4D). Most of the intrapancreatic cystic lesions resembled small branch-duct tumors, being rounded cystic lesions eccentric to the main duct, whereas one exhibited focal dilatation in the main duct (Figs. 3A, 3B). One patient had a small 1-cm intrapancreatic lesion that was solid. Four intrapancreatic lesions in three patients were remote from the resection margin, and two were not. No local intrapancreatic tumor recurrences were associated with vascular invasion. Three of the six patients with intrapancreatic neoplasms had invasive primary tumors.
|
|
|
|
|
|
Nine patients had distant metastases, three of whom also had concurrent local recurrence. Of these nine patients, eight patients (89%) had liver metastases, three (33%) had peritoneal involvement, and one (11%) each had nodal or pleural involvement (Figs. 5A, 5B).
|
|
The number of months from resection to intraductal papillary mucinous neoplasm recurrence was calculated using the first follow-up CT that showed recurrence at unblinded review. Local extrapancreatic tumors were seen on CT a mean of 38 months after resection (range, 4-148 months). Local intrapancreatic tumors were seen a mean of 15 months after resection (range, 3-28 months). Distant metastases were seen on CT a mean of 26 months after resection (range, 0-136 months). Recurrences arising from invasive tumors were first visualized on CT scans a median of 9 months after resection (range, 0-148 months), whereas noninvasive tumors recurred a median of 22 months after resection (range, 7-39 months). The single patient with residual low-grade dysplasia at the resection margin developed an extrapancreatic local recurrence 19 months after the initial resection of a noninvasive tumor.
Prospective Multiobserver Assessment
Three observers who were blinded to the results of all other imaging
studies interpreted one contrast-enhanced CT examination from every patient at
our institution who had undergone surgical resection for intraductal papillary
mucinous neoplasm of the pancreas and follow-up CT (28 patients without
recurrence and 17 patients with recurrent tumor).
The performance of the three observers in detecting recurrent intraductal papillary mucinous neoplasm is summarized in Table 1. Sensitivity for the detection of intraductal papillary mucinous neoplasm recurrence ranged from 76% (13/17) to 94% (16/17). Sensitivity for the detection of local intraductal papillary mucinous neoplasm recurrence ranged from 55% (6/11) to 82% (9/11). Sensitivity for distant metastases ranged from 89% (8/9) to 100% (9/9). Observer agreement between radiologists in detecting intraductal papillary mucinous neoplasm recurrence, whether local or distant, was also calculated using a kappa statistic (Table 2); agreement ranged from 0.51 to 0.65, indicating moderate to substantial interobserver agreement.
|
|
The specificity of our observers for detecting the absence of intraductal papillary mucinous neoplasm recurrence ranged from 79% (22/28) to 96% (27/28). The kappa statistic for interobserver agreement for detecting the absence of intraductal papillary mucinous neoplasm recurrence between observer 1 and observer 2 was 0.89, indicating near-perfect agreement. Kappa statistics for interobserver agreement involving observer 3 were not calculated because all cases with negative findings for disease were identified.
Kappa statistics for specific CT findings were also calculated (Table 3). There was moderate to near perfect agreement about the location of an intraductal papillary mucinous neoplasm recurrence as intrapancreatic or extrapancreatic. Although agreement for identifying recurrence adjacent to the resection margin was moderate to near perfect, agreement for determining that a recurrence was remote from the resection margin was only fair to moderate. The observers noted the presence of vascular invasion and hepatic and peritoneal metastases with moderate to substantial agreement, but they classified nodal metastases with only fair agreement.
|
Eight patients without proven tumor recurrence were misidentified by our observers as having recurrence. Six of these false-positives were identified by at least two of the three observers. Two patients had hepatic lesions thought to be potential recurrences. The remaining six patients all had masses identified adjacent to the site of prior surgical resection. Most of these lesions were extrapancreatic (5/6), being divided among purely cystic lesions (n = 2) and lesions with a solid component (n = 4).
False-negative interpretations occurred in five patients with documented recurrence visible on CT. One patient was classified as having no recurrence by all three observers. This recurrence was a small 1-cm solid mass located in the remaining pancreas. Two other patients, with false-negative interpretations by observers 1 and 2, had small solid recurrences within the extrapancreatic tissues. One patient had a small liver metastasis that was not seen by two of the three observers. The remaining patient had a remote nodal metastasis that was falsely interpreted as benign lymphadenopathy rather than metastatic disease by a single observer.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Extrapancreatic local recurrences are usually located adjacent to the previous resection margin (5/6) and have a solid component (5/6). One third of these recurrences showed vascular invasion. We find it helpful to think of extrapancreatic recurrences as being invasive intraductal papillary mucinous neoplasm, even though one third of these extrapancreatic recurrences occurred in patients who had noninvasive tumors at initial resection, because extrapancreatic recurrences often resemble recurrent ductal adenocarcinoma on CT. Interestingly, invasive intraductal papillary mucinous neoplasm can also appear similar to ductal adenocarcinoma at histologic examination.
Regarding the development of intraductal papillary mucinous neoplasm in the remaining pancreas after resection, four patients developed intrapancreatic cystic lesions that resembled primary intraductal papillary mucinous neoplasm in appearance. In three of these patients, the intrapancreatic cystic lesions resembled branch-duct intraductal papillary mucinous neoplasm in appearance (Figs. 4A, 4B, 4C, 4D), and one showed focal dilatation in the main duct near the resection margin (Figs. 3A, 3B). These lesions could have developed because residual epithelial dysplasia was left after surgical resection of the main tumor mass, because the initial disease was multifocal, or because a metachronous tumor developed in the remaining gland. Several authors have hypothesized that intraductal papillary mucinous neoplasm may result from a field effect on the pancreatic duct epithelium, giving rise to the observed multicentricity of these neoplasms in approximately 33% of patients [7, 9]. In this regard, some of the intrapancreatic "recurrences" in our study are likely separate metachronous lesions that have grown over time.
In our series we found no distinction between the appearances or types of recurrence between patients who had invasive and noninvasive tumors at the original resection. However, we did find that invasive tumors recurred sooner than noninvasive tumors (median, 9 vs 22 months, respectively), as has been previously reported [16].
Three blinded observers reviewed one follow-up CT scan from each patient who underwent contrast-enhanced CT after surgical resection for intraductal papillary mucinous neoplasm at our institution. The three observers in our study detected all types of intraductal papillary mucinous neoplasm recurrence with sensitivity ranging from 76% to 94%. The sensitivity for the detection local recurrence alone was poorer, ranging from 55% to 82%. Specificity for the absence of recurrent neoplasm ranged from 79% to 96%. We believe that the performance of our observers was hampered by the study design, which did not allow comparison with earlier studies to, for example, assist in the realization that new tissue was present adjacent to the resection margin, representing early extrapancreatic local recurrences. The observers may also have benefited in this regard from comparing images with earlier CT scans to distinguish between recurrent tumor and resolving postoperative changes. Our observers did not have false-negative interpretations for detecting the typically cystic lesions of recurrent intrapancreatic intraductal papillary mucinous neoplasm. False-positive examinations generally resulted from indeterminate hepatic lesions or from solid or cystic tissue adjacent to the resection margin.
Intraductal papillary mucinous neoplasms of the pancreas are classified as noninvasive or invasive. The epithelial dysplasia associated with noninvasive intraductal papillary mucinous neoplasm is classified according to World Health Organization criteria as adenoma, borderline mass, or carcinoma in situ [17]. Carcinoma in situ behaves like low-grade epithelial dysplasia in terms of its propensity to recur after resection [16]. Similar to ductal adenocarcinoma, invasive cancers are thought to recur more commonly because of the presence of micrometastases. Even when the invasive neoplasm is excised with negative surgical margins, recurrence is seen in 50-90% of the patients [9, 16], so postoperative surveillance of these patients is usually more intense. Surgical removal is also recommended for noninvasive intraductal papillary mucinous neoplasms, despite their generally more benign course, because these tumors can recur after surgical extirpation.
Surgical approaches for resection of intraductal papillary mucinous neoplasm are still evolving because of the unanswered question about the possibility of intraductal papillary mucinous neoplasm arising from a field defect that affects all the pancreatic duct epithelium [9]. Most surgeons perform an anatomic resection, such as a pancreaticoduodenectomy or distal pancreatectomy, and check the surgical margins for residual neoplasm. Total pancreatectomy is usually reserved for cases in which neoplastic involvement of the entire gland is evident as a result of imaging or intraoperative findings.
Our study has several weaknesses. It is retrospective, and the contrast-enhanced CT scans were obtained with different imaging protocols. An optimized pancreatic protocol with biphasic technique and narrow slice thickness would likely have improved the performance of our observers. In addition, our observers interpreted only one scan from each patient to avoid bias caused by a few patients having multiple scans. As we mentioned, this method may have decreased the performance of our observers because they could not compare the CT scan with earlier studies to detect interval growth of new tumor or assess for resolution of postoperative changes. Finally, some of the patients in our control group may have undetected recurrent intraductal papillary mucinous neoplasm that would potentially change our estimates of sensitivity and specificity. However, no period of clinical observation to estimate absence of recurrent disease appears practical because one of our patients had a recurrence 12 years after the initial resection.
In conclusion, intraductal papillary mucinous neoplasms of the pancreas exhibit specific CT patterns of recurrence after initial surgical resection. Local recurrences may be extrapancreatic or intrapancreatic. Extrapancreatic recurrences are usually located adjacent to the resection margin and almost always have a solid component. They may present with vascular invasion and often resemble a recurrence of ductal pancreatic adenocarcinoma. In contrast, recurrent intrapancreatic intraductal papillary mucinous neoplasm is usually cystic, mimicking the appearance of primary intraductal papillary mucinous neoplasm. Contrast-enhanced CT performs well in the detection of recurrent tumor but does less well in detecting local recurrence. In this regard, optimized pancreatic imaging should be performed when examining for tumor recurrence, and comparison with earlier scans may help to distinguish early recurrence from resolving postoperative changes. Knowing the patterns of recurrence for intraductal papillary mucinous neoplasm of the pancreas after surgical resection combined with optimal scanning technique will allow radiologists to detect recurrent disease with a high degree of accuracy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. F. Akisik, K. Sandrasegaran, A. A. Aisen, D. D. T. Maglinte, S. Sherman, and G. A. Lehman Dynamic secretin-enhanced MR cholangiopancreatography. RadioGraphics, May 1, 2006; 26(3): 665 - 677. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Kawamoto, K. M. Horton, L. P. Lawler, R. H. Hruban, and E. K. Fishman Intraductal Papillary Mucinous Neoplasm of the Pancreas: Can Benign Lesions Be Differentiated from Malignant Lesions with Multidetector CT? RadioGraphics, November 1, 2005; 25(6): 1451 - 1468. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
