- Email: [email protected]
Periductal hypoechoic sign: an endosonographic finding associated with pancreatic malignancy
ORIGINAL ARTICLE: Clinical Endoscopy
Periductal hypoechoic sign: an endosonographic finding associated with pancreatic malignancy Suck-Ho Lee, MD, Nuri Ozden, MD, Rishi Pawa, MD, Young Hwangbo, MD, Douglas K. Pleskow, MD, Ram Chuttani, MD, Mandeep S. Sawhney, MBBS, MS Seoul, Korea, Boston, Massachusetts, USA
Background: Despite advances in imaging, differentiating benign from malignant causes of pancreatic duct dilation is difficult. Objective: The aim of our study was to assess the accuracy of the periductal hypoechoic sign (PHS), defined as patchy hypoechoic areas adjacent to a dilated pancreatic duct, for the diagnosis of pancreatic malignancy. Design: Single-center, retrospective analysis. Setting: Tertiary care university hospital. Patients: All patients who underwent EUS from 2006 to 2008 for evaluation of pancreatic pathology were identified. Those with pancreatic duct dilation of 4 mm or more in the head of the pancreas or 3 mm or more in the body or tail were included. Digitally recorded EUS images were analyzed for PHS by 1 endoscopist blinded to final results. The final diagnosis was based on pathology results or clinical follow-up. Results: During the study period, 84 of 427 patients who underwent EUS for pancreas pathology had dilated pancreatic ducts. Of these, 42 patients had benign disease and 42 had pancreatic malignancy. The PHS was noted in 31 (73.8%) of 42 patients with malignancy compared with 6 (14.3%) of 42 patients with benign disease (P ! .001). The PHS had a sensitivity of 73.8%, a specificity of 85.7%, and an accuracy of 79.8% for the diagnosis of pancreatic malignancy. After adjusting for age, patients with the PHS were 17 times more likely to have a malignancy (odds ratio 16.66; 95% CI, 5.01-55.44). Pancreatic duct diameter or dilation of both bile and pancreatic ducts were not predictive of malignancy. Limitation: A retrospective design. Conclusions: The PHS was an accurate and independent predictor of pancreatic malignancy in patients with a dilated pancreatic duct. (Gastrointest Endosc 2010;71:249-55.)
See CME section; p. 365 Copyright ª 2010 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2009.08.023
difficult for several reasons. First, the pancreas is located in the retroperitoneum, an area that does not lend itself to detailed extracorporeal imaging. Second, pancreatic cancer can have the same density as normal pancreatic parenchyma and therefore does not appear as a discrete mass on imaging studies.2 Third, chronic pancreatitis and focal acute pancreatitis can radiographically mimic pancreatic cancer.3,4 Dilation of the main pancreatic duct is a well-recognized sign of pancreatic malignancy, but can also be seen in the benign conditions mentioned above.5 EUS and EUS-guided FNA (EUS-FNA) have high accuracy for the diagnosis of pancreatic cancer and are often used to differentiate malignant from benign causes of pancreatic duct dilation.6 Patients with chronic pancreatitis are at
www.giejournal.org
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 249
Approximately 38,000 patients were diagnosed with pancreatic cancer in the United States, and 34,000 died of this disease in 2008.1 The 5-year survival rate of patients with localized disease was 20%, with regional spread in 8%, and metastasis in 1.8%. Despite advances in imaging modalities, only 7% of cancers were diagnosed in the localized stage.1 Early diagnosis of pancreatic cancer is Abbreviations: PD, pancreatic duct; PHS, periductal hypoechoic sign. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
Periductal hypoechoic sign
increased risk of developing pancreatic cancer. These patients pose an especially difficult diagnostic problem because the presence of chronic pancreatitis substantially lowers the accuracy of EUS-FNA.7,8 Because of the possibility of sampling error, a benign aspirate does not exclude the presence of malignancy. Therefore, EUS features associated with malignancy are needed to improve diagnostic accuracy and to help select patients who are at increased risk of having malignancy despite a benign FNA finding.9 We observed on EUS imaging that several patients with malignant pancreatic duct obstruction demonstrated multiple, 1- to 2-mm patchy hypoechoic areas adjacent to the main pancreatic duct (Fig. 1), whereas most with benign obstruction did not (Fig. 2). These areas were located around the main pancreatic duct upstream from the level of duct obstruction. These findings could be readily imaged with standard B-mode US at 7.5-MHz frequency. The aim of our study was to assess the accuracy of this periductal hypoechoic sign (PHS) for the diagnosis of malignant pancreatic duct obstruction.
METHODS
Lee et al
Capsule Summary What is already known on this topic d
d
Dilation of the main pancreatic duct is a well-recognized sign of pancreatic malignancy, but it can also be seen in benign conditions. EUS-FNA can be useful for the diagnosis of pancreatic cancer, but chronic pancreatitis substantially lowers its accuracy.
What this study adds to our knowledge d
d
In a single-center, retrospective study of 427 patients who underwent EUS for pancreatic abnormalities, the periductal hypoechoic sign (PHS), defined as the presence of multiple patchy 1- to 2-mm hypoechoic areas around the dilated main pancreatic duct, had a sensitivity of 73.8%, a specificity of 85.7%, and an accuracy of 79.8% for the diagnosis of pancreatic malignancy. After adjusting for age, patients with the PHS were 17 times more likely to have a malignancy.
malignant surgical or radiological biopsy specimen, or decrease in size of the tumor mass after chemotherapy or radiation therapy. Patients were designated as having a benign lesion if they met any of the following criteria: surgical biopsy specimen showed benign tissue or there was no increase in size of the lesion or development of new lesions on radiologic imaging (CT or magnetic resonance imaging) obtained at least 6 months after the EUS procedure. Decisions regarding further evaluation and treatment of patients were made by the treating physicians. The study was approved by our institutional review board. Data were analyzed by using statistical software (SPSS version 13.0, SPSS Corp, Chicago, Ill). Categorical variables were compared by using the c2 or Fisher exact test. Continuous variables were analyzed by using the Student t test. A logistic regression model as created with the presence or absence of malignancy served as the outcome variable. The following were used as explanatory variables: PHS (present or absent), site of pancreatic duct dilation (head of the pancreas or body/tail of the pancreas), dilation of both pancreatic and biliary ducts (double duct sign present or absent), and maximum diameter of the pancreatic duct (continuous variable). Models were adjusted for patient age and sex.
All patients who underwent EUS for the evaluation of pancreatic pathology at Beth Israel Deaconess Medical Center in Boston, Massachusetts, between 2006 and 2008 by one endoscopist (M.S.S.) were identified. Endoscopy reports and digital US images were extracted from the electronic endoscopy database and reviewed. Patients with dilation of the main pancreatic duct were included in the study. Pancreatic duct dilation was defined as a main pancreatic duct diameter measured at EUS of 4 mm or more in the head of the pancreas or 3 mm or more in the body and tail of the pancreas. Digital EUS images were then separated from the EUS report and patient identifiers. These were coded and reviewed by an experienced endoscopist (S.H.L.) who had not performed the EUS procedures and was blinded to final diagnosis. EUS images that showed obvious malignant findings such as a mass lesion or obvious changes of chronic pancreatitis such as a pancreatic duct stone or pancreatic calcification were excluded. Image labels and images showing FNA were not included. An attempt was made to include only EUS images of the main pancreatic duct. Based on EUS images, the endoscopist recorded the presence or absence of the PHS for each patient. The PHS was defined as a multiple patchy, 1- to 2-mm hypoechoic areas seen adjacent to the dilated pancreatic duct for at least 25% of the length of the pancreatic duct or multiple patchy hypoechoic areas that were present around the entire circumference of the main pancreatic duct regardless of the length of duct involved. Patients’ medical records including surgery, pathology, and radiology reports were analyzed. The criterion standard was based on pathology findings or clinical followup. Patients were designated as having malignancy if they met any of the following criteria: malignant cytology,
During the study period, 427 patients underwent EUS for the assessment of pancreas pathology. Of these, 96 patients were noted to have dilation of the main pancreatic duct. Twelve patients were excluded from the analysis. This included 3 patients who were lost to follow-up.
250 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 2 : 2010
www.giejournal.org
RESULTS
Lee et al
Periductal hypoechoic sign
Figure 1. Dilated pancreatic ducts secondary to malignancy showing the presence of the PHS (arrows, patchy hypoechoic areas surrounding the dilated pancreatic duct).
Five patients were excluded because they had poor-quality EUS images, and determination of the PHS was not possible. Of these, 4 patients were finally diagnosed with benign disease and 1 patient with adenocarcinoma of the pancreas. Four patients were excluded because all EUS images showing the pancreatic duct also clearly demonstrated a mass in the pancreas. Of these, 3 were finally diagnosed with adenocarcinoma of the pancreas and 1 with a benign inflammatory mass. The remaining 84 patients constituted the study population. Forty-two of 84 patients met the study criteria for malignancy, whereas 42 of 84 patients met the study criteria for benign disease. Table 1 shows the clinical characteristics of the 2 groups. Patients with malignancy were on average older than those with benign disease (71 years vs 61.5 years, respectively; P ! .001). The final diagnoses were based on cytology (EUS-FNA, 36.9%, 31/84), surgical pathology (20.2%, 17/84), and clinical follow-up (42.9%, 36/84). In patients with pancreatic malignancy, 41 patients were diagnosed with adenocarcinoma of the pancreas and 1 patient was diagnosed with
a pancreatic lymphoma. Final diagnosis was based on clinical follow-up in 36 patients. Of these, 29 patients showed no increase in size of mass lesion or development of new lesions on radiologic imaging (CT or magnetic resonance imaging) obtained at least 6 months after the EUS procedure and met the study criteria for benign disease. In 7 patients, diagnosis of pancreas malignancy was made based on clinical information. These patients also met the study criteria for malignancy: 5 patients were noted to have a decrease in size of the mass lesion after radiation and/or chemotherapy, 1 patient showed an increase in size of the mass lesion, and 1 patient showed development of lymphadenopathy on follow-up imaging. Tumors were located in the head of the pancreas in 27 (64.3%) patients, the body of the pancreas in 10 (23.8%) patients, the uncinate pancreas in 4 (9.5%) patients, and in the tail of the pancreas in 1 (2.4%) patient. The average tumor size was 27.9 6.7 mm. In patients with benign pancreatic disease, 18 (42.9%) of 42 patients received a diagnosis of chronic pancreatitis. This diagnosis was based on EUS criteria for chronic pancreatitis, pancreatic calcifications on a CT
www.giejournal.org
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 251
Periductal hypoechoic sign
Lee et al
Figure 2. Dilated pancreatic ducts secondary to benign disease showing the absence of the PHS (note the absence of any patchy hypoechoic areas surrounding the dilated pancreatic duct).
scan, or positive results on the secretin stimulation test. In 17 of 42 patients, the pancreas was normal on EUS imaging other than dilation of the main pancreatic duct. No interval development of pathology was noted in the pancreas on follow-up imaging studies in these 17 patients. Seven patients were noted to have benign inflammatory mass lesions. All lesions were located in the head of the pancreas with a mean lesion size of 30.3 16.04 mm. All patients had benign cytology findings on EUS-FNA. Two patients underwent surgical resection with confirmation of benign disease. No change (n Z 1) or a decrease (n Z 4) in the size of the mass lesion was noted on follow-up imaging studies. The PHS was noted in 31 (73.8%) of 42 patients with pancreatic malignancy compared with 6 (14.3%) of 42 patients with benign pancreatic disease (P ! .001) (Table 2). Patients with the PHS were almost 17 times more likely to have malignancy compared with those without the PHS (odds ratio 16.9; 95% CI, 5.6–51.0). Table 2 also shows
that patients with malignancy were more likely to have dilation of the pancreatic duct in the body and tail of the pancreas and a discrete mass lesion in the pancreatic parenchyma. There was no difference in mean pancreatic duct diameter or dilation of both biliary and pancreatic ducts between patients with malignancy and those with benign disease. For the diagnosis of malignancy, the PHS had a sensitivity of 73.8%, a specificity of 85.7%, and an accuracy of 79.8%. The performance characteristics of the PHS and EUS-FNA are summarized in Table 3. Widely overlapping confidence intervals were noted between the accuracy of the PHS and EUS-FNA. Of the 6 patients with confirmed malignancy who had benign cytology findings on FNA (false-negative cytology results), 4 demonstrated the PHS. Logistic regression analyses were performed with final diagnosis (malignancy or benign disease) serving as the outcome variable. Table 4 shows the results of the final regression model. The presence of the PHS was found to be
252 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 2 : 2010
www.giejournal.org
Lee et al
Periductal hypoechoic sign
We defined PHS as the presence of multiple patchy 1- to 2-mm hypoechoic areas noted in the pancreatic parenchyma around the dilated main pancreatic duct. A positive sign required patchy hypoechoic areas to be present for at least 25% of the visible duct length or around the entire circumference of the duct. This was done to exclude imaging artifacts that may have mimicked a few patchy hypoechoic areas and resulted in possible misclassification. This finding was most prominent around the pancreatic duct just above the level of obstruction and became less prominent around the distal duct. We made several attempts to correlate this EUS finding with pathological specimens obtained after cancer resection. Unfortunately, we were unable to find adequate cross sections of the main pancreatic duct above the level of the tumor. In the absence of a tissue correlate for the PHS, we can only speculate on the cause and nature of this observation. We postulate that these hypoechoic areas represent mild, patchy acute pancreatitis from obstruction of the main pancreatic duct and, in turn, obstruction of major side branches. Because these changes are most prominent adjacent to the tumor obstructing the pancreatic duct, we think that there may be a pressure gradient along the
main pancreatic duct accounting for these changes. Despite a very large duct diameter, we did not see similar changes in patients with chronic pancreatitis. We attribute this to the fact that obstruction of the pancreatic duct in patients with benign disease occurs over a longer period and is perhaps less complete than in patients with pancreatic cancer. This may allow the pancreatic duct to dilate slowly without causing significant periductal changes. Further, atrophy and scarring of the parenchyma seen in chronic pancreatitis may prevent these changes from developing in patients. An alternative explanation for these findings is that the hypoechoic areas represent patchy infiltration of the pancreas parenchyma by tumor cells. Other authors have described pancreatic duct characteristics that are associated with malignancy: narrow stricture with marked dilation, double duct sign, pancreatic duct stenosis of 10 mm or more, and the absence of branch ducts.4,9,10 These changes have all been described during pancreatography and are less useful in clinical practice because endoscopic retrograde pancreatography is now rarely performed to make a diagnosis of pancreatic cancer. Newer imaging modalities such as EUS elastography have been proposed to help differentiate benign from malignant pancreatic lesions. Elastography requires specialized image processors and in one study had an accuracy of only 45% for the diagnosis of pancreatic malignancy.11 To our knowledge, our study is the first attempt to describe pancreatic duct characteristics associated with malignancy by using EUS imaging. Apart from the PHS, we also found that the main duct dilation involving the body and tail of the pancreas was predictive of malignancy. A likely explanation of this finding is that most pancreatic cancers in our study were found in the head of the pancreas. Because the pancreas head was mostly obscured by the tumor, dilation of the main pancreatic duct was observed mainly in the body and tail of the pancreas. In almost all patients with benign disease, such as chronic pancreatitis, the entire pancreatic duct was dilated. In our analysis, the presence of a double duct sign did not increase the likelihood of malignancy. Similar findings were also reported by Menges et al,12 who found that several patients with confirmed chronic pancreatitis also demonstrated dilation of both biliary and pancreatic ducts. The accuracy of EUS-FNA for the diagnosis of pancreatic cancer was lower in our study compared with that in other reports.6,13 We did not report on all EUS-FNAs performed at our institution, but only those in patients with a dilated pancreatic duct. Thus, a substantial proportion of patients included in our study had chronic pancreatitis. The low accuracy of EUS-FNA in the setting of chronic pancreatitis has been reported by other investigators as well.7,8 The PHS seems to be especially useful in patients with chronic pancreatitis. In our series, we did not observe any patient with chronic pancreatitis to demonstrate the PHS. We do not anticipate that the PHS would
www.giejournal.org
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 253
TABLE 1. Baseline characteristics of study patients
Age SD, y Sex, male, no. (%)
Benign (n Z 42)
Malignant (n Z 42)
61.5 13.21
71.98 10.51
19 (45.2)
21 (50)
Final diagnosis, no. (%) Adenocarcinoma
0 (0)
41 (97.6)
Lymphoma
0 (0)
1 (2.4)
Chronic pancreatitis
18 (42.9)
0 (0)
Inflammatory mass
7 (16.7)
0 (0)
17 (40.5)
0 (0)
Normal SD, Standard deviation.
a strong and independent predictor of malignancy (OR 16.66; 95% CI, 5.00–55.45). The only other variable that was independently predictive of malignancy was pancreatic duct dilation that extended to the body and tail portions (OR 4.71; 95% CI, 1.54–14.46). The pancreatic duct diameter and dilation of both biliary and pancreatic ducts did not predict malignancy. The presence of a mass lesion in the pancreas correlated strongly with other dependent variables and therefore could not be entered into the regression model.
DISCUSSION
Periductal hypoechoic sign
Lee et al
TABLE 2. EUS findings in patients with malignant and benign pancreatic duct dilation Benign (n Z 42)
Malignant (n Z 42)
OR (95% CI)
P value
PD dilation in body/tail
7 (16.7%)
20 (47.6%)
4.5 (1.7-12.5)
.003
Double duct sign
11 (26.2%)
13 (31%)
1.3 (0.5-3.3)
.63
6 (14.3)
31 (73.8)
16.9 (5.6-51)
.001
50.29 11.68
49.57 13.15
n/a
.79
9 (21.4%)
39 (92.9%)
47.7 (12.4-178.9)
.001
PHS PD size (mm SD) Mass lesion
OR, Odds ratio; CI, confidence interval; PD, pancreatic duct; n/a, not available; PHS, periductal hypoechoic sign.
TABLE 3. Test characteristics of the periductal hypoechoic sign and EUS–FNA Accuracy, % (95% CI)
Sensitivity, %
Specificity, %
PPV, %
NPV, %
PHS
79.8% (70.1-86.4)
73.8
85.7
83.8
76.6
EUS-FNA
85.8% (79.1-85.8)
71.4
100
100
78.0
CI, Confidence interval; PPV, positive predictive value; NPV, negative predictive value; PHS, periductal hypoechoic sign.
TABLE 4. Age-adjusted logistic regression model analyzing predictors of malignancy in patients with a dilated pancreatic duct
PD dilation in the body/tail DDS
Coefficient
SE
OR (95% CI)
P value
1.550
0.572
4.71 (1.54-14.46)
.007
0.536
0.95 (0.33-2.73)
.93
0.047
PHS
2.813
0.614
16.7 (5.01-55.44)
!.001
PD size (mm SD)
0.006
0.020
1.0 (0.97-1.05)
.78
SE, standard error; OR, odds ratio; CI, confidence interval; PD, pancreatic duct; DDS, dilation of both bile duct and main pancreatic ducts; PHS, periductal hypoechoic sign; SD, standard deviation.
replace EUS-FNA, but rather would be used as an adjunct. The PHS was noted in 4 of 6 patients with malignancy who had false-negative FNA cytology results. This again suggests that patients with the PHS are at increased risk of having malignancy and that a high suspicion for malignancy should be maintained in these patients despite negative EUS-FNA cytology. Our study had several limitations. First, the analysis was conducted retrospectively by using still EUS images that were not specifically obtained for the purposes of demonstrating the PHS. At no time during the acquisition of these images was the PHS specifically sought. We do not routinely obtain video recordings of EUS procedures and therefore could not conduct this study by using archived video images. We attempted to overcome these limitations by carefully selecting images of the main pancreatic duct that did not show obvious changes of malignancy or chronic pancreatitis. Further, we attempted to make the definition of the PHS precise so it could be objectively applied to images. Patients in whom optimal images could
not be obtained were excluded from the analysis. The retrospective nature of our analysis does not allow us to fully exclude the possibility that the endoscopist interpreting the images was influenced by factors other than periductal changes. Prospective validation of these findings is needed before the PHS can be widely used in clinical practice. Using a retrospective database had the advantage of allowing us to screen a large number of subjects (O400) to include in our study. Second, the final diagnosis was not based on pathology findings but on clinical followup in 42.9% of cases. Third, we were unable to correlate the PHS with histology because of the unavailability of cross sections of dilated pancreatic duct upstream from the level of obstruction. Our explanation of the possible pathophysiology of these findings thus remains speculative. The strengths of our analysis include careful image selection and that images were reviewed by one endoscopist who did not perform the EUS examinations and was blinded to the EUS report and final diagnosis.
254 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 2 : 2010
www.giejournal.org
Lee et al
In summary, we found patients with the PHS to be almost 17 times more likely to harbor malignancy compared with those who did not exhibit this sign. Determination of the PHS does not require additional equipment and can be performed during a routine EUS examination. We recommend that a high level of suspicion for malignancy be maintained in patients with a dilated pancreatic duct who demonstrate the PHS. REFERENCES 1. The SEER Cancer Statistics Review, National Cancer Institute, 19752005. Available at: http://seer.cancer.gov/csr/1975_2005/. Accessed January 5, 2009. 2. Prokesch R, Chow L, Beaulieu C, et al. Isoattenuating pancreatic adenocarcinoma at multi-detector row CT: secondary signs. Radiology 2002;224:764-8. 3. Kamisawa T, Imai M, Yui Chen P, et al. Strategy for differentiating autoimmune pancreatitis from pancreatic cancer. Pancreas 2008;37:62-7. 4. Adamek H, Albert J, Breer H, et al. Pancreatic cancer detection with magnetic resonance cholangiopancreatography and endoscopic retrograde cholangiopancreatography: a prospective controlled study. Lancet 2000;356:190-3. 5. Tanaka S, Nakaizumi A, Ioka T, et al. Main pancreatic duct dilatation: a sign of high risk for pancreatic cancer. Jpn J Clin Oncol 2002;32:407-11. 6. Wiersema MJ, Vilmann P, Giovannini M, et al. Endosonography-guided fine-needle aspiration biopsy: diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087-95. 7. Varadarajulu S, Tamhane A, Eloubeidi M. Yield of EUS-guided FNA of pancreatic masses in the presence or the absence of chronic pancreatitis. Gastrointest Endosc 2005;62:728-36. 8. Fritscher-Ravens A, Brand L, Kno¨fel W, et al. Comparison of endoscopic ultrasound-guided fine needle aspiration for focal pancreatic lesions in
www.giejournal.org
Periductal hypoechoic sign
9.
10.
11.
12.
13.
patients with normal parenchyma and chronic pancreatitis. Am J Gastroenterol 2002;97:2768-75. Saftoiu A, Vilmann P, Gorunescu F, et al. Neural network analysis of dynamic sequences of EUS elastography used for the differential diagnosis of chronic pancreatitis and pancreatic cancer. Gastrointest Endosc 2008;68:1086-94. Inoue K, Ohuchida J, Ohtsuka T, et al. Severe localized stenosis and marked dilatation of the main pancreatic duct are indicators of pancreatic cancer instead of chronic pancreatitis on endoscopic retrograde balloon pancreatography. Gastrointest Endosc 2003;58: 510-5. Hirche TO, Ignee A, Barreiros AP, et al. Indications and limitations of endoscopic ultrasound elastography for evaluation of focal pancreatic lesions. Endoscopy 2008;40:910-7. Menges M, Lerch M, Zeitz M. The double duct sign in patients with malignant and benign pancreatic lesions. Gastrointest Endosc 2000;52:74-7. Agarwal B, Abu-Hamda E, Molke K, et al. Endoscopic ultrasoundguided fine needle aspiration and multidetector spiral CT in the diagnosis of pancreatic cancer. Am J Gastroenterol 2004;99:844-50.
Received February 3, 2009. Accepted August 23, 2009. Current affiliations: Divisions of Gastroenterology (S.-H.L., N.O., R.P., D.K.P., R.C., M.S.S.), Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA, Soonchunhyang University Cheonan Hospital, (Y.H.), Seoul, Korea. Reprint requests: Mandeep S. Sawhney, MBBS, MS, Division of Gastroenterology, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, RABB-ROSE 101, Boston, MA 02215. If you would like to chat with an author of this article, you may contact Dr. Sawhney at [email protected].
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 255
Periductal hypoechoic sign: an endosonographic finding associated with pancreatic malignancy Suck-Ho Lee, MD, Nuri Ozden, MD, Rishi Pawa, MD, Young Hwangbo, MD, Douglas K. Pleskow, MD, Ram Chuttani, MD, Mandeep S. Sawhney, MBBS, MS Seoul, Korea, Boston, Massachusetts, USA
Background: Despite advances in imaging, differentiating benign from malignant causes of pancreatic duct dilation is difficult. Objective: The aim of our study was to assess the accuracy of the periductal hypoechoic sign (PHS), defined as patchy hypoechoic areas adjacent to a dilated pancreatic duct, for the diagnosis of pancreatic malignancy. Design: Single-center, retrospective analysis. Setting: Tertiary care university hospital. Patients: All patients who underwent EUS from 2006 to 2008 for evaluation of pancreatic pathology were identified. Those with pancreatic duct dilation of 4 mm or more in the head of the pancreas or 3 mm or more in the body or tail were included. Digitally recorded EUS images were analyzed for PHS by 1 endoscopist blinded to final results. The final diagnosis was based on pathology results or clinical follow-up. Results: During the study period, 84 of 427 patients who underwent EUS for pancreas pathology had dilated pancreatic ducts. Of these, 42 patients had benign disease and 42 had pancreatic malignancy. The PHS was noted in 31 (73.8%) of 42 patients with malignancy compared with 6 (14.3%) of 42 patients with benign disease (P ! .001). The PHS had a sensitivity of 73.8%, a specificity of 85.7%, and an accuracy of 79.8% for the diagnosis of pancreatic malignancy. After adjusting for age, patients with the PHS were 17 times more likely to have a malignancy (odds ratio 16.66; 95% CI, 5.01-55.44). Pancreatic duct diameter or dilation of both bile and pancreatic ducts were not predictive of malignancy. Limitation: A retrospective design. Conclusions: The PHS was an accurate and independent predictor of pancreatic malignancy in patients with a dilated pancreatic duct. (Gastrointest Endosc 2010;71:249-55.)
See CME section; p. 365 Copyright ª 2010 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2009.08.023
difficult for several reasons. First, the pancreas is located in the retroperitoneum, an area that does not lend itself to detailed extracorporeal imaging. Second, pancreatic cancer can have the same density as normal pancreatic parenchyma and therefore does not appear as a discrete mass on imaging studies.2 Third, chronic pancreatitis and focal acute pancreatitis can radiographically mimic pancreatic cancer.3,4 Dilation of the main pancreatic duct is a well-recognized sign of pancreatic malignancy, but can also be seen in the benign conditions mentioned above.5 EUS and EUS-guided FNA (EUS-FNA) have high accuracy for the diagnosis of pancreatic cancer and are often used to differentiate malignant from benign causes of pancreatic duct dilation.6 Patients with chronic pancreatitis are at
www.giejournal.org
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 249
Approximately 38,000 patients were diagnosed with pancreatic cancer in the United States, and 34,000 died of this disease in 2008.1 The 5-year survival rate of patients with localized disease was 20%, with regional spread in 8%, and metastasis in 1.8%. Despite advances in imaging modalities, only 7% of cancers were diagnosed in the localized stage.1 Early diagnosis of pancreatic cancer is Abbreviations: PD, pancreatic duct; PHS, periductal hypoechoic sign. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
Periductal hypoechoic sign
increased risk of developing pancreatic cancer. These patients pose an especially difficult diagnostic problem because the presence of chronic pancreatitis substantially lowers the accuracy of EUS-FNA.7,8 Because of the possibility of sampling error, a benign aspirate does not exclude the presence of malignancy. Therefore, EUS features associated with malignancy are needed to improve diagnostic accuracy and to help select patients who are at increased risk of having malignancy despite a benign FNA finding.9 We observed on EUS imaging that several patients with malignant pancreatic duct obstruction demonstrated multiple, 1- to 2-mm patchy hypoechoic areas adjacent to the main pancreatic duct (Fig. 1), whereas most with benign obstruction did not (Fig. 2). These areas were located around the main pancreatic duct upstream from the level of duct obstruction. These findings could be readily imaged with standard B-mode US at 7.5-MHz frequency. The aim of our study was to assess the accuracy of this periductal hypoechoic sign (PHS) for the diagnosis of malignant pancreatic duct obstruction.
METHODS
Lee et al
Capsule Summary What is already known on this topic d
d
Dilation of the main pancreatic duct is a well-recognized sign of pancreatic malignancy, but it can also be seen in benign conditions. EUS-FNA can be useful for the diagnosis of pancreatic cancer, but chronic pancreatitis substantially lowers its accuracy.
What this study adds to our knowledge d
d
In a single-center, retrospective study of 427 patients who underwent EUS for pancreatic abnormalities, the periductal hypoechoic sign (PHS), defined as the presence of multiple patchy 1- to 2-mm hypoechoic areas around the dilated main pancreatic duct, had a sensitivity of 73.8%, a specificity of 85.7%, and an accuracy of 79.8% for the diagnosis of pancreatic malignancy. After adjusting for age, patients with the PHS were 17 times more likely to have a malignancy.
malignant surgical or radiological biopsy specimen, or decrease in size of the tumor mass after chemotherapy or radiation therapy. Patients were designated as having a benign lesion if they met any of the following criteria: surgical biopsy specimen showed benign tissue or there was no increase in size of the lesion or development of new lesions on radiologic imaging (CT or magnetic resonance imaging) obtained at least 6 months after the EUS procedure. Decisions regarding further evaluation and treatment of patients were made by the treating physicians. The study was approved by our institutional review board. Data were analyzed by using statistical software (SPSS version 13.0, SPSS Corp, Chicago, Ill). Categorical variables were compared by using the c2 or Fisher exact test. Continuous variables were analyzed by using the Student t test. A logistic regression model as created with the presence or absence of malignancy served as the outcome variable. The following were used as explanatory variables: PHS (present or absent), site of pancreatic duct dilation (head of the pancreas or body/tail of the pancreas), dilation of both pancreatic and biliary ducts (double duct sign present or absent), and maximum diameter of the pancreatic duct (continuous variable). Models were adjusted for patient age and sex.
All patients who underwent EUS for the evaluation of pancreatic pathology at Beth Israel Deaconess Medical Center in Boston, Massachusetts, between 2006 and 2008 by one endoscopist (M.S.S.) were identified. Endoscopy reports and digital US images were extracted from the electronic endoscopy database and reviewed. Patients with dilation of the main pancreatic duct were included in the study. Pancreatic duct dilation was defined as a main pancreatic duct diameter measured at EUS of 4 mm or more in the head of the pancreas or 3 mm or more in the body and tail of the pancreas. Digital EUS images were then separated from the EUS report and patient identifiers. These were coded and reviewed by an experienced endoscopist (S.H.L.) who had not performed the EUS procedures and was blinded to final diagnosis. EUS images that showed obvious malignant findings such as a mass lesion or obvious changes of chronic pancreatitis such as a pancreatic duct stone or pancreatic calcification were excluded. Image labels and images showing FNA were not included. An attempt was made to include only EUS images of the main pancreatic duct. Based on EUS images, the endoscopist recorded the presence or absence of the PHS for each patient. The PHS was defined as a multiple patchy, 1- to 2-mm hypoechoic areas seen adjacent to the dilated pancreatic duct for at least 25% of the length of the pancreatic duct or multiple patchy hypoechoic areas that were present around the entire circumference of the main pancreatic duct regardless of the length of duct involved. Patients’ medical records including surgery, pathology, and radiology reports were analyzed. The criterion standard was based on pathology findings or clinical followup. Patients were designated as having malignancy if they met any of the following criteria: malignant cytology,
During the study period, 427 patients underwent EUS for the assessment of pancreas pathology. Of these, 96 patients were noted to have dilation of the main pancreatic duct. Twelve patients were excluded from the analysis. This included 3 patients who were lost to follow-up.
250 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 2 : 2010
www.giejournal.org
RESULTS
Lee et al
Periductal hypoechoic sign
Figure 1. Dilated pancreatic ducts secondary to malignancy showing the presence of the PHS (arrows, patchy hypoechoic areas surrounding the dilated pancreatic duct).
Five patients were excluded because they had poor-quality EUS images, and determination of the PHS was not possible. Of these, 4 patients were finally diagnosed with benign disease and 1 patient with adenocarcinoma of the pancreas. Four patients were excluded because all EUS images showing the pancreatic duct also clearly demonstrated a mass in the pancreas. Of these, 3 were finally diagnosed with adenocarcinoma of the pancreas and 1 with a benign inflammatory mass. The remaining 84 patients constituted the study population. Forty-two of 84 patients met the study criteria for malignancy, whereas 42 of 84 patients met the study criteria for benign disease. Table 1 shows the clinical characteristics of the 2 groups. Patients with malignancy were on average older than those with benign disease (71 years vs 61.5 years, respectively; P ! .001). The final diagnoses were based on cytology (EUS-FNA, 36.9%, 31/84), surgical pathology (20.2%, 17/84), and clinical follow-up (42.9%, 36/84). In patients with pancreatic malignancy, 41 patients were diagnosed with adenocarcinoma of the pancreas and 1 patient was diagnosed with
a pancreatic lymphoma. Final diagnosis was based on clinical follow-up in 36 patients. Of these, 29 patients showed no increase in size of mass lesion or development of new lesions on radiologic imaging (CT or magnetic resonance imaging) obtained at least 6 months after the EUS procedure and met the study criteria for benign disease. In 7 patients, diagnosis of pancreas malignancy was made based on clinical information. These patients also met the study criteria for malignancy: 5 patients were noted to have a decrease in size of the mass lesion after radiation and/or chemotherapy, 1 patient showed an increase in size of the mass lesion, and 1 patient showed development of lymphadenopathy on follow-up imaging. Tumors were located in the head of the pancreas in 27 (64.3%) patients, the body of the pancreas in 10 (23.8%) patients, the uncinate pancreas in 4 (9.5%) patients, and in the tail of the pancreas in 1 (2.4%) patient. The average tumor size was 27.9 6.7 mm. In patients with benign pancreatic disease, 18 (42.9%) of 42 patients received a diagnosis of chronic pancreatitis. This diagnosis was based on EUS criteria for chronic pancreatitis, pancreatic calcifications on a CT
www.giejournal.org
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 251
Periductal hypoechoic sign
Lee et al
Figure 2. Dilated pancreatic ducts secondary to benign disease showing the absence of the PHS (note the absence of any patchy hypoechoic areas surrounding the dilated pancreatic duct).
scan, or positive results on the secretin stimulation test. In 17 of 42 patients, the pancreas was normal on EUS imaging other than dilation of the main pancreatic duct. No interval development of pathology was noted in the pancreas on follow-up imaging studies in these 17 patients. Seven patients were noted to have benign inflammatory mass lesions. All lesions were located in the head of the pancreas with a mean lesion size of 30.3 16.04 mm. All patients had benign cytology findings on EUS-FNA. Two patients underwent surgical resection with confirmation of benign disease. No change (n Z 1) or a decrease (n Z 4) in the size of the mass lesion was noted on follow-up imaging studies. The PHS was noted in 31 (73.8%) of 42 patients with pancreatic malignancy compared with 6 (14.3%) of 42 patients with benign pancreatic disease (P ! .001) (Table 2). Patients with the PHS were almost 17 times more likely to have malignancy compared with those without the PHS (odds ratio 16.9; 95% CI, 5.6–51.0). Table 2 also shows
that patients with malignancy were more likely to have dilation of the pancreatic duct in the body and tail of the pancreas and a discrete mass lesion in the pancreatic parenchyma. There was no difference in mean pancreatic duct diameter or dilation of both biliary and pancreatic ducts between patients with malignancy and those with benign disease. For the diagnosis of malignancy, the PHS had a sensitivity of 73.8%, a specificity of 85.7%, and an accuracy of 79.8%. The performance characteristics of the PHS and EUS-FNA are summarized in Table 3. Widely overlapping confidence intervals were noted between the accuracy of the PHS and EUS-FNA. Of the 6 patients with confirmed malignancy who had benign cytology findings on FNA (false-negative cytology results), 4 demonstrated the PHS. Logistic regression analyses were performed with final diagnosis (malignancy or benign disease) serving as the outcome variable. Table 4 shows the results of the final regression model. The presence of the PHS was found to be
252 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 2 : 2010
www.giejournal.org
Lee et al
Periductal hypoechoic sign
We defined PHS as the presence of multiple patchy 1- to 2-mm hypoechoic areas noted in the pancreatic parenchyma around the dilated main pancreatic duct. A positive sign required patchy hypoechoic areas to be present for at least 25% of the visible duct length or around the entire circumference of the duct. This was done to exclude imaging artifacts that may have mimicked a few patchy hypoechoic areas and resulted in possible misclassification. This finding was most prominent around the pancreatic duct just above the level of obstruction and became less prominent around the distal duct. We made several attempts to correlate this EUS finding with pathological specimens obtained after cancer resection. Unfortunately, we were unable to find adequate cross sections of the main pancreatic duct above the level of the tumor. In the absence of a tissue correlate for the PHS, we can only speculate on the cause and nature of this observation. We postulate that these hypoechoic areas represent mild, patchy acute pancreatitis from obstruction of the main pancreatic duct and, in turn, obstruction of major side branches. Because these changes are most prominent adjacent to the tumor obstructing the pancreatic duct, we think that there may be a pressure gradient along the
main pancreatic duct accounting for these changes. Despite a very large duct diameter, we did not see similar changes in patients with chronic pancreatitis. We attribute this to the fact that obstruction of the pancreatic duct in patients with benign disease occurs over a longer period and is perhaps less complete than in patients with pancreatic cancer. This may allow the pancreatic duct to dilate slowly without causing significant periductal changes. Further, atrophy and scarring of the parenchyma seen in chronic pancreatitis may prevent these changes from developing in patients. An alternative explanation for these findings is that the hypoechoic areas represent patchy infiltration of the pancreas parenchyma by tumor cells. Other authors have described pancreatic duct characteristics that are associated with malignancy: narrow stricture with marked dilation, double duct sign, pancreatic duct stenosis of 10 mm or more, and the absence of branch ducts.4,9,10 These changes have all been described during pancreatography and are less useful in clinical practice because endoscopic retrograde pancreatography is now rarely performed to make a diagnosis of pancreatic cancer. Newer imaging modalities such as EUS elastography have been proposed to help differentiate benign from malignant pancreatic lesions. Elastography requires specialized image processors and in one study had an accuracy of only 45% for the diagnosis of pancreatic malignancy.11 To our knowledge, our study is the first attempt to describe pancreatic duct characteristics associated with malignancy by using EUS imaging. Apart from the PHS, we also found that the main duct dilation involving the body and tail of the pancreas was predictive of malignancy. A likely explanation of this finding is that most pancreatic cancers in our study were found in the head of the pancreas. Because the pancreas head was mostly obscured by the tumor, dilation of the main pancreatic duct was observed mainly in the body and tail of the pancreas. In almost all patients with benign disease, such as chronic pancreatitis, the entire pancreatic duct was dilated. In our analysis, the presence of a double duct sign did not increase the likelihood of malignancy. Similar findings were also reported by Menges et al,12 who found that several patients with confirmed chronic pancreatitis also demonstrated dilation of both biliary and pancreatic ducts. The accuracy of EUS-FNA for the diagnosis of pancreatic cancer was lower in our study compared with that in other reports.6,13 We did not report on all EUS-FNAs performed at our institution, but only those in patients with a dilated pancreatic duct. Thus, a substantial proportion of patients included in our study had chronic pancreatitis. The low accuracy of EUS-FNA in the setting of chronic pancreatitis has been reported by other investigators as well.7,8 The PHS seems to be especially useful in patients with chronic pancreatitis. In our series, we did not observe any patient with chronic pancreatitis to demonstrate the PHS. We do not anticipate that the PHS would
www.giejournal.org
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 253
TABLE 1. Baseline characteristics of study patients
Age SD, y Sex, male, no. (%)
Benign (n Z 42)
Malignant (n Z 42)
61.5 13.21
71.98 10.51
19 (45.2)
21 (50)
Final diagnosis, no. (%) Adenocarcinoma
0 (0)
41 (97.6)
Lymphoma
0 (0)
1 (2.4)
Chronic pancreatitis
18 (42.9)
0 (0)
Inflammatory mass
7 (16.7)
0 (0)
17 (40.5)
0 (0)
Normal SD, Standard deviation.
a strong and independent predictor of malignancy (OR 16.66; 95% CI, 5.00–55.45). The only other variable that was independently predictive of malignancy was pancreatic duct dilation that extended to the body and tail portions (OR 4.71; 95% CI, 1.54–14.46). The pancreatic duct diameter and dilation of both biliary and pancreatic ducts did not predict malignancy. The presence of a mass lesion in the pancreas correlated strongly with other dependent variables and therefore could not be entered into the regression model.
DISCUSSION
Periductal hypoechoic sign
Lee et al
TABLE 2. EUS findings in patients with malignant and benign pancreatic duct dilation Benign (n Z 42)
Malignant (n Z 42)
OR (95% CI)
P value
PD dilation in body/tail
7 (16.7%)
20 (47.6%)
4.5 (1.7-12.5)
.003
Double duct sign
11 (26.2%)
13 (31%)
1.3 (0.5-3.3)
.63
6 (14.3)
31 (73.8)
16.9 (5.6-51)
.001
50.29 11.68
49.57 13.15
n/a
.79
9 (21.4%)
39 (92.9%)
47.7 (12.4-178.9)
.001
PHS PD size (mm SD) Mass lesion
OR, Odds ratio; CI, confidence interval; PD, pancreatic duct; n/a, not available; PHS, periductal hypoechoic sign.
TABLE 3. Test characteristics of the periductal hypoechoic sign and EUS–FNA Accuracy, % (95% CI)
Sensitivity, %
Specificity, %
PPV, %
NPV, %
PHS
79.8% (70.1-86.4)
73.8
85.7
83.8
76.6
EUS-FNA
85.8% (79.1-85.8)
71.4
100
100
78.0
CI, Confidence interval; PPV, positive predictive value; NPV, negative predictive value; PHS, periductal hypoechoic sign.
TABLE 4. Age-adjusted logistic regression model analyzing predictors of malignancy in patients with a dilated pancreatic duct
PD dilation in the body/tail DDS
Coefficient
SE
OR (95% CI)
P value
1.550
0.572
4.71 (1.54-14.46)
.007
0.536
0.95 (0.33-2.73)
.93
0.047
PHS
2.813
0.614
16.7 (5.01-55.44)
!.001
PD size (mm SD)
0.006
0.020
1.0 (0.97-1.05)
.78
SE, standard error; OR, odds ratio; CI, confidence interval; PD, pancreatic duct; DDS, dilation of both bile duct and main pancreatic ducts; PHS, periductal hypoechoic sign; SD, standard deviation.
replace EUS-FNA, but rather would be used as an adjunct. The PHS was noted in 4 of 6 patients with malignancy who had false-negative FNA cytology results. This again suggests that patients with the PHS are at increased risk of having malignancy and that a high suspicion for malignancy should be maintained in these patients despite negative EUS-FNA cytology. Our study had several limitations. First, the analysis was conducted retrospectively by using still EUS images that were not specifically obtained for the purposes of demonstrating the PHS. At no time during the acquisition of these images was the PHS specifically sought. We do not routinely obtain video recordings of EUS procedures and therefore could not conduct this study by using archived video images. We attempted to overcome these limitations by carefully selecting images of the main pancreatic duct that did not show obvious changes of malignancy or chronic pancreatitis. Further, we attempted to make the definition of the PHS precise so it could be objectively applied to images. Patients in whom optimal images could
not be obtained were excluded from the analysis. The retrospective nature of our analysis does not allow us to fully exclude the possibility that the endoscopist interpreting the images was influenced by factors other than periductal changes. Prospective validation of these findings is needed before the PHS can be widely used in clinical practice. Using a retrospective database had the advantage of allowing us to screen a large number of subjects (O400) to include in our study. Second, the final diagnosis was not based on pathology findings but on clinical followup in 42.9% of cases. Third, we were unable to correlate the PHS with histology because of the unavailability of cross sections of dilated pancreatic duct upstream from the level of obstruction. Our explanation of the possible pathophysiology of these findings thus remains speculative. The strengths of our analysis include careful image selection and that images were reviewed by one endoscopist who did not perform the EUS examinations and was blinded to the EUS report and final diagnosis.
254 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 2 : 2010
www.giejournal.org
Lee et al
In summary, we found patients with the PHS to be almost 17 times more likely to harbor malignancy compared with those who did not exhibit this sign. Determination of the PHS does not require additional equipment and can be performed during a routine EUS examination. We recommend that a high level of suspicion for malignancy be maintained in patients with a dilated pancreatic duct who demonstrate the PHS. REFERENCES 1. The SEER Cancer Statistics Review, National Cancer Institute, 19752005. Available at: http://seer.cancer.gov/csr/1975_2005/. Accessed January 5, 2009. 2. Prokesch R, Chow L, Beaulieu C, et al. Isoattenuating pancreatic adenocarcinoma at multi-detector row CT: secondary signs. Radiology 2002;224:764-8. 3. Kamisawa T, Imai M, Yui Chen P, et al. Strategy for differentiating autoimmune pancreatitis from pancreatic cancer. Pancreas 2008;37:62-7. 4. Adamek H, Albert J, Breer H, et al. Pancreatic cancer detection with magnetic resonance cholangiopancreatography and endoscopic retrograde cholangiopancreatography: a prospective controlled study. Lancet 2000;356:190-3. 5. Tanaka S, Nakaizumi A, Ioka T, et al. Main pancreatic duct dilatation: a sign of high risk for pancreatic cancer. Jpn J Clin Oncol 2002;32:407-11. 6. Wiersema MJ, Vilmann P, Giovannini M, et al. Endosonography-guided fine-needle aspiration biopsy: diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087-95. 7. Varadarajulu S, Tamhane A, Eloubeidi M. Yield of EUS-guided FNA of pancreatic masses in the presence or the absence of chronic pancreatitis. Gastrointest Endosc 2005;62:728-36. 8. Fritscher-Ravens A, Brand L, Kno¨fel W, et al. Comparison of endoscopic ultrasound-guided fine needle aspiration for focal pancreatic lesions in
www.giejournal.org
Periductal hypoechoic sign
9.
10.
11.
12.
13.
patients with normal parenchyma and chronic pancreatitis. Am J Gastroenterol 2002;97:2768-75. Saftoiu A, Vilmann P, Gorunescu F, et al. Neural network analysis of dynamic sequences of EUS elastography used for the differential diagnosis of chronic pancreatitis and pancreatic cancer. Gastrointest Endosc 2008;68:1086-94. Inoue K, Ohuchida J, Ohtsuka T, et al. Severe localized stenosis and marked dilatation of the main pancreatic duct are indicators of pancreatic cancer instead of chronic pancreatitis on endoscopic retrograde balloon pancreatography. Gastrointest Endosc 2003;58: 510-5. Hirche TO, Ignee A, Barreiros AP, et al. Indications and limitations of endoscopic ultrasound elastography for evaluation of focal pancreatic lesions. Endoscopy 2008;40:910-7. Menges M, Lerch M, Zeitz M. The double duct sign in patients with malignant and benign pancreatic lesions. Gastrointest Endosc 2000;52:74-7. Agarwal B, Abu-Hamda E, Molke K, et al. Endoscopic ultrasoundguided fine needle aspiration and multidetector spiral CT in the diagnosis of pancreatic cancer. Am J Gastroenterol 2004;99:844-50.
Received February 3, 2009. Accepted August 23, 2009. Current affiliations: Divisions of Gastroenterology (S.-H.L., N.O., R.P., D.K.P., R.C., M.S.S.), Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA, Soonchunhyang University Cheonan Hospital, (Y.H.), Seoul, Korea. Reprint requests: Mandeep S. Sawhney, MBBS, MS, Division of Gastroenterology, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, RABB-ROSE 101, Boston, MA 02215. If you would like to chat with an author of this article, you may contact Dr. Sawhney at [email protected].
Volume 71, No. 2 : 2010 GASTROINTESTINAL ENDOSCOPY 255