- Email: [email protected]
The learning curve for EUS-guided FNA of pancreatic cancer
The learning curve for EUS-guided FNA of pancreatic cancer Howard Mertz, MD, Shiva Gautam, PhD Nashville, Tennessee
Background: EUS-guided FNA is an effective and safe method for tissue diagnosis of pancreatic cancer. However, EUS-guided FNA is technically challenging and requires special training. The number of cases required to become proficient and the technical steps required to achieve proficiency are unknown. Methods: The first 57 EUS-guided FNAs of pancreatic masses by a trained endosonographer were analyzed retrospectively. For 50 masses, the histopathologic diagnosis ultimately was cancer. The sensitivity for the EUS-guided FNA diagnosis of adenocarcinoma was compared in quintiles of 10 procedures. Results: Sensitivity for the diagnosis of pancreatic cancer from first to last quintile were, respectively, 50%, 40%, 70%, 90%, and 80%. There was a significant increase in sensitivity for the first 30 cases after improvement in specific technical skills: shortening of echoendoscope position, scrupulous maintenance of the US view of the needle tip at all times, swift jabbing punctures, sampling multiple areas of the mass in each pass, and performing more than 10 ‘‘jiggles’’ per needle pass. Sensitivity for the diagnosis of pancreatic cancer was greater than 80% for the last 20 of the 57 cases, a level that was maintained for cases 51 through 80. Conclusions: The current American Society for Gastrointestinal Endoscopy guideline of 25 supervised EUS-FNA procedures for the diagnosis of pancreatic adenocarcinoma is reasonable. (Gastrointest Endosc 2004;59:33-7.)
EUS-guided FNA (EUS-FNA) profoundly impacts the evaluation and treatment of patients with pancreatic cancer. EUS allows identification and local staging of the cancer with a high level of sensitivity and specificity. During the same procedure, a tissue diagnosis can be obtained conveniently, safely, and effectively.1 In centers where the volume of EUS procedures is high, EUS-FNA has a sensitivity and specificity, respectively, of at least 80% and 100%.2-6 EUS and EUS-FNA allow rapid allocation of patients with suspected pancreatic lesions into surgical or non-surgical treatment groups. In many cases, surgery can be avoided when it is unlikely to be curative and when the chance of resection is low and the patient is not an optimal candidate for an operation.
Received January 8, 2003. For revision May 5, 2003. Accepted July 23, 2003. Current affiliations: Departments of Medicine, Radiology and Radiologic Sciences, and Biostatistics, Vanderbilt University, Nashville, Tennessee. These data were presented in poster form at the 1999 American Society for Gastrointestinal Endoscopy poster session at Digestive Diseases Week, May 16-19, 1999, Orlando, Florida. Reprint requests: Howard Mertz, MD, 1501 TVC, Vanderbilt University, Nashville, TN 37232. Copyright Ó 2004 by the American Society for Gastrointestinal Endoscopy 0016-5107/$30.00 PII: S0016-5107(03)02028-5 VOLUME 59, NO. 1, 2004
EUS-guided FNA of pancreatic cancers is technically challenging and availability is limited to large centers. The number of training programs is relatively small, and experience is required to become proficient in EUS-FNA of the pancreas. However, the number of procedures needed to become proficient at EUS-FNA of pancreatic cancer is unknown. The steps necessary to acquire this skill have not been described formally. This study describes the learning curve for a single endosonographer without formal training in EUS-FNA of pancreatic cancer, and the technical steps associated with increased diagnostic accuracy. PATIENTS AND METHODS A database of all EUS procedures was maintained by the endosonographer (H.M.). The first 50 EUS-FNA procedures performed by this single examiner (H.M.) that confirmed the diagnosis pancreatic cancer were analyzed retrospectively. To obtain these 50 cases, the first 57 cases in which attempted sampling of a pancreatic mass by EUSFNA were identified. Six patients ultimately were determined to have benign disease based on assessment of cytologic samples obtained by EUS-FNA (n = 6), surgery (n = 5), CT-guided biopsy (n = 1), and a minimum clinical follow-up with symptom resolution of 36 months. One patient with cellular atypia identified by EUS-FNA was lost to follow-up and was excluded from the analysis. Pathology and clinical data were obtained from electronic medical records. When data were incomplete (n = 4), GASTROINTESTINAL ENDOSCOPY
33
H Mertz, S Gautam
Learning curve for EUS-guided FNA of pancreatic cancer
Table 1. Pancreatic cancer clinical features Age Gender (% male) Location (n) Head Uncinate Body/tail Diameter (cm) Time for EUS (min)
67.3 6 1.3 68% 38 6 6 3.4 6 0.2 64.8 6 2.6
follow-up information was obtained through telephone calls to referring physicians or patients. The final diagnosis of pancreatic cancer was established by the following methods: EUS-FNA (n = 33), CT-guided FNA (n = 3), biliary brush cytology (n = 1), surgical biopsy (n = 11), and autopsy (n = 2). No false-positive cytologic or histopathologic diagnosis was observed. Aspirated material was interpreted by one of 3 board certified cytopathologists as follows: adenocarcinoma, atypical cells not diagnostic of adenocarcinoma, and benign tissue. Cytopathologists shared responsibility for the interpretation of cytologic specimens on a rotational basis, so that the number of specimens pertaining to the study interpreted by each cytopathologist was equal. The tumors were most commonly in the head of the pancreas, most patients were men, and the mean patient age was 67.3 years (Table 1). The sensitivity of EUS-FNA for the diagnosis of adenocarcinoma was compared in blocks of 10 procedures in chronologic order with a trend test for proportion by using equally spaced scores. Because time intervals between blocks were unequal, an iso-chi-square test also was performed.7 An additional 30 EUS-FNA procedures for pancreatic cancer (cases 51-80) were reviewed to assess the stability of diagnostic sensitivity. These cases were not included in the statistical analysis. Before and during the study period, EUS-FNA was performed of masses or lymph nodes, including lesions extrinsic (n = 8) and intramural (n = 7) to the upper GI tract. Thus, together with the 6 cases of benign pancreatic disease, the total number of EUS-FNA procedures performed during the study period was 71. The total experience with EUS-FNA was the following: prior, two; first quintile, 18; second quintile, 30; third quintile, 46; fourth quintile, 59; and fifth quintile, 71. The study was approved by the institutional review board of our university.
radial and linear EUS (case experience, n = 542; EUS-FNA experience, n = 163) before performing EUS-FNA of pancreatic cancers. Of the proctored procedures, 32 were specifically for evaluation of the pancreas. The first 10 pancreatic EUS-FNA procedures also were proctored; the last 40 were not. After beginning EUS-FNA, additional training included attendance at informal presentations by expert endosonographers (‘‘Breakfast of champions: EUS,’’ Digestive Diseases Week), telephone consultation with experts, review of published studies, and assessment of practice outcomes by review of cases. EUS procedure A linear array echoendoscope (FG-32UA; Pentax Precision Instruments, Orangeberg, N.J.; and Phillips/Hitachi imaging system, Bothwell, Wash) was used to image the pancreatic tumors and perform EUS-FNA. The imaging frequency was 7.5 MHz. Three standard imaging stations were used in all patients: (1) gastric (views of celiac axis, pancreatic body/tail, splenic artery, and vein), (2) duodenal bulb (superior head of pancreas, portal confluence, and bile duct), (3) duodenal ‘‘C-loop’’ (inferior head and uncinate, major papilla, pancreatic duct, bile duct, superior mesenteric vein, and artery). All vascular structures were confirmed by Doppler US. Pancreatic masses were identified as discrete hypoechoic lesions within the gland. At each imaging station, vascular, lymph node, and gastroduodenal involvement were assessed. After imaging at each station, the optimal station for EUS-FNA was selected. The mass was lined up in the left center of the imaging field to allow appropriate targeting with the fine needle. A cytology technician was present, to smear and stain the aspirated material for immediate review, in the endoscopy room for all procedures. The time required for EUS-FNA was recorded from echoendoscope passage to removal. Fellows were involved in the majority of procedures, which increased procedure time. Procedure time also incorporated the time required for the cytology technician to arrive in the procedure room after being informed of the decision to perform EUS-FNA, the time required to smear and stain specimens on slides, to review the slides, and to offer a preliminary diagnosis, steps that were repeated until diagnostic material was obtained. The times required for these individual steps were not recorded separately. It is estimated that cytologic processing and assessment added 20 minutes or more to the total procedure time for each case.
Prior training
FNA technique
Before performing EUS-FNA of pancreatic masses, the endosonographer (H.M.) performed 15 supervised EUS procedures as a fellow and observed 80 EUS procedures performed by experts in centers with a high volume of EUS procedures (Maurits Wiersema, MD, St. Vincent’s Hospital, Indianapolis, Ind.; Frank Gress, MD, Indiana University, Indianapolis, Ind.), including 5 EUS-FNAs of pancreatic cancer. He was subsequently proctored during the performance of 117 EUS procedures by a radiologist ultrasonographer with experience in the performance of
Under US guidance, a 21-gauge needle with stylette (GIP, Mediglobe Inc, Achenmuble, Germany) was advanced to the mucosal surface of the stomach or duodenum. The stylette was withdrawn a few millimeters to expose the needle tip, and then the needle was advanced into the tumor under US guidance. The stylette was re-advanced to the needle tip to expel intestinal lining cells and then removed. Multiple to-and-fro ‘‘jiggles’’ within the mass were performed while simultaneously applying suction to the needle with a 10-mL glass syringe (made air tight with
34
GASTROINTESTINAL ENDOSCOPY
VOLUME 59, NO. 1, 2004
Learning curve for EUS-guided FNA of pancreatic cancer
H Mertz, S Gautam
silicone gel). After tumor sampling, the needle was withdrawn into its sheath, and the sheath/needle assembly was removed from the echoendoscope. The needle contents were expressed onto a slide, smeared, and stained. Excess material was placed in saline solution to make a cell block for later analysis. A minimum of 3 passes with the fine needle were made in each case. The maximum number of passes was 9. The decision to make more than 3 passes was based on the cytologic findings. Cytology evaluation Smears were stained immediately with H&E and then reviewed by the technician and endosonographer to determine the following: (1) cellularity, (2) presence of normal pancreatic tissue, and (3) presence of abnormal cells. If abnormal cells were identified, they were assessed by the technologist as likely malignant, possibly malignant, or probably not malignant. Additional smears were made when ‘‘possibly malignant’’ or ‘‘probably not malignant’’ cells were identified. If ‘‘possibly malignant’’ cells were identified after examination of specimens obtained by additional needle passes, an attending cytopathologist was summoned to interpret the slides while additional passes were being made with the fine needle. A final interpretation of the cytologic material was made by an experienced, board certified cytolopathologist within 24 hours. Criteria for malignancy were irregular size/shape of cells, irregular size/shape of nuclei, prominent nucleoli, nuclear hyperchromasia, and disorderly groups of cells with nuclear overlap. Specimens with cells that exhibited a smaller number of criteria for malignancy were interpreted as atypical but not diagnostic of malignancy.
RESULTS The sensitivity of EUS-FNA for the diagnosis of adenocarcinoma of the pancreas increased with increasing operator experience (p < 0.05). After the first 20 procedures, there was a reduction in the number of false-negative EUS-FNA results and a corresponding increase in the number of positive cytologic specimens from 40% to 50% to over 70% (Fig. 1). Variation in diagnostic sensitivity over time was not related to which cytopathologist interpreted the specimens. A sensitivity for the diagnosis of adenocarcinoma of 80% to 90% was achieved after 30 procedures. Review of subsequent EUS-FNA experience (cases 51-80) showed that sensitivity for the diagnosis of pancreatic adenocarcinoma remained constant at 87%. The diagnosis of atypia (suspicious but non-diagnostic of malignancy) was relatively constant at 10% to 20% of cases. Lesions for which EUS-FNA was positive for adenocarcinoma were slightly larger (3.6 cm diameter) than those in which the result was atypical or benign (3.0 cm), although the difference was not significant (p = 0.09). All tumors in the body and tail of the pancreas were VOLUME 59, NO. 1, 2004
Figure 1. Yield of 50 EUS-FNA procedures performed by one endosonographer for confirmed pancreatic adenocarcinoma by quintiles of 10 procedures in chronologic order.
diagnosed as adenocarcinoma (6/6), compared with 60% (26/44) of those in the head and uncinate (p = 0.07). Procedure time decreased from 73 minutes (cases 1-10) to 65 minutes (cases 41-50), although this was not significant (p = 0.3). Less time was required to arrive at a positive diagnosis of adenocarcinoma (mean 61.5 minutes) than a diagnosis of atypia (mean 69.4 minutes) or negative for malignancy (mean 73.4 minutes), in part because of the additional needle passes made when non-diagnostic aspirates were obtained, although statistical significance was not reached for these differences. The following technical changes implemented during the first 30 EUS-FNA procedures appeared to improve diagnostic sensitivity: Attempts to achieve a ‘‘short’’ echoendoscope position with minimal tip deflection and torquing of the insertion tube for a straight needle pass. For this reason, the preferred position for EUS-FNA of a lesion in the head of the pancreas was the duodenal bulb instead of the second or third portion. Scrupulous imaging of the needle tip, which can pass through the tumor and out of the imaging plane, resulting in a false impression of its position within the tumor. To identify the tip, the needle was pulled back into the tumor after insertion. Purging of gastroduodenal mucosa and normal pancreatic tissue from the needle with the stylette after tumor puncture. Making swift, jabbing needle advances to more efficiently puncture tumors. GASTROINTESTINAL ENDOSCOPY
35
H Mertz, S Gautam
Learning curve for EUS-guided FNA of pancreatic cancer
Repositioning the needle during each pass to sample multiple areas within a tumor. Multiple (>10) jiggling maneuvers within the tumor on each pass. Expressing the specimen from the needle with the stylette instead of a syringe, to avoid uncontrolled spraying onto the slide and stage area. There was no complication associated with the 50 EUS-FNA procedures. Specifically, intestinal perforation, pancreatitis, hemorrhage, and infection were not identified by review of medical records or reported directly by any patient. No patient was hospitalized because of a complication of the EUSFNA procedure. DISCUSSION Pancreatic adenocarcinoma is the second most common GI malignancy and the fourth leading cause of cancer-related mortality in the United States.8 Diagnosis can be difficult without surgery, but surgery rarely offers a cure. EUS-guided FNA substantially impacts the management of patients with pancreatic adenocarcinoma by preventing unnecessary surgery when the cancer is unresectable and for poor operative candidates. However, EUS-FNA of pancreatic cancer is technically challenging and time consuming, and expertise is required to obtain adequate specimens. Accordingly, training in EUSFNA is essential. Programs that teach EUS-FNA are limited in number and require a significant time commitment on the part of trainees. The number of procedures required to master the technique of pancreatic EUS-FNA is unknown, and the steps to achieve expertise have not been evaluated systematically. Training clearly improves the sensitivity of EUSFNA for the diagnosis of pancreatic cancer. Operator experience was identified as the primary factor in EUS-FNA sensitivity in a study by Harewood et al., 9 although this included only 20 procedures (9 before training, 11 after training). Moreover, increasing experience for the trained endosonographer was associated with a significant increase in the sensitivity for the diagnosis of extra-intestinal malignancies (mostly pancreatic cancer) by EUS-FNA from 79% to 95%.10 The results of the current study confirm that operator experience is a major factor with respect to diagnostic yield, whereas tumor size and location are less important, although this study was not adequately powered to assess the latter variables. Procedure time tended to decrease with experience, but the difference was not statistically significant. Based on the experience of a single endosonographer, it appears that at least 20 procedures are 36
GASTROINTESTINAL ENDOSCOPY
required before sensitivity for the diagnosis of pancreatic adenocarcinoma reaches 70%; and 30 procedures are needed to achieve a sensitivity of 80%, the yield obtained in most centers with a high volume of EUS procedures.2-6 Subsequent to the first 50 procedures, sensitivity for the diagnosis of cancer was constant at near 90%. This is the learning curve for a single endosonographer who had mastered the basic principles of EUS before attempting EUS-FNA of pancreatic masses. Although only the first 10 pancreatic EUS-FNA procedures were supervised, sensitivity continued to improve for the next 20 unsupervised procedures. Continuing education, a cornerstone of the practice of medicine, allows physicians to improve skills through practice, self-monitoring of outcomes, and ongoing education. Although this method may be less optimal than supervised and more prolonged training by an expert, the present study of the experience of one endosonographer indicates that self-education is effective. Conceivably, the learning curve could be accelerated with greater direct supervision and guidance by an endosonographer with experience in EUS-FNA. The endosonographer whose results were analyzed in the present study began pancreatic EUS-FNA after performing 132 proctored procedures, 32 of which were pancreaticobiliary. This is fewer than currently recommended by the American Society for Gastrointestinal Endoscopy (ASGE) (150 total cases, 75 pancreaticobiliary). Accordingly, the learning curve described in the present study may not be characteristic of endosonographers trained in dedicated EUS fellowship programs. However, because of the limited number of slots available in EUS training programs and the time and costs involved in the acquisition of additional training, some endosonographers in practice do perfect their EUS-FNA technique through experience and self-education after basic training in EUS. The current ASGE guideline for training in pancreatic EUS-FNA suggests that ‘‘the trainee be competent to perform diagnostic pancreaticobiliary EUS and have done at least 25 supervised FNA of pancreatic lesions.’’11 The present study demonstrates that a single endosonographer, competent in basic EUS imaging of the pancreas, achieved a sensitivity for the diagnosis of pancreatic cancer after performing 30 EUS-FNA procedures. This experience appears to confirm that the ASGE guideline for training in EUS-FNA of pancreatic masses is correct. REFERENCES 1. Kochman ML. EUS in pancreatic cancer. Gastrointest Endosc 2002;56:S6-S12. VOLUME 59, NO. 1, 2004
Learning curve for EUS-guided FNA of pancreatic cancer
2. Chang KJ, Nguyen P, Erickson RA, Durbin TE, Katz KD. The clinical utility of endoscopic ultrasound-guided fine-needle aspiration in the diagnosis and staging of pancreatic carcinoma. Gastrointest Endosc 1997;45:387-93. 3. Giovannini M, Seitz JF, Monges G, Perrier H, Rabbia I. Fineneedle aspiration cytology guided by endoscopic ultrasonography: results in 141 patients. Endoscopy 1995;27:171-7. 4. Faigel DO, Ginsberg GG, Bentz JS, Gupta PK, Smith DB, Kochman ML. Endoscopic ultrasound-guided real-time fineneedle aspiration biopsy of the pancreas in cancer patients with pancreatic lesions. J Clin Oncol 1997;15:1439-43. 5. Gress FG, Hawes RH, Savides TJ, Ikenberry SO, Cummings O, Kopecky K, et al. Role of EUS in the preoperative staging of pancreatic cancer, a large single-center experience. Gastrointest Endosc 1999;50:786-91. 6. Williams DB, Sahai AV, Aabakken ID, Penman ID, Van Velse A, Webb J, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: a large single centre experience. Gut 1999; 44:720-6.
H Mertz, S Gautam
7. Gautam S, Sampson A, Singh H. Iso-chi-squared testing of 2 3 K ordered tables. Can J Stat 2001;29:609-19. 8. Fernandez-del Castillo C, Jimenez RE. Pancreatic cancer, cystic pancreatic neoplasms and other nonendocrine pancreatic tumors. In: Feldman M, Friedman L, Sleisenger M, editors. Sleisenger and Fordtran’s gastrointestinal and liver disease. 7th ed. Philadelphia: Elsevier; 2002. 9. Harewood GC, Wiersema LM, Halling AC, Keeney GL, Salamao DR, Wiersema MJ. Influence of EUS training and pathology interpretation on accuracy of EUS-guided fine needle aspiration of pancreatic masses. Gastrointest Endosc 2002;55:669-73. 10. Wiersema MJ, Vilmann P, Giovannini M, Chang KJ, Wiersema LM. Endosonography-guided fine-needle aspiration biopsy: diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087-95. 11. ASGE guidelines for credentialing and granting privileges for endoscopic ultrasound. Gastrointest Endosc 2001;54: 811-4.
By the Numbers All of the previously published biostatistical articles are now available in a grouped format through a link, By the Numbers, under Features at www.mosby.com/gie.
VOLUME 59, NO. 1, 2004
GASTROINTESTINAL ENDOSCOPY
37
Background: EUS-guided FNA is an effective and safe method for tissue diagnosis of pancreatic cancer. However, EUS-guided FNA is technically challenging and requires special training. The number of cases required to become proficient and the technical steps required to achieve proficiency are unknown. Methods: The first 57 EUS-guided FNAs of pancreatic masses by a trained endosonographer were analyzed retrospectively. For 50 masses, the histopathologic diagnosis ultimately was cancer. The sensitivity for the EUS-guided FNA diagnosis of adenocarcinoma was compared in quintiles of 10 procedures. Results: Sensitivity for the diagnosis of pancreatic cancer from first to last quintile were, respectively, 50%, 40%, 70%, 90%, and 80%. There was a significant increase in sensitivity for the first 30 cases after improvement in specific technical skills: shortening of echoendoscope position, scrupulous maintenance of the US view of the needle tip at all times, swift jabbing punctures, sampling multiple areas of the mass in each pass, and performing more than 10 ‘‘jiggles’’ per needle pass. Sensitivity for the diagnosis of pancreatic cancer was greater than 80% for the last 20 of the 57 cases, a level that was maintained for cases 51 through 80. Conclusions: The current American Society for Gastrointestinal Endoscopy guideline of 25 supervised EUS-FNA procedures for the diagnosis of pancreatic adenocarcinoma is reasonable. (Gastrointest Endosc 2004;59:33-7.)
EUS-guided FNA (EUS-FNA) profoundly impacts the evaluation and treatment of patients with pancreatic cancer. EUS allows identification and local staging of the cancer with a high level of sensitivity and specificity. During the same procedure, a tissue diagnosis can be obtained conveniently, safely, and effectively.1 In centers where the volume of EUS procedures is high, EUS-FNA has a sensitivity and specificity, respectively, of at least 80% and 100%.2-6 EUS and EUS-FNA allow rapid allocation of patients with suspected pancreatic lesions into surgical or non-surgical treatment groups. In many cases, surgery can be avoided when it is unlikely to be curative and when the chance of resection is low and the patient is not an optimal candidate for an operation.
Received January 8, 2003. For revision May 5, 2003. Accepted July 23, 2003. Current affiliations: Departments of Medicine, Radiology and Radiologic Sciences, and Biostatistics, Vanderbilt University, Nashville, Tennessee. These data were presented in poster form at the 1999 American Society for Gastrointestinal Endoscopy poster session at Digestive Diseases Week, May 16-19, 1999, Orlando, Florida. Reprint requests: Howard Mertz, MD, 1501 TVC, Vanderbilt University, Nashville, TN 37232. Copyright Ó 2004 by the American Society for Gastrointestinal Endoscopy 0016-5107/$30.00 PII: S0016-5107(03)02028-5 VOLUME 59, NO. 1, 2004
EUS-guided FNA of pancreatic cancers is technically challenging and availability is limited to large centers. The number of training programs is relatively small, and experience is required to become proficient in EUS-FNA of the pancreas. However, the number of procedures needed to become proficient at EUS-FNA of pancreatic cancer is unknown. The steps necessary to acquire this skill have not been described formally. This study describes the learning curve for a single endosonographer without formal training in EUS-FNA of pancreatic cancer, and the technical steps associated with increased diagnostic accuracy. PATIENTS AND METHODS A database of all EUS procedures was maintained by the endosonographer (H.M.). The first 50 EUS-FNA procedures performed by this single examiner (H.M.) that confirmed the diagnosis pancreatic cancer were analyzed retrospectively. To obtain these 50 cases, the first 57 cases in which attempted sampling of a pancreatic mass by EUSFNA were identified. Six patients ultimately were determined to have benign disease based on assessment of cytologic samples obtained by EUS-FNA (n = 6), surgery (n = 5), CT-guided biopsy (n = 1), and a minimum clinical follow-up with symptom resolution of 36 months. One patient with cellular atypia identified by EUS-FNA was lost to follow-up and was excluded from the analysis. Pathology and clinical data were obtained from electronic medical records. When data were incomplete (n = 4), GASTROINTESTINAL ENDOSCOPY
33
H Mertz, S Gautam
Learning curve for EUS-guided FNA of pancreatic cancer
Table 1. Pancreatic cancer clinical features Age Gender (% male) Location (n) Head Uncinate Body/tail Diameter (cm) Time for EUS (min)
67.3 6 1.3 68% 38 6 6 3.4 6 0.2 64.8 6 2.6
follow-up information was obtained through telephone calls to referring physicians or patients. The final diagnosis of pancreatic cancer was established by the following methods: EUS-FNA (n = 33), CT-guided FNA (n = 3), biliary brush cytology (n = 1), surgical biopsy (n = 11), and autopsy (n = 2). No false-positive cytologic or histopathologic diagnosis was observed. Aspirated material was interpreted by one of 3 board certified cytopathologists as follows: adenocarcinoma, atypical cells not diagnostic of adenocarcinoma, and benign tissue. Cytopathologists shared responsibility for the interpretation of cytologic specimens on a rotational basis, so that the number of specimens pertaining to the study interpreted by each cytopathologist was equal. The tumors were most commonly in the head of the pancreas, most patients were men, and the mean patient age was 67.3 years (Table 1). The sensitivity of EUS-FNA for the diagnosis of adenocarcinoma was compared in blocks of 10 procedures in chronologic order with a trend test for proportion by using equally spaced scores. Because time intervals between blocks were unequal, an iso-chi-square test also was performed.7 An additional 30 EUS-FNA procedures for pancreatic cancer (cases 51-80) were reviewed to assess the stability of diagnostic sensitivity. These cases were not included in the statistical analysis. Before and during the study period, EUS-FNA was performed of masses or lymph nodes, including lesions extrinsic (n = 8) and intramural (n = 7) to the upper GI tract. Thus, together with the 6 cases of benign pancreatic disease, the total number of EUS-FNA procedures performed during the study period was 71. The total experience with EUS-FNA was the following: prior, two; first quintile, 18; second quintile, 30; third quintile, 46; fourth quintile, 59; and fifth quintile, 71. The study was approved by the institutional review board of our university.
radial and linear EUS (case experience, n = 542; EUS-FNA experience, n = 163) before performing EUS-FNA of pancreatic cancers. Of the proctored procedures, 32 were specifically for evaluation of the pancreas. The first 10 pancreatic EUS-FNA procedures also were proctored; the last 40 were not. After beginning EUS-FNA, additional training included attendance at informal presentations by expert endosonographers (‘‘Breakfast of champions: EUS,’’ Digestive Diseases Week), telephone consultation with experts, review of published studies, and assessment of practice outcomes by review of cases. EUS procedure A linear array echoendoscope (FG-32UA; Pentax Precision Instruments, Orangeberg, N.J.; and Phillips/Hitachi imaging system, Bothwell, Wash) was used to image the pancreatic tumors and perform EUS-FNA. The imaging frequency was 7.5 MHz. Three standard imaging stations were used in all patients: (1) gastric (views of celiac axis, pancreatic body/tail, splenic artery, and vein), (2) duodenal bulb (superior head of pancreas, portal confluence, and bile duct), (3) duodenal ‘‘C-loop’’ (inferior head and uncinate, major papilla, pancreatic duct, bile duct, superior mesenteric vein, and artery). All vascular structures were confirmed by Doppler US. Pancreatic masses were identified as discrete hypoechoic lesions within the gland. At each imaging station, vascular, lymph node, and gastroduodenal involvement were assessed. After imaging at each station, the optimal station for EUS-FNA was selected. The mass was lined up in the left center of the imaging field to allow appropriate targeting with the fine needle. A cytology technician was present, to smear and stain the aspirated material for immediate review, in the endoscopy room for all procedures. The time required for EUS-FNA was recorded from echoendoscope passage to removal. Fellows were involved in the majority of procedures, which increased procedure time. Procedure time also incorporated the time required for the cytology technician to arrive in the procedure room after being informed of the decision to perform EUS-FNA, the time required to smear and stain specimens on slides, to review the slides, and to offer a preliminary diagnosis, steps that were repeated until diagnostic material was obtained. The times required for these individual steps were not recorded separately. It is estimated that cytologic processing and assessment added 20 minutes or more to the total procedure time for each case.
Prior training
FNA technique
Before performing EUS-FNA of pancreatic masses, the endosonographer (H.M.) performed 15 supervised EUS procedures as a fellow and observed 80 EUS procedures performed by experts in centers with a high volume of EUS procedures (Maurits Wiersema, MD, St. Vincent’s Hospital, Indianapolis, Ind.; Frank Gress, MD, Indiana University, Indianapolis, Ind.), including 5 EUS-FNAs of pancreatic cancer. He was subsequently proctored during the performance of 117 EUS procedures by a radiologist ultrasonographer with experience in the performance of
Under US guidance, a 21-gauge needle with stylette (GIP, Mediglobe Inc, Achenmuble, Germany) was advanced to the mucosal surface of the stomach or duodenum. The stylette was withdrawn a few millimeters to expose the needle tip, and then the needle was advanced into the tumor under US guidance. The stylette was re-advanced to the needle tip to expel intestinal lining cells and then removed. Multiple to-and-fro ‘‘jiggles’’ within the mass were performed while simultaneously applying suction to the needle with a 10-mL glass syringe (made air tight with
34
GASTROINTESTINAL ENDOSCOPY
VOLUME 59, NO. 1, 2004
Learning curve for EUS-guided FNA of pancreatic cancer
H Mertz, S Gautam
silicone gel). After tumor sampling, the needle was withdrawn into its sheath, and the sheath/needle assembly was removed from the echoendoscope. The needle contents were expressed onto a slide, smeared, and stained. Excess material was placed in saline solution to make a cell block for later analysis. A minimum of 3 passes with the fine needle were made in each case. The maximum number of passes was 9. The decision to make more than 3 passes was based on the cytologic findings. Cytology evaluation Smears were stained immediately with H&E and then reviewed by the technician and endosonographer to determine the following: (1) cellularity, (2) presence of normal pancreatic tissue, and (3) presence of abnormal cells. If abnormal cells were identified, they were assessed by the technologist as likely malignant, possibly malignant, or probably not malignant. Additional smears were made when ‘‘possibly malignant’’ or ‘‘probably not malignant’’ cells were identified. If ‘‘possibly malignant’’ cells were identified after examination of specimens obtained by additional needle passes, an attending cytopathologist was summoned to interpret the slides while additional passes were being made with the fine needle. A final interpretation of the cytologic material was made by an experienced, board certified cytolopathologist within 24 hours. Criteria for malignancy were irregular size/shape of cells, irregular size/shape of nuclei, prominent nucleoli, nuclear hyperchromasia, and disorderly groups of cells with nuclear overlap. Specimens with cells that exhibited a smaller number of criteria for malignancy were interpreted as atypical but not diagnostic of malignancy.
RESULTS The sensitivity of EUS-FNA for the diagnosis of adenocarcinoma of the pancreas increased with increasing operator experience (p < 0.05). After the first 20 procedures, there was a reduction in the number of false-negative EUS-FNA results and a corresponding increase in the number of positive cytologic specimens from 40% to 50% to over 70% (Fig. 1). Variation in diagnostic sensitivity over time was not related to which cytopathologist interpreted the specimens. A sensitivity for the diagnosis of adenocarcinoma of 80% to 90% was achieved after 30 procedures. Review of subsequent EUS-FNA experience (cases 51-80) showed that sensitivity for the diagnosis of pancreatic adenocarcinoma remained constant at 87%. The diagnosis of atypia (suspicious but non-diagnostic of malignancy) was relatively constant at 10% to 20% of cases. Lesions for which EUS-FNA was positive for adenocarcinoma were slightly larger (3.6 cm diameter) than those in which the result was atypical or benign (3.0 cm), although the difference was not significant (p = 0.09). All tumors in the body and tail of the pancreas were VOLUME 59, NO. 1, 2004
Figure 1. Yield of 50 EUS-FNA procedures performed by one endosonographer for confirmed pancreatic adenocarcinoma by quintiles of 10 procedures in chronologic order.
diagnosed as adenocarcinoma (6/6), compared with 60% (26/44) of those in the head and uncinate (p = 0.07). Procedure time decreased from 73 minutes (cases 1-10) to 65 minutes (cases 41-50), although this was not significant (p = 0.3). Less time was required to arrive at a positive diagnosis of adenocarcinoma (mean 61.5 minutes) than a diagnosis of atypia (mean 69.4 minutes) or negative for malignancy (mean 73.4 minutes), in part because of the additional needle passes made when non-diagnostic aspirates were obtained, although statistical significance was not reached for these differences. The following technical changes implemented during the first 30 EUS-FNA procedures appeared to improve diagnostic sensitivity: Attempts to achieve a ‘‘short’’ echoendoscope position with minimal tip deflection and torquing of the insertion tube for a straight needle pass. For this reason, the preferred position for EUS-FNA of a lesion in the head of the pancreas was the duodenal bulb instead of the second or third portion. Scrupulous imaging of the needle tip, which can pass through the tumor and out of the imaging plane, resulting in a false impression of its position within the tumor. To identify the tip, the needle was pulled back into the tumor after insertion. Purging of gastroduodenal mucosa and normal pancreatic tissue from the needle with the stylette after tumor puncture. Making swift, jabbing needle advances to more efficiently puncture tumors. GASTROINTESTINAL ENDOSCOPY
35
H Mertz, S Gautam
Learning curve for EUS-guided FNA of pancreatic cancer
Repositioning the needle during each pass to sample multiple areas within a tumor. Multiple (>10) jiggling maneuvers within the tumor on each pass. Expressing the specimen from the needle with the stylette instead of a syringe, to avoid uncontrolled spraying onto the slide and stage area. There was no complication associated with the 50 EUS-FNA procedures. Specifically, intestinal perforation, pancreatitis, hemorrhage, and infection were not identified by review of medical records or reported directly by any patient. No patient was hospitalized because of a complication of the EUSFNA procedure. DISCUSSION Pancreatic adenocarcinoma is the second most common GI malignancy and the fourth leading cause of cancer-related mortality in the United States.8 Diagnosis can be difficult without surgery, but surgery rarely offers a cure. EUS-guided FNA substantially impacts the management of patients with pancreatic adenocarcinoma by preventing unnecessary surgery when the cancer is unresectable and for poor operative candidates. However, EUS-FNA of pancreatic cancer is technically challenging and time consuming, and expertise is required to obtain adequate specimens. Accordingly, training in EUSFNA is essential. Programs that teach EUS-FNA are limited in number and require a significant time commitment on the part of trainees. The number of procedures required to master the technique of pancreatic EUS-FNA is unknown, and the steps to achieve expertise have not been evaluated systematically. Training clearly improves the sensitivity of EUSFNA for the diagnosis of pancreatic cancer. Operator experience was identified as the primary factor in EUS-FNA sensitivity in a study by Harewood et al., 9 although this included only 20 procedures (9 before training, 11 after training). Moreover, increasing experience for the trained endosonographer was associated with a significant increase in the sensitivity for the diagnosis of extra-intestinal malignancies (mostly pancreatic cancer) by EUS-FNA from 79% to 95%.10 The results of the current study confirm that operator experience is a major factor with respect to diagnostic yield, whereas tumor size and location are less important, although this study was not adequately powered to assess the latter variables. Procedure time tended to decrease with experience, but the difference was not statistically significant. Based on the experience of a single endosonographer, it appears that at least 20 procedures are 36
GASTROINTESTINAL ENDOSCOPY
required before sensitivity for the diagnosis of pancreatic adenocarcinoma reaches 70%; and 30 procedures are needed to achieve a sensitivity of 80%, the yield obtained in most centers with a high volume of EUS procedures.2-6 Subsequent to the first 50 procedures, sensitivity for the diagnosis of cancer was constant at near 90%. This is the learning curve for a single endosonographer who had mastered the basic principles of EUS before attempting EUS-FNA of pancreatic masses. Although only the first 10 pancreatic EUS-FNA procedures were supervised, sensitivity continued to improve for the next 20 unsupervised procedures. Continuing education, a cornerstone of the practice of medicine, allows physicians to improve skills through practice, self-monitoring of outcomes, and ongoing education. Although this method may be less optimal than supervised and more prolonged training by an expert, the present study of the experience of one endosonographer indicates that self-education is effective. Conceivably, the learning curve could be accelerated with greater direct supervision and guidance by an endosonographer with experience in EUS-FNA. The endosonographer whose results were analyzed in the present study began pancreatic EUS-FNA after performing 132 proctored procedures, 32 of which were pancreaticobiliary. This is fewer than currently recommended by the American Society for Gastrointestinal Endoscopy (ASGE) (150 total cases, 75 pancreaticobiliary). Accordingly, the learning curve described in the present study may not be characteristic of endosonographers trained in dedicated EUS fellowship programs. However, because of the limited number of slots available in EUS training programs and the time and costs involved in the acquisition of additional training, some endosonographers in practice do perfect their EUS-FNA technique through experience and self-education after basic training in EUS. The current ASGE guideline for training in pancreatic EUS-FNA suggests that ‘‘the trainee be competent to perform diagnostic pancreaticobiliary EUS and have done at least 25 supervised FNA of pancreatic lesions.’’11 The present study demonstrates that a single endosonographer, competent in basic EUS imaging of the pancreas, achieved a sensitivity for the diagnosis of pancreatic cancer after performing 30 EUS-FNA procedures. This experience appears to confirm that the ASGE guideline for training in EUS-FNA of pancreatic masses is correct. REFERENCES 1. Kochman ML. EUS in pancreatic cancer. Gastrointest Endosc 2002;56:S6-S12. VOLUME 59, NO. 1, 2004
Learning curve for EUS-guided FNA of pancreatic cancer
2. Chang KJ, Nguyen P, Erickson RA, Durbin TE, Katz KD. The clinical utility of endoscopic ultrasound-guided fine-needle aspiration in the diagnosis and staging of pancreatic carcinoma. Gastrointest Endosc 1997;45:387-93. 3. Giovannini M, Seitz JF, Monges G, Perrier H, Rabbia I. Fineneedle aspiration cytology guided by endoscopic ultrasonography: results in 141 patients. Endoscopy 1995;27:171-7. 4. Faigel DO, Ginsberg GG, Bentz JS, Gupta PK, Smith DB, Kochman ML. Endoscopic ultrasound-guided real-time fineneedle aspiration biopsy of the pancreas in cancer patients with pancreatic lesions. J Clin Oncol 1997;15:1439-43. 5. Gress FG, Hawes RH, Savides TJ, Ikenberry SO, Cummings O, Kopecky K, et al. Role of EUS in the preoperative staging of pancreatic cancer, a large single-center experience. Gastrointest Endosc 1999;50:786-91. 6. Williams DB, Sahai AV, Aabakken ID, Penman ID, Van Velse A, Webb J, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: a large single centre experience. Gut 1999; 44:720-6.
H Mertz, S Gautam
7. Gautam S, Sampson A, Singh H. Iso-chi-squared testing of 2 3 K ordered tables. Can J Stat 2001;29:609-19. 8. Fernandez-del Castillo C, Jimenez RE. Pancreatic cancer, cystic pancreatic neoplasms and other nonendocrine pancreatic tumors. In: Feldman M, Friedman L, Sleisenger M, editors. Sleisenger and Fordtran’s gastrointestinal and liver disease. 7th ed. Philadelphia: Elsevier; 2002. 9. Harewood GC, Wiersema LM, Halling AC, Keeney GL, Salamao DR, Wiersema MJ. Influence of EUS training and pathology interpretation on accuracy of EUS-guided fine needle aspiration of pancreatic masses. Gastrointest Endosc 2002;55:669-73. 10. Wiersema MJ, Vilmann P, Giovannini M, Chang KJ, Wiersema LM. Endosonography-guided fine-needle aspiration biopsy: diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087-95. 11. ASGE guidelines for credentialing and granting privileges for endoscopic ultrasound. Gastrointest Endosc 2001;54: 811-4.
By the Numbers All of the previously published biostatistical articles are now available in a grouped format through a link, By the Numbers, under Features at www.mosby.com/gie.
VOLUME 59, NO. 1, 2004
GASTROINTESTINAL ENDOSCOPY
37