- Email: [email protected]
Endoscopic ultrasound–guided fine needle aspiration cytology of solid liver lesions: a large single-center experience
THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2003 by Am. Coll. of Gastroenterology Published by Elsevier Inc.
Vol. 98, No. 9, 2003 ISSN 0002-9270/03/$30.00 doi:10.1016/S0002-9270(03)00549-5
Endoscopic Ultrasound–Guided Fine Needle Aspiration Cytology of Solid Liver Lesions: A Large Single-Center Experience John DeWitt, M.D., Julia LeBlanc, M.D., Lee McHenry, M.D., Dan Ciaccia, M.D., Tom Imperiale, M.D., John Chappo, B.S., C.T., Harvey Cramer, M.D., Kathy McGreevy, R.N., Melissa Chriswell, R.N., and Stuart Sherman, M.D. Department of Gastroenterology and Hepatology and Department of Pathology and Laboratory Medicine, Indiana University Medical Center, Indianapolis, Indiana
OBJECTIVES: The aim of this study was to report the sensitivity, cytological diagnoses, endoscopic ultrasound (EUS) features, complications, clinical impact, and long term follow-up of a large single-center experience with endoscopic ultrasound– guided fine needle aspiration (EUS-FNA) of benign and malignant solid liver lesions. METHODS: A database of cytologic specimens from EUSFNA was reviewed to identify all hepatic lesions aspirated between January, 1997, and July, 2002. Procedural indications, prior radiographic data, patient demographics, EUS examination results, complications, and follow-up data were obtained and recorded. RESULTS: EUS-FNA of 77 liver lesions in 77 patients was performed without complications. Of these 77 lesions, 45 (58%) were diagnostic for malignancy, 25 (33%) were benign, and seven (9%) were nondiagnostic. A total of 22 lesions were confirmed as negative for malignancy by follow-up (mean 762 days, range 512–1556 days) or intraoperative examination; however, seven lesions could not be classified as benign or malignant. Depending on the status of the seven unclassified lesions, sensitivity of EUS-FNA for the diagnosis of malignancy ranged from 82 to 94%. When compared with benign lesions, EUS features predictive of malignant hepatic masses were the presence of regular outer margins (60% vs 27%; p ⫽ 0.02) and the detection of two or more lesions (38% vs 9%; p ⫽ 0.03). Of the 42 patients with malignancy identified by EUS-FNA and other available imaging records, EUS detected the malignancy in 41% of patients with previously negative examinations. For the 45 subjects with cytology positive for malignancy, EUSFNA changed management in 86% of subjects. CONCLUSION: EUS-FNA of the liver is a safe and sensitive procedure that can have a significant impact on patient management. Prospective studies comparing the accuracy and complication rate of EUS-FNA and percutaneous fine needle aspiration (P-FNA) for the diagnosis of liver tumors are needed. (Am J Gastroenterol 2003;98:1976 –1981. © 2003 by Am. Coll. of Gastroenterology)
INTRODUCTION Identification of hepatic metastases from systemic malignancy usually implies inoperable disease and a poor prognosis. Furthermore, the diagnosis of unresectable malignancy may obviate the need for surgery along with its inherent morbidity and mortality. The diagnosis of liver masses is traditionally accomplished by percutaneous fine needle aspiration (P-FNA) under fluoroscopic, computed tomographic, or ultrasonographic guidance (1–3). Although rare, complications from P-FNA may include severe bleeding and implantation metastases in up to 3% of patients (3– 8). Endoscopic ultrasound (EUS)– guided fine needle aspiration (EUS-FNA) has been established as a safe and accurate method for the diagnosis of benign or malignant lymphadenopathy and for intramural and extramural neoplasms (9, 10). The role of EUS-FNA in the diagnosis and staging of liver metastases or masses, however, is limited to a small single-center series (11), a large retrospective international survey (12), and several case series (13, 14). Furthermore, the sensitivity of EUS-FNA and endosonographic features of benign and malignant hepatic lesions have not been described. The aim of this study was to report the sensitivity, cytological diagnoses, EUS features, complications, and clinical impact of a large single-center experience with EUS-FNA of solid liver lesions.
MATERIALS AND METHODS From an existing EUS-FNA cytology database, we identified all patients between January, 1997, and July, 2002, in whom EUS-FNA of the liver was performed. The database used had been maintained and prospectively updated by one cytotechnologist (J.C.). Patients with hilar cholangiocarcinomas or cystic hepatic lesions were excluded from the analysis. Staging EUS examinations for known or suspected malignancy were performed initially with an Olympus GFUM20 or GFUM-130 radial echoendoscope (Olympus,
AJG – September, 2003
Melville NY). The right lobe of the liver was surveyed from the duodenum and distal stomach, whereas the left lobe was imaged from the proximal and mid-stomach. Endoscope rotation was used as necessary to visualize as much of the liver parenchyma as possible. EUS-FNA was performed using the Pentax 32-UA, Pentax 36-UX (Pentax Precision Instruments, Orangeburg, NY), Olympus GF-UC30P, or Olympus GF-UC140P curvilinear array echoendoscope. EUS-FNA was performed using a 22-gauge, 8-cm WilsonCook EUSN-1, EUSN-2, or EUSN-3 needle (Wilson-Cook Medical, Winston-Salem, NC) by one of four physicians (J.D., D.C., L.M., or J.L.). Doppler color angiography was used to ensure the absence of intervening vascular structures along the anticipated needle path. Depending on the amount of blood anticipated during tissue sampling, full, partial, or no suction was applied at the discretion of the endoscopist, cytotechnologist, or cytopathologist. A cytotechnologist or cytopathologist was available on-site for preliminary interpretations on all procedures. Samples aspirated were expressed onto a glass slide and two smear preparations were made. One slide was air-dried and stained with a modified Giemsa stain for rapid on-site interpretation. The other slide was alcohol-fixed and stained by the Papanicolaou method. EUS-FNA was repeated until a definitive diagnosis was made or the endoscopist believed that further sampling would not likely increase yield. Within several days of each EUS examination, a final cytological diagnosis was rendered by a staff cytopathologist. A true positive EUS-FNA for malignancy is defined as unequivocal cytological evidence of malignancy. A false negative aspirate is a nondiagnostic or benign specimen, which is subsequently found to be malignant by percutaneous FNA or intraoperative findings. A true negative EUSFNA for malignancy is defined as benign or nondiagnostic samples that are either confirmed as benign by alternative sampling, intraoperative examination, or appropriate clinical follow-up. Specimens with benign hepatocytes without evidence of atypical cells are categorized as benign aspirates. Aspirates from EUS-FNA that are interpreted as nondiagnostic, highly suspicious, suspicious, or atypical for malignancy are considered as negative for malignancy. Medical records of enrolled subjects were reviewed and procedural indications, prior radiographic data, patient demographics, EUS test results, clinical outcomes, procedural complications, and follow-up data were abstracted. When multiple liver lesions were noted, the endosonographic description of only the aspirated lesion was recorded. When charts were incomplete, written informed consent was sent and follow-up telephone calls were made to the subject, closest relative of the deceased, or referring physician for clarification or any required further information. The potential clinical impact of EUS-FNA of a liver metastasis or malignancy is defined as results that met the following criteria: 1) avoided surgery (results precluded resectable malignancy); 2) made diagnosis (results provided initial diagnosis of malignancy); 3) upstaged tumor (results up-
EUS-FNA Cytology of Solid Liver Lesions
1977
Table 1. Indications for EUS Examinations by Cytology Results of EUS-FNA of the Liver Cytology Result Indication Abnormal ERCP Pancreatic mass on CT Staging of known cancer Liver mass on CT Other abnormal CT findings Suspected recurrent cancer Chronic abdominal pain Other Total
All Malignant Benign/ND (n ⫽ 77) (n ⫽ 45) (n ⫽ 32) 21 (27) 19 (25) 10 (13) 9 (12) 4 (5) 2 (3) 2 (3) 10 (12) 77
11 (24) 12 (27) 7 (16) 9 (20) 1 (2) 1 (2) 0 (0) 4 (9) 45
10 (31) 7 (22) 3 (9) 0 (0) 3 (9) 1 (3) 2 (7) 6 (19) 32
ND ⫽ nondiagnostic
staged primary malignancy); 4) incurred no change in management (results did not change patient treatment). This study was approved by the Institutional Review Board at our institution. Statistical Analysis Assuming that the EUS-FNA diagnosis of malignancy is a true positive, sensitivity was calculated as the proportion of patients with cancer in whom EUS-FNA was positive for malignancy. For analysis, continuous variables were described as means and standard deviations, and dichotomous variables were expressed as simple proportions with or without 95% confidence limits. Student’s t test and 2 or Fisher’s exact tests were used to test for differences in comparisons between continuous and dichotomous variables, respectively.
RESULTS In total, EUS-FNA was performed on 77 different liver lesions in 77 patients (42 male and 35 female, 69 white and eight African American, with a mean age of 63 ⫾ 11 yr). Follow-up was available in all patients. Indications for EUS Examinations The indications for the 77 EUS procedures are recorded in Table 1. The two most common indications were abnormal ERCP findings, the majority of which were strictures, and a pancreatic mass on CT scan, which accounted for 27% and 25% of the procedures performed, respectively. Ten (13%) of 77 procedures were done for staging of a recently diagnosed malignancy (pancreas in five, esophageal in four, and lung in one). Of the 45 malignancies diagnosed by EUSFNA, the most common indications for the EUS were a CT finding of a pancreatic mass (27%) and abnormal ERCP findings (24%). Characteristics of EUS Examination EUS-FNA was performed in 77 patients (mean 3.4 ⫾ 1.8 passes, range 1– 8) into the left lobe (n ⫽ 66; 86%) and right lobe (n ⫽ 11; 14%) of the liver without procedural compli-
1978
DeWitt et al.
AJG – Vol. 98, No. 9, 2003
Table 2. Characteristics of EUS Examination All* Malignant† Benign‡ (n ⫽ 77) (n ⫽ 48) (n ⫽ 22)
p-value§
Site of FNA Left lobe 66 (86) 41 (85) 20 (91) Right lobe 11 (14) 7 (15) 2 (9) No. of passes 3.4 ⫾ 1.8 3.5 ⫾ 1.5 3.2 ⫾ 1.5 Range 1–8 1–8 1–8 Echogenicity Hypoechoic 52 (68) 33 (69) 13 (59) Hyperechoic 23 (30) 13 (27) 9 (41) Both 2 (2) 2 (4) 0 (0) Margins Regular 39 (51) 29 (60) 6 (27) Irregular 38 (49) 19 (40) 16 (73) Size (mm) 16.0 ⫾ 10.8 15.6 ⫾ 7.7 16.7 ⫾ 12.0 Range 3–40 3–35 4–40 No. of lesions seen 1 57 (74) 30 (62) 20 (91) ⬎1 20 (26) 18 (38) 2 (9)
0.57 0.44 0.36
0.02 0.70 0.03
* Includes seven patients unable to be classified as benign or malignant. † Includes 45 positive and three false negative EUS-FNA results. ‡ By follow-up or intraoperative findings. § Between malignant and benign lesions.
cations (upper 95% CI ⫽ 4.7%; see Table 2). A total of 45 (58%) aspirates were diagnostic for malignancy. Three false negatives were later discovered by intraoperative findings (n ⫽ 2) or P-FNA (n ⫽ 1). One patient with pancreatic adenocarcinoma (as confirmed by EUS-FNA of the pancreas) had benign cytology of a 6-mm left lobe mass that intraoperatively was confirmed as metastatic adenocarcinoma. A second patient with pancreatic adenocarcinoma (as confirmed by EUS-FNA of the pancreas) had benign cytology of a 6-mm right lobe mass, which was later confirmed by percutaneous ultrasound-guided biopsy as metastatic adenocarcinoma. The third patient with a false negative EUS-FNA had a 29-mm left lobe mass that was benign by EUS-FNA. Intraoperative examination, however, confirmed hepatocellular carcinoma. Overall, 48 of 77 (62%) of the liver masses were malignant. Among the remaining 29 subjects with nonmalignant aspirates, in 22 (76%) the masses were considered benign by clinical follow-up (n ⫽ 18; mean 762 days; range: 512–1556 days) or intraoperative evaluation (n ⫽ 4). Seven (24%) of the subjects with nonmalignant masses with pancreatic adenocarcinoma (as confirmed by EUS-FNA of the pancreas; n ⫽ 6) or lung cancer (n ⫽ 1) died (mean 154 days, range 14 – 424 days) without follow-up imaging, biopsy of the liver, or autopsy and therefore cannot be classified as benign or malignant. The 45 diagnostic aspirates for malignancy are considered true positives (Fig. 1). Assuming that the test results for these seven patients were all true negatives or all false negatives, the sensitivity of EUS-FNA for the diagnosis of malignancy would range from 82 to 94%. In comparison between benign and malignant lesions detected by EUS-FNA (Table 2), no statistically significant difference was found between the liver lobe aspirated (p ⫽ 0.57), number of passes (p ⫽ 0.44), lesion size (p ⫽ 0.70),
Figure 1. Sensitivity of EUS-FNA of the liver for the diagnosis of malignancy.
or echotexture (p ⫽ 0.36). Malignant masses, however, were more often accompanied by the presence of multiple hepatic lesions detected on EUS (38% vs 9%; p ⫽ 0.03; 2 analysis) and to have regular margins (60% vs 27%; p ⫽ 0.02; 2 analysis). Diagnoses from EUS-FNA of the Liver The final diagnoses for the 45 malignant liver aspirates are recorded in Table 3. Of these, 44 (98%) were metastatic and one (2%) was a hepatocellular carcinoma. The most common diagnosis was metastatic adenocarcinoma from the pancreas, which accounted for 34 of 45 (76%). The next most frequent diagnosis was metastatic neuroendocrine tumor from the pancreas (n ⫽ 5; 11%). Of the 32 nonmalignant aspirates, 25 (33%) were cytologically benign and seven (9%) were nondiagnostic. One (4%) benign aspirate demonstrated cytological features consistent with an abscess. Previous Imaging of Malignancy Diagnosed by EUS-FNA Prior radiographic imaging reports were available for 42 of 45 (93%) malignant hepatic masses diagnosed by EUSTable 3. Diagnoses of Malignant Aspirates of the Liver Diagnosis
No. (%)
Pancreatic adenocarcinoma Pancreatic neuroendocrine tumor Renal cell carcinoma Gallbladder adenocarcinoma Colon cancer Hepatocellular carcinoma Esophageal adenocarcinoma Total
34 (76) 5 (11) 2 (5) 1 (2) 1 (2) 1 (2) 1 (2) 45
AJG – September, 2003
FNA. All imaging had been performed within 45 days of the EUS examination. Of these 42 lesions, EUS-FNA detected malignancy in 17 (41%) when prior imaging of the liver by CT alone (n ⫽ 13), transabdominal ultrasound (US) alone (n ⫽ 1), or both (n ⫽ 3) were normal. For the 32 patients in whom CT alone was performed, EUS-FNA initially detected a malignancy in 13 (41%). For the nine patients in whom CT and US were both performed, EUS-FNA initially detected malignancy in three (33%). The mean size of the 17 malignant liver lesions detected initially by EUS-FNA was 12.6 mm (range 3–26 mm). Of these 17, six (35%) were ⬍1 cm in diameter. Clinical Impact of EUS-FNA of the Liver For the 45 patients with malignant cytology, EUS-FNA had (by protocol definition) a clinical impact in all patients. EUS-FNA changed management, however, in only 38 (84%) of these patients, because in seven cases (16%) tumor was upstaged but did not change overall patient management. EUS-FNA provided the initial diagnosis of malignancy in 35 patients (78%) and avoided surgery in 12 (27%). Of the 35 patients in whom EUS-FNA made the initial diagnosis of malignancy, 34 (97%) had metastases and one (3%) had a hepatocellular carcinoma. In 25 patients (55%), the FNA provided both the primary diagnosis and upstaged the malignancy. In nine subjects (20%), EUS-FNA made the initial diagnosis, upstaged the tumor, and prevented surgery. Three EUS-FNAs (7%) were diagnostic in which previous CT-FNA was benign (two of three) or nondiagnostic (one of three). In three patients with nonmalignant EUS-FNA aspirates, malignancy was later confirmed by surgery (n ⫽ 2) or percutaneous biopsy (n ⫽ 1). Follow-Up After EUS-FNA Of the 45 patients with malignant cytology from EUS-FNA, 40 have died, with a mean time to death after EUS-FNA of 135 days (range 7–754 days). The patient who died 7 days after EUS-FNA expired in his sleep outside the hospital without any apparent adverse consequence from EUS-FNA; however, no autopsy was performed. The patient who lived 754 days had metastatic pancreatic cancer and enrolled in an experimental chemotherapy protocol at another institution. The longest survival otherwise was 298 days. Five patients remain alive with a mean follow-up of 64 days (range 30 –140 days). The mean time to death from metastatic pancreatic adenocarcinoma to the liver was 113 days (n ⫽ 31; range 7–754 days).
DISCUSSION The results of this study demonstrate that EUS-guided FNA of solid hepatic masses is sensitive, safe, and has a significant impact on patient management when malignancy is diagnosed. In our series, the sensitivity of EUS-FNA for the diagnosis of malignancy ranged from 82% to 94%. Although it may be presumed that we obtained no false pos-
EUS-FNA Cytology of Solid Liver Lesions
1979
itive results when testing aspirates, the inability to histologically corroborate this does not allow for calculation of specificity. A previous report of 14 patients that underwent EUS-FNA of liver metastases described a sensitivity and specificity of 100% and 100% for the diagnosis of malignancy (11). To our knowledge, however, our study is the first report of the test characteristics of EUS-FNA of the liver that includes both malignant and benign hepatic lesions. The sensitivity of the EUS-FNA of the liver we report is similar to previous series that describe the sensitivity of EUS-FNA of pancreatic malignancy (9, 15–17) and P-FNA of liver tumors (2, 18 –20). Despite the good sensitivity of EUS-FNA of liver tumors, we obtained three false negative results. Therefore, follow-up percutaneous biopsy or intraoperative evaluation is warranted when the diagnosis of malignancy in the liver is strongly suspected and would significantly affect patient management. The mean number of overall passes performed for the diagnosis of hepatic lesions (n ⫽ 3.2; range 1– 8) is slightly higher in our series than has been previously reported (11). The exact reason for this difference is not clear, although it is not significantly different from the number required for ultrasound-guided P-FNA of small hepatic masses (21). In a series of 14 patients with EUS-FNA of a malignant liver lesion, Nguyen et al. reported a mean of 2.0 (range 1–5) passes for the diagnosis of malignancy (11). The presence of a cytopathologist on-site was implied, as interpretation of specimens was performed before making repeated passes. We also have a cytotechnologist or cytopathologist available on site to assess the adequacy of obtained specimens. Because the diagnosis of metastatic disease is clinically relevant and initial air-dried preparations are sometimes difficult to interpret, we will often repeat another pass to ensure that cytology is sufficient to render a diagnosis if the adequacy of a specimen is initially interpreted as “suspicious” but not diagnostic of malignancy. We found no statistically significant difference between the numbers of passes required for the diagnosis of benign (n ⫽ 3.2; range 1– 8) or malignant (n ⫽ 3.5; range 1– 8) lesions. In addition, no difference existed between the number of passes required during the first or second half of the study. This implies that increased experience with this technique may not decrease the number of passes needed to obtain an accurate cytological diagnosis. Despite the relatively higher number of passes performed in this study, we observed no complications after EUS-FNA of the liver. In a large international survey of 167 cases describing EUS-FNA of the liver, however, six complications (4%) were described (12). These complications included one death in a patient who underwent EUS-FNA of the liver despite a suspected obstructed biliary stent. Other complications noted were bleeding, abdominal pain, and fever. Although it is standard practice for gastroenterologists to observe patients in the right lateral decubitus position for several hours after percutaneous biopsy of the liver, we discharge stable patients within 60 –75 min after EUS-
1980
DeWitt et al.
FNA of the liver. Furthermore, we do not routinely place patients on their right side during recovery before discharge. A recent series (5) describing the use of P-FNA in the diagnosis of 216 liver tumors reported the occurrence of implantation metastases in seven patients (3%) a median 4 months (range 2– 49 months) after P-FNA of liver masses from colorectal cancer (n ⫽ 5), gallbladder carcinoma (n ⫽ 1), and a hepatoma (n ⫽ 1). In addition, the implantation metastases caused major problems locally and were fatal in four patients. This complication has not been described after EUS-FNA of the liver. The true incidence of implantation metastases after EUS-FNA of the liver, however, is difficult to estimate, inasmuch as hepatic metastases from pancreatic malignancy (which comprise 87% of malignancies in our study) usually carry a dismal prognosis and are managed nonoperatively. Therefore, as the overwhelming majority of patients diagnosed with of hepatic metastases after EUSFNA of the liver do not undergo intended curative resection, the true incidence of this complication is not known. The most common malignancies diagnosed by P-FNA (colorectal metastases and primary hepatomas) are often managed surgically with curative intent. We believe that these fundamental differences partially explain the disparity in reported incidence of implantation metastases between EUSFNA and P-FNA of liver masses. Studies comparing the diagnostic accuracy and complications EUS-FNA and PFNA of all types of suspected liver tumors are needed to resolve these issues. The overall sensitivity of transabominal US for the detection of liver metastases is reportedly between 80% and 90% (22, 23). Recent studies using helical CT and magnetic resonance imaging, however, show that noninvasive imaging is relatively insensitive for lesions ⬍1 cm (24 –26). By EUS and EUS-FNA, we detected 17 malignant hepatic lesions (mean 12.6 mm, range 3–26 mm) with a previously normal CT alone (n ⫽ 13), CT and US (n ⫽ 3), or US alone (n ⫽ 1). Of these, six (35%) were ⬍1 cm in diameter. These 17 lesions were found in 41% of the 42 patients with available radiographic imaging information performed before EUS. Nguyen et al. (11) reported that CT performed before EUS-FNA of the liver failed to detect liver lesions in 11 of 14 patients (79%). Collectively, these results demonstrate the utility of EUS-FNA for the detection of malignant liver lesions that have been missed by prior radiographic imaging. Further comparative studies are needed to determine whether other imaging modalities such as fluorine-18 fluorodeoxyglucose positron emission tomography (27), magnetic resonance imaging (28), harmonic ultrasound imaging (29), or ultrasound contrast agents (30) offer any advantages over EUS for detection of occult or subcentimeter liver metastases. When compared with benign lesions, we found that malignant lesions detected by EUS-FNA of the liver were more likely to have regular margins (60% vs 27%; p ⫽ 0.02) and to be accompanied by at least one other lesion detected on EUS (38% vs 9%; p ⫽ 0.03). Furthermore, no statistically
AJG – Vol. 98, No. 9, 2003
significant difference was found with regard to site of biopsy (p ⫽ 0.57), size (p ⫽ 0.70), echogenicity (p ⫽ 0.36), or number of passes performed (p ⫽ 0.44) for these lesions. Although EUS features of malignant lymphadenopathy have been described (31), to our knowledge the endosonographic features for malignant liver lesions have not been previously reported. These findings may help to guide decision making and assessment of the risk/benefit ratio of EUS-FNA of hepatic masses. In conclusion, the results of our study show that the sensitivity of EUS-FNA of the liver for the diagnosis of malignancy ranges from 82% to 94% and that, contrary to previous reports, it is a safe procedure. When malignancy is diagnosed, EUS-FNA of the liver significantly affects patient management and implies a poor overall prognosis. EUS features predictive of malignant hepatic masses are the presence of regular outer margins and the detection of two or more lesions. In this series, EUS and EUS-FNA detected malignant liver tumors that were not seen by CT or US (or both) in 41% of subjects. Therefore, surveillance of the entire liver is indicated during evaluation of known or suspected malignancy. Prospective studies are needed to compare the accuracy and complication rate of EUS-FNA and P-FNA for the diagnosis of liver tumors. Reprint requests and correspondence: John M. DeWitt, M.D., Department of Medicine, Division of Gastroenterology, Indiana University Medical Center, 550 N. University Boulevard, UH 4100, Indianapolis, IN 46202-5121. Received Nov. 11, 2002; accepted Jan. 30, 2003.
REFERENCES 1. Tatsuta M, Yamamoto R, Kasugai H, et al. Cytohistologic diagnosis of neoplasms of the liver by ultrasonically guided fine-needle aspiration biopsy. Cancer 1984;54:1682–6. 2. Buscarini L, Fornari F, Bolondi L, et al. Ultrasound-guided fine-needle biopsy of focal liver lesions: Techniques, diagnostic accuracy and complications. A retrospective study on 2091 biopsies. J Hepatol 1990;11:344 –8. 3. Edoute Y, Ben-Haim S, Brenner B, et al. Fatal hemoperitoneum after fine-needle aspiration of a liver metastases. Am J Gastroenterol 1992;87:358 –9. 4. Kowdley KV, Aggarwal A, Sachs PB. Delayed hemorrhage after percutaneous liver biopsy: Role of therapeutic angiography. J Clin Gastroenterol 1994;19:50 –3. 5. Ohlsson B, Nilsson J, Stenram U. Percutaneous fine-needle aspiration cytology in the diagnosis and management of liver tumors. Br J Surg 2002;89:757–62. 6. Goletti O, Chiarugi M, Buccianti P, et al. Subcutaneous implantation of liver metastasis after fine needle biopsy. Eur J Surg Oncol 1992;18:636 –7. 7. Scheele J, Alterndorf-Hofmann A. Tumor implantation from needle biopsy of hepatic metastases. Hepato-Gastroenterology 1990;37:335–7. 8. Andersson R, Andren-Sanberg A, Lundstedt C, et al. Implantation metastases from gastrointestinal cancer after percutaneous puncture or biliary drainage. Eur J Surg 1996;162:551–4. 9. Wiersema MJ, Vilmann P, Giovannini M, et al. Endosonography-guided fine needle aspiration biopsy: Diagnostic accu-
AJG – September, 2003
10.
11. 12.
13. 14.
15. 16. 17.
18. 19. 20.
racy and complication assessment. Gastroenterology 1997; 112:1087–95. Gress FG, Savides TJ, Sandler A, et al. Endoscopic ultrasonography, fine-needle aspiration biopsy guided by endoscopic ultrasonography, and computed tomography in the preoperative staging of non-small cell lung cancer: A comparison study. Ann Intern Med 1997;127:604 –12. Nguyen P, Feng JC, Chang KJ. Endoscopic ultrasound (EUS) and EUS-guided fine-needle aspiration (FNA) of liver lesions. Gastrointest Endosc 1999;50:357–61. tenBerge J, Hoffman BJ, Hawes RH, et al. EUS-guided fine needle aspiration of the liver. indications, yield, and safety based on an international survey of 167 cases. Gastrointest Endosc 2002;55:859 –62. Bentz J, Kochmann M, Faigel D, et al. Endoscopic ultrasoundguided real-time fine-needle aspiration: Clinicopathologic features of 60 patients. Diagn Cytopathol 1998;18:98 –100. Fritscher-Ravens A, Schirror L, Atay Z, et al. Endosonographically controlled fine needle aspiration-cytology-indications and results in routine diagnosis. Z Gastroenterol 1999; 37:343–51. Williams DB, Sahai AV, Aabakken L, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: A large single center experience. Gut 1999;44:720 –6. Hunerbein M, Dohmoto M, Haensch W, et al. Endosonography-guided biopsy of mediastinal and pancreatic tumors. Endoscopy 1998;30:32–6. DeWitt J, Devereaux B, Ciaccia D, et al. Prospective evaluation of endoscopic ultrasound fine-needle aspiration (EUSFNA) for the diagnosis of suspected pancreatic malignancy. Gastrointest Endosc 2002;55:AB246. Montali G, Solbati L, Croce F, et al. Fine-needle aspiration biopsy of liver focal lesions ultrasonically guided with a realtime probe. Report on 126 cases. Br J Radiol 1982;55:717–23. Bru C, Maroto A, Bruix J, et al. Diagnostic accuracy of fine-needle aspiration biopsy in patients with hepatocellular carcinoma. Dig Dis Sci 1989;34:1765–9. Bognel C, Rougier P, Leclere J, et al. Fine needle aspiration of the liver and pancreas with ultrasound guidance. Acta Cytol 1988;32:22–6.
EUS-FNA Cytology of Solid Liver Lesions
1981
21. Middleton W, Hiskes SK, Teffey SA, et al. Small (1.5 cm or less) liver metastases: US-guided biopsy. Radiology 1997; 205:729 –32. 22. Gore RM, Levine MS, Laufer I. Textbook of gastrointestinal radiology. Philadelphia: WB Saunders, 1994:1935. 23. Grainger RG, Allison DJ. Diagnostic radiology: A textbook of medical imaging, 3rd ed. New York: Churchill Livingstone, 1998:1162. 24. Ward J, Naid KS, Guthrie JA, et al. Hepatic lesion detection: Comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative-free response receiver operating characteristic analysis. Radiology 1999;210:459 –66. 25. Hagspiel KD, Neidl KFW, Eichenberger AC, et al. Detection of liver metastases: Comparison of superparamagnetic iron oxide-enhanced and unenhanced MR imaging at 1.5 T with dynamic CT, intraoperative US and percutaneous US. Radiology 1995;196:471–8. 26. Kuszyk BS, Bluemke DA, Urban BA, et al. Portal-phase contrast-enhanced helical CT for the detection of malignant hepatic tumors: Sensitivity based on intraoperative and pathologic findings. Am J Roentgenol 1996;166:91–5. 27. Nakamoto Y, Higashi T, Sakahara H, et al. Contribution of PET in the detection of liver metastases from pancreatic tumors. Clin Radiol 1999;54:248 –52. 28. Schima W, Fugger R, Schober E, et al. Diagnosis and staging of pancreatic cancer: Comparison of mangafodipir trisodiumenhanced MR imaging and contrast-enhanced helical hydroCT. Am J Roentgenol 2002;179:717–24. 29. Strobel D, Hahn EG, Becker D. Harmonic imaging for improving diagnosis of liver tumors—a preliminary report. Z Gastroenterol 2001;39:292–4. 30. Harvey CJ, Blomley MJ, Eckersley RJ, et al. Pulse-inversion mode imaging of liver specific microbubbles: Improved detection of subcentimetre metastases. Lancet 2000;355:807–8. 31. Catalano MF, Sivak MJ, Rice T, et al. Endosonographic features predictive of lymph node metastasis. Gastrointest Endosc 1994;40:442–6.
Vol. 98, No. 9, 2003 ISSN 0002-9270/03/$30.00 doi:10.1016/S0002-9270(03)00549-5
Endoscopic Ultrasound–Guided Fine Needle Aspiration Cytology of Solid Liver Lesions: A Large Single-Center Experience John DeWitt, M.D., Julia LeBlanc, M.D., Lee McHenry, M.D., Dan Ciaccia, M.D., Tom Imperiale, M.D., John Chappo, B.S., C.T., Harvey Cramer, M.D., Kathy McGreevy, R.N., Melissa Chriswell, R.N., and Stuart Sherman, M.D. Department of Gastroenterology and Hepatology and Department of Pathology and Laboratory Medicine, Indiana University Medical Center, Indianapolis, Indiana
OBJECTIVES: The aim of this study was to report the sensitivity, cytological diagnoses, endoscopic ultrasound (EUS) features, complications, clinical impact, and long term follow-up of a large single-center experience with endoscopic ultrasound– guided fine needle aspiration (EUS-FNA) of benign and malignant solid liver lesions. METHODS: A database of cytologic specimens from EUSFNA was reviewed to identify all hepatic lesions aspirated between January, 1997, and July, 2002. Procedural indications, prior radiographic data, patient demographics, EUS examination results, complications, and follow-up data were obtained and recorded. RESULTS: EUS-FNA of 77 liver lesions in 77 patients was performed without complications. Of these 77 lesions, 45 (58%) were diagnostic for malignancy, 25 (33%) were benign, and seven (9%) were nondiagnostic. A total of 22 lesions were confirmed as negative for malignancy by follow-up (mean 762 days, range 512–1556 days) or intraoperative examination; however, seven lesions could not be classified as benign or malignant. Depending on the status of the seven unclassified lesions, sensitivity of EUS-FNA for the diagnosis of malignancy ranged from 82 to 94%. When compared with benign lesions, EUS features predictive of malignant hepatic masses were the presence of regular outer margins (60% vs 27%; p ⫽ 0.02) and the detection of two or more lesions (38% vs 9%; p ⫽ 0.03). Of the 42 patients with malignancy identified by EUS-FNA and other available imaging records, EUS detected the malignancy in 41% of patients with previously negative examinations. For the 45 subjects with cytology positive for malignancy, EUSFNA changed management in 86% of subjects. CONCLUSION: EUS-FNA of the liver is a safe and sensitive procedure that can have a significant impact on patient management. Prospective studies comparing the accuracy and complication rate of EUS-FNA and percutaneous fine needle aspiration (P-FNA) for the diagnosis of liver tumors are needed. (Am J Gastroenterol 2003;98:1976 –1981. © 2003 by Am. Coll. of Gastroenterology)
INTRODUCTION Identification of hepatic metastases from systemic malignancy usually implies inoperable disease and a poor prognosis. Furthermore, the diagnosis of unresectable malignancy may obviate the need for surgery along with its inherent morbidity and mortality. The diagnosis of liver masses is traditionally accomplished by percutaneous fine needle aspiration (P-FNA) under fluoroscopic, computed tomographic, or ultrasonographic guidance (1–3). Although rare, complications from P-FNA may include severe bleeding and implantation metastases in up to 3% of patients (3– 8). Endoscopic ultrasound (EUS)– guided fine needle aspiration (EUS-FNA) has been established as a safe and accurate method for the diagnosis of benign or malignant lymphadenopathy and for intramural and extramural neoplasms (9, 10). The role of EUS-FNA in the diagnosis and staging of liver metastases or masses, however, is limited to a small single-center series (11), a large retrospective international survey (12), and several case series (13, 14). Furthermore, the sensitivity of EUS-FNA and endosonographic features of benign and malignant hepatic lesions have not been described. The aim of this study was to report the sensitivity, cytological diagnoses, EUS features, complications, and clinical impact of a large single-center experience with EUS-FNA of solid liver lesions.
MATERIALS AND METHODS From an existing EUS-FNA cytology database, we identified all patients between January, 1997, and July, 2002, in whom EUS-FNA of the liver was performed. The database used had been maintained and prospectively updated by one cytotechnologist (J.C.). Patients with hilar cholangiocarcinomas or cystic hepatic lesions were excluded from the analysis. Staging EUS examinations for known or suspected malignancy were performed initially with an Olympus GFUM20 or GFUM-130 radial echoendoscope (Olympus,
AJG – September, 2003
Melville NY). The right lobe of the liver was surveyed from the duodenum and distal stomach, whereas the left lobe was imaged from the proximal and mid-stomach. Endoscope rotation was used as necessary to visualize as much of the liver parenchyma as possible. EUS-FNA was performed using the Pentax 32-UA, Pentax 36-UX (Pentax Precision Instruments, Orangeburg, NY), Olympus GF-UC30P, or Olympus GF-UC140P curvilinear array echoendoscope. EUS-FNA was performed using a 22-gauge, 8-cm WilsonCook EUSN-1, EUSN-2, or EUSN-3 needle (Wilson-Cook Medical, Winston-Salem, NC) by one of four physicians (J.D., D.C., L.M., or J.L.). Doppler color angiography was used to ensure the absence of intervening vascular structures along the anticipated needle path. Depending on the amount of blood anticipated during tissue sampling, full, partial, or no suction was applied at the discretion of the endoscopist, cytotechnologist, or cytopathologist. A cytotechnologist or cytopathologist was available on-site for preliminary interpretations on all procedures. Samples aspirated were expressed onto a glass slide and two smear preparations were made. One slide was air-dried and stained with a modified Giemsa stain for rapid on-site interpretation. The other slide was alcohol-fixed and stained by the Papanicolaou method. EUS-FNA was repeated until a definitive diagnosis was made or the endoscopist believed that further sampling would not likely increase yield. Within several days of each EUS examination, a final cytological diagnosis was rendered by a staff cytopathologist. A true positive EUS-FNA for malignancy is defined as unequivocal cytological evidence of malignancy. A false negative aspirate is a nondiagnostic or benign specimen, which is subsequently found to be malignant by percutaneous FNA or intraoperative findings. A true negative EUSFNA for malignancy is defined as benign or nondiagnostic samples that are either confirmed as benign by alternative sampling, intraoperative examination, or appropriate clinical follow-up. Specimens with benign hepatocytes without evidence of atypical cells are categorized as benign aspirates. Aspirates from EUS-FNA that are interpreted as nondiagnostic, highly suspicious, suspicious, or atypical for malignancy are considered as negative for malignancy. Medical records of enrolled subjects were reviewed and procedural indications, prior radiographic data, patient demographics, EUS test results, clinical outcomes, procedural complications, and follow-up data were abstracted. When multiple liver lesions were noted, the endosonographic description of only the aspirated lesion was recorded. When charts were incomplete, written informed consent was sent and follow-up telephone calls were made to the subject, closest relative of the deceased, or referring physician for clarification or any required further information. The potential clinical impact of EUS-FNA of a liver metastasis or malignancy is defined as results that met the following criteria: 1) avoided surgery (results precluded resectable malignancy); 2) made diagnosis (results provided initial diagnosis of malignancy); 3) upstaged tumor (results up-
EUS-FNA Cytology of Solid Liver Lesions
1977
Table 1. Indications for EUS Examinations by Cytology Results of EUS-FNA of the Liver Cytology Result Indication Abnormal ERCP Pancreatic mass on CT Staging of known cancer Liver mass on CT Other abnormal CT findings Suspected recurrent cancer Chronic abdominal pain Other Total
All Malignant Benign/ND (n ⫽ 77) (n ⫽ 45) (n ⫽ 32) 21 (27) 19 (25) 10 (13) 9 (12) 4 (5) 2 (3) 2 (3) 10 (12) 77
11 (24) 12 (27) 7 (16) 9 (20) 1 (2) 1 (2) 0 (0) 4 (9) 45
10 (31) 7 (22) 3 (9) 0 (0) 3 (9) 1 (3) 2 (7) 6 (19) 32
ND ⫽ nondiagnostic
staged primary malignancy); 4) incurred no change in management (results did not change patient treatment). This study was approved by the Institutional Review Board at our institution. Statistical Analysis Assuming that the EUS-FNA diagnosis of malignancy is a true positive, sensitivity was calculated as the proportion of patients with cancer in whom EUS-FNA was positive for malignancy. For analysis, continuous variables were described as means and standard deviations, and dichotomous variables were expressed as simple proportions with or without 95% confidence limits. Student’s t test and 2 or Fisher’s exact tests were used to test for differences in comparisons between continuous and dichotomous variables, respectively.
RESULTS In total, EUS-FNA was performed on 77 different liver lesions in 77 patients (42 male and 35 female, 69 white and eight African American, with a mean age of 63 ⫾ 11 yr). Follow-up was available in all patients. Indications for EUS Examinations The indications for the 77 EUS procedures are recorded in Table 1. The two most common indications were abnormal ERCP findings, the majority of which were strictures, and a pancreatic mass on CT scan, which accounted for 27% and 25% of the procedures performed, respectively. Ten (13%) of 77 procedures were done for staging of a recently diagnosed malignancy (pancreas in five, esophageal in four, and lung in one). Of the 45 malignancies diagnosed by EUSFNA, the most common indications for the EUS were a CT finding of a pancreatic mass (27%) and abnormal ERCP findings (24%). Characteristics of EUS Examination EUS-FNA was performed in 77 patients (mean 3.4 ⫾ 1.8 passes, range 1– 8) into the left lobe (n ⫽ 66; 86%) and right lobe (n ⫽ 11; 14%) of the liver without procedural compli-
1978
DeWitt et al.
AJG – Vol. 98, No. 9, 2003
Table 2. Characteristics of EUS Examination All* Malignant† Benign‡ (n ⫽ 77) (n ⫽ 48) (n ⫽ 22)
p-value§
Site of FNA Left lobe 66 (86) 41 (85) 20 (91) Right lobe 11 (14) 7 (15) 2 (9) No. of passes 3.4 ⫾ 1.8 3.5 ⫾ 1.5 3.2 ⫾ 1.5 Range 1–8 1–8 1–8 Echogenicity Hypoechoic 52 (68) 33 (69) 13 (59) Hyperechoic 23 (30) 13 (27) 9 (41) Both 2 (2) 2 (4) 0 (0) Margins Regular 39 (51) 29 (60) 6 (27) Irregular 38 (49) 19 (40) 16 (73) Size (mm) 16.0 ⫾ 10.8 15.6 ⫾ 7.7 16.7 ⫾ 12.0 Range 3–40 3–35 4–40 No. of lesions seen 1 57 (74) 30 (62) 20 (91) ⬎1 20 (26) 18 (38) 2 (9)
0.57 0.44 0.36
0.02 0.70 0.03
* Includes seven patients unable to be classified as benign or malignant. † Includes 45 positive and three false negative EUS-FNA results. ‡ By follow-up or intraoperative findings. § Between malignant and benign lesions.
cations (upper 95% CI ⫽ 4.7%; see Table 2). A total of 45 (58%) aspirates were diagnostic for malignancy. Three false negatives were later discovered by intraoperative findings (n ⫽ 2) or P-FNA (n ⫽ 1). One patient with pancreatic adenocarcinoma (as confirmed by EUS-FNA of the pancreas) had benign cytology of a 6-mm left lobe mass that intraoperatively was confirmed as metastatic adenocarcinoma. A second patient with pancreatic adenocarcinoma (as confirmed by EUS-FNA of the pancreas) had benign cytology of a 6-mm right lobe mass, which was later confirmed by percutaneous ultrasound-guided biopsy as metastatic adenocarcinoma. The third patient with a false negative EUS-FNA had a 29-mm left lobe mass that was benign by EUS-FNA. Intraoperative examination, however, confirmed hepatocellular carcinoma. Overall, 48 of 77 (62%) of the liver masses were malignant. Among the remaining 29 subjects with nonmalignant aspirates, in 22 (76%) the masses were considered benign by clinical follow-up (n ⫽ 18; mean 762 days; range: 512–1556 days) or intraoperative evaluation (n ⫽ 4). Seven (24%) of the subjects with nonmalignant masses with pancreatic adenocarcinoma (as confirmed by EUS-FNA of the pancreas; n ⫽ 6) or lung cancer (n ⫽ 1) died (mean 154 days, range 14 – 424 days) without follow-up imaging, biopsy of the liver, or autopsy and therefore cannot be classified as benign or malignant. The 45 diagnostic aspirates for malignancy are considered true positives (Fig. 1). Assuming that the test results for these seven patients were all true negatives or all false negatives, the sensitivity of EUS-FNA for the diagnosis of malignancy would range from 82 to 94%. In comparison between benign and malignant lesions detected by EUS-FNA (Table 2), no statistically significant difference was found between the liver lobe aspirated (p ⫽ 0.57), number of passes (p ⫽ 0.44), lesion size (p ⫽ 0.70),
Figure 1. Sensitivity of EUS-FNA of the liver for the diagnosis of malignancy.
or echotexture (p ⫽ 0.36). Malignant masses, however, were more often accompanied by the presence of multiple hepatic lesions detected on EUS (38% vs 9%; p ⫽ 0.03; 2 analysis) and to have regular margins (60% vs 27%; p ⫽ 0.02; 2 analysis). Diagnoses from EUS-FNA of the Liver The final diagnoses for the 45 malignant liver aspirates are recorded in Table 3. Of these, 44 (98%) were metastatic and one (2%) was a hepatocellular carcinoma. The most common diagnosis was metastatic adenocarcinoma from the pancreas, which accounted for 34 of 45 (76%). The next most frequent diagnosis was metastatic neuroendocrine tumor from the pancreas (n ⫽ 5; 11%). Of the 32 nonmalignant aspirates, 25 (33%) were cytologically benign and seven (9%) were nondiagnostic. One (4%) benign aspirate demonstrated cytological features consistent with an abscess. Previous Imaging of Malignancy Diagnosed by EUS-FNA Prior radiographic imaging reports were available for 42 of 45 (93%) malignant hepatic masses diagnosed by EUSTable 3. Diagnoses of Malignant Aspirates of the Liver Diagnosis
No. (%)
Pancreatic adenocarcinoma Pancreatic neuroendocrine tumor Renal cell carcinoma Gallbladder adenocarcinoma Colon cancer Hepatocellular carcinoma Esophageal adenocarcinoma Total
34 (76) 5 (11) 2 (5) 1 (2) 1 (2) 1 (2) 1 (2) 45
AJG – September, 2003
FNA. All imaging had been performed within 45 days of the EUS examination. Of these 42 lesions, EUS-FNA detected malignancy in 17 (41%) when prior imaging of the liver by CT alone (n ⫽ 13), transabdominal ultrasound (US) alone (n ⫽ 1), or both (n ⫽ 3) were normal. For the 32 patients in whom CT alone was performed, EUS-FNA initially detected a malignancy in 13 (41%). For the nine patients in whom CT and US were both performed, EUS-FNA initially detected malignancy in three (33%). The mean size of the 17 malignant liver lesions detected initially by EUS-FNA was 12.6 mm (range 3–26 mm). Of these 17, six (35%) were ⬍1 cm in diameter. Clinical Impact of EUS-FNA of the Liver For the 45 patients with malignant cytology, EUS-FNA had (by protocol definition) a clinical impact in all patients. EUS-FNA changed management, however, in only 38 (84%) of these patients, because in seven cases (16%) tumor was upstaged but did not change overall patient management. EUS-FNA provided the initial diagnosis of malignancy in 35 patients (78%) and avoided surgery in 12 (27%). Of the 35 patients in whom EUS-FNA made the initial diagnosis of malignancy, 34 (97%) had metastases and one (3%) had a hepatocellular carcinoma. In 25 patients (55%), the FNA provided both the primary diagnosis and upstaged the malignancy. In nine subjects (20%), EUS-FNA made the initial diagnosis, upstaged the tumor, and prevented surgery. Three EUS-FNAs (7%) were diagnostic in which previous CT-FNA was benign (two of three) or nondiagnostic (one of three). In three patients with nonmalignant EUS-FNA aspirates, malignancy was later confirmed by surgery (n ⫽ 2) or percutaneous biopsy (n ⫽ 1). Follow-Up After EUS-FNA Of the 45 patients with malignant cytology from EUS-FNA, 40 have died, with a mean time to death after EUS-FNA of 135 days (range 7–754 days). The patient who died 7 days after EUS-FNA expired in his sleep outside the hospital without any apparent adverse consequence from EUS-FNA; however, no autopsy was performed. The patient who lived 754 days had metastatic pancreatic cancer and enrolled in an experimental chemotherapy protocol at another institution. The longest survival otherwise was 298 days. Five patients remain alive with a mean follow-up of 64 days (range 30 –140 days). The mean time to death from metastatic pancreatic adenocarcinoma to the liver was 113 days (n ⫽ 31; range 7–754 days).
DISCUSSION The results of this study demonstrate that EUS-guided FNA of solid hepatic masses is sensitive, safe, and has a significant impact on patient management when malignancy is diagnosed. In our series, the sensitivity of EUS-FNA for the diagnosis of malignancy ranged from 82% to 94%. Although it may be presumed that we obtained no false pos-
EUS-FNA Cytology of Solid Liver Lesions
1979
itive results when testing aspirates, the inability to histologically corroborate this does not allow for calculation of specificity. A previous report of 14 patients that underwent EUS-FNA of liver metastases described a sensitivity and specificity of 100% and 100% for the diagnosis of malignancy (11). To our knowledge, however, our study is the first report of the test characteristics of EUS-FNA of the liver that includes both malignant and benign hepatic lesions. The sensitivity of the EUS-FNA of the liver we report is similar to previous series that describe the sensitivity of EUS-FNA of pancreatic malignancy (9, 15–17) and P-FNA of liver tumors (2, 18 –20). Despite the good sensitivity of EUS-FNA of liver tumors, we obtained three false negative results. Therefore, follow-up percutaneous biopsy or intraoperative evaluation is warranted when the diagnosis of malignancy in the liver is strongly suspected and would significantly affect patient management. The mean number of overall passes performed for the diagnosis of hepatic lesions (n ⫽ 3.2; range 1– 8) is slightly higher in our series than has been previously reported (11). The exact reason for this difference is not clear, although it is not significantly different from the number required for ultrasound-guided P-FNA of small hepatic masses (21). In a series of 14 patients with EUS-FNA of a malignant liver lesion, Nguyen et al. reported a mean of 2.0 (range 1–5) passes for the diagnosis of malignancy (11). The presence of a cytopathologist on-site was implied, as interpretation of specimens was performed before making repeated passes. We also have a cytotechnologist or cytopathologist available on site to assess the adequacy of obtained specimens. Because the diagnosis of metastatic disease is clinically relevant and initial air-dried preparations are sometimes difficult to interpret, we will often repeat another pass to ensure that cytology is sufficient to render a diagnosis if the adequacy of a specimen is initially interpreted as “suspicious” but not diagnostic of malignancy. We found no statistically significant difference between the numbers of passes required for the diagnosis of benign (n ⫽ 3.2; range 1– 8) or malignant (n ⫽ 3.5; range 1– 8) lesions. In addition, no difference existed between the number of passes required during the first or second half of the study. This implies that increased experience with this technique may not decrease the number of passes needed to obtain an accurate cytological diagnosis. Despite the relatively higher number of passes performed in this study, we observed no complications after EUS-FNA of the liver. In a large international survey of 167 cases describing EUS-FNA of the liver, however, six complications (4%) were described (12). These complications included one death in a patient who underwent EUS-FNA of the liver despite a suspected obstructed biliary stent. Other complications noted were bleeding, abdominal pain, and fever. Although it is standard practice for gastroenterologists to observe patients in the right lateral decubitus position for several hours after percutaneous biopsy of the liver, we discharge stable patients within 60 –75 min after EUS-
1980
DeWitt et al.
FNA of the liver. Furthermore, we do not routinely place patients on their right side during recovery before discharge. A recent series (5) describing the use of P-FNA in the diagnosis of 216 liver tumors reported the occurrence of implantation metastases in seven patients (3%) a median 4 months (range 2– 49 months) after P-FNA of liver masses from colorectal cancer (n ⫽ 5), gallbladder carcinoma (n ⫽ 1), and a hepatoma (n ⫽ 1). In addition, the implantation metastases caused major problems locally and were fatal in four patients. This complication has not been described after EUS-FNA of the liver. The true incidence of implantation metastases after EUS-FNA of the liver, however, is difficult to estimate, inasmuch as hepatic metastases from pancreatic malignancy (which comprise 87% of malignancies in our study) usually carry a dismal prognosis and are managed nonoperatively. Therefore, as the overwhelming majority of patients diagnosed with of hepatic metastases after EUSFNA of the liver do not undergo intended curative resection, the true incidence of this complication is not known. The most common malignancies diagnosed by P-FNA (colorectal metastases and primary hepatomas) are often managed surgically with curative intent. We believe that these fundamental differences partially explain the disparity in reported incidence of implantation metastases between EUSFNA and P-FNA of liver masses. Studies comparing the diagnostic accuracy and complications EUS-FNA and PFNA of all types of suspected liver tumors are needed to resolve these issues. The overall sensitivity of transabominal US for the detection of liver metastases is reportedly between 80% and 90% (22, 23). Recent studies using helical CT and magnetic resonance imaging, however, show that noninvasive imaging is relatively insensitive for lesions ⬍1 cm (24 –26). By EUS and EUS-FNA, we detected 17 malignant hepatic lesions (mean 12.6 mm, range 3–26 mm) with a previously normal CT alone (n ⫽ 13), CT and US (n ⫽ 3), or US alone (n ⫽ 1). Of these, six (35%) were ⬍1 cm in diameter. These 17 lesions were found in 41% of the 42 patients with available radiographic imaging information performed before EUS. Nguyen et al. (11) reported that CT performed before EUS-FNA of the liver failed to detect liver lesions in 11 of 14 patients (79%). Collectively, these results demonstrate the utility of EUS-FNA for the detection of malignant liver lesions that have been missed by prior radiographic imaging. Further comparative studies are needed to determine whether other imaging modalities such as fluorine-18 fluorodeoxyglucose positron emission tomography (27), magnetic resonance imaging (28), harmonic ultrasound imaging (29), or ultrasound contrast agents (30) offer any advantages over EUS for detection of occult or subcentimeter liver metastases. When compared with benign lesions, we found that malignant lesions detected by EUS-FNA of the liver were more likely to have regular margins (60% vs 27%; p ⫽ 0.02) and to be accompanied by at least one other lesion detected on EUS (38% vs 9%; p ⫽ 0.03). Furthermore, no statistically
AJG – Vol. 98, No. 9, 2003
significant difference was found with regard to site of biopsy (p ⫽ 0.57), size (p ⫽ 0.70), echogenicity (p ⫽ 0.36), or number of passes performed (p ⫽ 0.44) for these lesions. Although EUS features of malignant lymphadenopathy have been described (31), to our knowledge the endosonographic features for malignant liver lesions have not been previously reported. These findings may help to guide decision making and assessment of the risk/benefit ratio of EUS-FNA of hepatic masses. In conclusion, the results of our study show that the sensitivity of EUS-FNA of the liver for the diagnosis of malignancy ranges from 82% to 94% and that, contrary to previous reports, it is a safe procedure. When malignancy is diagnosed, EUS-FNA of the liver significantly affects patient management and implies a poor overall prognosis. EUS features predictive of malignant hepatic masses are the presence of regular outer margins and the detection of two or more lesions. In this series, EUS and EUS-FNA detected malignant liver tumors that were not seen by CT or US (or both) in 41% of subjects. Therefore, surveillance of the entire liver is indicated during evaluation of known or suspected malignancy. Prospective studies are needed to compare the accuracy and complication rate of EUS-FNA and P-FNA for the diagnosis of liver tumors. Reprint requests and correspondence: John M. DeWitt, M.D., Department of Medicine, Division of Gastroenterology, Indiana University Medical Center, 550 N. University Boulevard, UH 4100, Indianapolis, IN 46202-5121. Received Nov. 11, 2002; accepted Jan. 30, 2003.
REFERENCES 1. Tatsuta M, Yamamoto R, Kasugai H, et al. Cytohistologic diagnosis of neoplasms of the liver by ultrasonically guided fine-needle aspiration biopsy. Cancer 1984;54:1682–6. 2. Buscarini L, Fornari F, Bolondi L, et al. Ultrasound-guided fine-needle biopsy of focal liver lesions: Techniques, diagnostic accuracy and complications. A retrospective study on 2091 biopsies. J Hepatol 1990;11:344 –8. 3. Edoute Y, Ben-Haim S, Brenner B, et al. Fatal hemoperitoneum after fine-needle aspiration of a liver metastases. Am J Gastroenterol 1992;87:358 –9. 4. Kowdley KV, Aggarwal A, Sachs PB. Delayed hemorrhage after percutaneous liver biopsy: Role of therapeutic angiography. J Clin Gastroenterol 1994;19:50 –3. 5. Ohlsson B, Nilsson J, Stenram U. Percutaneous fine-needle aspiration cytology in the diagnosis and management of liver tumors. Br J Surg 2002;89:757–62. 6. Goletti O, Chiarugi M, Buccianti P, et al. Subcutaneous implantation of liver metastasis after fine needle biopsy. Eur J Surg Oncol 1992;18:636 –7. 7. Scheele J, Alterndorf-Hofmann A. Tumor implantation from needle biopsy of hepatic metastases. Hepato-Gastroenterology 1990;37:335–7. 8. Andersson R, Andren-Sanberg A, Lundstedt C, et al. Implantation metastases from gastrointestinal cancer after percutaneous puncture or biliary drainage. Eur J Surg 1996;162:551–4. 9. Wiersema MJ, Vilmann P, Giovannini M, et al. Endosonography-guided fine needle aspiration biopsy: Diagnostic accu-
AJG – September, 2003
10.
11. 12.
13. 14.
15. 16. 17.
18. 19. 20.
racy and complication assessment. Gastroenterology 1997; 112:1087–95. Gress FG, Savides TJ, Sandler A, et al. Endoscopic ultrasonography, fine-needle aspiration biopsy guided by endoscopic ultrasonography, and computed tomography in the preoperative staging of non-small cell lung cancer: A comparison study. Ann Intern Med 1997;127:604 –12. Nguyen P, Feng JC, Chang KJ. Endoscopic ultrasound (EUS) and EUS-guided fine-needle aspiration (FNA) of liver lesions. Gastrointest Endosc 1999;50:357–61. tenBerge J, Hoffman BJ, Hawes RH, et al. EUS-guided fine needle aspiration of the liver. indications, yield, and safety based on an international survey of 167 cases. Gastrointest Endosc 2002;55:859 –62. Bentz J, Kochmann M, Faigel D, et al. Endoscopic ultrasoundguided real-time fine-needle aspiration: Clinicopathologic features of 60 patients. Diagn Cytopathol 1998;18:98 –100. Fritscher-Ravens A, Schirror L, Atay Z, et al. Endosonographically controlled fine needle aspiration-cytology-indications and results in routine diagnosis. Z Gastroenterol 1999; 37:343–51. Williams DB, Sahai AV, Aabakken L, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: A large single center experience. Gut 1999;44:720 –6. Hunerbein M, Dohmoto M, Haensch W, et al. Endosonography-guided biopsy of mediastinal and pancreatic tumors. Endoscopy 1998;30:32–6. DeWitt J, Devereaux B, Ciaccia D, et al. Prospective evaluation of endoscopic ultrasound fine-needle aspiration (EUSFNA) for the diagnosis of suspected pancreatic malignancy. Gastrointest Endosc 2002;55:AB246. Montali G, Solbati L, Croce F, et al. Fine-needle aspiration biopsy of liver focal lesions ultrasonically guided with a realtime probe. Report on 126 cases. Br J Radiol 1982;55:717–23. Bru C, Maroto A, Bruix J, et al. Diagnostic accuracy of fine-needle aspiration biopsy in patients with hepatocellular carcinoma. Dig Dis Sci 1989;34:1765–9. Bognel C, Rougier P, Leclere J, et al. Fine needle aspiration of the liver and pancreas with ultrasound guidance. Acta Cytol 1988;32:22–6.
EUS-FNA Cytology of Solid Liver Lesions
1981
21. Middleton W, Hiskes SK, Teffey SA, et al. Small (1.5 cm or less) liver metastases: US-guided biopsy. Radiology 1997; 205:729 –32. 22. Gore RM, Levine MS, Laufer I. Textbook of gastrointestinal radiology. Philadelphia: WB Saunders, 1994:1935. 23. Grainger RG, Allison DJ. Diagnostic radiology: A textbook of medical imaging, 3rd ed. New York: Churchill Livingstone, 1998:1162. 24. Ward J, Naid KS, Guthrie JA, et al. Hepatic lesion detection: Comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative-free response receiver operating characteristic analysis. Radiology 1999;210:459 –66. 25. Hagspiel KD, Neidl KFW, Eichenberger AC, et al. Detection of liver metastases: Comparison of superparamagnetic iron oxide-enhanced and unenhanced MR imaging at 1.5 T with dynamic CT, intraoperative US and percutaneous US. Radiology 1995;196:471–8. 26. Kuszyk BS, Bluemke DA, Urban BA, et al. Portal-phase contrast-enhanced helical CT for the detection of malignant hepatic tumors: Sensitivity based on intraoperative and pathologic findings. Am J Roentgenol 1996;166:91–5. 27. Nakamoto Y, Higashi T, Sakahara H, et al. Contribution of PET in the detection of liver metastases from pancreatic tumors. Clin Radiol 1999;54:248 –52. 28. Schima W, Fugger R, Schober E, et al. Diagnosis and staging of pancreatic cancer: Comparison of mangafodipir trisodiumenhanced MR imaging and contrast-enhanced helical hydroCT. Am J Roentgenol 2002;179:717–24. 29. Strobel D, Hahn EG, Becker D. Harmonic imaging for improving diagnosis of liver tumors—a preliminary report. Z Gastroenterol 2001;39:292–4. 30. Harvey CJ, Blomley MJ, Eckersley RJ, et al. Pulse-inversion mode imaging of liver specific microbubbles: Improved detection of subcentimetre metastases. Lancet 2000;355:807–8. 31. Catalano MF, Sivak MJ, Rice T, et al. Endosonographic features predictive of lymph node metastasis. Gastrointest Endosc 1994;40:442–6.