Endoscopic ultrasound has proven to be a highly sensitive tool for diagnosing lesions in and adjacent to the gastrointestinal tract [
FNA, originally used by Martin and Ellis [
EUS-FNA has proven very useful diagnostically, obviating unwarranted procedures, and reduction of cost, all of which lead to improvements in overall patient care [
The focus of this research was to continue the progress of the aforementioned studies of EUS-FNA affecting patient management in a study not limited to specific GI tract lesions. This large single-center quality control study investigated how EUS-FNA impacted patient care at one hospital and how its implementation might be improved to further increase its diagnostic accuracy. Such results might help to persuade the medical community at large to utilize EUS-FNA to efficiently obtain accurate diagnoses that can lead to speedier patient recovery, fewer unnecessary operations, reduced patient and hospital medical expenses, and, most important of all, lead to better patient care.
A retrospective clinical analysis was performed, followed by statistical analyses.
Subjects for this research consisted of two hundred sixty-eight consecutive patients from computerized case files from 2008–2010 at Rambam Healthcare Campus in Haifa, Israel. These subjects were chosen from the hospital's gastrointestinal and cytology departments, all of whom had undergone EUS-FNA by Rambam gastroenterologists and have had cytological analysis performed. This population is representative of the population of Haifa, its immediate surroundings, and various communities of northern Israel.
Patient files were analyzed for management and diagnosis before and after EUS-FNA. Any change in diagnosis and/or treatment was noted, such as more aggressive or more conservative, more or fewer tests being performed, with chemotherapy and surgery among the various possibilities. Moreover, methods to improve the diagnostic accuracy of EUS-FNA were considered as to limit potential errors such as inadequate FNA samples or morbidities associated with EUS-FNA. Demographics including age and gender were noted.
Data was collected from patient files and results were charted according to the target region of EUS-FNA aspiration and also according to overall results. Data was arranged into pre-EUS-FNA and postaspiration groups as described in Hirdes et al. [
A total of 268 patients' files comprised the study sample. The mean patient age was 66.6 years old. The majority (68%) of FNA samples were taken from the pancreas, with other frequent targets being the stomach, mediastinum, and abdominal lymph nodes (Figure
EUS-FNA by region and diagnostic accuracy.
Region | Frequency | FNA nondiagnostic or false negative | FNA diagnostic (%) |
---|---|---|---|
Pancreas | 201 | 16 | 92.04 |
Abdominal lymph nodes | 27 | 7 | 74.07 |
Stomach | 20 | 0 | 100.00 |
Mediastinum | 21 | 2 | 90.48 |
Esophagus | 13 | 0 | 100.00 |
Liver | 5 | 0 | 100.00 |
Duodenum | 4 | 0 | 100.00 |
Adrenal | 3 | 0 | 100.00 |
Ascites/other | 4 | 1 | 75.00 |
| |||
Total | 268 | 20 | 92.54 |
Various regions sampled by EUS-FNA cytology, showing frequency of FNA, number of false negatives or nondiagnostic samples, and overall diagnostic accuracy are displayed.
Total includes several cases of overlap in patients where multiple regions were sampled by FNA. For example, a patient had FNA biopsies taken from his pancreas and abdominal lymph nodes.
The chart represents the frequency EUS-FNA biopsy at analyzed regions. The pancreas was the most frequently biopsied organ.
226 of the total 268 patients (84.3%) lacked a definitive diagnosis prior to performing EUS-FNA. Examples commonly encountered were obstructive jaundice or a widened Wirsung duct that was found on computerized tomography (CT). After EUS-FNA, 134 of the cases were determined to be benign (59.3%), 67 cases of malignancy were found (29.6%), and 7.5% of cases still lacked a definitive diagnosis following the procedure (Figure
Charted is a breakdown of the diagnoses after EUS-FNA, cases of which prior diagnoses initially were undetermined. This population represented the majority of patients (84.3%, 226/268).
EUS-FNA cytology proved useful in establishing the diagnosis in 248 out of 268 patients (92.54%). Sensitivity and specificity were established by evaluating diagnoses for changes prior to and after EUS-FNA. Prior to FNA, 27 of the 268 patients (10.1%) were determined to have benign lesions, 12 (4.5%) patients had malignant conditions, and 3 patients (1.1%) had lesions of chronic inflammation designed as “other.” Positive values were indicated as malignant, and negative values were labeled as benign.
EUS-FNA diagnosed 10 out of 10 malignant cases for 100% positive predictive value (PPV). The PPV value indicates that 100% of cases having known malignancy tested positive for malignancy after EUS-FNA, indicating the importance of a positive result that EUS-FNA is diagnostic. This also indicates 100% specificity in ensuring no false positives, meaning that a lesion found to be malignant by EUS-FNA had a 100% of being malignant. EUS-FNA did not detect two other cases of malignancy, instead giving false negatives results that resulted in 83.3% (10/12) sensitivity. False negatives represent lesions that EUS-FNA cytology determined as benign, but were soon after diagnosed as malignant. Therefore, the 83.3% sensitivity indicates that a negative result by EUS-FNA has an 83.3% chance of being benign, but does not always rule out a malignant condition (see Table
The greatest percentage of false negatives or nondiagnostic FNAs was found in analysis of abdominal lymph nodes. 27 cases were identified in which EUS-FNA from abdominal lymph nodes was performed, two-thirds of which involved additional sites of EUS-FNA. Of these 27 patients, seven cases of nondiagnostic or false negative were found (25.9%), The majority of such nondiagnostic or false negative cases (6/7) of abdominal lymph nodes either involved FNA the nodes alone or with accompanying FNA from the pancreas (Table
FNA results for abdominal lymph nodes.
Region | Cases | FNA nondiagnostic or false negative |
Percent of FNA diagnostically useful |
---|---|---|---|
Nodes alone |
9 | 3 | 66.67 |
With pancreas |
8 | 3 | 62.50 |
With stomach |
2 | 0 | 100.00 |
With liver |
2 | 0 | 100.00 |
With adrenal |
1 | 0 | 100.00 |
With mediastinum |
3 | 0 | 100.00 |
With mediastinum |
1 | 1 | 0.00 |
With other (ascites) |
1 | 0 | 100.00 |
| |||
Total | 27 | 7 | 70.47 |
Displayed in this chart are the various regions from which EUS-guided FNA was performed on abdominal lymph nodes either alone or in combination with other organs. Alongside each value is the number of cases that the FNA proved unhelpful (nondiagnostic or false negative) in establishing the patient’s diagnosis.
nondiagnostic EUS-FNA cases.
Reason | Frequency | Percent of total |
---|---|---|
Number of FNA passes |
||
3 passes |
2 |
10.00 |
2 passes |
5 |
25.00 |
1 pass |
3 |
15.00 |
Not listed |
10 |
50.00 |
| ||
Total |
20 |
100.00 |
| ||
Other reasons |
||
Unidentifiable cause |
10 |
50.00 |
Minimal fluid withdrawn |
8 |
40.00 |
Difficult to pass needle* | 3 |
15.00 |
Suspected autoimmune or chronic inflammation |
2 |
10.00 |
False negatives |
8 |
40.00 |
| ||
Total |
20 |
100.00 |
| ||
Sensitivity |
83.3% | |
Specificity |
100.00% | |
Positive predictive value |
100.00% | |
Negative predictive value |
91.6% | |
Accuracy |
94.1% |
Listed is a summary of suspected causes of nondiagnostic and false negative FNA cases, along with sensitivity, specificity, PPV, NPV, and overall accuracy of EUS-FNA.
A total of 20 cases (7.5% of total) nondiagnostic and false negative FNA cases were identified. Eight of the twenty cases were false negatives; zero false positives were found. The column labeled “FNA nondiagnostic or false negative” includes the cases in which EUS-FNA and/or subsequent cytology could not effectively diagnose a suspected lesion or resulted in a false negative (Table
EUS-guided FNA can have a profound influence on patient management. Its diagnostic ability is one of its greatest assets. Patients with benign conditions do not require intensive treatment and usually routine patient followup is needed, depending on the nature of the pathology. In contrast, the patients diagnosed with malignancies are referred for surgical resection when possible, chemo/radiation therapy, and/or appropriate palliative care. For both benign and malignant diseases, quality assessments of care may point to ways to improve the service provided to our patients. In providing accurate diagnoses, EUS-FNA helps to establish proper patient care while avoiding futile, costly, and potentially risky procedures and operations.
In regard to location, EUS-FNA was most accurate in the esophagus, stomach, and adrenal and peri-GI tract areas (marked as “other” in the results section), demonstrating 100% accuracy. In the pancreas, which was the region of most frequent EUS-guided FNA usage (see Figure
Despite being a highly sensitive and selective method, EUS-FNA cytology was not equally accurate for several analyzed regions. One such area was in the assessment of abdominal lymph nodes, where the accuracy was 74.07%. Accuracy in assessing mediastinal lesions was 90.48% (see Table
Several factors can be responsible for the diminished accuracy of these selected areas. First, the number of passes made into the lesion of interest comes into question. In two out of the 20 cases (10.0%) of FNA being nondiagnostic or false negative, it was documented that three passes were made from the lesion by FNA. Five cases (25.0%) involved two passes and three cases (15.0%) were documented where only one pass was attempted to aspirate contents from suspicious lesions. This can be due to several reasons. Minimal fluid was reported to be withdrawn from eight of the 20 cases (40.0%), which likely made it difficult for cytologists to delineate or exclude a specific pathology if present (see Table
Lesions with very viscous mucous contents often are unsuitable for obtaining adequate fluid for chemistry (CEA and amylase) evaluation. Data from samples taken from EUS-FNA biopsies of pancreatic cysts have shown that a CEA level above 192 ng/mL is indicative of mucinous neoplasms in 79% of cases [
Issues that infrequently complicate EUS-FNA procedures are the vicinity of the lesion for biopsy and the condition of the patient. It was documented in three nondiagnostic FNA cases (15.0%) that fewer passes were taken of suspected lesions because the operator found it difficult to insert the needle into the lesion for biopsy. In one of these cases, multiple vessels surrounded the suspected mass, resulting in a narrow window to insert the needle for FNA. This may be remedied by using newer small 25 gauge or more flexible needles that can more easily penetrate hard lesions ensuring the ease of puncture and a greater quantity of aspirate extracted [
Occasionally, the documentation on patient case files was insufficient to determine a specific cause of nondiagnostic FNAs. There are two junior echoendoscopists currently working in Rambam Healthcare Campus, to whom may be attributed several of the nondiagnostic FNAs. One senior operator performed or attended greater than 90% of the EUS-FNA procedures. However, it cannot be confirmed if the nondiagnostic FNAs were due to operator inexperience because the patient files investigated did not list the name of the doctor who performed the respective EUS-FNA. Reports may be written by either the attending or the in-training echoendoscopist. Furthermore, recent literature showed as good results among newly trained echoendoscopists as compared to experienced self-trained ones. The number and type of needles used were rarely documented, factors which may impact the amount of material for cytology and chemical analyses. Moreover, local compensation for EUS is nationally standardized and strongly discourages usage of more than one needle. One additional method to ensure diagnostic accuracy of EUS-FNA samples is to have a cytologist present during the procedure to ensure the adequacy of the aspirated samples and has been shown in several centers to increase the accuracy of cytological diagnosis by 10–15% [
In analyzing the 268 patient files from 2008–2010, there was one reported case of FNA-related complication in which a minor hemorrhage occurred near the site of aspiration. This required the patient to be hospitalized for several additional hours before being released. No other cases of EUS-FNA morbidity or mortality caused by EUS-FNA were found; therefore, the morbidity rate is 1/268, or 0.37% with a mortality of 0.0%. This is in comparison to published benchmark data, which showed that the morbidity rate from EUS-guided FNA was 0.98% overall, including postprocedural pancreatitis, pain, and mortality [
As previously noted, accurate diagnoses were made possible by EUS-FNA in 92.54% of evaluated cases, which minimized the overall health and financial costs brought about by the amount of extra tests, chemotherapy, and fruitless operations that patients may undergo as a result of misdiagnosis. A previous report found that EUS-FNA markedly decreased the frequency of futile surgery for pancreatic cancers while additionally aiding in the tailoring of optimal individualized patient treatment according to the stage of the patient's cancer [
Limitations of this study included its retrospective nature at a single medical center. Although many of the computerized data files investigated contained fairly complete data, analysis was limited in a few files that consisted of incompletely documented reports. The study was also performed at a single healthcare center with its equipment and personnel. Therefore, it should be noted that the diagnostic accuracy of EUS-FNA may vary by several factors, including the experience of the center's echoendoscopists, cytologists, and lab personnel, and also by the sensitivity of its equipment. Rambam Healthcare Campus is also a training facility for many young doctors, and as such, there exist wide gaps in experience of the performing echoendoscopists, although a senior endoscopist supervises or consults on most procedures.
EUS-FNA was found to have 94.1% accuracy. Quality control analysis and literature suggest that this may be further improved by more thorough reporting of procedural information (perhaps with forcedfields in the electronic medical report), optimizing needle selection, and by having a senior echo-endoscopist and a cytologist present during each procedure.
Because patient files are their own personal property, this work underwent the formal approval of the institutional review board (IRB). In conducting this study, patients were not approached, nor were any adverse effects or harm brought upon them.
The authors would like to acknowledge the substantial contribution of the following staff members for involvement with/performing the clinical work: Varda Moscovitch, and Farres Tahibash.