Abbreviations
EBUS:
endobronchial ultrasound
EUS:
esophageal ultrasound
FNA:
fine needle aspiration
ROSE:
rapid on-site cytologic evaluation
Introduction
In 2005, Dr. Vilmann, a gastroenterologist and leader in the field of esophageal ultrasound
(EUS), who did some of the pioneering work on esophageal access of mediastinal structures
for the diagnosis and staging of thoracic diseases, stated: “But how can we proceed
to implement EUS-FNA as a routine procedure in respiratory medicine? Now most groups
performing EUS-FNA in the chest are still gastroenterologists, because the method
was originally developed in this specialty” [1]. This quote is as relevant today as it was then; it has subsequently been shown
that pulmonologists can effectively use the esophagus alone or simultaneously with
airway access using the endoscope originally designed for bronchoscopic ultrasound
and that pulmonologists can be trained to use the endoscope used by gastroenterology
[2]
[3]
[4]
[5]
[6]
[7], but EUS for the diagnosis of mediastinal disease is not widely performed by pulmonologists
and is, in fact, discouraged by some of them [8]. There have been few studies of performance characteristics of EUS vs. EBUS. In
our practice, we perform both EBUS and EUS and the approach chosen is based upon patient
anatomy. We performed a retrospective review of our database specifically to contrast
EBUS with EUS and to investigate relationships between route of access and other procedural
parameters.
Methods
After institutional review board approval, all ultrasound-guided endoscopies performed
by Interventional Pulmonary at a single institution from August 1, 2012 until April
30, 2013 were retrospectively reviewed. To allow head-to-head comparison of EUS and
EBUS procedures, patients requiring sedation for additional procedures such as regular
bronchoscopy or chest tube insertion were excluded. All procedures were performed
by one or both of the authors, both of whom are pulmonologists. At our institution,
the endoscope that was originally designed for EBUS (BF-UC180F, Olympus Medical Supply
Corporation, Melville, New York, United States) was used for both EBUS and EUS; however,
if indicated, crossover from one route of access to the other was readily possible.
(We limited our use of both techniques to the diagnosis of thoracic disease; endosonography
for primary gastrointestinal processes was performed by gastroenterology.) Positioning
for both (EBUS and EUS) was the same. Patients were supine on a stretcher with the
head elevated to about 30 degrees. All sampling was performed with the 21-gauge needle
made by Olympus for the BF-UC180F endoscope. Suction was routinely applied. The number
of punctures was determined by on-site adequacy evaluation and by the level of suspiction
for malignant involvment, and if so, the suspect lesions were sampled up to ten times.
However, a standard of at least three passes was performed on all lesions. Rapid on-site
cytologic evaluation (ROSE) was available for all procedures.
The following parameters were selected for comparison: type of procedure (EUS, EBUS,
EUS + EBUS), sedation dosing, number of sites biopsied, specific sites biopsied, yield,
procedure time, maximal oxygen flow during the procedure, time on oxygen post-procedure,
and total time from procedure termination (scope out) until discharge. Student’s t-tests were used to determine significance with P < 0.05 considered significant.
Results
One hundred and sixty-five ultrasound-guided needle aspiration biopsy procedures were
performed over the study interval. For three of these procedures, propofol was used
for sedation, making it impossible to do comparative analysis. For seven procedures,
the relevant data were not available on retrospective chart review. These ten procedures
were eliminated from the analysis, leaving 155 patient procedures for review. There
were 61 patients who underwent EUS alone, 73 patients who underwent EBUS alone, and
21 patients who underwent combined procedures. Age distribution, drug doses administered,
procedure times, oxygen flow rates, and recovery times are presented in [Table 1]. One hundred and fifty of the procedures were performed for both diagnosis and staging
of lung cancer. The remaining five procedures were performed to evaluate for possible
mediastinal metastases of cancers previously diagnosed. These procedures were performed
on four patients diagnosed with adenocarcinoma of the lung previously diagnosed using
CT-guided biopsy and one patient diagnosed with adenocarcinoma of the breast. The
midazolam dosing for EUS alone was lower than for EBUS alone (P < 0.0002) and for the combined procedure (P < 0.001). There was no significant midazolam dosing difference between the EBUS and
EUS + EBUS groups. The same pattern was found for fentanyl dosing; fentanyl dosing
for EUS alone was lower than for EBUS alone (P < 0.0001) and for the combined procedure (P < 0.002), although there was no significant fentanyl dosing difference between the
EBUS and EUS + EBUS groups. Mean number of sites sampled varied significantly between
groups and for EUS vs. the other two groups (P < 0.0001). The mean procedure time for EUS was approximately one-half the mean procedure
time required for EBUS (P < 0.0001). Procedure times for EBUS and EBU + EUS were not significantly different.
Mean maximal oxygen flow rate was lower for the EUS group. Time to discharge was shortest
for patients who underwent EUS (P < 0.0002).
Table 1
Data for all treatment groups: EUS, EBUS, and EUS + EBUS.
|
EUS (n = 61)
|
EBUS (n = 73)
|
EBUS + EUS (n = 21)
|
P value
EUS vs. EBUS
|
P value
EUS vs. EBUS + EUS
|
P value EBUS vs. EBUS + EUS
|
|
Age (years)
|
59.4 ± 14.5
|
61.7 ± 13.6
|
61.7 ± 13.7
|
0.36
|
0.78
|
0.71
|
|
Total sites
|
1.4 ± 0.6
|
2.0 ± 0.8
|
2.5 ± 0.9
|
< 0.0001
|
< 0.0001
|
0.03
|
|
Midazolam (mg)
|
3.4 ± 1.4
|
4.6 ± 1.9
|
4.3 ± 1.8
|
0.0002
|
0.001
|
0.59
|
|
Fentanyl (µg)
|
73.6 ± 27
|
98.3 ± 39.7
|
97.6 ± 35.3
|
0.0001
|
0.002
|
0.94
|
|
Procedure time (min)
|
13.1 ± 5.1
|
24.1 ± 8.4
|
26.9 ± 8.4
|
< 0.0001
|
< 0.0001
|
0.19
|
|
O2 flow (L/min)
|
5.3 ± 2.6
|
6.7 ± 3.4
|
5.8 ± 2.8
|
0.009
|
0.43
|
0.27
|
|
Time to room air (min)
|
18.9 ± 19.7
|
26.5 ± 23.9
|
35.5 ± 28.8
|
0.05
|
0.005
|
0.15
|
|
Time to discharge (min)
|
42.7 ± 15.9
|
55.7 ± 20.6
|
60.8 ± 21.9
|
0.0002
|
0.0002
|
0.33
|
Abbreviations: EUS, esophageal ultrasound; EBUS, endobronchial ultrasound.
Whereas the significant differences in characteristics could have been related to
the fact that fewer sites were sampled with EUS, we compared all procedures that involved
sampling of two or fewer sites with each approach (none of the combined approach procedures
involved sampling of fewer than two sites). Data are presented in [Table 2]. There were 56 such EUS procedures and 52 such EBUS procedures, with a mean of 1.4
± 0.6 sites for EUS and 1.6 ± 0.6 for EBUS (NS, P = 0.18). All data apart from maximal oxygen flow rates were significantly different
when the results from these two groups were compared (P < 0.001). Time to discharge remained shorter for the EUS group (P = 0.001).
Table 2
Data for sampling of fewer than two sites.
|
EUS (n = 56)
|
EBUS (n = 52)
|
P value
|
|
Total sites
|
1.4 ± 0.6
|
1.6 ± 0.6
|
0.18
|
|
Procedure time (min)
|
12.1 ± 5.4
|
21.8 ± 8.3
|
< 0.0001
|
|
Midazolam (mg)
|
3.4 ± 1.4
|
4.5 ± 1.7
|
0.0003
|
|
Fentanyl (µg)
|
72.1 ± 27.1
|
98.2 ± 34.6
|
< 0.0001
|
|
Maximal O2 flow (L/min)
|
5.3 ± 2.7
|
6.4 ± 3.4
|
0.06
|
|
Time to room air (min)
|
18.6 ± 16.9
|
27.2 ± 25.8
|
0.05
|
|
Time to discharge (min)
|
42.4 ± 15.8
|
55.1 ± 22.3
|
0.001
|
Abbreviations: EUS, esophageal ultrasound; EBUS, endobronchial ultrasound.
Patients in which both EUS and EBUS were performed (n = 21) were reviewed. In twelve
of these patients the sequence was EUS followed by EBUS, and in the remaining nine
patients it was EBUS followed by EUS. The most common reason for a combined procedure
(n = 13, 62 %) was lack of diagnosis from the first approach based upon ROSE; negative
findings from one approach led to accessing other sites, or (particularly with station
7) other areas of a nodal station to be certain that a significant pathologic process
had not been missed. In six of the 21 patients (28 %) the change in approach led to
a positive diagnosis that was originally missed using the first approach (2/12 EUS
→ EBUS, 4/9 EBUS → EUS). In one patient, initial on-site EUS cytology was negative,
however, a slide obtained by EUS before EBUS but processed after the change to EBUS
was found to be positive for malignancy. In one patient, we switched from EBUS to
EUS to obtain additional (station 7) material for flow cytometry because of patient
discomfort with the endoscope in the airways.
Sites accessed are listed in [Table 3]. As expected, the most frequently biopsied area was station 7, which was followed
by 4 R, 4 L, and 11 R. 4 R was occasionally accessible via the esophagus. A mass/structure
other than a node was accessed 13 % of the time (36/282 needle aspiration biopsies).
Mean nodal sizes are also listed in [Table 3] although nodal size did not impact the capacity to sample it. An exception to this
generalization is in 4 R from the esophagus; it is not generally accessible from the
esophagus, but in some instances it was so enlarged that esophageal access was possible.
The mean number of passes was 3.42 (minimum of 3 and maximum of 10).
Table 3
Nodal stations and masses sampled; listed by approach.
|
Single Approach
|
Combined Approach
|
|
Site
|
EBUS
|
Size (cm)
|
EUS
|
Size (cm)
|
EBUS
|
Size (cm)
|
EUS
|
Size (cm)
|
|
Station 1
|
|
|
1
|
1
|
|
|
2
|
1 ± 0.7
|
|
Station 2 R
|
4
|
2 ± 0.5
|
4
|
1 ± 0.5
|
|
|
1
|
0.8
|
|
Station 2 L
|
1
|
1
|
|
|
|
|
|
|
|
Station 3
|
4
|
1 ± 0.5
|
|
|
|
|
1
|
0.8
|
|
Station 4 R
|
41
|
1.7 ± 0.6
|
3
|
3 ± 1.5
|
9
|
0.8 ± 3
|
|
|
|
Station 4 L
|
10
|
0.87 ± 0.25
|
19
|
1.35 ± 0.9
|
|
|
4
|
0.7 ± 0.2
|
|
Station 7
|
34
|
1.98 ± 1.2
|
45
|
2.3 ± 1.15
|
5
|
2.6 ± 1.1
|
13
|
2.3 ± 1.9
|
|
Station 8
|
|
|
2
|
2 ± 1.5
|
|
|
|
|
|
Station 10 R
|
1
|
1.2
|
|
|
1
|
0.8
|
|
|
|
Station 10 L
|
2
|
1.25 ± 0.4
|
|
|
1
|
0.5
|
|
|
|
Station 11 R
|
18
|
1.18 ± 0.8
|
|
|
4
|
1.25 ± 0.5
|
|
|
|
Station 11 L
|
8
|
1.3 ± 0.6
|
|
|
3
|
1.4 ± 0.8
|
|
|
|
Station 12 R
|
3
|
0.6 ± 0.2
|
|
|
1
|
0.5
|
|
|
|
Station 13 R
|
1
|
0.5
|
|
|
|
|
|
|
|
Mass
|
19
|
|
8
|
|
7
|
|
2
|
|
|
Totals
|
142
|
|
86
|
|
31
|
|
23
|
|
Abbreviations: EUS, esophageal ultrasound; EBUS, endobronchial ultrasound.
Seventy EUS procedures were performed (61 as the initial procedure, nine as crossover
studies). There were two false negatives, which were documented by crossover to EBUS. In
one patient, station 7 was negative from the esophagus and positive from the airway,
and in a second patient, station 7 was not visualized from the esophagus, but was
observed and was positive from the airway. EUS was diagnostic in the remaining 97 %
of patients: pathologic diagnoses among 73 % and no pathology found among 23 % of
patients (sensitivity, 0.96; 95 %CI 0.86 – 0.99; specificity, 1; 95 %CI 0.76 – 1).
Negatives were shown to be negative for pathology by stability or regression over
a two-year follow-up or, occasionally, by surgical resection. Eighty-two EBUS studies
were performed. Four false negative results were documented by crossover to EUS, which
yielded three station 7 nodes and one 4 L node. EBUS was diagnostic in the remaining
81 procedures examined, with 74 % pathologic and 20 % benign diagnoses (sensitivity,
0.92; 95 %CI 0.84 – 0.98; specificity, 1; 95 %CI 0.74 – 1). Once again, a benign designation
was substantiated by two-year follow-up and occasionally by surgery. If the crossover
studies were examined as a separate subset, diagnostic accuracy was 100 % (74 % pathologic,
26 % benign, confirmed with two-year follow-up or with surgical resection).
Complications were rare. Two patients experienced severe post-procedural hypoxia (one
EBUS and one EBUS + EUS). One patient had mediastinal bleeding related to the procedure
(EBUS + EUS). All three of these patients were admitted for observation and discharged
the following day. No interventions were required.
The primary objective of this study was to analyze differences in procedure characteristics
related to differences in approach, but the data were also analyzed with respect to
whether or not a specific diagnosis was achieved. Achieving a specific diagnosis did
have a significant impact upon study characteristics; for all procedures combined,
studies for which a diagnosis was achieved involved sampling of 1.8 ± 0.8 stations
over 19.3 ± 9.7 min, whereas studies without specific diagnoses involved sampling
of 2.4 ± 0.8 stations over 27.3 ± 7.1 min (P = 0.007 and P = 0.017, respectively).
Discussion
Endoscopic ultrasound has revolutionized the diagnosis of mediastinal abnormalities
and the nodal staging of lung cancer. The first ultrasound endoscope to be developed
was for the gastrointestinal tract [9]. EUS was initially used for the diagnostic evaluation of diseases of the gastrointestinal
tract, but its application to lung cancer diagnosis and staging was rapidly conceptualized
and brought to fruition [10]. The first clinical use of the smaller-diameter convex curvilinear ultrasound bronchoscope
(EBUS scope) was reported in 2004 [11], and EBUS as a diagnostic tool has been widely adopted by the pulmonary community.
It was subsequently shown both that the EBUS scope can be used for EUS and that pulmonologists
can be trained to use the gastrointestinal endosonoscope [2]
[3]
[4]
[5]
[7]. EUS with the EBUS endoscope has not, however, been widely adopted by the pulmonary
community, and prominent pulmonologists have questioned the true value of adding EUS
to EBUS and stated that EBUS is the initial procedure of choice: “We agree that EBUS-TENA
(their term for EUS-FNA with EBUS scope) should be used only in circumstances when
lymph node stations are difficult or are not accessible by TBNA.” [8] Their bottom line was that EUS using either endoscope might be better tolerated
but was rarely indicated for the diagnosis of thoracic disease.
Two prospective studies have examined EBUS vs. EUS. Kang et al investigated the value
of EBUS vs. EUS in the diagnosis of potentially resectable lung cancers [12]. They used the same endoscope for both procedures. Their study randomized initial
approach (EBUS vs. EUS) but mandated a crossover. They found that if they did EUS
first, adding EBUS contributed significantly to diagnostic yield, but if they did
EBUS first there was not a significant increase, leading to the conclusion that there
is no complementary role for EUS. Kang et al. were unable to comment on parameters
such as sedation differences between EBUS and EUS given the mandated crossover. They
concluded that EBUS is the appropriate initial approach, and the lack of complementary
data from EUS would lead to the conclusion that EBUS is the only approach that should
be used. They did comment in their discussion that, “EUS-FNA is generally well tolerated,
and procedure tolerance may affect the selection procedure.”
In a second prospective study, Oki et al randomized all candidates for endosonography
with lesions accessible from both airway and esophagus and performed either EBUS-TBNA
alone or EUS-FNA alone while monitoring performance characteristics [13]. With both study groups sedated to the extent that there was equal patient procedure
tolerance, diagnostic EUS-FNA was associated with lower doses of lidocaine (P < 0.001) and sedatives (P = 0.02), shorter procedure times (P < 0.001), and fewer oxygen desaturations (P < 0.001). Also noted was that endoscopists preferred the esophageal approach (P < 0.001).
Our study has the disadvantage of not being prospective/randomized and the advantage
of reflecting the application of both approaches to routine clinical practice. In
our laboratory, we use the same scope for both EBUS and EUS; there is no gradient
of experience that leads us to choose one route over the other. It has been recurrently
suggested in the literature that EUS is better tolerated than EBUS [3]
[8]
[12]
[14]
[15]
[16]. Our data combined with those of Oki et al provide objective confirmation for this
impression. When EUS alone sufficed to achieve our clinical goals, we were able to
do so with statistically significant economies of sedation, oxygen requirements, and
recovery time. This was true when we looked at all diagnostic studies regardless of
required sampling and remained true when we limited analysis to sampling of an equivalent
number of nodes. When a crossover study was performed, total procedure time, sedation,
oxygen requirements, and recovery for the combined study were significantly higher
than for EUS alone but not significantly different from those for EBUS alone; there
was no negative impact upon these parameters from having started with EUS.
In our study with no mandated crossover, we based the initial approach upon anatomy
as defined by computed tomography scans. In so doing, we did more EBUS than EUS, reflecting
the fact that more nodes were accessible via EBUS than via EUS. The numbers were not,
however, dramatically different; 71 patients were diagnosed by EBUS alone and 61 by
EUS alone. The major differences in sedation for the two routes occurred 1) at the
time of insertion across the vocal cords and, subsequently, 2) when cough or discomfort
made us wait for additional medication (systemic or endobronchial) to take effect,
or 3) when patient desaturation made us pause. However, nodes are not more difficult
to “see” or to reach from the airway vs. the esophagus, and identification of any
specific node takes a matter of seconds. In a calm patient, one can reach 4 R from
the airway just as fast as 8 R from the esophagus. We would thus say that nodal anatomy
was not a factor. Crossover to from EUS to EBUS (or vice versa) is always an option
in our lab, and in our study EBUS was just as likely as EUS to require a crossover
procedure. The most important finding from the crossover data are that the two approaches
were complementary; in patients not diagnosed by the initial approach, crossover led
to a diagnosis 28 % of the time. As noted, Kang et al. did not find EUS and EBUS to
be complementary, [12] but there is a significant body of literature that, like our data, suggests that
they are [4]
[5]
[17]
[18]
[19].
This study examined needle aspiration of solid tissue structures with EUS and EBUS,
and we alternated between the two approaches in some patients in both “directions”
(EUS to EBUS and EBUS to EUS). The published infectious complication rate was extremely
low by both routes [20]
[21]. Neither EUS nor EBUS is a “clean” procedure; the “dirtiest” structure involved
in these procedures is the mouth, and both EBUS and EUS involve passing the endoscope
through this bacteria-laden environment. We consider this to be the dominant etiology
of risk and have felt that entry into the esophagus before the trachea does not increase
that risk. Notably, aspiration of cystic structures has been associated with a higher
risk [21]; we did encounter one of these situations. Based upon our understanding, we have
routinely performed EBUS and EUS-FNA interchangeably, with the first procedure performed
to a) maximize yield (diagnosis, staging if relevant) and b) minimize risk and discomfort.
EUS-FNA is frequently the leading procedure. To date we have performed over 1000 combined
procedures, and we have yet to see an infectious complication.
In summary, Villman’s vision of EUS becoming a routine pulmonary procedure has not
been realized; only a small fraction of the pulmonary community has incorporated EUS,
some using the EBUS endoscope and an even smaller minority using the gastrointestinal
endoscope. The data of Kang et al would lead to the conclusion that that EBUS alone
is all that is required for evaluation of diseases involving the mediastinum [12]. The data of Oki et al coupled with our data leads to different conclusions [22]. The impression that EUS is better tolerated is now supported by data demonstrating
efficiencies of sedation, time, and oxygenation with EUS vs. EBUS. This study demonstrates
advantages of EUS, regardless of whether the operator is a pulmonologist or a gastroenterologist.
The advantages of pulmonary alone performing both with the same endoscope include
economies of time, sedation, and equipment and a seamless transition between modalities.
In an institution with separate physicians performing EBUS and EUS, we would suggest
careful pre-review of the radiologic data and a collaborative procedure if both routes
of access are likely to yield important data. Based upon our data, we conclude that
if EUS might adequately establish a diagnosis (and, if relevant, a stage), then EUS
should be the first procedure. This is particularly applicable to individuals with
marginal physiologic reserve. Finally, EUS and EBUS do indeed have complementary roles.
Integration of the two procedures can be achieved either via cross-training by pulmonologists
or by collaborative procedures including both pulmonology and gastroenterology.
