Introduction
Endoscopic ultrasound (EUS) allows one to identify the presence of lymph nodes in
the mid and posterior mediastinum, the perigastric area and the celiac axis region
[1]. EUS, by means of endoscopic ultrasound guided fine needle aspiration biopsy (EUS
FNA), may also be employed for procurement of a tissue diagnosis in a minimally invasive
fashion [1]. This has been proven to be useful in a number of gastrointestinal and pulmonary
malignancies. This review article focuses on nodal staging of esophageal cancer and
attempts to answer the following questions: 1. Number or site of nodes for nodal staging?
How to improve accuracy of nodal staging? 2. Is FNA always necessary? 3. Junctional
tumors - what's N and what's M?
Preoperative staging of esophageal carcinoma
Patients diagnosed with esophageal carcinoma should be first evaluated for presence
of comorbid conditions (mainly cardiac or pulmonary diseases) that may preclude the
patient from undergoing surgery [2]
[3]
[4]
[5]. Those patients who are fit for surgery should undergo preoperative tumor staging.
First thing required in fit patients is to exclude the presence of distant metastasis
or unresectable disease. For such purpose, helical CT scan and more recently Positron
Emission Tomography (PET) scanning are typically employed (diagnostic accuracy 70
% to 80 %) [6]. Those patients that appear to be resectable based on those imaging techniques should
pursue a more detailed local-regional staging (T and N stage) (Table [1]) [6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]. EUS has been proven to be most accurate imaging technique for such purpose [6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20] (Table [2]).
Table 1 TNM and Stage Grouping for Esophageal Carcinoma (A.J.C.C. classification) [20]
|
Primary Tumor (T):
|
Distant Metastases (M):
|
| Tx: |
Primary tumor cannot be assessed |
Mx: |
Distant metastases cannot be assessed. |
| T0: |
No evidence of primary tumor. |
M0: |
No distant metastases. |
| Tis: |
Carcinoma in situ |
M1: |
Distant metastases. |
| T1: |
Tumor invades lamina propria or submucosa. |
• |
Tumors of the lower thoracic esophagus
:
|
| T2: |
Tumor invades muscularis propria. |
|
M1a : Metastases in celiac lymph nodes. |
| T3: |
Tumor invades adventitia. |
|
M1b: Other distant metastases. |
| T4: |
Tumor invades adjacent structures. |
• |
Tumors of the mid-thoracic esophagus:
|
|
|
|
M1a: Not applicable. |
|
Regional Lymph Nodes (N):
|
|
M1b: Non-regional lymph nodes and/or |
| Nx: |
Regional lymph nodes cannot be assessed. |
|
other distant metastases. |
| N0: |
No regional lymph node metastases. |
• |
Tumors of the upper thoracic esophagus:
|
| N1: |
Regional lymph node metastases. |
|
M1a: Metastases in cervical nodes. |
|
|
|
M1b: Other distant metastases. |
|
|
|
|
|
Stage Group:
|
T stage
|
N stage
|
M stage
|
| Stage 0: |
Tis |
N0 |
M0 |
| Stage I: |
T1 |
N0 |
M0 |
| Stage IIA: |
T2/T3 |
N0 |
M0 |
| Stage IIB: |
T1/T2 |
N1 |
M0 |
| Stage III: |
T3 |
N1 |
M0 |
| Stage IV: |
Any T |
Any N |
M1 |
| Stage IVA: |
Any T |
Any N |
M1a |
| Stage IVB: |
Any T |
Any N |
M1b |
Table 2 Preoperative TN staging accuracy of CT and EUS in esophageal carcinoma patients [19]
|
Patients
(n) |
T Stage (range)
(%) |
N Stage (range)
(%) |
| CT |
1 154 |
45
(40 - 50) |
54
(48 - 71) |
| EUS |
1 035 |
85
(59 - 92) |
77
(50 - 90) |
1. Number or site of nodes for nodal staging? How to improve accuracy of nodal staging?
Lymph node staging of esophageal cancer patients by EUS has classically relied on
4 EUS features that are based on size (lymph node width greater than 10 mm), shape
(round), echogenicity (echopoor pattern) and lymph node border (smooth), that have
been proposed to be suggestive of malignancy [21]
[22]. Although these criteria allow one to differentiate benign from malignant lymph
nodes in esophageal carcinoma patients with an accuracy close to 80 %, none of them
is diagnostic of malignancy alone [21]
[22]. It has been reported that when all four features suggestive of malignancy are present,
there is an 80 - 100 % probability that the lymph node is malignant; unfortunately,
only 25 % of malignant nodes will present all 4 criteria diagnostic of malignancy
[21]
[22]. These results demonstrate limitations of current EUS criteria for preoperative
determination of lymph node staging.
Efforts have been made to improve EUS accuracy for preoperative lymph node staging
of esophageal cancer, and to provide a more reliable diagnosis [23]
[24]
[25]. It has been proposed that by adding some additional criteria to the classical EUS
lymph node criteria, one may better predict which lymph nodes are benign or malignant
in nature [23]. These ”modified EUS malignant lymph node criteria” consist in the 4 classical ones
(Lymph node ≥ 5 mm in width; round shape; hypoechoic pattern; smooth border) plus
3 additional criteria that include lymph node location (celiac region), number of
lymph nodes identified on EUS exam (≥ 5) and presence of an advanced T stage tumor
based on EUS exam (Tumor was T3/4) [23]. These additional features included in the modified EUS criteria arose from prior
investigations performed in lymph node staging of esophageal carcinoma [26]
[27]
[28]. Performance characteristics (sensitivity, specificity and overall accuracy) of
the Standard and Modified EUS malignant lymph node criteria were assessed in a prospective
cohort of 144 patients (84 % of patients were N1, and 16 % were N0), with results
being compared by using the Receiver Operating Characteristics (ROC) curves, with
the Area Under the Curve (AUC) representing the overall diagnostic accuracy of EUS
lymph node criteria (standard and modified). ROC curves allow one to measure the sensitivity
and specificity of each set of criteria at different cut off points, determining the
most accurate cut off point to differentiate between benign and malignant lymph nodes
(knowing that an AUC of 0.90 - 1.0 is excellent, AUC of 0.80 - 0.90 is good, AUC of
0.70 - 0.80 is fair, AUC of 0.60 - 0.70 is poor, and AUC of 0.50 - 0.60 implies the
test fails). According to the gold standard adopted in this study, modified EUS criteria
for lymph node staging were found to be more accurate than standard criteria (ROC
curve/area under the curve 0.88 vs 0.78, respectively (Fig. [1]). In other words, larger numbers of lymph nodes, celiac lymph nodes, and deeper
mural invasion all portend greater likelihood of N1 disease (26 - 28). The maximum
accuracy of modified EUS criteria (86 %) was observed when presence of 3 or more of
the 7 modified EUS criteria were required to diagnose lymph node malignancy (Fig.
[1]). Although no single criteria was predictive of malignancy, the cut off number of
6 or more of the modified criteria (cut off at that point) provided a specificity
of 100 % and in that cohort of patients (prevalence of malignant lymph nodes: 84 %)
had a positive predictive value of 100 % (40/40, 95 % CI: 90 - 100 %) (Fig. [1]). Thus all patients with one or more lymph nodes with ≥ 6 modified EUS criteria
had N1 stage disease in that cohort of patients. Additionally, the cut off of 1 or
more of the modified EUS criteria (cut off at that point) provided a sensitivity of
100 % and in that cohort of patients (prevalence of benign lymph nodes: 16 %) the
negative predictive value was 100 % (21/21, 95 % CI: 82 - 100 %) if zero criteria
were present (Fig. [1]). That is, none of the patients included in that study who had lymph nodes with
< 1 positive modified EUS criteria presented with N1 stage disease. A detailed analysis
of single criteria employed for lymph node diagnosis (multiple logistic regression
analysis) disclosed that lymph node width ≥ 5 mm, round shape, ≥ 5 lymph nodes identified
on EUS exam, and T3/4 tumor as per EUS assessment were the criteria that best predicted
malignancy in that cohort of patients.
EUS accuracy depends on operator's experience [29]
[30]
[31]
[32]. Several controlled studies have demonstrated that experienced (> 50/75 EUS exams
in esophageal cancer cases), but not inexperienced endosonographers (< 20 EUS exams
in esophageal cancer) have good agreement and high accuracy for N staging of esophageal
cancer [29]
[30]
[31]
[32]. A number of technical factors (balloon overinflation, oblique scanning, and inadequate
use of higher scanning frequencies) appear to be responsible for errors among inexperienced
endosonographers [29]
[30]
[31]
[32]. Mistakes occurred in lymph node staging of esophageal cancer have been associated
with benign lymph nodes larger that 10 mm (overstaging them as N1), and understaging
of lymph nodes as N0 in cases of microscopic infiltration of the node by tumor that
is beyond the resolution capabilities of the echoendoscopes.
Fig. 1 Performance charateristics for preoperative lymph node staging of esophageal carcinoma
[23].
2. Is FNA always necessary?
Some investigators have suggested that EUS FNA may help improve the accuracy of N
staging by providing cytologic confirmation of malignant spread of the disease [1]
[23]
[24]
[25]
[26]. This hypothesis is based on prior studies that have clearly demonstrated EUS FNA
has an elevated sensitivity, specificity and accuracy for peri-intestinal lymph nodes
[1]
[23]
[24]
[25]
[26]. Retrospective studies have shown EUS FNA accuracy for esophageal cancer lymph node
staging is superior to EUS alone (accuracy EUS vs EUS FNA 70 versus 93 per cent, p
< 0.05) [23]
[24]
[25]
[26]
[33]. These results were confirmed by a later study conducted at the Mayo Clinic comparing
in a prospective and blinded fashion the performance characteristics of helical CT,
EUS and EUS FNA for preoperative lymph node staging of esophageal carcinoma [34]. Authors followed a strict algorithm for lymph node selection, starting by sampling
celiac lymph nodes (if present), continuing by lymph nodes located in the perigastric
space and finally sampling lymph nodes located in the periesophageal space in a non
peritumoral location (Fig. [2]). The endosonographers were blinded to CT findings, and committed to a N stage prior
to performing the EUS-FNA part (EUS-FNA was performed with the asistance of an on
site cytopathologist). Table [3] demonstrates the superior accuracy of EUS FNA over EUS and helical CT.
In another study (previously mentioned in this review), the authors aimed to identify
a newly defined combination of EUS lymph node criteria that have sufficient sensitivity
and specificity to preclude the need for EUS FNA in certain clinical situations [23]. For such purpose, the modified set of criteria previously described were designed
and evaluated. Results of the study showed that all patients with one or more lymph
nodes with ≥ 6 modified EUS criteria had N1 stage disease, and none of the patients
included in this study who had lymph nodes with < 1 positive modified EUS criteria
presented with N1 stage disease. Therefore, in those patients with either ≥ 6 or <
1 criteria, EUS FNA may be avoided as we may predict if the lymph node is malignant
or benign in nature, and EUS FNA results are unlikely to change lymph node stage in
those patients. When direct costs assotiated with a routine EUS FNA approach for esophageal
cancer staging vs a selective EUS FNA approach (EUS FNA is avoided in those patients
with ≥ 6 or < 1 lymph node criteria, 42 % of patients included in that study) was
compared, the selective EUS FNA practice was associated with a cost saving of $117.84
per patient ($16,968.37 in the total cohort of patients evaluated) (2004 US Medicare
reimbursement fees employed for calculations). Although these numbers may differ depending
on the payer, procedure reimbursement, clinical setting and country, avoiding EUS
FNA in 42 % of esophageal carcinoma patients will reduce costs without diminishing
test accuracy.
Fig. 2 Algorithm for selection of lymph nodes undergoing EUS FNA in preoperative staging
of esophageal cancer [34].
Table 3 Prospective lymph node staging of esophageal carcinoma: CT, vs EUS vs EUS FNA [34]
|
Sensitivity |
Specificity |
Accuracy |
|
n % (95 % C.I.)
|
|
|
|
| CT |
14/48
29 % (17 %, 44 %) |
25/28
89 % (72 %, 98 %) |
39/76
51 % (40 %, 63 %) |
| EUS |
34/48
71 % (56 %, 83 %) |
22/28
79 % (59 %, 92 %) |
56/76
74 % (62 %, 83 %) |
| EUS FNA |
40/48
83 % (70 %, 93 %) |
26/28
93 % (77 %, 99 %) |
66/76
87 % (77 %, 94 %) |
|
p-value
|
|
|
|
| CT vs EUS |
< 0.001* |
0.257 |
0.003* |
| CT vs EUS FNA |
< 0.001* |
0.655 |
< 0.001* |
| EUS vs EUS FNA |
0.058 |
0.102 |
0.012* |
3. Junctional tumors - what's N and what's M?
Tumors located at the distal esophagus and/or invading the gastroesophageal junction
are usually adenocarcinoma, and present a characteristic and distinct pathway for
lymph node spread of the disease. This type of tumor tends to spread to the periesophageal
ganglia first, and then into the abdomen (subcardial area, gastrohepatic ligament
and celiac axis region). More recent TNM classifications have considered that lymphatic
spread of the disease into the celiac axis region represents metastatic disease, and
therefore those patients have been classified as M1a disease, as this implies a prognosis
similar to patiens with liver metastasis [20]. Patients with M1 disease are not considered candidates for cure, and palliative
measures are typically indicated. For this reason, it seems important to look carefully
at the celiac axis area to investigate for presence of metastatic spread of the disease
to distant lymph nodes. Furthermore, it appears to be important as well, to confirm
the nature of celiac lymph nodes by means of cytology/histology, as important treatment
decisions will depend on this. Although prevalence of celiac lymphadenopathy depends
on the type of esophageal cancer patients evaluated at each institution, some reports
have found that 40 of 198 esophageal cancer patients undergoing EUS examination for
preoperative staging may present malignant appearing lymph nodes in the celiac axis
area [35]. From those 40 patients, EUS-FNA was able to prove malignancy in the nodes in 31
patients (78 percent). From the remaining nine patients, 8 had no malignant lymph
(correctly assessed by EUS-FNA and confirmed by surgery), and one patient was a false
negative by EUS-FNA. These excellent results have been confirmed by other studies
conducted at different institutions, in which EUS was shown to have a sensitivity
for celiac lymph node detection of 77 % (95 % CI: 67 - 88 %), and a specificity of
85 % (95 % CI: 74 - 96 %), with an overall accuracy for EUS-FNA of celiac lymph nodes
as high as 98 % (95 % CI: 90 - 100 %) (25). These results support the use of EUS and
EUS FNA for the evaluation of the celiac axis region in search for metastatic nodes.
Summary
In summary, EUS and EUS FNA are the most accurate techniques for preoperative local-regional
staging of esophageal carcinoma, once CT and/or PET scan have excluded the presence
of distant metastasis. An experienced endosonographer, familiar with the EUS technique,
and the inclusion of lymph node location, number of nodes and EUS T stage as diagnostic
criteria may help improve preoperative lymph node staging accuracy in esophageal cancer.
EUS FNA helps improve diagnostic accuracy in esophageal cancer staging. However, in
patients with a certain number of EUS lymph node criteria, EUS FNA may be avoided
without affecting diagnostic accuracy. In tumors of the distal esophagus detection
and biopsy of celiac lymph nodes may be successfully perdormed by EUS and FNA.
