Predicting Malignancy Using Endoluminal Ultrasound Characteristics in Mediastinal Lymph Nodes
Lung cancer is the leading cause of cancer-related death in the United States with a 5-year
survival rate of 15,7%.1 In lung cancer and other malignancies, such as esophageal cancer,
accurate staging is essential to establish prognosis and for patient management. Staging
will help determine if surgery, chemotherapy, radiation therapy, a combination of these, or
a palliative approach is the most adequate.2,3
Assessment of lymph nodes (LNs) for invasion by metastasis is undertaken if patient
management is influenced by a positive or negative result. In non-small cell lung cancer
(NSCLC), mediastinal LN invasion by malignant cells is a critical factor to determine if a
patient is considered as a surgical candidate or not.2 In esophageal cancer, LN metastasis
may be found in the mediastinum, the neck and the abdomen.4 Invasion of these LNs must be
considered in the selection of treatment of esophageal cancer.3,4
There is no single method to investigate mediastinal LN invasion. Hence, a patient may have
to undergo several tests and procedures. Noninvasive and invasive approaches are employed.
Within the invasive techniques, endoscopic ultrasonography with needle aspiration (EUS-FNA)
and endobronchial ultrasonography with transbronchial needle aspiration (EBUS-TBNA) are
gaining importance in mediastinal staging.5 They provide ultrasonographic images and permit
needle aspiration under direct vision for cytology specimen analysis. As more evidence is
being accumulated on these staging approaches, the number of cervical mediastinoscopies,
considered as the gold-standard for mediastinal staging, is diminishing.6
Characteristics of benign and malignant LNs using ultrasonography have been described in
various contexts. Cervical and axillary LNs provided accurate characteristics (eg. shape,
short- to long-axis ratio, number of blood vessels in LN, vascularization pattern, etc.) in
predicting malignancy with linear superficial transducers.7-11 LN characteristics using EUS
or EBUS also have established criteria predicting mediastinal LN malignancy (eg. size,
shape, borders and echogenicity).12 Depending on the studies, it has been suggested to
obtain specimens of all accessible LNs without considering their characteristics. It has
also been suggested to proceed to a selective LN aspiration based on LN characteristics,
with the advantage of cost reduction (fewer needles used, less time to carry out procedure,
and specimens needed to be analysed).6
Color Doppler LN characteristics with endoluminal ultrasound (US) is only mentioned in a
small number of studies and needs to be further investigated.13,14 With the good results
obtained with superficial US, it seems reasonable to believe that color Doppler
characteristics would increase accuracy in detecting malignancy of mediastinal LNs with
New processors and probes are more accurate than the ones used in some of the past studies.
Hence, the investigators can possibly obtain more precise findings than the ones the
investigators can find in the literature.14
This study will consist of a prospective assessment of LN characteristics in patients
undergoing EUS-FNA and/or EBUS-TBNA for diagnostic purposes. Cytology results will be
considered as the gold-standard. By comparing the LN characteristics with the cytology
results, the investigators will establish a scoring system to accurately predict malignancy.
The impact of this study is significant. If clinicians can accurately predict malignancy in
mediastinal lymph nodes using ultrasound characteristics, biopsy of these lymph nodes could
be avoided which would decrease the risk to the patient and the cost to the system
(disposable needle costs, cytological preparation and analysis).
With a combination of endoluminal ultrasound characteristics, a simple scoring system can be
established to help predict malignancy in mediastinal lymph nodes.
1. Establish a scoring system to accurately predict mediastinal lymph node malignancy with
2. Retrospectively evaluate the impact of the established scoring system on the number of
lymph nodes aspirations required for accurate staging.
Prospective, cohort study design. Clinical trial: Intervention - endoluminal ultrasound.
This study will consist of a prospective non-randomized evaluation of endoluminal ultrasound
characteristics in mediastinal lymph nodes using endoluminal ultrasound to predict
malignancy within these nodes. All patients undergoing EUS-FNA and/or EBUS-TBNA for
mediastinal LN staging for any suspected or confirmed malignancy at the CHUM - Hôpital
Notre-Dame will be approached in an attempt to enroll them in the study. Patients will be
enrolled and consented for the study during the pre-operative evaluation in the
pre-operative clinic. The investigators will aim to recruit 100 consecutive patients.
Consented patients will undergo EUS with or without FNA and/or EBUS with or without TBNA in
the same way as planned or decided during the procedure by the surgeon. During the
procedure, data will be collected, including lymph node size, shape, border, echogenicity,
homogeneity, location, color Doppler characteristics, number of nodes, etc. This will not
add additional steps or time to the procedure. All this data and the cytology results
obtained from EUS-FNA and EBUS-TBNA will be entered in a prospective database.
The cytology results will be considered as the gold-standard and with the collected
characteristics univariate and multivariate logistic regressions will be used to establish a
scoring system to predict malignancy. Sensitivity, specificity, positive predictive value,
negative predictive value and accuracy will be determined based on the developed model.
Observational Model: Cohort, Time Perspective: Prospective
With a combination of endoluminal ultrasound characteristics, a simple scoring system can be established to help predict malignancy in mediastinal lymph nodes.
collected characteristics univariate and multivariate logistic regressions will be used to establish a scoring system to predict malignancy. Sensitivity, specificity, positive predictive value, negative predictive value and accuracy will be determined based on the developed model.
Moishe Liberman, MD, PhD
Centre Hospitalier de l'Université de Montréal
Canada: Ethics Review Committee