Trial Information

Minimally Invasive, Trans-Luminal Diagnosis and Staging of Lung Cancer: A Prospective Head-to-Head Comparison With Traditional Gold Standard Diagnostic Techniques

1. Problem:

The gold standard techniques to stage and diagnose the mediastinum in patients with
lung cancer consist of cervical mediastinoscopy (CM). CT, PET and PET-CT scans have
improved the radiological staging of the mediastinum; however these techniques are
unable to provide a tissue diagnosis and are inaccurate. Endobronchial ultrasound
(EBUS) and endoscopic ultrasound (EUS) have been available for the past few years and
are proving invaluable in the diagnosis and staging of the mediastinum. These
techniques rely on imaging of mediastinal lymph nodes (LN) through the airway (EBUS)
and the esophagus (EUS) using miniaturized ultrasound probes mounted on the tip of
flexible bronchoscopes (EBUS) and esophagoscopes (EUS). Once a LN is visualized and
identified using either of these techniques, a needle is passed through the airway
(EBUS) or esophageal (EUS) wall into the LN and biopsies are performed. These minimally
invasive trans-luminal techniques do not require general anesthesia and can be
performed safely, rapidly, accurately and at low cost.

Due to the cost, inconvenience and risks associated with traditional surgical
mediastinal diagnosis and staging, many physicians are foregoing tissue diagnosis of
the mediastinum, resulting in incomplete pre-operative staging for patients. The
endoscopic techniques are quickly replacing traditional techniques in some centers;
however, when replacing a gold standard test with a new modality, one must first prove
equivalence in safety and efficacy. Until such a study is completed, we do not have
compelling data to be able to offer these techniques as equivalent alternatives to
patients. The proposed project will be the first head-to-head comparison of both
endobronchial (EBUS) and endoscopic ultrasound (EUS) in addition to transluminal biopsy
with gold standard invasive surgical staging techniques in the diagnosis and staging of
NSCLC.

2. Objectives:

- Define and compare the sensitivity, specificity, positive predictive value,
negative predictive value, likelihood ratio and accuracies between traditional and
minimally invasive mediastinal diagnostic and staging techniques for lung cancer.

- Compare morbidity, convenience, cost and efficiencies between traditional and
minimally invasive diagnostic strategies.

- Examine the impact on decisions regarding changes in treatment plans (neoadjuvant
chemoradiotherapy, prevention of thoracotomy) based on minimally invasive
mediastinal staging compared to conventional staging strategies.

3. Methodology:

The study design consists of a prospective comparison of diagnostic test findings on a
consecutive case series with each subject serving as his or her own control. In order
to do this, we will enroll N subjects (N=166), and will subject each of the subjects to
K procedures (K=3). One of those procedures, the "traditional" one (K-1), is considered
to be the gold standard for comparison. We will compare the diagnostic findings from
the K-2 and K-3 alternative procedures (EBUS and EUS, respectively) with the gold
standard across the N subjects. We will compare them singly and in combination with the
gold standard. All patients with a new or suspected diagnosis of lung cancer will be
prospectively enrolled in the study. All patients will undergo standard mediastinal
staging, EBUS and EUS in the operating room, during one procedure. Patients will be
included in the study if they meet one of the following criteria: (A) Lung lesion with
mediastinal lymphadenopathy and/or positive PET scan in the mediastinum, (B) Lung
lesion without mediastinal lymphadenopathy or positive PET scan in the mediastinum.
Sample size has been calculated to be 166 patients.

Results of traditional staging techniques will be considered as the "gold standard"
tests for mediastinal diagnosis and staging. These tests will be used as the standard
to which other modalities and combinations of modalities are compared. Data will be
collected and analyzed as a whole (all patients with mediastinal LN sampling) and
separately based on pre-biopsy diagnosis, post-biopsy diagnosis, mediastinal LN size
(CT scan), LN station and PET avidity. Sensitivity, specificity, positive and negative
predictive values, likelihood ratios and accuracies will be calculated and used to
compare EUS and EBUS (transluminal strategy) to traditional staging, EUS to EBUS, EUS
to traditional staging and EBUS to traditional staging. Subanalysis based on pre-biopsy
patient and imaging (CT and PET) characteristics as well as ultrasound imaging
characteristics during biopsy will be used to develop predictive models for minimally
invasive mediastinal diagnosis, in order to better define patients at increased or
decreased likelihood of successful diagnosis using these techniques. ROC curves will be
constructed and used to discriminate results between diagnostic tests.

The primary focus of this project and the hypothesis for which the sample size
calculations were designed around is whether or not minimally invasive trans-luminal
mediastinal LN biopsy is as accurate as traditional "gold standard" techniques in the
diagnosis and staging of the mediastinum. Sensitivity, specificity, positive predictive
value, negative predictive value and accuracy will be calculated for both strategies
(minimally invasive and traditional) and compared. Receiver operating curves will be
used to compare the two strategies at differing sensitivity and 1-specificity values.
Relative operating characteristic curves will be used to compare the two strategies at
different true positive and false positive rates. The study is estimated to be
completed in three years. Sample size calculation has determined that 166 patients are
required for the study. The thoracic surgery division at the CHUM estimates that
accrual will be approximately 1.5 patients per week (6 patients/month).

4. Expected Results:

We expect to find that minimally invasive mediastinal LN staging and diagnosis is not only
equivalent in terms of accuracy to traditional staging strategies, but is actually superior.
This superiority is possible, secondary to the additional LN basins accessible via these
techniques.