Oxybuprocaine

Background: Bronchoscopy is frequently used to assess invasion of esophageal cancer into the tracheobronchial tree. Prospective studies evaluating the role of bronchoscopy in pretherapeutic staging of esophageal cancer are lacking.

Study objectives: To evaluate the diagnostic utility of fiberoptic bronchoscopy for the assessment of airway involvement by esophageal carcinoma and its resectability.

Patients and methods: In a prospective study, we analyzed 150 bronchoscopies in 116 consecutive patients with potentially operable esophageal carcinoma, and correlated the findings with other staging modalities, intraoperative evaluation, and histopathologic data.

Results: One unknown additional bronchial cancer was found. In 32% of bronchoscopies performed in patients with esophageal cancer located above the tracheal bifurcation, some macroscopic abnormality was detected in the trachea and main bronchi, with mobile protrusion of the posterior tracheal wall being the most frequent abnormality (20.7%). When compared with histologic results, normal macroscopic appearance of the trachea and main bronchi had a negative predictive value of 98.5%, but the positive predictive value of all macroscopic abnormalities for the diagnosis of airway involvement was low, particularly after radiation therapy. The overall accuracy of bronchoscopy with multiple brush cytology and biopsy sampling in proving or excluding airway invasion in patients with otherwise operable conditions was 95.8% (95% confidence interval, 88.3 to 99.1%). Bronchoscopy was the sole decisive staging procedure, resulting in exclusion from surgery because of airway invasion, in 9.7% of patients with otherwise potentially operable conditions. The results of bronchoscopy and CT were discordant in 40% of the patients; the specificity and positive predictive value were higher for bronchoscopy than for CT.

Conclusions: When performed as the last investigation in the staging workup, bronchoscopy with biopsy and brush cytology is a very accurate procedure in evaluating possible airway invasion of esophageal cancer; macroscopic findings alone are not reliable. (CHEST 1998; 113:687-95)

Key words: bronchial neoplasms; bronchoscopy; esophageal neoplasms; esophagectomy; staging; trachea

Abbreviations: CI = confidence interval; RO = complete tumor resection

One reason that esophageal cancer carries a poor prognosis is its tendency to infiltrate surrounding structures.[1] Locally advanced esophageal cancer located at or above the level of the tracheal bifurcation is associated with early invasion of the airways.[2] En bloc resection for cure is, in this situation, not possible. Preoperative assessment to exclude tracheobronchial tree infiltration is therefore essential to identify those patients who will benefit from a radical surgical approach. Most important is to assess whether the tumor is completely resectable, because only patients undergoing complete tumor resection (all visible tumor removed, margins microscopically tumor free, ie, an R0 resection according to the Union Internationale Contre le Cancer 1992 classification[3]) benefit from surgery in terms of their long-term prognosis.[4-8]

Bronchoscopy has been advocated as a valuable tool for the assessment of possible airway involvement. Choi et al[9,10] reported abnormalities in 33.9% of patients, 17.3% with tracheal impingement (ie, extrinsic compression by tumor), and 16.6% with tracheal invasion. Melissas et al[11] found frank tumor growth or an esophagotracheal fistula in 30.3% of their group of 145 patients. In a study by lmdahl et al,[12] bronchoscopy revealed an unknown additional bronchial carcinoma in 5 of 95 patients. Argyros and Torrington[13] stated that bronchoscopy is useful only in patients who have pulmonary symptoms or radiographic findings of atelectasis, pulmonary nodules, or mediastinal adenopathy. Baisi et al[14] matched bronchoscopic results with intraoperative findings in 52 patients and found that impingement with normal motility of the tracheobronchial wall does not preclude a radical resection.

Prospective studies evaluating the role of bronchoscopy for pretherapeutic staging of large and homogeneous patient groups with esophageal cancer are lacking. Therefore, the aim of this study was to prospectively assess the usefulness of bronchoscopy in diagnosing esophageal cancer invasion into the airways, with consequent avoidance of unnecessary surgery.

MATERIALS AND METHODS

In a prospective protocol, a total of 116 patients with carcinoma of the cervical and thoracic esophagus ([is greater than] 15 cm from the incisors) diagnosed between January 1, 1995, and September 30, 1996, by esophagoscopy with biopsy underwent preoperative staging with esophageal endosonography, swallow esophagography, posteroanterior and lateral chest radiography, thoracoabdominal CT scanning with IV and oral contrast application, and fiberoptic bronchoscopy. Patients with a frank esophagotracheal fistula or systemic metastases at presentation were excluded. The level and length of the primary tumor were determined by taking the measurements on swallow esophagography, esophagoscopy, and CT.[15] The esophageal tumor was considered "suprabifurcal" if its proximal end was at or above the tracheal bifurcation on radiographic studies. All radiographs and CTs were interpreted at the time of acquisition by a radiologist without prior knowledge of each patient's clinical status or outcome. All these readings were confirmed by subsequent blinded review by a pulmonologist. A questionnaire was designed that asked for detailed information about the tumor invasion into the tracheobronchial tree from the bronchoscopist before the final therapeutic decision, and from the surgeon and pathologist postoperatively.

Informed written consent for bronchoscopy was obtained from each patient. Fiberoptic bronchoscopy using video endoscopes (Olympus Optical; Hamburg, Germany) was performed via a transnasal or transoral approach after premedication with 0.5 mg atropine and 7.5 mg hydrocodone subcutaneously 30 min before the investigation. Local anesthesia with benoxinate hydrochloride (oxybuprocaine) was also used. Midazolam (2.5 to 10 mg) was administered IV as needed for sedation. Cardiac rate and rhythm as well as oxygen saturation were continuously monitored by ECG and ear oximetry, respectively. Supplemental oxygen at 1 to 2 L/min was administered throughout the investigation.[16]

The complete tracheobronchial tree was inspected; laryngeal structures were included in the examination. All direct tumor signs (exophytic intraluminal growth or necrosis, esophagotracheal fistula, and infiltration of the airway wall) and indirect tumor signs (distortion or compression of normal structures, disappearance of mucosal corrugations, altered mucosal structure, erythema with prominent vascular markings, teleangiectatic blood vessels, protrusion at the posterior wall of the trachea or a major bronchus, widened and immobile bifurcation, or a rigid and fixed pars membranacea during breathing or coughing maneuvers)[16] were recorded and three to five biopsy specimens as well as brush cytology samples were taken from these areas. Washings and needle aspiration for cytologic study were not used. In cases in which no abnormalities could be seen, three to four biopsy specimens were routinely taken from the pars membranacea of the right main bronchus, left main bronchus, and the distal, medial, and proximal part of the trachea, respectively. The quality of biopsy specimens as assessed by the pathologist was considered optimal when at least two pieces [is greater than] 2 min were obtained from the region of interest, and adequate when at least two specimens were [is greater than] 1 mm. Possible complications of bronchoscopy (respiratory depression, significant bleeding, bronchospasm, pneumothorax, important arrhythmia) were specifically sought and recorded.

The bronchoscopic findings were classified according to the classification of Choi et al[9,10] modified by Baisi et al:[14] category 1, no abnormalities; category 2a, impingement with normal motility during breathing and coughing and without mucosal invasion (ie, mobile protrusion); category 2b, impingement with reduced motility but without mucosal invasion (ie, rigid protrusion); and category 3, mucosal invasion.

After the initial staging, 26 patients (18 with suprabifurcal and 8 with infrabifurcal lesions) proceeded directly to esophageal resection, on average 12.6 [+ or -] 9.6 days after the first bronchoscopy. Forty-three patients with locally advanced cancer (40 suprabifurcal and 3 infrabifurcal) received neoadjuvant therapy with 30 to 40 Gy radiation and fluorouracil after completion of the staging procedures. Nine thereafter proceeded directly to surgery and 34 underwent a second restaging bronchoscopy. Thirty of these 34 patients subsequently underwent surgery, on average 11.9 [+ or -] 5.3 days after the restaging bronchoscopy.

Altogether, 65 patients (54 with suprabifurcal and 11 with infrabifurcal tumors) staged preoperatively as having tumors potentially resectable for cure underwent surgery. The criteria used to judge local resectability were established prior to the commencement of' the study and included exclusion of distant metastases and exclusion of airway invasion by bronchoscopy with microscopic proof of cancer. Subtotal transthoracic en bloc esophagectomy with two-field lymphadenectomy was the procedure of choice.[17] Reconstruction was performed with a gastric tube anastomosed to the cervical esophagus via a left cervical incision. Tumor extent and tracheobronchial invasion were determined intraoperatively and by histopathologic examination of the resection specimens. All operated-on patients were subsequently followed up; evidence suggestive of cancer invasion into the trachea or main bronchi was specifically looked for.

For the assessment of validity of bronchoscopy in the diagnosis of airway invasion, frank tumor growth or microscopic proof of cancer in samples taken at bronchoscopy was considered a gold standard for the presence of airway invasion. An RO resection with survival of [is less than] 6 months and no pulmonary complications was considered proof of the absence of airway invasion. In patients who underwent two bronchoscopies, only the results of the restaging bronchoscopy were correlated with the surgical findings. Sensitivity, specificity, predictive values, and accuracy were calculated from the results. A binomial distribution was used to compute 95% confidence intervals (CIs) for the validity indexes of the bronchoscopic abnormalities. The variables are expressed as mean [+ or -] (continous variables) or as a percentage of the group from which they were derived (categorial variables).

RESULTS

One hundred fifty bronchoscopies were performed during the study period in a total of 116 patients. Baseline characteristics for the patients are given in Table 1. Sixty-nine percent of the patients were smokers, 19% were past smokers, and 12% were nonsmokers. Four patients had dyspnea, six had dry cough, nine had hoarseness, and one had hemoptysis; the remaining patients had no pulmonary symptoms.

(*) Values are mean [+ or -] SD.

In 16 (13.8%) of the 116 patients, the proximal esophageal cancer site was below the level of the tracheal bifurcation as assessed by swallow esophagography and CT. The remaining 100 patients had suprabifurcal cancer of the esophagus.

There were no complications of the bronchoscopy. The biopsy specimen quality was considered optimal in 72.4%, adequate in 22.4%, and insufficient in 5.2%. The insufficient biopsy specimens were not used for analysis of the results.

Patients With Infrabifurcal Esophageal Cancer

Twelve of the 16 patients with infrabifurcal cancer had no macroscopic abnormality in the trachea and main bronchi at the first bronchoscopy. Positive macroscopic findings were obtained in four patients (mobile protrusion in two, widened carina in two, and altered structure of the mucosa in one patient). None of the histology and cytology samples taken from these abnormalities were cancer positive. None of the routine biopsy specimens taken from macroscopically normal areas of the trachea and main bronchi were cancer positive.

One case of coexistent asymptomatic bronchial cancer (squamous cell type), which was not seen on CT or plain chest radiograph, was found at bronchoscopy. There were no other endobronchial abnormalities precluding an esophagectomy. Thoracic CT showed no abnormalities of the trachea or main bronchi in any of the 16 patients.

Five of the 16 patients were ultimately treated palliatively; 1 because of the coexistent bronchial cancer, 1 due to distant metastases, and 3 because of poor general condition. Eight patients underwent surgery directly after the initial staging (on average 11.3 [+ or -] 6.8 days after the bronchoscopy) and three patients with locally advanced Barrett's cancer were operated on after neoadjuvant chemotherapy, without a restaging bronchoscopy. Of the 11 ultimately operated-on patients, 9 underwent an RO resection and 2 an R1/R2 resection because of tumor invasion into organs other than the airways. Nine of the 11 operated-on patients have survived [is greater than] 6 months after surgery without any pulmonary problems and 2 died postoperatively of causes unrelated to the airways.

Patients With Suprabifurcal Esophageal Cancer

Initial Bronchoscopy: Table 2 lists the macroscopic findings in the trachea and main bronchi at the first bronchoscopy. Only three patients showed a direct tumor sign (one an endoluminal mass and two frank infiltration). Of the indirect tumor signs, mobile protrusion of the pars membranacea was the most frequent, occurring in 22% of the patients. Thirty-two percent of patients showed some abnormality at bronchoscopy, whereas in 68% of patients, the trachea and main bronchi were macroscopically normal. Taking bronchoscopic biopsy as gold standard, the positive predictive value for all abnormalities (ie, the probability that a given macroscopic abnormality was due to cancer) was 3.6% (95% CI, 0.1 to 18.4%). Of the patients with no macroscopic abnormality, only one routine biopsy sample was cancer positive (predictive value of a negative result, 98.5%; 95% CI, 91.8 to 100%). This patient, who had no pulmonary symptoms or findings suggestive of tracheal involvement on CT and swallow esophagography, was excluded from surgery on the basis of this positive biopsy specimen. A subsequent bronchoscopy performed 3 weeks later showed frank tumor growth in the trachea. In three patients with some macroscopic abnormality (one endoluminal mass, one mucosal infiltration, and one rigid protrusion), the brush cytology sample from the abnormal areas was cancer positive.

Table 2--Macroscopic Findings at Bronchoscopy and Their Correlation With Biopsy Specimens in Patients With Suprabifurcal Cancer

Clinical Course After the Initial Bronchoscopy: Figure 1 shows the fate of the patients after the first bronchoscopy. Six of the 46 ultimately palliatively treated patients were excluded from surgical resection because of airway invasion diagnosed by bronchoscopy (confirmed by biopsy specimen in 2 and by brush cytology findings in the other 4 patients). In these six patients, bronchoscopy was decisive for the therapeutic decision and avoided thoracotomy. Of the other 40 palliatively treated patients, 9 were rejected from surgery because of extensive local tumor mass, 14 due to general inoperability, 11 due to distant metastases, and 6 for other reasons. Of these 40 patients, 27 had normal bronchoscopic findings and 13 had some macroscopic abnormalities with normal microscopy in samples taken from these areas. Validation of the bronchoscopic results was not possible in these patients.

[Figure 1 ILLUSTRATION OMITTED]

Restaging Bronchoscopy After Neoadjuvant Therapy: Forty patients (8 of them with some macroscopic abnormality at the first bronchoscopy but no histologic proof of cancer) received neoadjuvant therapy because of locally advanced tumor. Six of them thereafter proceeded directly to surgery, and 34 underwent a second bronchoscopy (Fig 1 and 2). In four of them, the bronchoscopic findings were downstaged, with normalization of the macroscopic appearance and normal histologic samples. Nine patients showed macroscopic progression at the second bronchoscopy, with cytologic proof of cancer in one and strongly cancer-suspect brush cytology findings in another patient. The latter patient thereafter underwent an RO resection and is well 7 months after surgery; the cytologic diagnosis of airway invasion at bronchoscopy was thus falsely positive in this patient. Of the other seven patients with macroscopic progression of the tracheobronchial findings but normal results of microscopic examination, four showed a frank intraluminal mass (white soft consistency tissue) and three showed mucosal irregularities with telangiectasia. In all these patients, airway invasion was excluded at surgery.

[Figure 2 ILLUSTRATION OMITTED]

Taking bronchoscopic biopsy as the gold standard (Table 2), the positive predictive value of all macroscopic abnormalities at the second bronchoscopy was 0%; specifically, all histologic samples taken from endoluminal masses and from areas suggestive of tumor infiltration showed only necrosis but no cancer cells. The predictive value of a negative result was 100% (95% CI, 84.6 to 100%).

Esophageal Resection: All 54 patients were operated on with curative intent. In 12, complete tumor resection could not be achieved due to airway infiltration in 2 patients and infiltration into other structures in the remainder. Both patients with airway infiltration at surgery had no macroscopic abnormalities at bronchoscopy (performed 12 and 16 days before the operation, respectively) and routine histology and brush cytology samples had been normal.

Of the 52 patients without tracheal infiltration at surgery, 41 are alive [is greater than] 6 months after resection without any pulmonary problems. Eight patients died of causes unrelated to the tracheobronchial tree and three are alive and well 3 to 6 months after surgery.

The correlation between bronchoscopic results and operative findings in the 54 operated-on patients is shown in Table 3. Taking the macroscopic abnormalities without microscopic proof of cancer (Baisi et al[14] categories 2a and 2b) as indexes of airway involvement, the specificity, negative predictive value, and accuracy of bronchoscopy were 73.1%, 95.0%, and 70.4%, respectively. When only microscopic proof of cancer at bronchoscopy was taken as evidence of tracheobronchial involvement, the specificity, negative predictive value, and accuracy of bronchoscopy increased to 98.1% (95% CI, 89.7 to 100%), 96.2% (95% CI, 87 to 99.5%), and 94.4% (95% CI, 84.6 to 98.8%), respectively. The difference in the specificity of these two modes of bronchoscopic evaluation was statistically significant (p [is less than] 0.01).

Table 3-Comparison of Bronchoscopic Results With Operative Findings in Patients With Suprabifurcal Esophageal Cancer

(*) Baisi et al[14] category 2a and 2b.

Correlations With Other Staging Procedures: Most patients had either a normal chest radiograph or a radiograph consistent with an obstructed esophagus, ie, an air-fluid level in the esophagus, as the only abnormality. Specifically, no patient had atelectasis, pulmonary nodules, or mediastinal lymphadenopathy on chest radiography. Two patients showed compression of the trachea (tracheal infiltration was confirmed at bronchoscopy in both) and one patient had a pleural effusion that proved to contain malignant cells; these patients were treated palliatively.

Swallow esophagography showed compression of the trachea in one patient, who was then successfuly operated on after neoadjuvant radio-chemotherapy. According to the selection criteria, there were no cases of esophagotracheal fistula on esophagography.

Endosonography showed possible invasion outside the esophagus (T4 category)[18] in five patients. In one, direct tracheal infiltration was suspected. This patient was eventually treated palliatively due to liver metastases; bronchoscopy in this case showed mobile protrusion of the trachea and vocal cord paralysis. The positive predictive value of endosonography in suggesting possible airway invasion was only 50% in the 53 patients in whom the analysis could be performed.

Fifty-two of the 100 patients with suprabifurcal tumor had no airway abnormalities on thoracic CT; one CT was not available. Of the 47 patients with some airway abnormalities on CT, 28 showed the tumor mass abutting the trachea or bronchus with no intervening tissue plane, 30 showed tumor compression of the trachea or bronchus, and 3 cases were suspicious for tumor infiltration of the airway. When all these findings were taken as evidence of tracheobronchial invasion and compared with definitive diagnosis obtained at bronchoscopy or surgery (Table 4), the sensitivity of CT in diagnosing airway involvement was 62.5%, specificity was 73.1%, positive predictive value was 26.3%, negative predictive value was 92.7%, and accuracy was 71.7%. When only tracheal compression was taken as evidence of airway involvement, the specificity increased to 82.7% and accuracy to 78.3%, at the cost of sensitivity (50%). In 40% of the patients, the results of bronchoscopy and CT were discordant. Forty percent of patients with and 23% of patients without CT abnormalities had some macroscopic abnormalities at bronchoscopy; conversely, 61% of patients with and 41% of patients without any macroscopic abnormality at bronchoscopy had CT signs suggestive of involvement or close contact of the esophageal tumor with the airways. The validity indexes of the combination of concordant results of macroscopic bronchoscopy and CT (n = 38) were worse than that of the bronchoscopy with biopsy sampling alone.

Table 4--Comparison of CT Findings With the Definitive Diagnosis of Airway Invasion in Patients With Suprabifurcal Esophageal Cancer(*)

(*) Includes 54 operated-on patients and 6 palliatively treated patients with histologic proof of cancer invasion at bronchoscopy.

The incidence of respiratory tree involvement did not correlate with T stage at endosonography or with the esophageal tumor length at esophagoscopy and swallow esophagography.

Overall Validity of Bronchoscopy

Table 5 summarizes the results of bronchoscopy in the 72 patients (including those with suprabifurcal as well as those with infrabifurcal esophageal cancer) for whom a definitive diagnosis was obtained. Included are all 65 surgically treated patients and the 7 palliatively treated patients with tracheobronchial infiltration or bronchial cancer diagnosed by bronchoscopy and confirmed microscopically. Excluded from this analysis are the 44 patients who were rejected for surgery because of reasons other than tracheal infiltration, reasons found by examinations other than bronchoscopy, ie, patients in whom the bronchoscopy could have been omitted if it had been done as the last procedure in the staging process. The gold standard for the presence of airway invasion was positive microscopy at bronchoscopy or surgery, and for the absence of invasion an RO resection. The sensitivity of bronchoscopy was 77.8% (95% CI, 40 to 97.2%), specificity was 98.4% (95% CI, 91.5 to 100%), positive predictive value was 87.5% (95% CI, 47.4 to 99.7%), and negative predictive value was 96.9% (95% CI, 89.2 to 99.6%). The overall accuracy of bronchoscopy in proving or excluding tracheobronchial invasion in these potentially operable patients was 95.8% (95% CI, 88.3 to 99.1%). By excluding from surgery patients in whom curative resection was not possible, bronchoscopy was the sole decisive staging investigation in seven patients (9.7% of potentially operable and 6.0% of all patients).

Table 5--Validity of Bronchoscopy in All Patients With Potentially Operable Conditions

DISCUSSION

The esophagus has no serosal barriers; consequently, tumors of the esophagus can grow extensiVely before producing symptoms. The posterior wall of the trachea and main bronchi is in close contact with the esophagus and offers no barrier to tumor extension.[4] Airway invasion is therefore a common route of extraesophageal spread in cancer of the esophagus; an autopsy study showed tracheal involvement in 32% of cases and bronchial involvement in 16%.[2]

Because tracheobronchial invasion usually indicates that curative resection cannot be performed, it is critical to determine whether such spread has occurred. Bronchoscopy has been advocated as a valuable tool for the assessment of possible airway involvement;[9,10,11,12,14,19] the value of CT in proving or excluding tracheobronchial infiltration seems questionable, at least with older CT equipment.[20] If invasion of the airways is taken as a sign of far advanced and unresectable cancer, then interpretation of the bronchoscopic abnormalities as to whether this has, indeed, occurred becomes crucial.

The value of bronchoscopy to diagnose invasion of esophageal cancer into the airways has been assessed by several authors. Choi et al[9,10] proposed a classification of bronchoscopic findings based on their experience with rigid instruments in 525 patients. Category 1 included patients with no macroscopic abnormalities, category 2 included those with evidence of impingement (in the form of a bulge at the posterior wall, deviation of the passage, or widening of the carina) but without mucosal invasion, and category 3 included those with macroscopic evidence of mucosal invasion. The authors found impingement in 17.3% and invasion in 16.6%, and noted that impingement may be due only to the bulky mass in the vicinity, a situation in which the tumor may yet be removed by careful resection. They did not correlate their macroscopic findings either with the histology of the lesions or with the stage of the esophageal. tumor by examination of the surgical specimens, and thus the proposed classification was not validated. For their patients, bronchoscopic findings did not have an impact on the therapeutic decision making, as resection was attempted even when tracheobronchial invasion was evident. Nevertheless, the authors stated that bronchoscopy should be performed in all patients regardless of the size and location of the tumor in the esophagus.

Melissas et al[11] performed rigid bronchoscopy in 145 patients and found gross infiltration of the airways in 30.3% and early invasion in 18.6%. Most of their patients (77.2%) already had metastases to distant sites and therefore had a priori inoperable conditions. Watanabe and colleagues[19] performed bronchoscopy in 14 out of 58 patients and found that the likelihood of airway involvement with tumor increased when respiratory symptoms were present. Brush cytology of reddened or protruding mucosa revealed malignant cells in two of six cases.

Imdahl et al[12] compared bronchoscopy with operative findings in 59 out of 95 patients and reported a specificity of 93%, sensitivity of 54%, and accuracy of 70%. A histologic or cytologic evaluation of the bronchoscopic abnormalities, however, was not attempted. In 5 of their 95 patients, bronchoscopy revealed an unknown additional bronchial carcinoma.

Baisi et al[14] evaluated 91 patients by bronchoscopy and in 52 of them correlated the findings with operative results. They found that impingement with normal motility during coughing and breathing (category 2a) does not mean infiltration by esophageal. cancer and radical resection is still possible, whereas when impingement is associated with reduced motility (category 2b), tracheobronchial infiltration must be suspected and a curative resection is unlikely. They also found that by neoadjuvant regimens, a downstaging of the 2b-bronchoscopic category with resectability of the tumor could be obtained. The authors did not report the biopsy results of the macroscopic abnormalities found at bronchoscopy.

Our study was undertaken to assess the value of bronchoscopy for determining the resectability of esophageal cancer. The best design for this aim would have been a prospective randomized study in which patients staged with standard methods were compared with these methods in addition to bronchoscopy. However, because bronchoscopy is already considered a standard staging procedure by our surgeons[15] and elsewhere,[1,15] it would be unethical not to perform this examination. The results of this study (exclusion from surgery of 9.7% of patients with otherwise potentially operable conditions based solely on the results of bronchoscopy) underline this point of view. Therefore, we conducted the next best approach with prospective case series of all patients undergoing staging to decide on the possibility of curative radical resection. Only patients with evidence of esophagotracheal fistula or with systemic metastases on presentation were excluded. Our patient group is therefore representative of patients with potentially operable conditions. In addition to macroscopic evaluation, we took brush cytology and biopsy samples from all areas of macroscopic abnormalities as well as routine biopsy specimens from several macroscopically normal areas. Needle aspiration from areas of protrusion or from the widened carina was not used, because a positive result could come from the esophageal tumor outside the trachea or from a resectable precarinal lymph node, and would not indicate airway infiltration per se. Washings were not used because of the possibility of aspiration of malignant cells from the obstructed esophagus into the trachea, yielding a false-positive bronchoscopic cytology.

Macroscopic abnormalities in the trachea or main bronchi were observed in 32% of our patients with suprabifurcal tumor at initial bronchoscopy. However, proof of cancer through biopsy could be obtained in 3.6% of these abnormalities only. Fifteen of the patients with macroscopic abnormalities and without microscopic proof of cancer eventually underwent an R0 resection; they would have been rejected for curative surgery if the diagnosis of airway invasion had been based on macroscopic findings only. Therefore, we believe that the diagnosis of airway invasion must be based on microscopic proof of cancer from suspected areas, at least from areas with indirect tumor signs only. The assessment of indirect tumor signs is subjective; in an ongoing study, we are now videorecording all bronchoscopies to better define the macroscopic abnormalities and to establish the interobserver variability of indirect tumor signs.

The positive macroscopic findings at bronchoscopy in 4 of 16 patients with infrabifurcal tumor could be due to extraesophageal spread of the tumor to high levels of the mediastinum, to tumor invasion from involved mediastinal lymph nodes, or, as noted above, to overestimation of indirect tumor signs. In these cases, the microscopic proof of malignancy may be even more important than in suprabifurcal tumors for the exclusion of false-positive results.

We agree with Choi et al[10] and Baisi and colleagues[14] that protrusion with normal motility of the tracheal wall during breathing and coughing may simply be due to the presence of tumor in the vicinity and does not preclude radical resection.

Only one of the routine biopsy specimens taken in patients with no macroscopic abnormality was cancer positive. The decision to exclude this patient from surgery was certainly correct, because a subsequent bronchoscopy 3 weeks later showed frank tumor growth in the trachea. It is difficult to assess from our results the value of routine biopsy specimens from macroscopically normal areas. The taking of three to four biopsy specimens from each of five regions, even in the absence of visible abnormalities, represents more sampling than in the usual clinical setting. Even the few abnormal bronchoscopic biopsy specimens might be missed under less rigorous testing conditions. The utility of such routine biopsies might be higher and the performance easier if the biopsy specimens were taken only from areas nearest to the esophageal cancer, as assessed by CT. In an ongoing study, we are currently evaluating such targeted routine biopsies.

Only one case of synchronous unsuspected bronchial cancer was found. In this patient with an otherwise operable condition with infrabifurcal esophageal cancer, curative resection of both cancers was considered unlikely and palliative therapy was instituted. Patients with upper aerodigestive tract cancers develop secondary primary tumors at a rate of 4% of patients per year.[1] The low incidence of bronchial cancer in our study (0.9%), compared with previous studies (5.3%,[12] 4.6%,[14] and 3.2%[21]), can be explained by the thorough evaluation of our patients aimed at exclusion of patients with inoperable conditions prior to bronchoscopy. We believe that bronchoscopy to exclude an unsuspected bronchial cancer should be performed in all patients with esophageal cancer, including those with infrabifurcal tumor; curative resection of an associated esophageal and lung cancer can be performed in approximately 50% of the patients, with a low mortality rate, and is associated with real long-term survival.[21]

Neoadjuvant chemotherapy and radiation before surgery appears promising in improving long-term and disease-free survival in patients with locally advanced esophageal cancer.[1,7,8] It was surprising at the restaging bronchoscopy to find direct tumor signs in 17.7% of patients, with microscopic proof of cancer in only one of them. This indicates that macroscopic abnormalities after radiation must be interpreted even more carefully than at baseline staging, because a white soft intraluminal mass and altered mucosal structure (pallor, contracting scars, and telangiectasia) could be the result of radiation-induced peritumorous inflammatory or fibrous reactions alone and not of tumor invasion. The one false-positive cytology in this subgroup (interpreted as "strongly cancer-suspect") documents the problems with cytology and raises the question whether only biopsy specimen-proven cases should be considered as true positives.

The total incidence of proved airway invasion of esophageal cancer was lower in our study (8 patients, ie, 6.9% of the total 116 patients, or 11.1% of the 72 patients with potentially operable conditions) than has been reported previously,[9-11] probably reflecting the thorough prescreening of our patients. Bronchoscopy identified six of these eight patients; in two cases the bronchoscopy was falsely negative.

A postoperative survival of [is greater than] 6 months without any pulmonary problems in most of the patients is further proof of an R0 resection with regard to the airways, making the "gold standard" with which our bronchoscopic findings were correlated even better. It is impossible to ascertain whether the exclusion of airway invasion by bronchoscopy was correct in the patients treated palliatively for other reasons. We did not investigate the survival of the palliatively treated patients, because the natural tumor growth and the palliative radiotherapy therapy in these patients could have influenced tumor invasion in the airways, making any later evaluation of the accuracy of the bronchoscopy useless.

The chest radiography, swallow esophagography, and esophageal endosonography did not indicate airway involvement in the vast majority of patients. In our and others' experience, these investigations are not useful in assessing airway infiltration in the absence of a fistula. Esophageal endosonography is currently the most accurate means to assess the depth of wall penetration of esophageal cancer. However, due to limitations of the currently available ultrasonographic probes, the method is not able to reliably assess tumor invasion into the airways.[15]

CT cannot differentiate the individual layers of the esophageal wall; it is also difficult to differentiate the tracheal wall from the closely abutting tumor on CT.[22] In our study, CT proved helpful in revealing the tumor's relation to the tracheobronchial tree. We believe that, the bronchoscopist should have a CT scan available for review before the procedure in order to be able to take the biopsy specimens from the areas most suspicious for tumor invasion. In 40% of our patients with suprabifurcal esophageal cancer, the results of bronchoscopy and CT were discordant. Both methods showed comparable sensitivity and negative predictive value in the diagnosis of airway involvement; however, the specificity, positive predictive value, and the overall accuracy were substantially higher for bronchoscopy than CT. In our opinion, patients with normal bronchoscopy results should not be excluded from potentially curative neoadjuvant therapy simply because of CT signs of tumor protrusion into the airways.

In our study, a significant number of patients (44 of 116, ie, 37.9%) were excluded from curative resection because of systemic metastases or poor physiologic status documented only after bronchoscopy was performed. If this information had been available prior to bronchoscopy, the procedure could have been avoided. We believe that bronchoscopy should be performed as the last investigation in the staging workup, after inoperability has been excluded by chest radiography, CT, swallow esophagography, esophageal sonography, and assessment of the physiologic status.[6] The results of bronchoscopy in patients with inoperable conditions without pulmonary symptoms will not modify their palliative treatment.

In conclusion, our study showed a high overall accuracy (95.8%) for bronchoscopy with biopsy and brush cytology in proving or excluding airway invasion of esophageal cancer in patients with potentially operable conditions. Macroscopic findings alone were not reliable. Bronchoscopy with biopsy and brush cytology was the sole decisive staging procedure, enabling exclusion from surgery because of airway invasion, in 9.7% of patients with otherwise potentially operable conditions.

ACKNOWLEDGMENT: We thank Prof. Dr. K. Ulm for statistical assistance with this project, and Dr. C. Duvernoy, Department of Internal Medicine, University of Michigan, for review of the manuscript.

REFERENCES

[1] Roth JA, Lichter AS, Putnam JB, et al. Cancer of the esophagus. In: DeVita VT, Hellman S, Rosenberg SA, eds. Cancer: principles and practice of oncology. Philadelphia: JB Lippincott, 1993; 776-817

[2] Mandard AM, Chasle J, Marnay J, et al. Autopsy findings in 111 cases of esophageal cancer. Cancer 1981; 48:329-35

[3] UICC. Osophagus. In: Hermanek P, Scheibe O, Spiessl B, et al, eds. TNM classification of malignant tumours. Berlin: Springer, 1992; 44-46

[4] Siewert JR. Osophagus and Tracheobronchialsystem--eine schicksalhafte Beziehung [editorial]. Langenbecks Arch Chir 1990; 375:325

[5] Siewert JR, Dittler HJ. Esophageal carcinoma: impact of staging on treatment. Endoscopy 1993; 25:28-32

[6] Holscher AH, Dittler HJ, Siewert JR. Staging of squamous esophageal cancer: accuracy and value. World J Surg 1994; 18:312-20

[7] Fink U, Stein HJ, Wilke H, et al. Multimodal treatment for squamous cell esophageal cancer. World J Surg 1995; 19:198-204

[8] Fink U, Stein HJ, Lukas P, et al. Combined modality therapy for locally advanced squamous cell esophageal cancer at or above the level of the tracheal bifurcation: final report of a phase 11 study. In: Peracchia A, Rosati R, Bonavina L, et al, eds. Recent advances in diseases of the esophagus. Bologna: Monduzzi Editore, 1996; 463-69

[9] Choi TK, Siu KF, Lam KH, et al. Bronchoscopy and carcinoma of the esophagus: I. Findings of bronchoscopy in carcinoma of the esophagus. Am J Surg 1984; 147: 757-59

[10] Choi TK, Siu KF, Lam KH, et al. Bronchoscopy and carcinoma of the esophagus: II. Carcinoma of the esophagus with tracheobronchial involvement. Am J Surg 1984; 147:760-62

[11] Melissas J, Minnaar R, Mannell A. Bronchoscopic findings in patients with oesophageal carcinoma. S Afr J Surg 1986; 24:24-26

[12] Imdahl A, Kirchner R, Swoboda L, et al. Ist die Bronchoskopie beim Osophaguskarzinom eine sinvolle praoperative Zusatzuntersuchung? Langenbecks Arch Chir 1990; 375: 326-29

[13] Argyros GJ, Torrington KG. Fiberoptic bronchoscopy in the evaluation of newly diagnosed esophageal carcinoma. Chest 1995; 107:1447-49

[14] Baisi A, Bonavina L, Fumagalli U, et al. Bronchoscopic staging of esophageal carcinoma correlated to operative findings. In: Peracchia A, Rosati R, Bonavina L, et al, eds. Recent advances in diseases of the esophagus. Bologna: Monduzzi Editore, 1996; 155-59

[15] Bumm R, and Panel of Experts. Staging and risk analysis in esophageal carcinoma: results of a consensus conference held at the sixth World Congress of the International Society for Diseases of the Esophagus. Dis Esophagus 1996; 9:20-29

[16] Hauck RW, Emslander HP. Bronchoscopy of lung cancer. Diagn Ther Endosc 1994; 1:9-18

[17] Siewert JR, Bartels H, Bollschweiler E, et al. Plattenepithelcarcinom des Oesophagus: Behandungskonzept der Chirurgichen Klinik der Technischen Universitat Munchen. Chirurg 1992; 63:693-700

[18] Dittler HJ, Siewert JR. Role of endoscopic ultrasonography in esophageal carcinoma. Endoscopy 1993; 25:156-61

[19] Watanabe A, Saka H, Sakai S, et al. Bronchoscopic and cytopathological findings of tracheobronchial involvement in esophageal carcinoma. Endoscopy 1990; 22:273-75

[20] Lehr L, Rupp N, Siewert JR. Assessment of resectability of esophageal cancer by computed tomography and magnetic resonance imaging. Surgery 1988; 103:344-50

[21] Fekete F, Sauvanet A, Kaisserian G, et al. Associated primary esophageal and lung carcinoma: a study of 39 patients. Ann Thorac Surg 1994; 58:837-42

[22] Inculet RI, Keller SM, Dwyer A, et al. Evaluation of noninvasive tests for the preoperative staging of carcinoma of the esophagus: a prospective study. Ann Thorac Surg 1985; 40:561-65

From the Departments of Internal Medicine I (Drs. Riedel, Hauck, Schulz, and Schomig, and Mr. Mounyam) and Surgery (Drs. Stein and Siewert), Klinikum rechts der Isar, Techniscbe Universitat Manchen, Munich, Germany.

Manuscript received May 29, 1997; revision accepted September 24, 1997.

Reprint requests: Martin Riedel, MD, 1. Medizinische Klinik und Poliklinik, Klinikum rechts der Isar, Ismaninger Strasse 22, D-81675 Munchen, Germany

COPYRIGHT 1998 American College of Chest Physicians
COPYRIGHT 2000 Gale Group