the Harwood Academic Publishers imprint, part of The Gordon and Breach Publishing Group. Printed in Singapore. Results of Two Years Experience with Fluorescence Bronchoscopy in Detection of Preinvasive Bronchial Neoplasia

The aim of the study was to determine whether use of fluorescence bronchoscopy improves the detection of preinvasive neoplastic bronchial lesions. The data of all patients who underwent fluorescence bronchoscopy and in whom bronchial biopsies were taken, were analyzed. Most patients were at risk for preinvasive lesions. A total of 174 bronchoscopies were performed in 95 patients. Of the 681 representative biopsies, 31 were found to be moderate dysplastic, 39 were found to be severe dysplastic and 9 exhibited carcinoma in situ. These 79 preinvasive lesions were found in 34 patients. The respective results of fluorescence bronchoscopy in addition to conventional bronchoscopy and of conventional bronchoscopy alone for detection of preinvasive lesions were: sensitivity 85% (67/79) and 59% (47]79); specificity 60% (351/581) and 85% (493/581); positive predictive values of 23% (67/297) and 35% (47/135); negative predictive values of 97% (3511363) and 94% (493]525). A separate analysis of only the first bronchoscopy of each patient showed similar results. Results of fluoresence bronchoscopy were better in the second part of the patient group. We conclude that after a learning period fluorescence bronchoscopy can increase the yield of finding preinvasive neoplastic lesions when used in addition to conventional bronchoscopy.


INTRODUCTION
Squamous cell carcinoma of the bronchus is thought to arise after a series of progressive histo-logic changes in the bronchial epithelium [1].
These preinvasive neoplastic lesions are, in order of increasing severity: basal cell hyperplasia, metaplasia, mild, moderate and severe dysplasia, and 78 B.J.W. VENMANS et al. carcinoma in situ. When the bronchial mucosa is illuminated by a blue light with a wavelength of 405-442nm, such as light from a Krypton ion laser or a Helium-Cadmium laser, there is a progressive reduction in the green wavelength band of the autofluorescence spectrum as the tissue becomes more neoplastic [2,3]. Based on this difference in fluorescence intensity, the lung imaging fluorescence endoscope (LIFE) system (Xillix Technologies Corp; Richmond, BC, Canada) was developed to detect preinvasive bronchial lesions [4,5]. LIFE provides true fluorescence images in real time. Normal tissue which fluoresces green, can be distinguished from neoplastic tissue which appears to be brown or brownish red in color depending on the pathology grade. Experience with the LIFE system has shown that it can improve the detection rate of preinvasive bronchial lesions compared with conventional bronchoscopy. Lam et al. [5] found the sensitivity of LIFE to be twice the sensitivity of conventional bronchoscopy for the detection of preinvasive bronchial lesions. Fluorescence bronchoscopy could therefore be clinically useful in detecting and localizing early lung cancer, in staging the extent of endobronchial spread of bronchial cancer and in detecting synchronous roentgenographically occult cancer. Early detection may improve the prognosis of lung cancer patients, since it was shown that early treatment as surgical resection or bronchoscopic treatment for roentgenographically occult intraluminal tumors gives the best chance for cure [6][7][8][9][10][11][12][13][14]. We started using the LIFE system in clinical practice in November 1995. The aim of this study was to determine whether use of the LIFE system improves the detection of preinvasive bronchial lesions. Preinvasive lesions were defined as lesions that were scored by the pathologist as moderate dysplasia, severe dysplasia or carcinoma in situ. These preinvasive lesions were considered clinically important in the population at risk, and were considered to be the interest of using a LIFE system for early tumor detection.

Bronchoscopy
All the patients from the Department of Pulmonary Medicine of the University Hospital Vrije Universiteit in Amsterdam who underwent a LIFE bronchoscopy in the period from November 1995 to December 1997 and in whom bronchial biopsies were taken were included in this study. The results of 33 of these patients were reported before [15]. In these 33 patients the modality of the initial bronchoscopy to be performed, conventional bronchoscopy or LIFE, was detemined by randomization. Indications for LIFE bronchoscopy in most patients of the present study were: suspected lung cancer, staging of newly diagnosed lung cancer or follow-up of lung or head and neck cancer. All bronchoscopic procedures were performed by the same bronchoscopist (TS). Standard bronchoscopy procedure was followed. Local anesthesia with lidocaine 2% spraying was administered to all patients. In the decubitus position, the bronchoscope (Olympus BF-20D) was introduced transorally. The fiberoptic bronchoscope was coupled to either the conventional or the LIFE video camera and bronchoscopy was performed. Careful bronchoscopy is obligatory to prevent unnecessary mucosal contact, as hemorrhagic bruising may undermine imaging. Digitized images were videotaped and frozen images of abnormal or suspicious lesions were captured. All locations were documented. After switching the camera, bronchoscopic examination was repeated in the same manner using the alternate device. All abnormal and suspicious lesions were separately biopsied, using a new biopsy forceps for each location to prevent contamination. In all cases at least one biopsy was taken of an area considered not suspicious during both LIFE and conventional bronchoscopy. Abnormalities detected by the LIFE were biopsied using the LIFE camera to ensure the exact spot, unless images were too blurred because of hemorrhagic bruising of the RESULTS OF FLUORESCENCE BRONCHOSCOPY 79 mucosa during manipulation. Abnormalities detected only during conventional bronchoscopy were subsequently biopsied.

Histological Examination
The biopsies were processed in a routine fashion. After fixation for at least 24h in 4% buffered formalin, the tissue was blocked in paraffin and 4gm slides were mounted on glass and stained with hematoxylin and eosin using standard procedure. The pathologists reported the results of microscopy of all specimens in separate jars, by indicating them to be: unsatisfactory, normal, inflammation, hyperplasia, mild dysplasia, moderate dysplasia, severe dysplasia, carcinoma in situ, and carcinoma.

Analysis
The results of the histologic examination were compared with the results of the visual assessment using the conventional and the LIFE cameras.
Sensitivity, specificity, positive and negative predictive values for detection of the total number of moderate dysplasias, severe dysplasias and carcinomas in situ were calculated. The relative sensitivity, or the ratio of the sensitivity of conventional bronchoscopy combined with LIFE, as compared with conventional bronchoscopy alone, along with the 95% confidence interval (CI), was calculated to evaluate the contribution of fluorescence examination to detect preinvasive lesions.
The results of all bronchoscopies using the LIFE system were analyzed. In a subgroup of patients several LIFE bronchoscopies were done. Because patient selection may influence the results, a separate analysis was performed using only the first bronchoscopy data of each patient. This second analysis was also used to examine if there is a learning effect in using the LIFE system. We therefore compared the results of the first 48 patients with the results obtained in the next 47 consecutive patients.

RESULTS
Patients demographics and bronchoscopy data are shown in Table I. A total of 174 bronchoscopies were performed in 95 patients. All procedures were completed under local anesthesia and no complications were observed during or after bronchoscopic examination. The median duration of the additional LIFE examination was 15 min (range: 5-25 min). Of the total number of biopsies taken, 92% (681/742) was judged by the pathologists to be satisfactory, enabling adequate assessment of the morphology. Three-hundred-and-two of these 681 biopsies were taken based on suspicious findings during LIFE bronchoscopy. Onehundred-and-fourty of these 302 biopsies were of the biopsies taken were judged to be satisfactory for histological examination. Fourty preinvasive lesions were found in 25 patients (female/male 5/20) and 16 carcinomas were found in 13 patients (female/male 1/12). A total of 3 preinvasive lesions were found in 3 of these 13 patients. Two of these 3 preinvasive lesions were found in other bronchi     than the one where the carcinoma was localized. The bronchoscopy data, the results of the histologic examination and the performance of conventional bronchoscopy and LIFE in detecting preinvasive lesions are shown in Tables III and V.
The relative sensitivity of conventional bronchoscopy combined with LIFE vs conventional bronchoscopy alone was 1.14 (95% CI 0.89-1.48). The results are comparable to those of all bronchoscopies. There was a trend that conventional bronchoscopy combined with LIFE was more sensitive in detecting preinvasive changes in the bronchial mucosa than conventional bronchoscopy alone. However, at cost of a lower sensitiv-ity and positive predictive value. In Table VI the performance of LIFE in detecting preinvasive lesions in the first half of the patient group is compared with the results in the second half. As can be seen from the table there seems to be a learning effect using the LIFE system. Sensitivity, specificity, positive predictive value and negative predictive value all were higher in the second half of the patient group.

DISCUSSION
Finding more early stage cancer lesions in the central airways which can be treated with curative intent using currently available treatment modalities such as photodynamic therapy or electrocautery may improve the poor prognosis of patients with lung cancer [9][10][11][12][13][14]. Efforts have been made in applying conventional fiberoptic bronchoscopy to detect and localize early stage lung cancer [16][17][18]. Even in patients with positive sputum cytology repeated bronchoscopic examinations were necessary and this was despite the fact that only a proportion of the lesions were preinvasive. Therefore the routine use of conventional fiberoptic bronchoscopy seems impractical for screening high risk individuals among the high numbers of smokers with COPD [19], those with a previous history of curatively treated head and neck cancer [20] and those at risk for multiple or subsequent lung cancers [21]. The  When LIFE was added to conventional bronchoscopy the sensitivities improved to 85%. These results are comparable to those mentioned above. In recent studies sensitivities of 90% and 67% in detecting (pre)invasive lesions were reported [22,23]. However, the specificities when using the LIFE system were lower in the present study (60-69%). This resulted in low positive predictive values for detection of preinvasive lesions when LIFE was added to conventional bronchoscopy (23% and 27%). However, in the only other study reporting a positive predictive value this was found to be 23% [23]. The [22]. Dysplasia and carcinoma in situ are difficult to detect during a standard fiberoptic bronchoscopic examination, because these lesions are superficial and minute in size [21,24]. However, subtle changes may be seen such as increased redness, changes in vascularization, local swelling, relatively thickened mucosa, loss of luster etc. [25], as can be learned by experience while performing careful bronchoscopic examination of the entire tracheobronchial tree. Some preinvasive lesions and even two carcinomas were seen by conventional bronchoscopy but were missed by LIFE. This study therefore confirms that fluorescence bronchoscopy cannot replace conventional bronchoscopy but should be performed in addition to it [23]. Several preinvasive lesions were missed by both conventional bronchoscopy and LIFE. However, histologic examination of these lesions only showed moderate dysplasias. So only the lesions with the lowest degree of preinvasive changes, that were considered to be the interest of using the LIFE system in this study, were missed. The reason of the failure of LIFE to localize some (pre)invasive lesions is not clear.
Moreover the meaning of suspicious LIFE findings with negative histology also is unknown. Whether these lesions represent very early changes at the molecular level is unclear. In biopsies taken by others, using the LIFE system, molecular genetic changes such as cumulative gene losses in the progression of preinvasive epithelial lesions have been found [26][27][28]. The fluorescence pattern RESULTS OF FLUORESCENCE BRONCHOSCOPY 83 of the bronchial mucosa may therefore reflect [4] molecular genetic changes which are beyond the threshold of the macroscopic and microscopic [5] abilities of the bronchoscopist and the pathologist.
By using current technique and histological criteria as the golden standard for diagnosing preinva- [6] sive lesions, the clinical impact of molecular genetic changes and its relation to tissue auto- [7] fluorescence remains elusive. A subgroup of patients underwent several LIFE bronchoscopies. [8] To exclude that patient selection bias may influence the results, a separate analysis was performed [9] using only the first bronchoscopy of each patient. The results of this analysis were comparable to those of all the bronchoscopies. The number of [10] preinvasive lesions found with LIFE was higher and the best results were obtained when both [111 modalities were combined. Both analyses showed that specificity and positive predictive were lower when LIFE was used and that a high sensitivity [12] can be obtained by conventional bronchoscopy alone. To study if" there was a learning effect in using the LIFE system the results of the first bronchoscopies of the first half of the patient [13] group were compared with the results obtained in the second half. Sensitivity, specificity, positive [141 predictive value and negative predictive value all were higher in the second half of the patient group. This indicates a learning effect in using the LIFE system. We conclude that fluorescence [15] bronchoscopy can be performed safely and that after a learning period it can increase the yield of finding preinvasive neoplastic lesions when used [16] in addition to conventional bronchoscopy.