Is Autofluorescence Bronchoscopy Needed to Diagnose Early Bronchogenic Carcinoma?

Abstract

Once roentgenographically occult lung cancer or severe dysplasia is detected by sputum cytology, the lesion must be localized by bronchoscopy because central-type early-stage lung cancer shows no abnormality on chest radiograph, even on helical computed tomography. However, it is sometimes difficult to locate these lesions by conventional bronchoscopy alone, even with careful inspection, because early-stage lung cancer, especially carcinoma in situ, shows only subtle mucosal changes. 1 Thus, repeated examinations are sometimes necessary to localize the lesion. 2,3 Therefore, autofluorescence bronchoscopy (e.g., the lung imaging fluorescence endoscopy [LIFE] system) has been developed for these patients and has been applied clinically. Many investigators in the world have demonstrated the utility of this system for the early detection of endobronchial malignancies. 4,5 Recently, the autofluorescence endoscope system (SAFE-1000) has been developed. This system is equipped with a conventional xenon lamp with a special bandpass filter as an excitation light source instead of laser light. We discuss the usefulness of this simple device as well as the LIFE lung system in the detection of premalignant and malignant lesions, and the usefulness of autofluorescence bronchoscopy for the screening of abnormal findings in sputum cytology. DEVELOPMENTS OF FLUORESCENCE BRONCHOSCOPY Fluorescence diagnosis can be classified into two groups. One is photodynamic diagnosis using light and a tumor-specific photosensitizer. The other is autofluorescence diagnosis. After the pioneering work on photodynamic diagnosis by Profio and Doiron, 6 Hayata and others 7,8 also examined the possibility of photodiagnosis for central-type early-stage lung cancer since 1978. We have developed several kinds of laser diagnostic systems using a krypton laser, 7 excimer dye laser, 8 and diode laser. 9 However, these systems were hampered by problems including skin photosensitization, false-positive fluorescence, and autofluorescencent interference. Therefore, autofluorescence from endogenous chromophores attracted our attention. The concept of autofluorescence bronchoscopy is based on the fact that the autofluorescence of abnormal areas is different from that of normal areas. The normal area of the bronchus shows green autofluorescence when excited by blue light, but abnormal areas such as cancer or a precancerous lesion show cold spots as a result of decreased green autofluorescence. 10,11 In 1991, the LIFE lung system using a He-Cd laser, without any photosensitizer, was developed by Palcic et al. 10 and was applied clinically by Lam et al. 10 Thereafter, Asahi Pentax developed the SAFE-1000 system, and we started to apply it clinically since 1994. 12 Autofluorescence diagnosis has been reported to improve the detection rate of intraepithelial lesions in recent years. Lam et al. 5 reported favorable results in their multicenter clinical trial using the LIFE system. Because fluorescence diagnosis is advantageous, we need to make an inexpensive, simple, and effective device. One advantage of the SAFE-1000 system is that it uses a standard xenon lamp with a filter, rather than a laser. 12,13 PRINCIPLES OF TISSUE AUTOFLUORESCENCE Aubin 14 demonstrated that autofluorescence arises from intracellular nicotinamide adenine dinucleotide (NADH), riboflavin, and flavin coenzymes, and Barenboim 15 showed that collagen and elastin in connective tissue have strong autofluorescence in the blue–green and yellow spectral regions respectively. Using an excimer dye laser diagnostic system, autofluorescence showed two emission peaks of autofluorescence appeared at around 450 and 500 nm, which are compatible with the fluorescence spectra of collagen, elastin, NADH, and flavin. Autofluorescence is decreased after tissue transformation from normal to carcinoma. The basis of this phenomenon can be explained by three factors as follows: tissue structure, intracellular metabolism, and increase of microvasculature and blood flow. The thickening of the bronchial membrane decreases the autofluorescence from the submucosal layer. In our study the autofluorescence was detected from the submucosal layer by fluorescence microscopy, which is composed of collagen and elastin. 4 On the other hand, tumor tissue does not show autofluorescence, and the matrix between cancer cells shows very low autofluorescence. Accelerated intracellular metabolism in cancer cells decreases riboflavin and flavin coenzymes and NADH caused by overproduction of lactic acid through glycolysis. Adachi et al. 13 showed a decrease of autofluorescence intensity after an increase of the ratio of lactic acid to flavin mononucleotide. Also, flavin adenine dinucleotide in tumor tissue was lower than in normal tissue in our study. Recently, Keith et al. 16 demonstrated an increase of microvasculature in dysplastic lesions that may cause a decrease of autofluorescence because hemoglobin absorbs the excitation light. This angiogenic squamous dysplasia is likely a key step in the transition of intraepithelial preneoplasia to submucosal invasion. AUTOFLUORESCENCE ENDOSCOPE SYSTEM (SAFE-1000) A recently developed autofluorescence endoscope system (SAFE-1000; Asahi Pentax Co., Tokyo, Japan) is equipped with a conventional xenon lamp for excitation light instead of a laser. 12,13 This system is composed of a camera unit, flexible bronchoscope, excitation light source, PVE filing system, and peripheral equipment. The optical system of the SAFE-1000 is shown in Fig. 1. White light from the 75-W xenon lamp is passed through an infrared cut filter and then through an excitation filter that specifically passes 420 to 480 nm excitation light. The excitation light is then transmitted via a light guide to the target area. Images obtained with an object lens are transmitted via a fiberoptic image guide back to the eyepiece of the endoscope and are guided through the prism to the fluorescence filter, which specifically passes 490 to 590-nm fluorescence signals. The selected signal is then amplified by the image intensifier, and, by means of a video camera, it appears as a fluorescent image on the monitor. Abnormal mucosa shows a cold image because of the lack of autofluorescence. It is easy to alternate between white light and excitation blue light by the changeover switch on the camera unit. This system is also compact and easy to handle.FIG. 1.: The white light from the 75-W xenon lamp is passed through an infrared cut filter and then through an excitation filter (420–480 nm). Images obtained with an object lens are transmitted via a fiberoptic image guide and are guided through the prism to the fluorescence filter (490–590 nm). The selected signal is then amplified by the image intensifier.COMPARISON OF PERFORMANCE OF SAFE-1000 WITH LIFE SYSTEM The subjects examined from January 1997 through December 1999 were classified into three categories: cases with lung cancer, abnormal sputum findings, and smokers with current symptoms. Conventional white-light bronchoscopy (WLB) was performed followed by the SAFE-1000. Bronchial biopsy specimens were performed at the site of abnormal findings discovered by WLB or the SAFE-1000, and specimens were investigated histopathologically. A total of 108 patients were enrolled (Table 1). The 56 cases of lung cancer included 24 squamous cell carcinomas, 26 adenocarcinomas, and 4 small cell carcinomas. Thirty-two cases of abnormal sputum cytology findings included 1 invasive cancer, 5 early cancers, 17 dysplasias, and 9 of unknown etiology. The 20 cases of symptomatic smokers included 6 dysplasias and 14 normal or inflammation cases.TABLE 1: Characteristics of enrolled patients (SAFE, N = 108 cases)A total of 234 sites of abnormal findings discovered by WLB or the SAFE-1000, or both, were sampled. A comparison of endoscopic findings and pathologic diagnosis of the biopsy specimens is shown in Table 2. Invasive cancer was detected at 32 sites, all of which were diagnosed correctly by both WLB and the SAFE-1000 examinations. Early cancer was detected at 12 sites, and both WLB and the SAFE-1000 examinations recognized 10 of them. One site was not recognized by WLB but was recognized by the SAFE-1000 alone. The other was recognized by WLB only. Dysplasia was detected at 87 sites, 43 of them were recognized by both WLB and the SAFE-1000, however 31 cases (36%) were recognized by the SAFE-1000 alone. Of normal or inflammation cases, 48% were false positive on the SAFE-1000 examinations resulting from increased cell layers as a result of chronic inflammation.TABLE 2: Results of endoscopic findings and pathology (SAFE, N = 234 sites)The summary of clinical data from the SAFE-1000 examination is shown in Table 3. The results of sensitivity for cancer plus dysplasia were 74% by WLB and 89% by the SAFE-1000 (relative sensitivity, 1.20). For dysplasia alone they were 64% by WLB and 85% by the SAFE-1000 (relative sensitivity, 1.33). The positive predictive value was 64% by WLB and 70% by the SAFE-1000. The SAFE-1000 data were similar to those obtained by the LIFE system in a previous study by the authors (sensitivity, 95% for cancer plus dysplasia, 90% for dysplasia; positive predictive value, 73%). 4TABLE 3: Summary of clinical data (SAFE)SCREENING OF ABNORMAL FINDINGS IN SPUTUM CYTOLOGY We also investigated subjects with abnormal sputum cytology findings by WLB and additional autofluorescence bronchoscopy using the LIFE system or the SAFE-1000 in 78 cases. Class C (moderate dysplasia), class D (severe dysplasia), and class E (carcinoma) according to the Japanese classification were recognized in 28, 22, and 28 subjects respectively (Table 4). The results of localization of abnormal sputum findings are shown in Fig. 2. Fifty percent of the lesions in class C, 73% in class D, and 93% in class E were localized by both conventional and autofluorescence bronchoscopy. In class D, 14% were found to be carcinomas. Therefore, if severe dysplasia is detected by sputum cytology, careful inspection should be performed and additional autofluorescence bronchoscopy should improve the diagnostic rate by 14% in class C, 18% in class D, and 4% in class E (Table 5). Therefore, autofluorescence bronchoscopy is especially useful for the subjects in class D. However, in 27% of the subjects in class D, the lesions could not be localized by WLB and additional autofluorescence bronchoscopy. These lesions may be located at peripheral sites or may be false negative using both methods.TABLE 4: Abnormal findings of sputum cytology (N = 78)TABLE 5: Localization of abnormal sputum findings by endoscopyFIG. 2.: Fifty percent of the lesions in class C (moderate dysplasia), 73% in class D (severe dysplasia), 93% in class E (carcinoma) according to the Japanese classification of sputum cytology were localized by both conventional and autofluorescence bronchoscopy.ARGUMENTS IN SUPPORT OF AUTOFLUORESCENCE BRONCHOSCOPY Lung cancer is a solid tumor with poor prognosis, mainly because of its aggressive nature as well as the fact that most lung cancer cases are detected at an advanced stage. However, there is no question that a favorable prognosis can be expected if lung cancer is detected and treated during an early stage. These types of early cancers can be treated by minimally invasive treatment, such as photodynamic therapy, brachytherapy, or electrocautery. Also, detection of precancerous lesions such as dysplasia is important because previous studies demonstrated that approximately 11% of subjects with moderate dysplasia and 19 to 46% with severe dysplasia subsequently develop invasive cancer. 17–19 Therefore, detection of dysplasia is meaningful to prevent the progression of malignancy. Advances in bronchoscopy and prevalence of sputum cytology have helped to increase numerically the detection of centrally located cancerous or precancerous lesions, but these lesions usually show only subtle changes of bronchial mucosa, which are sometimes difficult to recognize by conventional bronchoscopy. To improve the detection rate of early bronchial lesions, fluorescence diagnosis has been investigated and used clinically in several facilities. Several clinical trials have proved the usefulness of autofluorescence bronchoscopy in the detection of early cancer and dysplastic lesions. 5 However, Kurie et al. 20 demonstrated that autofluorescence bronchoscopy using the LIFE system did not improve the detection rate of dysplasia. Lam and Palcic 21 and Venmans et al. 22 commented on this report that the reasons why they failed to find dysplasia included notable variations among pathologists' subjective interpretation of the fluorescence images and differences in the study population. Lam et al. 5 reported that the relative sensitivity of WLB and the LIFE system versus WLB alone was 6.3 for intraepithelial neoplastic lesions. Venmans et al. 23 reported that the LIFE system is slightly more sensitive (89%) than WLB alone (78%) in the diagnosis of dysplasia and carcinoma in situ. However, the specificity and positive predictive value of the LIFE system was lower (61% and 14%) than those of WLB (88% and 32%). In our study the diagnostic rate for invasive cancer was the same by WLB and by the SAFE-1000. Although the objective of fluorescence evaluation was not to diagnose advanced cancer, the extent of the lesion could be observed objectively by the SAFE-1000 system, which is useful in the preoperative determination of the resection line or to determine the indications of endoscopic local treatment, such as photodynamic therapy. Thirty-one cases of dysplasia (36%) were recognized only by the SAFE-1000 system. This enhanced detection of dysplasia using autofluorescence bronchoscopy may be meaningful in preventing progression to neoplastic lesions. Because only 1 of 12 early cancers (8.3%) was recognized by the SAFE-1000 alone, but not by WLB, the rate of early cancer detection by the SAFE-1000 alone is low. Also, our relative sensitivities of cancer plus dysplasia and dysplasia were lower (1.20 and 1.33) than those of multicenter clinical trials performed in North America using the LIFE system reported by Lam et al. 5 (6.3 and 2.71). One reason may be that we have several expert bronchoscopists and pathologists at our institution who have encountered a relatively large number of early-stage lung cancers treated by photodynamic therapy since 1979. Therefore, our detection rate of early cancer by WLB is usually higher than other institutions'. We also detected 13 false-negative dysplasias (15%) that were recognized by WLB alone, but not by the SAFE-1000. The fluorescence image may reflect some molecular events that cannot be observed microscopically. Molecular genetic change such as cumulative gene losses during the progression of precancerous lesions have been reported. 24,25 Further investigations are needed to determine the reason for the false negatives. The overall sensitivity for cancer plus dysplasia was 74% by WLB and 89% by the SAFE-1000, and dysplasia only was 64% by WLB and 85% by the SAFE-1000. The positive predictive values of WLB and the SAFE-1000 were 64% and 70%. Therefore, fluorescence bronchoscopy using the SAFE-1000 in addition to WLB should increase the diagnostic rate by 15 to 21%. This technique is promising for the early detection of premalignant and malignant lesions. The increasing prevalence of sputum cytology has helped to increase the detection of centrally located cancerous or precancerous lesions, but localization by WLB is sometimes difficult. Sato et al. 3 performed 527 bronchoscopic sessions in 180 patients with 200 occult cancers (positive on sputum cytology and negative on chest radiographic film) to confirm the diagnosis, and demonstrated the difficulty in detecting and localizing carcinoma in situ with WLB alone. Shibuya et al. 26 performed fluorescence bronchoscopy in the detection of preinvasive bronchial lesions in 64 patients with sputum cytology suspicious or positive for malignancy, and demonstrated that the diagnosis of preinvasive bronchial lesions was greatly enhanced in the LIFE group compared with the WLB group (45 versus 7). The percentage of participants with preinvasive bronchial lesions was also notably higher in the LIFE group than in the WLB group. They concluded that the use of autofluorescence bronchoscopy in addition to conventional WLB could greatly enhance the detection and localization of preinvasive bronchial lesions in patients with abnormal sputum cytology findings. Also, our data on abnormal findings of sputum cytology showed enhancement of the detection and localization of the lesions by autofluorescence bronchoscopy, and suggested the usefulness of autofluorescence bronchoscopy especially in cases showing severe dysplasia in sputum cytology. Roentgenographically occult lung cancers are characterized by a high incidence of synchronous lesions (ranging from 7–14%) and metachronous lesions (approximately 5% per year). 27 The frequency of cases of synchronous lung cancer eligible for thoracotomy was approximately 8.8 to 9.3% of all lung cancers. 27,28 Also, fluorescence bronchoscopy surveillance using the LIFE system after curative surgical resection for nonsmall cell cancer identified intraepithelial or invasive lesions in 12%. The LIFE system is three times more sensitive than conventional WLB in identifying these early mucosal lesions. 29 Therefore, autofluorescence bronchoscopy may be a useful adjunct in the perioperative evaluation of lung cancer. Autofluorescence endoscopy is useful for early detection of premalignant and malignant lesions. The newly developed SAFE-1000 system appears to be equally capable for the diagnosis of premalignant lesions as the LIFE system. Our study suggests that autofluorescence diagnosis of subtle bronchial lesions is possible with the simple SAFE-1000 system. Acknowledgment: The authors thank Prof. J. P. Barron, International Medical Communications Center, Tokyo Medical University, for his excellent support in reviewing this manuscript.