Published ahead of print on October 25, 2007, doi:10.1164/rccm.200704-598OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200704-598OC
Molecular Predictive Factors for Progression of High-Grade Preinvasive Bronchial Lesions1 Clinique Pneumologique and Groupe LITIS–Quantif EA 4108, Rouen University Hospital, and Faculté de Médecine-Pharmacie, Rouen France; 2 INSERM U.614, Faculté de Médecine-Pharmacie, Rouen, France; and 3 Service d'Anatomie et Cytologie Pathologiques, and Laboratoire de Génétique Somatique des Tumeurs, Rouen University Hospital, Rouen, France Correspondence and requests for reprints should be addressed to Luc Thiberville, M.D., Clinique Pneumologique, Hôpital Charles Nicolle–CHU de Rouen, 1 rue de Germont, 76031 Rouen Cedex, France. E-mail: luc.thiberville{at}univ-rouen.fr
Rationale: The outcome of precancerous bronchial lesions is not well known, and their management is subject to controversy. Many molecular alterations are present in preinvasive lesions, but none has been assessed to predict the evolution of the lesions. Objectives: To analyze the outcome of high-grade precancerous lesions according to their molecular profile. Methods: Twenty-three severe dysplasia and 31 carcinoma in situ (CIS) lesions in 37 patients were monitored using repeated autofluorescence bronchoscopy over a 12-year period. Microdissection and polymerase chain reaction analysis were performed on paraffin tissue sections to assess loss of heterozygosity (LOH) and microsatellite instability on chromosome 3p, 5q, and 9p. Histology and molecular status at baseline were compared between 7 lesions that became invasive, 11 that relapsed after treatment, 17 that were eradicated with local treatment, and 19 that spontaneously regressed. Measurements and Main Results: Ninety-four percent of lesions that progressed or relapsed were CIS at baseline, whereas 79% of spontaneously regressing lesions were severe dysplasia (P < 0.0001). 3p and 9p LOH was more frequent in CIS than in severe dysplasia (P = 0.03). In the whole group of lesions as well as in the CIS group, 3p LOH was strongly associated with progression (P < 0.0001 and P = 0.02, respectively). Microsatellite instability was not associated with the outcome of the lesions. A therapeutic strategy based on the presence of 3p or 9p LOH would have led to overtreatment of six lesions but would have missed only 1 among the 18 progressing lesions. Conclusions: Baseline histology and 3p LOH analysis appear to be useful in predicting the outcome of high-grade precancerous lesions.
Key Words: precancerous conditions • gene deletion • follow-up studies • bronchoscopy • disease progression
The early detection of lung cancer appears to be crucial for the treatment efficacy and prognosis of this devastating disease (1). In the past decade, modern endoscopic methods of centrally located cancers, such as autofluorescence endoscopy (2, 3), have been developed that are able to detect and localize more effectively the early stages of epithelial bronchial cancerization, such as severe dysplasia (SD) and carcinoma in situ (CIS). However, because bronchoscopy is too invasive to be considered as a screening tool for bronchial cancers, a prescreening method must be used to select the highest risk patients who could benefit from an endoscopic early-detection program. A practical approach is to select the patients on the basis of clinical predictive factors, such as occupational exposure to lung carcinogens, synchronous lung cancers, and active smoking, that are also associated with a high frequency of proximal high-grade lesions (4). Despite this prescreening selection, the efficacy of the early cancer endoscopic detection approach is currently limited by our restricted knowledge on the natural history of the preinvasive bronchial epithelium (5–7). Invasive squamous cell carcinoma of the bronchus is believed to develop from precancerous lesions, beginning with hyperplasia, through the increasing degrees of dysplasia and CIS (8). All the published series using autofluorescence bronchoscopy in individuals at high risk for lung cancer have found a very high frequency of preinvasive lesions, exceeding 50% of the cases (5, 6, 9, 10). Therefore, a major issue is to more accurately differentiate lesions that are at high risk to progress into invasive cancers from those that are not (5, 7). Using fluorescence endoscopy to monitor the evolution of precancerous lesions, we have previously shown that the CIS lesions have the highest probability to persist, progress, or relapse (5), as compared with SD or lower grade lesions. However, although the presence of an SD or a CIS lesion has been clearly associated with the occurrence of a synchronous or a metachronous lung cancer (4, 10, 11), studies have currently failed to find reliable predictive factors of progression of high-grade precancerous lesions at the individual level. Particularly, despite a greater probability for the highest grade lesions to progress, it currently appears impossible to make an accurate prediction of the long-term evolution of a given lesion on the basis of the histology results alone (5, 6). With advances in molecular biology and the advent of microdissection techniques, the genomic changes that are associated with the malignant bronchial transformation have been better characterized (12–17). The main molecular alterations found in invasive lung cancer consist in inactivating mutations of tumor suppressor genes, activating mutations of oncogenes, loss of heterozygosity (LOH), and amplifications of chromosomal regions (18). Recently, many of the abnormalities characteristic of the invasive tumors were also found in bronchial precancerous lesions (19), although with an overall lower frequency as compared with invasive cancers. Particularly, LOH of specific chromosomal markers, such as in chromosomes 3 and 9, was found to be a very early event in bronchial cancerization, present as early as in reserve cell hyperplasia in both proximal and distal bronchi (12, 19–21). In a previous article, we evaluated chromosomal changes in a range of premalignant bronchial lesions of increasing severity, and found evidence of accumulation of gene losses with the progression of the lesions from one step to the next (12). We also performed a limited molecular follow-up study, which found a good correlation between the persistence of the genomic abnormality over time and the evolution of the dysplastic lesion on follow-up (12). In this article, we consider the hypothesis that the molecular abnormalities existing in high-grade precancerous lesions, such as SD and CIS, could be linked to their progression, and therefore could be used for their clinical management. The objective of this study was to compare the molecular abnormalities present in CIS and SD with progressing lesions and regressing lesions.
Study Design This study is part of the early lung cancer detection program, which was initiated at Rouen University Hospital in 1995. In this program, individuals at high risk of developing lung cancer underwent a fluorescence bronchoscopy to detect early lung cancer and precancerous lesions in their bronchial tree. The evolution of preinvasive lesions found at baseline was assessed using repeated fluorescence bronchoscopy as described elsewhere (5). In this study, high-grade lesions (SD and CIS) were monitored at 3 months and CIS or SD lesions that persisted at two consecutive endoscopies underwent conservative endobronchial treatment, including cryotherapy, electrocautery, or photodynamic therapy. The patients and the lesions (both locally treated and untreated) were followed up until death or the endpoint of the study (September 1, 2007) with a length of follow-up for the surviving patients of 115 to 144 months. This made it possible to separate both patients and high-grade lesions on the basis of their long-term evolution. We used microdissection and polymerase chain reaction (PCR) analysis of 3p, 5q, and 9p microsatellites to assess LOH in high-grade lesions, and correlated the results to the long-term evolution of the lesions and patients.
Patients and Lesions The selection criteria for the molecular study were as follows: a baseline autofluorescent endoscopy showing the presence of at least one CIS or SD lesion, a clinical follow-up of a minimum of 9 years or until death, and sufficient archived material to achieve the molecular study. Thirty-seven patients who underwent the baseline autofluorescence endoscopy from February 1995 to October 1998 were selected for the molecular and follow-up study. From these patients, 26 were part of the cohort explored in our previous work on short-term evolution of precancerous bronchial lesions (5).
Histologic and Molecular Evaluation of Bronchial Biopsies The three other cuts adjacent to the hematoxylin-and-eosin–stained material were used for microdissection as described elsewhere (12, 23). DNA was purified from microdissected cells (200–300 cells) in 50 µl of proteinase K buffer (Tris-HCl 50 mM, pH 8.8; ethylenediaminetetraacetic acid, 0.5 mM; Tween 20, 0.5%; proteinase K, 500 µg/ml) for 48 hours at 37°C, followed by heating at 100°C for 10 minutes. Reference genomic DNA was extracted for each subject from blood and extracted by standard techniques involving phenolchloroform extraction, ethanol precipitation, and solubilization in 50 µl distilled water. Primers flanking polymorphic DNA motifs located at chromosomal regions 3p21–25, 5q21–22, and 9p22 were used to evaluate LOH as described (12). PCR products on 3p and 5q were analyzed by electrophoresis in an 8% polyacrylamide gel followed by ethidium bromide or silver staining. LOH and microsatellite instability were studied on two 9p22 and two 3p21 markers in labeling the reverse primer 5' end with C6-FAM (Eurogentec, Angers, France). 9p and 3p PCR products were analyzed using the automated ABI Prism sequencer model 310 (Applied Biosystems, Foster City, CA) according to the supplier's protocol. Microsatellite instability was defined as the appearance of additional bands that differed in size compared with normal tissue DNA, according to established techniques (24).
Definition of Progression, Stability, and Regression
Inclusion Period and Follow-up
Statistical Analysis
Regression/Progression Rate and Patients' Characteristics Thirty-seven patients and 54 lesions (31 CIS, 23 SD) were included in the molecular study. The main characteristics of the patients and high-grade lesions are summarized in Table 1, in which each patient is classified as CIS or SD according to the highest grade bronchial lesion found in his or her bronchial tree at baseline. The CIS and SD groups did not differ in age, sex, asbestos exposure, smoking status, or history of ENT or lung cancer. According to our restrictive definition of regression, stability, and progression, 19 lesions were classified as regressive and 17 lesions were considered as stable (also referred to as treatment sensitive), whereas 18 lesions progressed during the long-term monitoring period, including 11 lesions that relapsed with the same histology after local treatment (also referred to as treatment resistant) and 7 lesions that became locally invasive. From these progressing lesions, 17, including all the lesions that progressed to an invasive cancer, were initially classified as CIS. Conversely, 7 of 17 of the stable lesions and 15 of the 19 spontaneously regressive lesions were classified as SD at baseline (P < 0.0001, SD vs. CIS).
Long-Term Follow-up Details of long-term follow-up for each lesion and patient, including molecular analysis, local evolution of the lesions, treatment administered, occurrence of an invasive cancer and its location, and the cause of death, are given in the online supplement (Table E1). A mean of seven biopsies per lesion were taken during follow-up (range, 2–15). At the study endpoint (September 1, 2007), nine patients were still alive without lung cancer with a follow-up of 115 to 144 months and two patients without cancer were lost to follow-up at 102 and 120 months. Twenty-five patients died during follow-up, including nine of lung cancer; of these nine patients, four deaths could be directly attributable to the progression of the initial high-grade lesion. Other deaths from cancer were as follows: five patients with ENT cancer, one patient with bladder cancer, and one patient with esophageal cancer.
Molecular Abnormalities According to Histology
Progression/Regression Status and Molecular Abnormalities Table 3 shows the molecular abnormalities according to the progression/regression rate in the entire group of lesions and the CIS group. In the entire group of lesions, the existence of at least one LOH and 3p LOH were strongly associated with progression of the considered lesion (P < 0.01 and P < 0.0001, respectively). The association of 3p deletion with 5q or 9p deletion was also associated with progression of the lesion.
In the CIS group, 3p deletion and the occurrence of 3p or 5q deletion were significantly associated with progression (P < 0.02 and P < 0.04, respectively; Fisher's exact test). The microsatellite instability was not statistically associated with progression in the whole group or in the CIS group.
Overall and Lung Cancer–specific Survival Curves
Figure 1 shows the overall survival curve (median, 67 mo) and the lung cancer–specific survival curves (median not reached) for the whole group of patients. Survival curves of the patients according to the progression/regression status of the high-grade lesions are displayed in Figure 2. The median survival was not reached at the study endpoint in the group of patients with lesions that regressed, whereas median survival was 65 months in the stabilization (or treatment-sensitive) group, 65 months in the treatment-resistant group, and 44 months in the group of patients with lesions that became invasive, respectively.
Figure 3 displays the overall survival according to the histology of the highest grade lesion, and according to the presence or absence of 3p LOH in the highest grade lesion of a given patient. This figure shows a significantly better survival in the 3p LOH–free group (P = 0.04) and a tendency for a longer survival in the SD group as compared with CIS groups.
Influence of Lung or ENT Cancer History on the High-Grade Lesion Outcome Twenty of the 37 patients included in the study had a previous history of lung or ENT cancer (Table E1). The comparison of the lesions found in these patients and in individuals without previous malignancy did not show any difference in histology (SD vs. CIS), 3p LOH frequency (P = 1), or high-grade lesion evolution during the long-term monitoring period (P = 0.63).
During the past decade, molecular studies of precancerous bronchial lesions have shown a relationship of progressive increase of 3p and 9p LOH frequency within bronchial preinvasive lesions to increasing severity (12, 19). This suggested that specific LOH or the association of several LOH could be used as a marker of advanced bronchial disease, and thus could be tested as predictive of progression for a given lesion. In the present work, we postulated that the use of 3p, 5q, and 9p LOH analysis might be helpful to predict the aggressiveness of bronchial SD and CIS lesions. To take into account the length of progression of the premalignant epithelium, this study required a long-term (up to 12 yr) follow-up of the patients using fluorescence bronchoscopy and repeated bronchial sampling over time. Although our series contains a large proportion of patients with a history of ENT or lung cancer, who are known to be at very high risk of developing a second cancer in the respiratory tract, we did not find any difference in the behavior of the high-grade lesions between the patients with cancer history in their respiratory tract and the ones without. Therefore, it is likely that our results could be generalized to patients without previous malignancy. With a maximum of 144 months' monitoring period in 54 high-grade bronchial lesions, we found that 3p LOH and histology (SD vs. CIS) were strong predictive factors of progression. We also found a tendency for the association of 3p and 9p LOH, or the association of 3p and 5q LOH with progression, whereas there was no significant association between microsatellite instability and the lesion outcome. In this series, both prognostic values of histology and 3p LOH had a similar level of significance. However, 3p LOH was significantly associated with progression when the CIS group was analyzed separately, and was constantly found in CIS lesions that progressed to invasive cancer during the monitoring period. This demonstrated that the use of molecular analysis adds useful information over the use of histology alone for the prediction of progression. The results presented here are concordant with our previous findings on LOH frequency in precancerous bronchial lesions (12). In particular, 3p LOH appears to be the more frequent alteration in high-grade lesions, closely followed by 9p LOH, whereas 5q LOH appears to be a late event in bronchial carcinogenesis, more often associated with invasive lesions (12), but without prognosis value when found isolated from 3p or 9p deletion. On the other hand, 5q LOH may represent a molecular marker of clinical short-term progression when associated with 3p LOH, indicating a fully transformed epithelium. The 3p and 5q LOH allelotype was found in the present series in only six CIS lesions, from which two progressed to invasive cancer at 4 and 21 months' follow-up, three were treatment resistant, and only one was treatment sensitive. Similar findings were observed in our previous molecular study where the only CIS lesions displaying 3p, 5q and 9p alterations progressed into an invasive cancer despite endobronchial photodynamic therapy (12). The same observation was recently made in Foster and colleagues' study, which followed up the spontaneous evolution of two CIS lesions in the same patient and found that the CIS lesion that rapidly progressed into an invasive lesion harbored the association of 5q, 3p, and 9p LOH (25). To our knowledge, only three other studies on predictive molecular markers of preinvasive lesion progression have been published to date, and all focused exclusively on preinvasive lesions of the oral cavity (26–28). These studies analyzed 3p alterations in precancerous epithelium (3p LOH or fragile histidine triad [FHIT] gene alterations at 3p14), and all found a significant association between the molecular alteration and the progression from precursor oral lesions to oral squamous cell carcinoma (27, 28). These studies and ours confirm the major role of chromosome 3p alterations in the progression of the respiratory epithelium to invasive cancer, and their possible use as a reliable data that predict cancer progression. The present study, which represents the first published work on the predictive value of molecular markers for high-grade bronchial lesion outcome, may have important implications in clinical practice. The few studies that have implemented a bronchoscopy follow-up procedure for the evaluation of precancerous lesion outcome have found a variable progression rate of the highest grade lesions (i.e., SD and CIS), ranging from rates of 100 and 87% for CIS (5, 6) to 37% for SD (5) and 17% for SD or CIS (11). These differences appear to be mainly related to the interseries differences regarding the definition of progression, the precocity of the treatment approaches of CIS and SD lesions (5, 29), the possible interference of systemic cancer treatment (11), as well as the follow-up duration of the patients. Therefore, there is still a debate on whether or not all high-grade preinvasive lesions are truly premalignant, and how they should be managed (30). In this context, the finding that 3p LOH frequency is linked to the long-term progression/regression of lesions appears to be valuable because it would provide objective and reproducible data to determine whether a high-grade bronchial lesion should be immediately treated or not. In our series, only two spontaneously regressing SD lesions and none of the four long-term spontaneously regressing CIS lesions harbored a 3p LOH at baseline. Therefore, a 3p LOH–based treatment decision approach would have led to only two "overtreated" lesions. On the other hand, a treatment approach based on the presence of 3p or 9p LOH would have resulted in the treatment of six lesions that would have spontaneously regressed but would have missed only 1 among the 18 progressing lesions. We believe that this study also lends strong support to the World Health Organization classification for premalignant bronchial lesions, which differentiates bronchial CIS from SD, because it confirms the different outcomes of these lesions, as we previously observed (5), but also provides evidence of differences at the molecular level. In the present work, after a classification of the high-grade lesions on the basis of careful, double-histology analysis, we found a significantly lower proportion of long-term regression in CIS as compared with SD lesions and a trend for a shorter survival in patients presenting with CIS lesions. We also demonstrated the clinical value of both molecular analysis and regression status, because the patients harboring a high-grade bronchial lesion that regressed and those without 3p LOH appear to have the longer survival. Finally, we have clearly shown that the molecular abnormalities existing in SD (i.e., 3p and 9p LOH) are significantly less frequent as compared with those in CIS lesions. The distinction between SD and CIS was not always recognized in the literature as clinically relevant (11), presumably because of the difficulties encountered in the histologic classification of high-grade lesions, leading to interobservational variation in the interpretation of histology (11). The significant differences in 3p and 9p LOH frequency between CIS and SD observed in our study and their association with different outcomes suggest that the molecular analysis of the highest grade lesions could also help in their histoprognostic classification. The study presented here has several limitations. It only considers four molecular events, encompassing a relatively small number of chromosomal loci that correspond to putative tumor suppressor genes that are not yet identified. In particular, it is recognized that several LOH hot spots exist on the short arm of chromosome 3 in lung cancer and premalignant lesions (31). Each of these specific hot spots and/or the size of the 3p allele loss regions (31) may not have the same predictive value in terms of lesion aggressiveness. However, a more extensive study of 3p LOH in addition to 9p and 5q analysis would have been impossible in our series because of the small amount of the material that is available from paraffin-embedded archived samples. Another possible limitation is related to our treatment decision tree, which includes a conservative endobronchial treatment of the high-grade lesions that persisted unchanged at 3 months, and which obviously interferes with the natural evolution of the bronchial disease. To overcome this known limitation, we used in this study a very restrictive definition of "progression," including local progression to invasive cancer and local recurrence after the endobronchial treatment of the lesion during the monitoring period. This made it possible to accurately classify the more aggressive lesions in the progression group. We also selected in our database the lesions for which we had a very long-term monitoring period, up to 12 years, to ascertain that the untreated, regressing lesions were truly benign. Besides the length of the monitoring period, the strength of our study also relies on the robustness and the simplicity of the molecular methods that were used, which could be easily implemented in the clinical practice. The microdissection of paraffin-embedded tissue sections using laser capture is currently a simple and reliable technique that is routinely used in our somatic genetic tumor laboratory. In this series, the microdissection of paraffin-embedded tissue sections appeared to be mandatory for several reasons: the very small size of the sampled preinvasive epithelium (frequently a few hundred cells), the need to avoid contamination from normal or lower grade adjacent precancerous epithelium, as well as the need to analyze a very precise epithelial area. Coupled with PCR analysis, the method is inexpensive and the results can be available in less than a week's time, which assures that they could be effectively integrated in a treatment decision tree. In the future, more sophisticated methods, such as quantitative multiplex PCR of short fluorescent fragments (QMPSF), may replace the single PCR method, in allowing simultaneous amplification under quantitative conditions of multiple dye-labeled targets from both tumor and nonmalignant tissues (32). The QMPSF technique, which also appears to be very robust and is rapidly performed, has been successfully used in our laboratory to better characterize quantitative genetic alterations in colorectal cancers (32). However, it needs to be adapted to very small, fixed, and paraffin-embedded samples, which was not possible at the time of this study. A prospective study of QMPSF on fresh or frozen tissue could, however, be performed in the future. Taken together, these data indicate that molecular analysis can be added to the histopathologic grading of precancerous lesions to better classify lesions and to identify the more aggressive ones. They also suggest the possible use of a simple, molecular-based management decision tree of high-grade precancerous lesions. In this decision tree, the high-grade lesions that harbor a 3p LOH would be more aggressively treated, because their probability of progression appears to be very high (76% in our series), whereas the lesions without 3p or 9p molecular alterations, whose probability of progression is only 5%, would be submitted to a monitoring procedure. In the future, lower grade lesions, such as mild and moderate dysplasia, could also be assessed using an equivalent molecular approach. In previous work, we demonstrated the low probability of progression of low-grade precancerous bronchial lesions to high-grade lesions at 24-month monitoring, as well as the association of their progression with a field of cancerization effect (12). On the basis of this observation, our group is currently conducting a prospective, multicenter, 3-year bronchoscopic follow-up study of low-grade precancerous lesions (NCT00213603; http://clinicaltrials-nccs.nlm.nih.gov/ct/show/NCT00213603). The results of this trial involving 350 high-risk individuals will be available in 2009, and will help to assess the probability of progression of the lowest grade lesions as well as their relationship with molecular alterations. Future studies will also analyze the cost-effectiveness of an approach integrating molecular analysis into the management decision tree of high-grade preinvasive bronchial lesions.
Supported by a research grant from the Comités Départementaux de la Ligue contre le Cancer (Eure et Seine Maritime), and the French Cancéropole Nord Ouest. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200704-598OC on October 25, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form April 18, 2007; accepted in final form October 25, 2007
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
2 tests or Fisher's exact test when required. A P value less than 0.05 was considered significant. 
