Natural history of bronchial preinvasive lesions

Abstract

Preinvasive bronchial lesions defined as dysplasia and carcinoma in situ (CIS) have been considered as precursors of squamous cell carcinoma of the lung. The risk and rate of progression of preinvasive lesions to invasive squamous cell carcinoma as well as the mechanism of progression or regression are incompletely understood. While the evidence for the multistage, stepwise progression model is weak with relatively few documented lesions that progress through various grades of dysplasia to CIS and then to invasive carcinoma, the concept of field carcinogenesis is strongly supported. The presence of high-grade dysplasia or CIS is a risk marker for lung cancer both in the central airways and peripheral lung. Genetic alterations such as loss of heterozygosity in chromosome 3p or chromosomal aneusomy as well as host factors such as the inflammatory load and levels of anti-inflammatory proteins in the lung influence the progression or regression of preinvasive lesions. CIS is different than severe dysplasia at the molecular level and has different clinical outcome. Molecular analysis of dysplastic lesions that progress to CIS or invasive cancer and rare lesions that progress rapidly from hyperplasia or metaplasia to CIS or invasive cancer will shed light on the key molecular determinants driving development to an invasive phenotype versus those associated with tobacco smoke damage.

1 Introduction

Lung cancer is the most common cause of cancer deaths worldwide [1, 2]. Less than 16% of patients with lung cancer patients survive 5 years or more [2], owing to late diagnosis and a paucity of effective therapies. In contrast to the poor survival of patients with advanced disease, the survival of those with stage 0 (carcinoma in situ, CIS) or stage 1A disease (tumor ≤2 cm without metastatic spread) has excellent prognosis with 5-year survival >70% [3–5].

In contrast to the peripheral airways and lung parenchyma where the majority of adenocarcinomas arise, squamous cell carcinomas usually arise in the central airways (first five generations of bronchi). Changes in the bronchial epithelium in the central airways are readily accessible to bronchoscopic biopsy under local anesthesia and conscious sedation. The development of autofluorescence bronchoscopy that makes use of blue/violet light to induce tissue fluorescence provides a highly sensitive method to localize preinvasive lesions (moderate/severe dysplasia and carcinoma in situ) in the central airways for biopsy confirmation [6–8]. A better understanding of the make-up and natural history of preinvasive lesions will lead to identification of molecular targets and pathways for early detection and chemopreventive intervention [9]. It will also assist in clinical decision regarding management of these lesions. Development of centrally located squamous cell carcinoma and peripherally located adenocarcinoma proceeds through different pathways [10]. In this review, we will focus on preinvasive lesions in the squamous cell carcinoma development pathway because recent advances in bronchoscopic imaging methods make it possible to study these lesions in vivo [8, 11]. The recent information on the natural history of these lesions and the factors that predict progression of these lesions to invasive squamous cell carcinoma will be reviewed.

2 Squamous carcinogenesis

Squamous cell carcinoma is thought to develop through a stepwise process where the epithelium changes from normal to hyperplasia, metaplasia, mild, moderate, and severe dysplasia and then carcinoma in situ [12–15]. The recent histological classification by the World Health Organization includes squamous dysplasia and CIS as a main morphological form of preinvasive lung lesions [16]. In mild dysplasia, there is disarray in the lower third of the epithelium and mild cytological atypia. Mitoses are absent or rare. Moderate dysplasia is associated with disarray in the lower two thirds of epithelium and more significant cytological atypia. Mitotic figures are confined to the lower third. In severe dysplasia, the disarray extends into the upper third of the epithelium but does not reach the surface. Mitoses are confined to the lower two third. CIS is associated with extension of the disarray to the epithelial surface with malignant cytological features and mitotic figures present through the full thickness. Atypical or malignant cytological features are characterized by variations in nuclear size, shape, and staining properties such as hyperchromatism, multiplicity of nucleoli, irregularities of nuclear membrane, coarsening of chromatin distribution, and discordance of maturation between nucleus and cytoplasm (dyskaryosis). In a subset of squamous dysplasia, there is budding in the subepithelial tissues that result in papillary protrusions of the epithelium. This has been termed angiogenic squamous dysplasia [17].

There is considerable interobserver variation in the classification of preinvasive lesions [18, 19]. Comparing the histopathology grades scored by two experienced pulmonary pathologists with quantitative nuclear morphometry showed that although the objective nuclear morphometry index was significantly different between invasive carcinoma and CIS as well as between CIS and various grades of dysplasia by visual examination, there was no significant difference between moderate and mild dysplasia or between mild dysplasia and metaplasia [20]. The study illustrates the uncertainty in classification of the intermediate grades by visual examination even by highly experienced pulmonary pathologists. The same study also showed a better correlation between allelic losses in several chromosomal regions with nuclear morphometry than with histopathology [20]. The limitation of conventional histopathology classification in terms of predicting the malignancy potential of preinvasive lesions will be further discussed below.

The stepwise progression model of squamous cell carcinogenesis is based on several lines of evidence in animal models [21] and in humans [12–14, 21–26]. Serial sputum cytology examinations in uranium miners and in smokers showed that invasive lung cancer develops through a series of stages from mild, moderate, and severe atypia, carcinoma in situ, and then invasive cancer [22, 23]. However, sputum cells can come from different parts of the tracheobronchial tree. Progression of sputum atypia to cancer may reflect field cancerization rather than progression from the same lesion. The concept of stepwise progression of preinvasive lesions to invasive carcinoma was recently challenged [27].

3 Prevalence of squamous preinvasive bronchial lesions

The prevalence of preinvasive lesions has been studied in vivo using autofluorescence bronchoscopies in heavy smokers with and without lung cancer. In the largest study involving 511 volunteer smokers 40 to 74 years of age (mean age 56 years) who had smoked for at least 20 pack-years (number of packs of cigarettes smoked per day × number of years of smoking), the prevalence of mild, moderate, or severe dysplasia or CIS was 44%, 13%, 6%, and 1.6%, respectively [28, 29]. Similar results were found in autofluorescence bronchoscopy studies conducted around the same period of time in the 1990s [30, 31]. The significant findings in the volunteer smoker studies are (a) women had a lower prevalence of moderate/severe dysplasia and CIS (14% versus 31%) compared with men, (b) women had fewer of these high-grade preinvasive lesions, and (c) the prevalence of preinvasive lesions did not change substantially for more than 10 years after smoking cessation [28, 29]. The prevalence of preinvasive lesions appears to have decreased recently probably reflecting the worldwide trend of a change in the proportion of squamous cell carcinoma versus adenocarcinoma. Among 1,581 smokers with a similar demographics as our previous studies [28, 29], the prevalence of moderate or severe dysplasia or CIS has become 9%, 1.9%, and 0.8%, respectively, a decade later (Lam et al., unpublished data). Fewer high-grade preinvasive lesions were also observed in more recent multicenter autofluorescence bronchoscopy studies [32–34].

4 Longitudinal studies of preinvasive lesions

In an attempt to clarify the natural history of preinvasive lesions, longitudinal studies using serial autofluorescence bronchoscopy and biopsy were performed in patients with dysplasia or CIS [35–41].

With rare exceptions, these studies either had a small number of subjects (≤50) and/or a short duration (≤3 years) of follow-up [42]. A significant proportion of the subjects were patients with previous lung or head and neck cancer. Some of the studies combined severe dysplasia with CIS [43] making it difficult to determine the natural history of these lesions separately. The study by Satoh et al. illustrates the importance of long-term follow-up [44]. In a patient with mild dysplasia, the pathology of the follow-up biopsies fluctuates between squamous metaplasia to different grades of dysplasia before development of an invasive carcinoma 6 years later [44].

Overall, 59% of the lesions with severe dysplasia regressed spontaneously [27, 36–41, 45] while 41% persisted or progressed to CIS or invasive cancer. The progression rate of CIS to invasive cancer is difficult to evaluate as all except one center [43] treated patients with CIS at the time of diagnosis or when the lesion did not resolve within 3 months. Overall, only 13% of the CIS lesions regressed spontaneously without treatment while the majority (87%) progressed to invasive cancer, recurred despite endobronchial therapy, or persisted [27, 35–39]. Over 50% of patients with CIS were found to progress to invasive cancer within 30 months [35, 37, 46].

In a study where patients with severe dysplasia or CIS were not treated until development of invasive cancer, 16 patients had seven severe dysplasia and 29 CIS [43]. Among the 18 lesions with severe dysplasia/CIS that were evaluable without the influence of treatment to the lesion or to a synchronous lesion in the same area, five patients with six CIS progressed to invasive cancer within 15 months. Three of these five patients had progressive disease despite radical radiotherapy or photodynamic therapy [43]. Since the majority of CIS lesions do not resolve spontaneously and when these lesions were allowed to progress to invasive cancer, they can become incurable by local therapy, treatment is preferable instead of observation with repeated bronchoscopies and biopsies.

In contrast to CIS or severe dysplasia, with the exception of the study by Breuer et al. [27], the spontaneous regression rate is higher and the progression rate is lower for lesions with mild/moderate dysplasia. For example, in the study by George et al. [43], none of the lesions with mild or moderate dysplasia progressed to CIS or invasive cancer during the period of follow-up of 12 to 85 months (median 23 months). Hoshino et al. reported only one of the 88 (1.1%) lesions with mild or moderate dysplasia progressed to invasive carcinoma [40]. The study by Breuer et al. [27] found the rate of progression to CIS or invasive cancer from metaplasia to be 9% which is the same as the progression rate of mild/moderate dysplasia. Progression of metaplasia to CIS or invasive cancer occurred within 4 to 7 months in three of the four lesions studied [27]. Occasional progression of hyperplasia or metaplasia to CIS/invasive cancer within a relatively short time has also been reported by other investigators [37]. This led to the concept of a nonstepwise development of squamous cell carcinoma. Alternatively, the intermediate lesions are not always detectable by periodic bronchoscopy and biopsy because of their rapid progression. It should be noted that autofluorescence bronchoscopy devices used in many of the longitudinal studies to localize preinvasive lesions were optimized to detect lesions that are moderate dysplasia or worse [8]. Very few control biopsies were performed from areas with normal or slightly abnormal fluorescence. Lesions with abnormal fluorescence but a benign pathology have been shown to have more frequent DNA alterations compared to those with normal fluorescence [47]. The combined effect of identifying lesions with more genetic alterations and fewer biopsies probably led to an overestimation of the malignancy potential of areas of hyperplasia, metaplasia, or mild dysplasia.

5 British Columbia Lung Health Study

As part of several chemoprevention trials sponsored by National Cancer Institute, 2,154 heavy volunteer smokers above 45 years of age with ≥20 pack-years smoking history without previous cancer in the upper aerodigestive tract were screened by autofluorescence bronchoscopy prior to enrollment onto a chemoprevention trial [48–51]. Repeat bronchoscopy and biopsy of the same sites was performed in 3 to 6 months for those with severe or moderate dysplasia, respectively. Those with CIS were treated with electrocautery ablation or cryotherapy. Seven hundred fifty-four subjects had two or more bronchoscopies. All subjects were followed up in person or through the Cancer Registry. One hundred one subjects were found to have lung cancer after a mean follow-up of 9.2 years (range 1.7 to 20.9 years). A case-control study was performed by matching the cancer and noncancer subjects by age, gender, smoking history, and ethnicity (Table 1). The pattern of progression of various histopathology grades of the bronchial biopsies obtained in the first (baseline) bronchoscopy was compared between the two groups (Fig. 1). For biopsies that were normal, hyperplasia, metaplasia, mild dysplasia, or moderate dysplasia, progression was defined by a change in the pathology by two or more grades (e.g., mild dysplasia to at least severe dysplasia). For severe dysplasia, progression was defined as a change by one grade higher or more (e.g., severe dysplasia to CIS or invasive cancer). The progression rates are higher in the cancer group versus the cancer-free group (Fig. 1; Table 2). The difference reached statistical significance for lesions that were normal or moderate dysplasia at baseline (P=0.04 and 0.045, respectively; one-tailed Fisher exact test). Progression to CIS/invasive cancer in the same site where the initial biopsy showed severe dysplasia was 5.6% and 4.9% for moderate dysplasia. In contrast, the progression rate for metaplasia or mild dysplasia and normal or hyperplasia was 0.8% and 0.9%, respectively. As discussed above, the progression rates of lesions that were normal, hyperplasia, metaplasia, or mild dysplasia were probably falsely high as only one or two control biopsies were taken from areas with normal autofluorescence. Although some lesions appear to progress from hyperplasia/metaplasia to CIS/invasive cancer within 2 years in a fashion similar to that observed by Breuer et al. [27] which is much shorter than the traditional thinking of 10 to 20 years, the progression rate appears to correlate with the severity of the pathological grade in general.

Table 1.

Case-control study of smokers who did or did not develop lung cancer

	No lung cancer	Lung cancer
No. of subjects	101	101
Men/women	59:42	59:42
Age (mean ± SD)	61±7	61±8
Ethnicity White/Asian	98:3	98:3
Current/former smokers	57:44	57:44
Pack-years (mean ± SD)	47±16	57±25
Smoking duration (years; mean ± SD)	41±9	40±8
Years quit (former smokers; mean ± SD)	11±7	10±7
Duration of follow-up (years, mean ± SD)	9.8±3.9	9.2±3.9
No. of biopsies	1,759	2,051
Development of
Squamous cell carcinoma		33
Adenocarcinoma		42
Other cell types		26

Fig. 1.

Outcome of lesions that were normal, hyperplasia, metaplasia, mild dysplasia, moderate dysplasia, or severe dysplasia at baseline. Subjects free of lung cancer on follow-up for each histology grade are shown in a, c, e, g, i, and k, and those who developed CIS or invasive cancer in the same site or a distant site are shown in b, d, f, h, j, and l. The progression rates are higher in the cancer group and reach statistical significance for lesions that were normal or moderate dysplasia at baseline (P=0.04 and 0.045, respectively; 2 by 2 comparison, one-tailed Fisher exact test)

Table 2.

Progression rates according to histopathology grade

	Cancer group (%)	Noncancer group (%)	P value
Normal	22	4	0.04
Hyperplasia	18	10	0.16
Metaplasia	15	9	0.11
Mild dysplasia	6	2	0.10
Moderate dysplasia	24	4	0.045
Severe dysplasia	11	0	0.50

For biopsies that were normal, hyperplasia, metaplasia, mild dysplasia, or moderate dysplasia, progression was defined by a change in the pathology by two or more grades (e.g., mild dysplasia to at least severe dysplasia). For severe dysplasia, progression was defined as a change by one grade or more (e.g., severe dysplasia to CIS or invasive cancer)

The second observation from the BC Lung Health Study is that a significantly higher progression rate occurred in those who developed squamous cell carcinoma compared with those who developed adenocarcinoma. An example of this is shown in Fig. 2. The progression rate of mild dysplasia was 30.9% for those who developed squamous cell carcinoma versus 3.3% in those who developed adenocarcinoma (P=0.041). The underlying mechanism has not been identified yet. The third important finding from this study is that there are more cancers developed from a separate site in the same individual than progression from the initially biopsied site. This is in agreement with previous studies [43, 45]. Thus, the presence of dysplasia/CIS is a risk marker for lung cancer developing elsewhere in the lung.

Fig. 2.

Outcome of lesions with mild dysplasia in subjects who developed squamous cell carcinoma versus those who developed adenocarcinoma. A significantly higher progression rate was observed in those who developed squamous cell carcinoma (P=0.041; P=0.002 for lesions that progressed to moderate dysplasia or worse)

Taken together, the results of this and other studies suggest that bronchial preinvasive lesions progress to invasive squamous carcinoma in a stepwise fashion although some lesions appear to progress rapidly from hyperplasia or metaplasia. The risk of progression to CIS or invasive cancer is higher for severe dysplasia than for metaplasia or mild dysplasia. The majority of CIS progresses to invasive carcinoma. High-grade dysplasia and CIS are markers of lung cancer development for the entire epithelium at risk. Close follow-up with autofluorescence bronchoscopy and radiological imaging should be considered in patients with these preinvasive lesions.

6 Genetic alterations in preinvasive bronchial lesions

Lung cancer development is characterized by sequential accumulation of epigenetic and genetic alterations in the bronchial epithelial cells [15]. In general, mutations follow a sequence, with allelic losses at chromosomes 3p (3p21, 3p14, 3p22–24, and 3p12), 9p21 (p16^INK4a), and 8p21–23 occurring relatively early, losses and inactivation of the 13q14 (RB) and 17q13 (TP53) genes are intermediate, and losses at 5q are late events [10, 15, 52–54]. The losses at 3p are progressive, and advanced lesions and tumors often have lost most of the chromosomal arm or the entire arm, while early lesions have more focal lesions [54]. Abnormal gene methylation is also a frequent event in squamous carcinogenesis [55–57]. For example, there is an increasing frequency of methylation of p16^INK4a during disease progression from hyperplasia (17%) to squamous metaplasia (24%) to CIS and invasive carcinoma (50–75%) [56].

These studies indicate that tobacco smoke exposure results in widespread changes in the bronchial tree. Because not all preinvasive lesions develop into invasive tumors, it is important to identify the molecular determinants driving to an invasive phenotype versus those associated with tobacco smoke damage.

In a recent study, 37 patients with 31 CIS and 23 lesions with severe dysplasia were followed up by repeated autofluorescence bronchoscopy and biopsy for up to 12 years [38, 39]. Loss of heterozygosity (LOH) of chromosome 3p was found to be significantly more frequent in CIS compared to severe dysplasia. All CIS lesions that progressed to invasive cancer and 91% of CIS lesions that were resistant to treatment had 3p LOH. In four of the 31 CIS lesions (13%) that regressed spontaneously, only 12% had 3p LOH [39]. In a study by Jonsson and coworkers using a four-color fluorescence in situ hybridization (FISH) probe set encompassing the chromosome 6 centromere, 5p15.2, 7p12 (EGFR), and 8q24 (v-myc), the proportion of subjects with chromosomal aneusomy was found to increase from moderate dysplasia (22.2%) to severe dysplasia (41.7%) and CIS lesions (75%)[58]. The odds of the presence of chromosomal aneusomy in preinvasive lesions were 4.5 times greater in those with invasive lung cancer in an adjacent site or at a distant site (different lobe). Using two additional FISH probes (TP63 and CEP3) in the same specimen set, the sensitivity and specificity of predicting lung cancer development were 82% and 58%, respectively, using four probes (TP63, CEP3, CEP6, MYC) [59]. Whether LOH or chromosomal aneusomy in preinvasive lesions predicts the subsequent development of invasive lung cancer will require a prospective longitudinal study. More specific markers are also needed to improve the accuracy.

The significance of preinvasive lesions in lung cancer development has also been characterized by immunohistochemistry. Aberrant expression of proteins playing key roles in cell cycle control, apoptosis, and DNA repair such as p53, Ki-67, cyclin D1, cyclin E, Bax, Bcl2, phosphor-Akt, p65/RELA, and cIAP-2 have been investigated. Overexpression of p53 (especially suprabasal staining), cyclin D1/cyclin E, and Ki-67 were detected by immunohistochemistry in bronchial dysplasia, increasing in frequency with their histology grade [60–63]. The acquisition of anti-apoptotic signals is critical for the survival of cancer cells. Bronchial epithelial protein levels of the phosphorylated (active) form of AKT kinase and the caspase inhibitor cIAP-2 were found to be increased in more advanced grades of bronchial IEN lesions than in normal bronchial epithelium [64]. Additionally, the percentage of biopsies with nuclear localization of p65/RELA in epithelial cells increases with advancing pathology grade, suggesting that NF-κB transcriptional activity is induced more frequently in advanced IEN lesions [64].

7 Factors which predict lung cancer development

Telomerase expression in nonmalignant epithelium in high-risk patients with treated lung cancer is associated with an increased subsequent development of second bronchial malignancies [65]. Recent data on longitudinal observation, reported by Hoshino et al., showed that high telomerase activity, increased Ki-67 labeling index, and p53 positivity in dysplasia tend to remain as dysplasia and might have the potential to progress to squamous cell carcinoma [40]. Host factors can also influence progression or regression of preinvasive lesions. High plasma levels of C-reactive protein predict progression of bronchial dysplasia [66]. Surfactant protein D, a large multimeric, calcium-dependent collagenous glycoprotein secreted by type II pneumocytes is an important regulator of innate immunity, inflammation, and oxidative stress. Decreased level of SPD in bronchoalveolar lavage fluid was found to correlate with progression of bronchial dysplasia [67]. Higher levels of another anti-inflammatory protein, CC10, secreted by bronchiolar Clara cells, were found to correlate with regression of bronchial dysplasia and improvement in sputum cytometry [68]. Thus, the inflammatory load in the local environment and anti-inflammatory proteins may determine progression or regression of preneoplastic lesions in addition to genetic and epigenetic alterations.

8 Conclusion

Among bronchial preinvasive lesions, CIS is a strong predictor of progression to invasive squamous cell carcinoma. CIS is different than severe dysplasia at the molecular level and has different clinical outcome. The spontaneous regression rate of dysplasia is significantly higher than CIS (>50% versus ~13%) even for high-grade dysplasia. The addition of nuclear morphometry or molecular analysis to histopathologic grading allows more accurate classification of preinvasive lesions and better identification of lesions that are biologically more aggressive. Host factors such as the inflammatory load and anti-inflammatory proteins can influence the progression or regression of preinvasive lesions. More specific biomarkers are needed to improve the accuracy of predicting which dysplastic lesion will progress to carcinoma. While the evidence for stepwise progression model is relatively weak, the concept of field carcinogenesis is strongly supported. The presence of high-grade dysplasia or CIS is a risk marker for lung cancer both in the central airways and the peripheral lung. Close follow-up with bronchoscopy and radiological imaging are indicated in patients with these lesions.

9 Key unanswered questions

The key molecular determinants driving development of preinvasive bronchial lesions to an invasive phenotype versus those associated with tobacco smoke damage have yet to be defined. Molecular analysis of rare lesions that progress rapidly from hyperplasia or metaplasia to CIS or invasive cancer as well as dysplasias that progress to CIS or invasive cancer will shed light on this very important issue.