Progressive Endobronchial Premalignancy: Marked by Original CIN

In the United States, over 220,000 new cases of lung cancer were diagnosed in 2010, with over 157,000 deaths (1). The 5-year survival of lung cancer has improved from 9% to 16% over the past 50 years, a hard-fought but profoundly disappointing advance. Early detection and prevention are likely to lead to the majority of the decrease in cancer mortality in the future. In this respect, the recent publication of the results of the National Lung Screening Trial, demonstrating a 20.3% reduction in lung cancer mortality in the low-dose computed tomography (CT) arm, is a landmark that may lead to significant improvement in lung cancer survival, provided we are able to implement screening in a cost-effective manner and minimize harm from diagnostic workup (2).

While there is now a clear lung cancer mortality benefit to low-dose CT based early detection strategies in high-risk groups, bronchoscopic surveillance of the central airways has lagged behind. It is widely believed that the detection and treatment of preneoplastic lesions in accessible epithelial organs (the skin, cervix, and colon) reduces cancer mortality. In the lung, the complex airway anatomy and the undifferentiated appearance of preneoplasia have slowed progress on the clinical detection of this biomarker. Autofluorescence bronchoscopy significantly improves the detection rate of preneoplastic lesions, raising sensitivity by 3- to 4-fold over white light bronchoscopy and making a focus on treatment of preneoplasia more feasible (3). However, the natural history of airway preneoplasia is poorly defined, and it has not been clear which lesions warrant detection and treatment (4, 5). Histologic dysplasia (at least short of carcinoma in situ [CIS]) is not a strong predictor of lung cancer risk at a specific bronchial site. While this finding supports the American College of Chest Physician guidelines of bronchoscopic follow-up of severe dysplasia or CIS in the central airways, many would advocate local treatment for CIS that does not regress after initial biopsy (6).

Molecular analysis of premalignant lesions has been ongoing since the early 1990s, and overall patterns of progressive chromosomal instability (CIN) in lung squamous cell carcinogenesis have been proposed (4). Salaun and coworkers reported that, in a series of patients with severe dysplasia or CIS, chromosome 3p loss of heterozygosity, as demonstrated by PCR analysis of microdissected formalin fixed paraffin-embedded lesions, is significantly associated with progression to invasive cancer (7). However, these studies offer only a limited analysis of possible changes to the premalignant genome.

In this issue of the Journal, van Boerdonk and colleagues (pp 948) report a search for CIN in the genome of microdissected dysplastic epithelium obtained from squamous metaplasia lesions. Multiple regions of genome copy number gain and loss, the indelible stains of CIN, were found in five of the six subjects with progressive bronchial lesions and subsequently diagnosed invasive squamous cell lung cancer (8). There are several remarkable aspects to this report. First, to obtain these six biopsies, the investigators performed 1,238 autofluorescence bronchoscopies over 12 years in a very high-risk cohort of 474 subjects followed at their institution. In addition, at each of the 4,170 endobronchial sites biopsied in these subjects, the investigators acquired and banked a frozen biopsy in addition to the routine formalin-fixed, paraffin-embedded specimen. Each of the six nested case and 23 control biopsies were carefully microdissected, leaving only the genomes of dysplastic epithelial cells to contribute to the assay output. Finally, they employed array comparative genomic hybridization (CGH), using a platform that quantifies over 40,000 chromosomal loci, to analyze these precious specimens. In this way van Boerdonk and colleagues signifcantly build on and extend the work of Salaun and coworkers, in which three genetic loci were analyzed. DNA copy number alterations, particularly at chromosomes 3p, 3q, 9p, and 17p, were significantly more frequent among cases than among control subjects. Control subjects were selected to exhibit similar squamous metaplasia at baseline, but no cancer diagnosis on follow up.

There are several limitations to this study. First, 5 of the 6 cases (and 9 of 23 control subjects) had a history of respiratory tract cancer, so it is possible that the lesions analyzed represented recurrent, rather than incident, lung cancer. This concern is particularly salient for two subjects who had been treated with radiotherapy for invasive cancer at the site of the new squamous cell lung cancer. In the remaining four cases, the new tumor arose at a distant site, making recurrence less likely. This distinction between CIN originating in the neoplastic field and CIN inherited from the parent tumor is important for models of the natural history of lung carcinogenesis, but the wages of CIN seem apparent: four of the five subjects exhibiting cytogenetically abnormal lesions (and five of the six with incident lung cancer) died of lung cancer over the next 21 to 68 months. This highlights the question of how effective bronchoscopic early detection and local treatment were in this setting.

The work described above is focused on the likelihood of progression to invasive cancer at a specific site. We are clearly a long way from a clinically applicable approach, except perhaps for highly advanced dysplasias. There is, however, significant evidence that global damage to the respiratory epithelium is a biomarker of lung cancer risk. Slaughter and colleagues first described the field cancerization concept in 1953, when two or more independent squamous cell carcinomas were described in 88 of 783 patients with oral cancer (9). In addition, widespread dysplastic epithelial changes were found in these subjects. These diffuse premalignant changes in patients with lung cancer are likely due to a combination of genetic susceptibility, exposure of the entire respiratory epithelium to the multiple carcinogens present in tobacco smoke, and the expansion and dispersal of clones of mutant respiratory epithelial cells, as has been demonstrated for both p53 and EGFR mutant preneoplastic cells (10, 11). The hypothesis that the presence of premalignant epithelial cells harboring DNA copy number alterations reflects risk of lung cancer at some site (but not necessarily the site in which the analysis was undertaken) is supported by previous studies using fluorescence in situ hybridization (FISH) to detect chromosomal aneusomy in sputum and dysplastic biopsies, as well as spectral karyotyping of cultured bronchial epithelial cells, in which clonal chromosomal abnormalities parallel lung cancer risk factors (12–15). These approaches are limited by the small number of chromosomal locations that can be interrogated by FISH. We expect that the use of higher-resolution methods, including CGH, will improve our ability to define lung cancer risk, with potential application to both bronchoscopic and low-dose CT early detection strategies.