Because of CT’s high sensitivity and lack of specificity, it would be desirable to develop a minimally invasive test for genetic susceptibility that may assist in identifying those individuals at the highest risk for developing lung cancer. However, conventional cytologic sputum screening lacks sensitivity for various reasons, including its inability to detect small atypical squamous cells, the fact that abnormal cells may not be shed from peripherally based lesions, the patient’s inability to produce an adequate cough specimen, and contamination of the specimen by neutrophils and bacteria.
The uninduced sputum from healthy subjects contains for the most part squamous cells from oral contamination and upper airways, a few alveolar macrophages, and a few bronchial epithelial cells.14
In patients who have smoked and are at risk to develop bronchogenic carcinoma, the precursor lesions are squamous metaplasia, squamous dysplasia and carcinoma in situ
that occur in the bronchi and large airways. Sputum cytology is the only noninvasive way to assess these lesions. The sputum classification schemes consist of gradations of microscopic abnormalities similar to those observed in histologic sections from lower airways of smokers. The cytologic manifestations of dysplasia occur as increasingly severe cellular changes involving both: cell size and nuclear size, chromatin density, and nuclear membrane irregularity.13
Molecular changes occur early in neoplastic development including LOH at one or more 3p regions and 9p21, and may be found in histologically normal epithelium15
similar to the molecular abnormalities of 3p22.1 and SP-A
, and are the manifestations of a field cancerization effect.6,7
Recently, Wang et al16
detailed four types of bronchial epithelial dysplasia (basal cell dysplasia, columnar cell dysplasia, bronchial epithelial dysplasia with transitional differentiation, and squamous dysplasia) that occurred throughout the bronchial tree in patients who had their lungs resected for bronchogenic cancer. For both squamous and adenocarcinomas, there existed large numbers of various types of bronchial epithelial dysplasias regardless of the histological subtype of the bronchogenic carcinoma. Hence, it is not surprising that sputum from patients with adenocarcinoma should manifest cellular atypias in the form of squamous metaplasia, mild, moderate, and severe dysplasia, representative of the field effect of atypias secondary to tobacco usage, passive smoking, or industrial or asbestos exposure.15
In a recent prospective study of smokers with COPD, risk for incident lung cancer was increased among those with sputum cytologic atypia graded as moderate or worse.17
Because of the weaker association between adenocarcinoma and sputum atypia, it was suggested that other biomarkers in sputum such as FISH assays of chromosomal changes may be complementary to diagnosis.17
The findings in this study showing squamous metaplasia and all degrees of dysplasia to be present in patients with lung cancer regardless of histologic subtype are consistent with the findings of recent investigators.12,16,17
If only the abnormalities detected on FISH analysis were used, the sensitivity was higher but the specificity was lower. However, a combination of FISH and sputum cytology resulted in a considerably higher sensitivity rate than obtained with morphology alone. We concluded that subjects in a high-risk group, with high probability scores derived from cytologic and FISH analyses according to our model, should undergo CT scanning. If the CT findings were negative, these subjects would be ideal candidates to undergo fluorescence bronchoscopy to exclude the presence of central airway preinvasive malignant lesions.
In this study, there were six high-risk patients without CT evidence of lung cancer that might qualify for bronchoscopy based on their probability scores. The finding of a low probability score in several of the lung cancer patients no doubt reflected an inadequate sputum sample. In the future, use of induced sputa and stricter criteria for adequacy, such as the presence of bronchial epithelial cells and a greater number of histiocytes may improve the accuracy of the test.
The results of this study validated our approach to measuring and quantitating molecular abnormalities in consecutive fields of epithelial cells that were not necessarily cytologically abnormal. The presence of chromosomal abnormalities in both normal and atypical epithelial cells exfoliated in sputum confirmed our previous findings that cellular genetic abnormalities of 3p22.1 and 10q22.3 reflect a field cancerization effect within the bronchial cells of individuals at high risk for developing cancer.6,7
Other investigators tested epithelial cells in sputum by FISH using a commercial probe set for four different chromosomal regions (5p15, 6p11- q11, 7p12 (including epidermal growth factor receptor) and 8q24 (including C-myc
), and required a positive sputum diagnosis to be based on DNA copy number gains for at least two probes in a minimum of two or three cells.18,19
When using this approach, the sensitivity (50%) and specificity (81%) of FISH did not exceed the sensitivity of sputum cytology to detect lung cancer.18
Furthermore, heavy tobacco smokers and asbestos-exposed workers had FISH results similar to those seen in never smokers, suggesting that the composition of these probes was not optimal for detecting early lung cancer in high-risk populations.18
In contrast, others reported that FISH combined with cytology led to an improved diagnosis of malignancy.20
In a recent study, combined genetic aberrations for genes HYAL2 and FHIT were found with FISH in 76% of sputa from patients with cancer, but in only 47% of cases that considered positive on cytology, demonstrating that with an appropriate choice of probes, FISH can detect abnormal cells that may be undetectable by cytology.
Our sputum probe set of 3p22.1 and SP-A
was selected on the basis of results of high-resolution comparative genomic hybridization analysis of cDNA microarrays in adenocarcinomas and squamous cell carcinomas that showed significantly higher levels of these deletions, relative to those in normal human bronchial epithelial cells, in almost all tumors tested.21
These probes were subsequently tested by FISH in adenocarcinomas and squamous cell carcinomas and found to correlate significantly with the results of the comparative genomic hybridization.21
Deletion of SP-A
are frequent in lung cancer cells as well as in adjacent bronchi, normal lung, and bronchial cells from main stem bronchi on the normal and tumor sides.5–7
Deletions of SP-A
in lung cancer cells are associated with polysomy of the genes for centromeric 10/10q22.3, EGFR, 5p15, 6p11-q11, and C-MYC
, which are genes that drive cell cycle and proliferation.22
deletions also correlated with lymph node metastases, higher stage and poor prognosis.5,22
Chromosome 3p deletion is currently the most common finding in lung cancer. Allelic losses at one or more chromosome 3p21.3 locus are the most frequent chromosomal abnormalities detected in the bronchial epithelia of smokers and are detected even in normal bronchial mucosae of smokers.15
The use of FISH for diagnostic purposes has increased considerably in the last few years, primarily because FISH permits visualization and examination of genetic aberrations as rare events in a large number of cells that may have normal genetic composition. FISH is ideally suited for cytologic specimens such as sputum, which may be obtained spontaneously by coughing or induced by inhalation of nebulized saline. The major value of sputum biomarkers is to identify patients at high risk for cancer-related events, such as the development of premalignant lesions or early cancers, so that these patients may be subject to intense surveillance either by fluorescent bronchoscopic examination with removal of neoplastic lesions23,24
or by regular helical CT scanning of lungs to detect peripheral carcinomas. Additionally, this is an ideal population to benefit from the use of chemopreventive agents and smoking-cessation counseling.
In summary, we used an automated quantitative system to score FISH abnormalities in epithelial cells from noninduced sputum specimens, which resulted in a gallery of cells that could then be interactively classified in conjunction with morphologic findings. Software programs specific to the sputum application and the size of the probes and specific filter sets were used to maximize the accuracy of the testing. Correlating our sputum findings with disease state per individual, we discovered that epithelial cells in sputum from patients with NSCLC were cytologically and genetically abnormal relative to those from a high-risk control group with no CT evidence of lung cancer as well as healthy controls. On the basis of the results of our model for risk of lung cancer, which should be validated in a larger study, we conclude that the best predictive sputum assay for lung cancer will be a combination of morphologic characteristics and molecular abnormalities in both atypical cells and morphologically normal cells.