Carcinoma of the lung is the most common cause of cancer death worldwide and in the United States (1–3). In 2002, it was estimated that 169,400 Americans would be diagnosed with lung cancer and 154,900 will die (4). Using current standards of diagnosis and treatment, fewer than 15% of patients with lung cancer will survive five years after diagnosis (5). The prognosis for patients with lung cancer is strongly correlated with the stage of disease at the time of diagnosis. Over two-thirds of the patients have regional lymph node involvement or distant disease at the time of presentation, and the prognosis for these patients is poor, with five-year survival rates ranging from 16% (regional stage) to 2% (distant disease) (5). However, survival is markedly improved for patients with early stage disease, with five-year survival rates of 49% for localized disease (5) (staging according to the Surveillance, Epidemiology, and End Results (SEER) Program staging scheme (6, 7)). Thus, development of effective, noninvasive, diagnostic methods for early detection of lung cancer is clearly important to reduce lung cancer mortality.
Historically, chest radiography has been the principal diagnostic test for early detection of lung cancer. A recent review by Etzioni et al. (8) shows that relative survival (five- and 10-year) for lung cancer is highest among patients diagnosed with localized disease compared to regional or distant disease. In the 1970s, controlled trials sponsored by the U.S. National Cancer Institute (NCI) were undertaken to determine whether screening with sputum cytology and annual chest radiography would reduce lung cancer mortality. Although these trials showed that lung cancer was detected at an earlier stage and resectability and survival rates were higher in the study groups than in the controls, overall lung cancer mortality was not significantly improved (9–13).
These findings have stimulated a search for improved diagnostics for early detection of lung cancer. Interest in sputum as an inexpensive, noninvasive source of respiratory epithelial cells has been revived by several studies reporting detection of molecular abnormalities in sputum samples of patients with, or at risk for, lung cancer (14–27). However, a major limitation to the use of sputum samples for detailed molecular studies is that only a small proportion of sputum cell populations consist of respiratory epithelial cells. The contaminating inflammatory cell populations can obscure detection of important molecular abnormalities in the neoplastic respiratory epithelial cells.
Detailed analysis of induced sputum has been adopted as a relatively noninvasive method for the evaluation of airway inflammation in diseases such as asthma and chronic obstructive pulmonary disease (COPD). Two basic techniques for processing sputum have been described. The first consists of selecting all viscid portions from the expectorated sample to reduce squamous cell contamination, whereas the second processes the whole expectorate, containing sputum plus saliva (28). Studies comparing both techniques directly on samples from the same patient with either asthma or COPD found no significant differences in total or differential cell counts in the nonsquamous cell population (29, 30).
The cellular content of induced sputum samples have recently been established in healthy adults. Spanevello et al. (31) published reference values based on 96 healthy volunteers and reported that epithelial cells constituted only 1.5 ± 1.8% of cells in induced sputum samples. Belda et al. (32) also evaluated induced sputum of nonsmoking healthy adults and reported a mean of 1.6% respiratory epithelial cells. Differential cell counts in induced sputum have shown that the proportion of inflammatory cells in induced sputums is comparable between smokers and nonsmokers. For example, Lensmar et al. (33) found no significant differences in total cell counts and the percentages of macrophages, lymphocytes, neutrophils, and eosinophils in induced sputums of smokers and nonsmokers. Similar results were reported by Domagala-Kulawik et al. (34). These combined results indicate that respiratory epithelial cells, the targets of field carcinogenesis in the lung, constitute a very small percentage of cells in induced sputum samples.
We, and others, have previously shown that flow cytometric cell sorting can be used to identify and enrich epithelial cells from heterogeneous cell populations in fresh/frozen and formalin-fixed, paraffin-embedded human tissues, including breast cancer and Barrett's esophagus (35–43). Our goal was to develop a flow cytometric technique to enrich sputum respiratory epithelial cells from the heterogeneous mixture of sputum cellular elements. In the present study, we examined induced sputum specimens from current and former smokers who are at high risk for lung cancer and are enrolled in a low-dose spiral CT scan screening trial for lung cancer.
- Top of page
- MATERIALS AND METHODS
- LITERATURE CITED
Cellular heterogeneity in sputum samples, which typically include only 1% respiratory epithelial cells, impairs biomarker analysis for early detection of lung cancer because the large excess of inflammatory cells limits detection and quantitation of neoplastic changes occurring in the epithelial cell populations. We describe a multiparameter flow cytometric approach to enrich the minority population of 1% respiratory epithelial cells in sputum samples to 42%, on average, using an AE1AE3 anti-cytokeratin/DNA content staining protocol, which is a 38-fold enrichment of respiratory epithelial cells relative to unsorted sputum samples. Although the absolute number of respiratory epithelial cells varies widely in different sputum samples, the mean predicted yield in the samples evaluated in this study was more than 32,000, sufficient for many types of quantitative biomarker analyses.
Previous studies have used flow cytometric cell sorting to enrich for defined cellular elements in sputum. For example, Alexis et al. (45) used flow cytometry to investigate whether sputum phagocytes have phenotypes indicative of increased functional activation and inflammation compared to phagocytes from the alveolar airways and peripheral blood in healthy subjects. In their study, gating of the different cell populations (macrophages, monocytes, polymorphonuclear neutrophils, eosinophils, and lymphocytes) in sputum was based on light-scatter properties, including forward-scatter for cell size, side-scatter for cell density/granularity, and antibody detection of positive/negative expression of relevant antigens such as CD45, a pan-leukocyte marker. This report is another example of this approach, in which respiratory epithelial cells can be selected based on similar criteria.
Accurate molecular analysis of human cancers and their precursor lesions requires identification and enrichment of subpopulations of cells from a composite background of cellular heterogeneity (35–43, 46–61). Multiparameter flow cytometry is a useful tool to purify cells of interest from human tissue and other human samples that are typically heterogenous. For our purposes, we adapted a multiparameter AE1AE3/DNA-content flow cytometric cell-sorting protocol that had been previously published for breast cancer (39), to enrich respiratory epithelial cells from sputum samples. Our results support the use of induced sputum specimens as an easily obtained and noninvasive source of biological material for flow cytometric detection and enrichment of rare respiratory epithelial subpopulations from a heterogeneous cellular background.
Recent biomarker studies of unprocessed sputum samples have detected abnormalities associated with lung cancer, thus demonstrating the feasibility of detecting molecular lesions in sputum (14–27). However, the ability to extend these results to early detection programs has not been established and may be limited in many cases. For example, some results required prior knowledge of specific p53 mutations in the tumor, a condition that cannot be readily adapted to mass screening (21). In other cases, positive signals have been detected in sputum, but it is not clear whether the abnormalities are in the epithelial cell targets of tobacco-induced field carcinogenesis or in other cells in the heterogeneous sputum mixtures (19, 20, 62, 63). Many biomarker tests, such as determination of loss of heterozygosity (LOH) or DNA sequencing, are inaccurate if the positive cells are a small minority of the total cells in a sample.
Although much work remains to be done to extend our results and those of others to early detection of lung cancer, our study shows that highly enriched populations of respiratory epithelial cells can be obtained from induced sputum samples by flow cytometric cell sorting. This helps overcome a major limitation to the study of neoplasia in sputum samples. Given the yield of enriched respiratory epithelial cells in these sputum samples, it should be possible to quantify biomarker abnormalities by fluorescence in situ hybridization (FISH), monoclonal antibodies, and other techniques. Purity may be further enhanced by adding a pan-leukocyte antibody (CD45) conjugated to FITC to further eliminate contaminating macrophages, or by depleting the samples with immunomagnetic beads to CD45 before flow sorting, which might make possible additional molecular testing, including DNA sequencing (for mutation analysis) and genotyping (for loss of heterozygosity analysis). For example, we have routinely used DNA content and multiparameter flow cytometric cell sorting to purify proliferating epithelial cells from Barrett's esophagus for molecular assays, including mutation detection, LOH analysis, and methylation detection (35–38, 40–43, 46, 47, 53, 64).
In summary, we report a technique to enrich sputum respiratory epithelial cells to 42% purity, with predicted average yields of greater than 32,000 cells. These proof-of-principle results overcome a major limitation to quantitative molecular analysis of induced sputum specimens as an easily obtained and noninvasive source of biological material in persons at increased risk for developing lung cancer. Development of quantitative biomarker panels that can be validated as predictors of progression to lung cancer in prospective studies may improve early detection as well as offering the possibility of validated surrogate endpoints in prevention trials (8, 65).