Although many current cancer therapies are based on their ability to kill most cells within a tumor, it has been recognized for more than 30 years that not all cells within a tumor are alike. Studies have shown that only a fraction of cells within a tumor can be propagated in patients, mouse transplants, or cell cultures.⁷,⁸,⁹,¹⁰ These rare clonogenic cancer cells were hypothesized to arise in a stochastic fashion: although the overall occurrence is rare, any cell within the tumor is equally likely to exhibit clonogenic activity. An alternative hypothesis to explain these findings is that a rare subpopulation exists within tumors, and that these rare tumor cells have unique biological characteristics that provide clonogenic activity. In support of the latter hypothesis, it has been recently demonstrated that several types of hematopoietic and solid tumors harbor a distinct subpopulation of cells called CSCs that can propagate the tumor phenotype in vivo.¹,²,³,⁴,⁵,⁶,¹¹ These cells are called CSCs because the same molecular markers used to isolate the normal tissue stem cells could be used to isolate clonogenic tumor cells in some tissues, the cells could be passaged serially through mice (demonstrating their ability to self-renew, a hallmark property of stem cells), and the isolated tumor cell population gave rise to a heterogeneous tumor (suggesting an ability to differentiate, a second hallmark property of stem cells).¹²

The CSC hypothesis has also emerged as an attractive explanation for tumor resistance to chemotherapy, recurrence, and metastasis. It has been hypothesized that CSCs have distinct biological mechanisms that render them more resistant to chemotherapy than other cancer cells, explaining the refractory nature of many tumors to treatment.¹²,¹³,¹⁴ Specifically, CSCs are hypothesized to be resistant to chemotherapy because they may be quiescent and may efficiently export drugs, like stem cells in normal tissues.¹⁵ CSCs may also be the cells that are required to generate metastases. For example, the pathways responsible for the dissemination and homing of normal stem cells during development may be aberrantly upregulated in CSCs. Therapeutic strategies that specifically eliminate the CSC population may therefore be more effective than standard means of therapy.¹⁶,¹⁷,¹⁸,¹⁹

Methods used in identification and characterization of stem cell populations from normal adult tissues have proven to be useful in uncovering CSC populations. The most widely used technique for isolating stem cells has been fluorescent-activated cell sorting (FACS) using a combination of cell-surface markers that select cells with markers of more primitive cells and exclude cells of differentiated cell lineages, followed by transplantation of sorted cancer cells into immunodeficient mice (Table 11.1). For example, CSCs were first identified in human acute myeloid leukemias as the cancer cells that had the same surface marker status as human hematopoietic stem cells (CD34+ CD38−).¹ CD133, a positive marker of hematopoietic stem cells and neural stem cells, has been used as a marker of CSCs from brain and colon cancer.³,⁴,⁵ CSCs have also been shown to exhibit similarities to normal stem cells
with regard to their ability to self-renew in serial-plating experiments in culture⁵,²⁰,²¹,²²,²³,²⁴ as well as their demonstrated activation of developmental pathways known to function in normal stem cells. For example, chronic myelogenous leukemia CSCs exhibit Wnt pathway activation as determined by elevated levels of nuclear β-catenin,²⁵ and mixed-lineage leukemia CSCs are granulocyte-macrophage progenitors that share a gene-expression program with hematopoietic stem cells.²⁴ Several pathways known to be crucial for development, such as the Wnt, Hedgehog, Notch, and Polycomb-group protein pathways, have been implicated in adult stem cell selfrenewal, and dysregulation of these pathways contributes to many types of cancer.²⁶,²⁷

Lung cancer remains the major cause of death from cancer worldwide,²⁸ and relatively little is known about the molecular heterogeneity of the cells within lung cancers. Lung cancer can be divided into two histopathological groups: 80% are non-small cell lung cancers (NSCLCs), and 20% exhibit neuroendocrine features. NSCLCs can be further subdivided into adenocarcinomas (50% to 60%), squamous cell carcinomas (20% to 25%), and large cell carcinomas. The average 5-year survival rate for NSCLC is only 16% because most lung cancers are refractory to chemotherapeutics or quickly become resistant to therapeutic response.²⁹ Also contributing to lung cancer morbidity, most NSCLC patients already have advanced diseases at the time of diagnosis; 21% of diagnosed cases have distant metastases in the brain, bone, liver, or adrenal glands.²⁹,³⁰,³¹ Surgery or therapies that treat primary lung tumors rarely prevent metastases. For example, 72% of patients who had NSCLC tumors surgically removed eventually develop distant metastases, most commonly in the bone or brain.³² Lung CSCs may be responsible for these observations, and an improved understanding of the cellular mechanisms operating in lung cancers should elucidate new therapeutic approaches.

Therapeutic Resistance of Lung Adenocarcinomas Harboring Epidermal Growth Factor Receptor Mutations The recent treatment success of gefitinib (Iressa) and erlotinib (Tarceva), two small molecule inhibitors of EGFR, in a fraction of patients with NSCLC has solidified the premise that EGFR is an important molecule in the pathogenesis of lung cancer (see Chapter 49). Several groups have independently identified frequent somatic mutations in the kinase domain of the EGFR gene in lung adenocarcinoma. These occurred in up to 10% of lung adenocarcinoma specimens sequenced in the United States and up to 30% of those sequenced in Asia. The mutations are associated with sensitivity to both gefitinib and erlotinib, explaining in part the rare and dramatic clinical responses to treatment with these agents.³³,³⁴,³⁵ Subsequent studies by multiple groups have now identified EGFR kinase domain mutations from more than 600 lung cancer patients. These mutations cluster in four groups or regions: exon 19 deletions, exon 20 insertions, and point mutations at G719S and L858R. Exon 19 deletions and exon 21 L858R point mutations account for more than 85% of all EGFR kinase domain gefitinib-and erlotinib-sensitizing mutations.

In the clinical setting, although these EGFR-mutant NSCLCs initially respond rapidly and dramatically to gefitinib and erlotinib, tumors eventually become refractory to treatment, and nearly all patients who initially respond to these drugs subsequently relapse.³⁶,³⁷,³⁸ Three studies identified EGFR T790M mutations in approximately 50% of the tumors from patients who relapsed.³⁹,⁴⁰,⁴¹ These mutants, when combined with sensitizing EGFR kinase domain mutation, permit the continued growth of tumor cells in the presence of erlotinib and gefitinib. Structural studies suggest that the T790M mutation introduces a bulky methionine residue in the EGFR kinase domain, which sterically hinders tyrosine kinase inhibitor (TKI) binding.³⁶,³⁷,³⁸ Whereas gefitinib and erlotinib are reversible inhibitors that mimic adenosine triphosphate (ATP), irreversible inhibitors such as HKI-272 or BIBW2992 mimic ATP and covalently bind to EGFR, enabling them to inhibit EGFR
kinase activity even in the presence of T790M.⁴²,⁴³,⁴⁴ Although irreversible inhibitors are currently being tested in clinical trials, animal models suggest that tumors will eventually become refractory to these treatments as well. 44 Thus, it is crucial to determine the cellular basis of lung adenocarcinoma resistance to treatment and develop new therapeutic strategies that will not be susceptible to resistance mechanisms.

Evidence for Non-Small Cell Lung Cancer Stem Cells Several pieces of evidence suggest that NSCLC tumors, including adenocarcinomas, contain a rare population of cells with stem cell characteristics. The initial sensitivity of human adenocarcinomas with activating EGFR mutations to EGFR TKIs and the acquired resistance to these treatments suggest that drug-resistant CSCs may be present in these tumors.³⁶,³⁷,³⁸ Side population cells, isolated by their ability to efflux Hoechst dye, were identified in six human NSCLC cell lines. These cells exhibit several stem cell characteristics, including increased drug-exporting transporter expression, enriched tumorinitiating capacity, and resistance to multiple chemotherapies.⁴⁵ Additionally, CD133+ cells from human lung tumors were recently shown to form self-renewing spheres in culture that could propagate tumors when transplanted subcutaneously into immunodeficient mice.⁴⁶ Importantly, although these studies support the likelihood of CSCs in lung cancers, the isolation and characterization of a population of human lung cancer cells that can serially passage the lung tumor phenotype in the lung microenvironment has not been reported, and the operation of pathways that regulate stem cells in advanced lung tumors has not been understood.