Preclinical studies suggest that cyclooxygenase (COX)-2 may be involved in the molecular pathogenesis of some types of lung cancer. Most of the available studies point to its involvement in non-small cell lung cancer. Survival of patients with non-small cell lung cancer expressing high levels of COX-2 is markedly reduced. Treatment of humans with the selective COX-2 inhibitor celecoxib augments the antitumor effects of chemotherapy in patients with non-small cell lung cancer. COX-2 has been shown to regulate some aspects of tumor-associated angiogenesis. Most of the results we have published point to effects on the regulation of vascular endothelial growth factor. However, prostaglandins derived from COX-2 affect other signaling pathways as well, such as the epidermal growth factor and its receptor. Others have recently shown that non-small cell lung cancer exhibits a COX-2 downstream enzyme expression pattern that is altered in lung tumor cells and tumor-supplying vessels. Therefore, COX-2 and prostaglandins may have a major impact on lung tumor progression and tumor-associated inflammation. Clinical trials currently underway are exploring the potential of targeting COX-2 in lung cancer.
Lung cancer results in an enormous clinical burden for health care providers and very high patient mortality rates. Despite extensive research, the overall 5-year survival rate is only 8–14% and has improved only marginally during the past 25 years. Novel approaches for the management of lung cancer are urgently needed. Hence, enormous efforts have been made to identify risk factors associated with the development of lung cancer and to explore more effective strategies for treatment and prevention of this deadly disease. One such potential target is the cyclooxygenase (COX) pathway. Recent evidence suggests a potential role of the inducible COX-2 isoform in the development of some lung cancers. COX-2 expression is associated with a poor prognosis, and preinvasive lesions contain elevated levels of this enzyme. This review will focus on our current understanding of the role of COX-2 in lung carcinogenesis, and how selective COX-2 targeting may add benefit for the treatment or prevention of lung cancer.
COX EXPRESSION IN LUNG CANCER
COX is the key enzyme in the conversion of arachidonic acid to prostaglandins (PGs) and other bioactive lipids, which are involved in the regulation of normal growth responses and in aberrant cellular growth. Since the identification of two COX enzymes (COX-1 and COX-2), the mitogen-inducible COX-2 isoform has been the focus of considerable attention in epithelial tumorigenesis (1) . COX-1, the constitutive isoform, is present in most tissues and has been postulated to function as a “housekeeping” gene, responsible for physiological functions, particularly in the gastrointestinal tract. Conversely, COX-2 is dramatically up-regulated by a wide variety of stimuli such as interleukin (IL)-1, tumor necrosis factor α, platelet-derived growth factor, epidermal growth factor, lipopolysaccharide, and tumor promoters. COX-2 levels are elevated in gastric, hepatocellular, esophageal, pancreatic, head and neck, colorectal, breast, bladder, cervical, endometrial, skin, and lung cancers, when compared with nonmalignant tissue controls (2) . In lung cancer, COX-2 expression is found at most stages of tumor progression. Almost all non-small cell lung cancer preinvasive precursor lesions, as well as invasive lung carcinomas, express COX-2 at higher levels than normal lung tissue (3, 4, 5, 6, 7, 8) . Increased COX-2 expression is also associated with increased levels of downstream enzymes required for prostanoid synthesis, such as prostaglandin E2 synthase (PGE-S), prostaglandin D2 synthase (PGD-S), and thromboxane A2 synthase (TXA-S) (9) . Also, markedly higher COX-2 expression has been observed in lung cancer metastatic to lymph nodes (3) . The prognostic significance of increased COX-2 expression has been reported by multiple groups. A number of studies have shown a correlation between COX-2 expression and poor prognosis: COX-2 protein levels are elevated in stage I disease and confer a poor prognosis (10) ; and increased COX-2 mRNA levels portend a worse overall survival rate (11) and aggressive disease (12) in non-small cell lung cancer. These reports imply a potential role of COX-2 in the pathogenesis of lung cancer and that COX-2 is an independent poor prognostic indicator; some groups indicate the converse situation, in which elevated COX-2 expression is a positive prognostic indicator. The precise mechanisms responsible for elevated COX-2 expression in lung cancer are not completely understood; however, COX may directly impact on lung carcinogenesis because it can activate environmental carcinogens (13) . Conversely, benzo(a)pyrene itself can induce COX-2 expression and PGE2 production (14) . Many other stimuli present in the pulmonary microenvironment that are associated with risk of lung cancer development can also induce COX-2 expression.
COX-2 ACTIVITY AND ITS MULTIFACETED ROLE IN LUNG CANCER
The role of COX-2 in carcinogenesis is thought to involve modulation of apoptosis, stimulation of angiogenesis, modulation of the immune response, and promotion of tumor invasion.
Early tumor growth is dependent on the balance between increased proliferation and decreased cell death. In the hostile and often hypoxic tumor environment, genetic mutations may allow cells to develop resistance to apoptosis and increased metastatic potential. A number of studies have demonstrated the role of COX-2 in this process. Forced COX-2 expression in normal intestinal epithelial cells results in increased bcl-2 expression, increased avidity in binding to extracellular matrix components, reduced transforming growth factor β receptor expression (15) , and a prolongation in the cell cycle G1 phase (16) . Increased cell survival has also been observed in lung cancer cells forced to express COX-2 (17) . COX-2-selective inhibitors have been shown to induce apoptosis in lung carcinoma cells (18, 19, 20) . The ability of selective COX-2 inhibitors to induce apoptosis is of particular interest when considering their potential use in combination with other treatment modalities such as radiation therapy because preclinical studies have indicated improved responses when treatment with celecoxib and radiation therapy are combined.
Angiogenesis and Invasiveness.
Within the tumor microenvironment, the maintenance of a vascular supply is required for tumors to progress beyond a small size. Therefore, angiogenesis is a prerequisite for successful tumor growth (21) . Growth factors, such as vascular endothelial growth factor (VEGF) (22) , basic fibroblast growth factor (23) , and transforming growth factor β (24) , as well as cytokines, such as IL-8 (25) , have already been implicated in promoting a sustained vascular supply in lung cancer. Genetic mutations of various oncogenes and tumor suppressor genes and dysregulated immune responses are associated with regulation of these angiogenic mediators (26) . The angiogenic demand of tumors has the potential to serve as an important therapeutic target. PGs can contribute to tumor development in part via their role in the regulation of angiogenesis. Recent studies indicate a role of COX-2 in the regulation of angiogenesis. For example, COX-2-expressing colorectal cancer cells can alter the behavior of endothelial cells, whereas endothelial cell-derived COX-1 may also play a significant role in the angiogenic response (27) . However, experimental models of lung cancer-induced angiogenesis demonstrate that COX-2 can play a dominant role. In a model of syngeneic lung tumor growth, COX-2 expression correlated with tumor-associated neoangiogenesis, such that selective COX-2 inhibition significantly inhibited tumor growth (28) . Host-derived COX-2 products were shown to be instrumental for sustained tumor growth because lung cancer cells injected into syngeneic mice bearing a genetic background of COX-2 (but not COX-1) deficiency had a markedly reduced growth rate (29) .
COX-2 can also affect cell invasion, which is vital for the dissemination of metastatic cells across extracellular matrices and spread to distant organ sites. Lung cancer cells forced to express COX-2 (with elevated PGE2) also display concomitant increased expression of CD44, a cell receptor for selected matrix components (30) . In fact, antibody-mediated blockade of this COX-2-derived PGE2 was sufficient to decrease both CD44 and matrix metalloproteinase 2 expression as well as invasion in a EP4-dependent manner (31) . Exposure of non-small cell lung cancer cells to PGE2 up-regulated CD44, EP4, and matrix metalloproteinase 2 expression and enhanced cell invasion.
Cell-mediated immunity can serve an important role in the elimination of transformed cells. Unfortunately, the lung cancer microenvironment makes it difficult for the immune system to recognize and destroy transformed lung epithelial cells (32) . Of particular importance are T-cell-derived cytokines. The Th subsets are reciprocally inhibitory to each other via the cytokines they produce. Th1-derived cytokines can produce antitumor effects, and forced expression of the Th1-type cytokine gene in tumor cells elicits antitumor immunity. When cell-mediated immunity is influenced by the balance between Th1 and Th2 activity, IL-10 facilitates suppression of antitumor immunity. High levels of IL-10 may be immunostimulatory and thus activate T cells, preventing tumor growth, and a repertoire of IL-10-responsive immunocompetent cells may correlate with the effect on tumor growth. IL-10 can down-regulate COX-2 in host inflammatory cells; however, this capacity is lost in human non-small cell lung cancer. Lung cancer cells elaborate immunosuppressive mediators including PGE2 and transforming growth factor β, which may interfere directly with cell-mediated antitumor immune responses. IL-10 overproduction at the tumor site has been implicated in tumor-associated immunosuppression and enhanced angiogenesis and appears to be an indicator of poor prognosis. Selective inhibition of COX-2 abrogates the capacity of IL-1β-stimulated A549 cells to induce IL-10 in lymphocytes and macrophages. Furthermore, treatment of A549 cells with a selective COX-2 inhibitor reversed the tumor-derived PGE2-dependent inhibition of macrophage IL-12 production (33) . These results indicate that lung tumor-derived PGE2 plays a pivotal role in promoting lymphocyte and macrophage IL-10 production, while simultaneously inhibiting macrophage IL-12 production. Studies indicate that COX-2 metabolites can play a major role in tumor-induced inhibition of dendritic cell differentiation. Thus, inhibition of COX-2 expression or activity can prevent tumor-induced suppression of dendritic cell activities (34) .
THE FUTURE FOR NONSTEROIDAL ANTI-INFLAMMATORY DRUGS (NSAIDs) IN THE TREATMENT OR PREVENTION OF LUNG CANCER
The clinical association between reduced epithelial cancer risk and NSAIDs has been recognized for over 20 years (35) . Increased PGE2 levels have been noted in bronchoalveolar lavage fluid from patients with primary lung cancer with or without advanced metastatic disease (36) . Regular aspirin and NSAID use has been associated with an approximate 61–68% reduction in risk and incidence of lung cancer in some epidemiological studies (37, 38, 39) . However, these results are not conclusive because other epidemiological studies failed to show a protective effect of regular NSAID use (40) . NSAIDs have been shown to be effective in treating mice with carcinogen-induced lung cancer (41 , 42) . More recently, selective COX-2 inhibitors have also displayed antitumor activity in models of lung cancer (43) , although the mechanism of these antineoplastic effects may involve modulation of inflammatory cytokines (44) . The currently available preclinical studies provide encouragement for some evaluation of cyclooxygenase-2 specific inhibitors (Coxibs) in humans with lung cancer.
Dr. Thomas Lynch: We are seeing trial designs in which celecoxib is being added to standard chemotherapy. What surrogate markers should we be including in these studies?
Dr. Raymond N. DuBois: I think that the urine PGE-M (prostaglandin E2 metabolite) assay is going be very helpful in looking at COX-2 as a target. There are other targets of COX-2-selective inhibitors. Some of these are only affected at very high doses, but there could be others, specifically mitogen-activated protein, the p38 or the PTEN, which are affected as well. But the primary target is COX, and the urine PGE-M assay is probably worth evaluating.
Dr. Paul Bunn: How much of what is in the urine is coming from peripheral blood mononuclear and other cells, and how much is coming from the tumor?
Dr. DuBois: We have some pilot data on patients before and after resection, and it looks like the tumor is responsible for the major contribution.
Dr. Alan Sandler: We are doing a study with docetaxel in the second-line setting. We are getting serum for VEGF and urine metabolites. It seems to show that the inhibitor is hitting the target, although our study doesn’t have a tissue component.
Dr. Rafael Rosell: In our experience in patients with metastatic non-small cell lung cancer treated with paclitaxel/carboplatin, there were striking differences in survival in favor of those who were overexpressing COX-2 in the tumor. In second-line docetaxel treatment, we also analyzed 40 tumors, and those patients who overexpress COX-2 in their tumors had a much better response. In a first-line study of docetaxel/cisplatin, we looked at the expression of COX-2 and the peripheral lymphocyte and also observed that the patients who overexpress COX-2 have better time to progression and better survival. There are some preclinical models, primarily in liver cancer, in which there is a coexpression of inducible nitric oxide synthase at the same time as COX-2, and it is thought that this is a chemosensitizer.
Dr. Dubois: Looking at COX-2 in human neoplasia, several groups have shown that it is highly inducible by taxanes as well as some other chemotherapeutic agents. It is really important to know exactly what the patients were treated with and whether or not COX-2 expression is a primary event due to the biology of the tumor or due to induction by treatment with chemotherapy or radiation therapy. I suspect that some data may be confounded, depending on whether the expression is a primary effect of tumorigenesis or due to some secondary effect.
Dr. Ramaswamy Govindan: We completed a neoadjuvant COX-2 study with chemoradiation in esophageal cancer in 31 patients. The pathologic clinical response (CR) was no different than what is seen with chemotherapy and radiation alone. One fundamental issue is that in vivo we haven’t optimized the antitumor dose of COX-2 inhibitor. Second, COX-2 has been shown to do everything and anything you want in tumor cells, and there are some very early provocative studies showing that the inhibitor can provoke resistance. Third, preclinically there is a very interesting non-COX mechanism of COX-2 inhibitors at high doses. So, there are a lot of interesting issues that unfortunately have not been very well worked out.
Dr. Lynch: Along those same lines, we’ve done a study of CPT-11/platinum/radiation/COX-2 in esophageal cancer, and our pathologic CR rate is about 10% higher than it is with 5-fluorouracil/platinum, but it is no different than it is with 5-fluorouracil/platinum and paclitaxel—it is all within the same range. Again, this gets to the point that we need better surrogate markers of antitumor activity than just pathologic CR rates.
Dr. Alex Adjei: The early results with COX-2 inhibitor haven’t been good. Given what we know of the biology, is this target going to be important in advanced disease or is it maybe more relevant early on?
Dr. DuBois: We have focused most of our attention on prevention issues. In the clinical trials of colon polyp prevention, there is about a 50% reduction in polyp burden and polyp growth. It clearly has activity in these trials, and that is where most of our emphasis has been. Lung cancers are a very heterogeneous disease, and as we see with gefitinib, you can hit these targets and still not see a lot of effect. There may be a role for COX-2 inhibitors in a subset of these patients, and I think by doing these studies and looking at the effects on some of these biomarkers, we might be able to tease out a subset where it would be more effective.
Dr. Lynch: Dr. Heymach, you spent a lot of your time worrying about VEGF inhibitors. What are your thoughts on good surrogate end points in this setting to determine whether or not there is antitumor activity?
Dr. John Heymach: From the reductions in VEGF I’ve seen in different studies with COX-2 inhibitors, I doubt that a reduction in VEGF expression is responsible for the entire antiangiogenic effect. So we may be fooling ourselves if we just titrate a dose based on seeing, for example, 30% or 40% reduction in VEGF. This degree of inhibition of VEGF expression in a tumor may be a clinically imperceptible effect altogether. I suspect there are different or additional mechanisms responsible for the antiangiogenic effects of COX-2 inhibitors, and further preclinical and in vivo studies will be essential for understanding these mechanisms and taking advantage of them.
Presented at the First International Conference on Novel Agents in the Treatment of Lung Cancer, October 17–18, 2003, Cambridge, Massachusetts.
Requests for reprints: Raymond N. DuBois, PRB 694, 2300 Pierce Avenue, The Vanderbilt-Ingram Cancer Center, Nashville, TN 37232-6838. Phone: (615) 343-0527; Fax: (615) 936-1790; E-mail: raymond.