Simultaneously Targeting Epidermal Growth Factor Receptor Tyrosine Kinase and Cyclooxygenase-2, an Efficient Approach to Inhibition of Squamous Cell Carcinoma of the Head and Neck

Approximately 40,000 new cases of squamous cell carcinoma of the head and neck (SCCHN) occur in the United States each year, with a death rate of >12,000 patients annually (1). The survival rate for the patients with SCCHN remains poor despite advances in diagnosis and treatment (2). Head and neck cancers usually develop in area of the carcinogen-exposed epithelium and likely result from the accumulation of cellular and genetic alterations, leading to aberrant expression of many proteins involved in cell growth regulation (3, 4, 5). Blockade or modification of the function of one or several of these proteins may impede or delay the development of cancer. An effective preventive approach that could be implemented before the development of invasive cancer to reduce the incidence of SCCHN would be highly desirable. Among the potential chemopreventive agents, retinoids have been investigated extensively as treatment for premalignant lesions of the upper aerodigestive tract (6, 7). However, toxicity is significant, and the duration of responses has been limited. Thus, a new strategy is needed for chemoprevention of upper aerodigestive tract malignancies.

As a potential investigational approach for prevention and treatment of SCCHN, we studied a combination of two types of molecular targeted agents; epidermal grow factor receptor (EGFR)-selective tyrosine kinase inhibitors (TKIs) AG1478 and ZD1839 (Iressa or gefitinib; AstraZeneca Pharmaceuticals, Cheshire, United Kingdom), and the cyclooxygenase-2 inhibitor (Cox-2I) celecoxib (Celebrex; Pharmacia Corporation/Pfizer, Inc., G. D. Searle & Co., Chicago, IL). These two types of agents act on different biological targets: tyrosine phosphorylated EGFR (p-EGFR) and Cox-2, respectively. Both targets have been shown to contribute to SCCHN carcinogenesis. EGFR is a 170-kDa transmembrane protein with intrinsic tyrosine kinase activity that regulates cell growth in response to binding of its ligands such as EGF and transforming growth factor α. EGFR expression has been documented extensively in a wide variety of malignant tumors, including SCCHN. Overexpression of EGFR and its ligand transforming growth factor α was observed in 80 to 100% of SCCHN specimens (8, 9, 10, 11). Use of EGFR-selective TKIs has been one of the approaches to block EGFR activity in both preclinical and clinical studies (12).

Cox catalyzes the synthesis of prostaglandins (PGs) from arachidonic acid. Two members of the Cox family have been identified. Cox-1 is constitutively expressed in most tissues and responsible for the synthesis of PGs that mediate normal physiologic functions (13). In contrast, Cox-2 expression is not detected in most normal tissues. It is induced by inflammatory or mitogenic stimuli such as cytokines, growth factors, tumor promoters, and viral infection, resulting in increased synthesis of PGs in inflamed or neoplastic tissues (14). Cox-2 is overexpressed in many human cancers (for review, see ref. 15). In SCCHN, Cox-2 expression is found to be up-regulated at both mRNA and protein levels (16, 17). Treatment using Cox-2Is in cancer chemopreventive trials reduced the risk of developing familial adenomatous polyposis, some of which will inevitably progress to full-fledged colon cancer (18). Currently, there are >20 ongoing cancer chemopreventive trials using Cox-2Is, including celecoxib (19).

In general, combination therapies have proven to be more effective than single agents in the prevention and treatment of cancer. They not only enhance clinical response but also diminish the probability of developing drug resistance. There have been some promising results with combination chemoprevention strategies for cancer (20, 21). In the current study, we evaluated in vitro antitumor activities of a combined regimen of EGFR-selective TKIs (i.e., AG1478 or ZD1839) and a Cox-2I (i.e., celecoxib) on SCCHN cells. We also examined protein levels of primary targets of EGFR TKI and Cox-2I, p-EGFR and Cox-2, respectively, and downstream signaling molecules of EGFR and Cox-2–mediated pathways after the combined treatment.

Several SCCHN cell lines were used in this study. Cell line Tu177 was established from the larynx. Tu212 and 212LN were established from a primary hypopharyngeal tumor and a lymph node metastasis, respectively, from a single patient. Cell lines 686LN and 886LN were established from lymph node metastases of SCC of the tongue and larynx, respectively. These cell lines were obtained from Dr. Gary L. Clayman (University of Texas M. D. Anderson Cancer Center, Houston, TX) and Dr. Peter G. Sacks (New York University College of Dentistry, New York, NY; refs. 22, 23). They were grown in DMEM:Ham’s F-12 (1:1) with supplemented 10% fetal bovine serum.

Of the EGFR-selective TKIs, AG1478 was purchased from Calbiochem (San Diego, CA), and ZD1839 (Iressa or gefitinib) was provided by AstraZeneca Pharmaceuticals. Both AG1478 and ZD1839 are competitive inhibitors for ATP binding in the TK domain of EGFR. The Cox-2 inhibitor celecoxib was provided by Pharmacia Corporation/Pfizer, Inc., G. D. Searle & Co. All three drugs can be dissolved in DMSO in appropriate concentrations and stored at −20°C until use.

SCCHN cell lines were plated at a concentration of 5 × 10³ cells/well into 96-well plates in quadruplicate. Twenty-four hours later, the drugs were added in a range of concentrations as single agents [AG1478 (0–30 μmol/L); ZD1839 (0–10 μmol/L); and celecoxib (0–100 μmol/L)]. In another experiment, cells were treated with two drugs in fixed concentrations (AG1478 at 10 μmol/L or ZD1839 at 0.5 μmol/L plus celecoxib at 25 μmol/L). Cell growth inhibition was measured by determining cell density with sulforhodamine B assay (24) at 72 hours after addition of the drugs. Percentage of inhibition was determined by comparison of cell density in the drug-treated cells with that in the untreated cell controls in the same incubation period (percentage of inhibition = 1 − cell density of a treated group per cell density of the control group). All experiments were repeated three times.

To study effect of the EGFR-selective TKIs and the Cox-2I on clonogenicity of SCCHN cells, exponentially growing cells from two cell lines, Tu177 and 686LN, were seeded into 6-well plates at concentration of 2 × 10³/well in triplicates. After 24 hours, ZD1839 and celecoxib were added as either single or double agents in concentrations of 0.2 and 20 μmol/L, respectively. An equivalent amount of DMSO used to dissolve both agents was added to the control cells. The cells were incubated in the presence of the drugs for 10 to 14 days to form colonies. The cell colonies were stained in crystal violet (0.5%). They were counted under a microscope using the standard definition of a colony that should contain at least 50 cells. The experiment was repeated twice.

Soft agar clonogenic assay was also performed. SCCHN cells from two cell lines, Tu177 and 686LN, were plated into 6-well plates at concentration of 10 × 10³/well in 0.5% agarose with 1% agarose underlay in triplicates. ZD1839 and celecoxib were mixed with the top agarose as either single or double agents at final concentration of 0.1 and 10 μmol/L, respectively, before plating of the cells. The equivalent amount of DMSO was also mixed with the top agar and SCCHN cells as the control. Cells were incubated for 21 days, and colonies > 0.02 mm were counted. The total numbers of the colonies were recorded as average of three counts with a SD as indicated in Table 1. The experiment was repeated twice.

The effects of the EGFR-selective TKIs and Cox-2I were analyzed for both cell cycle and apoptosis using flow cytometry. Tu177 cells were incubated with AG1478 (5 μmol/L), ZD1839 (0.5 μmol/L), and celecoxib (25 μmol/L), alone, as single agents, or with combinations of either AG1478/celecoxib or ZD1839/celecoxib for 72 and 96 hours. The cells were stained with propidium iodide (200 μg/mL) and RNase A (10 μg/mL) and then analyzed by flow cytometry (Epics XL-MCL; Beckman Coulter, Miami, FL). Cell cycle distribution was determined using Modfit-LT software (Verity, Topsham, ME).

The same samples used for cell cycle distribution were analyzed for apoptosis by terminal deoxynucleotidyl transferase-mediated nick end labeling assay that was performed by using an APO-BRDU Apoptosis kit (The Phoenix Flow Systems, Inc., San Diego, CA) according to the manufacturer’s protocol. All experiments were performed at least three times.

To perform the capillary tube formation assay (25), 24-well plates were coated with Matrigel (250 μL/well; BD Bioscience, Bedford, MA). Human umbilical vein endothelial cells (HUVECs; Clonetics, Walkersville, MD) were pretreated with DMSO (control), ZD1839 (0.5 μmol/L), celecoxib (12.5 μmol/L), or combination of ZD1839 (0.5 μmol/L), and celecoxib (12.5 μmol/L) for 12 hours. Forty thousand HUVECs suspended in EGM-2 medium (Clonetics Co., San Diego, CA) were added to each Matrigel-coated well. Four wells were used for each treatment with DMSO, ZD1839, celecoxib, and the combination. After 18 hours of incubation at 37°C and 5% CO₂, the status of capillary tube formation by HUVECs was recorded using an Olympus inverted microscope (CKX40; Olympus, New York, NY) connected to a SPOT insight quantum efficiency digital camera, at ×40 magnification in five randomized fields.

Immunoblotting analyses were used to study expression levels of the relevant proteins potentially modulated by AG1478, ZD1839, celecoxib, or the combinations. These proteins include direct targets of EGFR TKIs and Cox-2I and those that are downstream of EGFR and Cox-2 pathways. Monoclonal antibodies against p-extracellular signal-regulated kinase (ERK)^1/2, as well as polyclonal antibodies against p-EGFR, total EGFR, and total ERK^1/2 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Polyclonal antibodies against phosphorylated and total Akt (p-Akt and Akt) were obtained from Cell Signaling Technology (Beverly, MA). Anti-Cox-2 monoclonal antibody was purchased from Research Diagnostics, Inc. (Flanders, NJ). Antibody to β-actin for an equal loading control was obtained from Sigma Chemicals (St. Louis, MO).

Whole cell lysates (50 μg) were used for immunoblotting analyses that were performed following a standard procedure. The antibody binding signals were detected using an enhanced chemiluminescence detection reagent (Amersham Pharmacia Biotech, Inc., Piscataway, NJ).

Immunoprecipitation of EGFR was performed using a standard procedure. Whole cell lysates (150 μg) were incubated with anti-EGFR monoclonal antibody (Upstate Biotechnology, Inc., Lake Placid, NY) and rProtein G-agarose (Invitrogen, Carlsbad, CA) at 4°C overnight. After washing, the precipitates were analyzed by blotting with anti-phospho-tyrosine antibody PY99 (Santa Cruz Biotechnology). The same membrane was then blotted with another anti-EGFR monoclonal antibody purchased from Cell Signaling.

To measure prostaglandin E₂ (PGE₂) concentration in cell culture media, Tu177 and 686LN cells were seeded at 0.4 × 10⁶ cells/well in 6-well plates and allowed to grow for 24 hours. The media were then replaced by 2 mL of fresh media containing ZD1839 (0.5 μmol/L) and/or celecoxib (25 μmol/L). The media were collected, and the total cell number in each group was counted at following time points; 24, 48, and 72 hours. PGE₂ levels in the media were measured by PGE₂ EIA kit following the manufacture’s protocol (Cayman Chemical, Ann Arbor, MI). The PGE₂ concentrations were calculated using a standard curve that was generated from PGE₂ standards provided by the manufacturer. All experiments were repeated twice.

Two-sided tests with a Bonferroni adjustment were used to compare the mean inhibition of the two-drug combination to the sum of the mean inhibitions of each of the drugs separately. Because in several cases, the percent inhibition with the two-drug combination was greater than the sum for the two drugs separately, we performed formal statistical tests to determine whether there was a synergistic effect. To do this, we subtracted the sum of the mean inhibitions for each of the drugs separately from the mean inhibition for the two-drug combination (divided by the square root of the sum of their variances) and tested to see if this difference was statistically significantly >0. We used two-sided tests with a Bonferroni adjustment because 10 comparisons were performed, i.e., we only declared statistical significance when P < 0.005.

To study the sensitivity of SCCHN to EGFR TKIs and Cox-2Is, we initially tested five SCCHN cell lines Tu177, Tu212, 212LN, 686LN, and 886LN using AG1478 (0 to 30 μmol/L), ZD1839 (0 to 10 μmol/L), and celecoxib (0 to 100 μmol/L). The results showed that all three compounds inhibited SCCHN cell growth in a dose-dependent manner (Fig. 1 A–C). However, two cell lines, 686LN and 886LN, established from lymph node metastases were less sensitive to both AG1478 and ZD1839 than other cell lines at the concentrations < 10 and 5 μmol/L of AG1478 and ZD1839, respectively, reflecting heterogeneity among SCCHN cell lines.

We then examined whether expression levels of p-EGFR, total EGFR and Cox-2 determine the sensitivity of SCCHN cells to EGFR TKIs and Cox-2I. Immunoblotting analyses showed that all SCCHN cell lines expressed similar levels of p-EGFR, an immediate target of EGFR-TKI, though cell lines 686LN and 886LN showed different sensitivities to AG1478 or ZD1839 compared with other cell lines in cell growth assay. Furthermore, 686LN expressed higher Cox-2 level than the other cell lines, but demonstrated the same sensitivity to celecoxib as others, suggesting that the growth inhibitory effects of these TKIs and Cox-2I may not completely dependent on expression levels of p-EGFR or Cox-2 in these SCCHN cell lines (Fig. 2).

To test whether a combination of EGFR TKIs and Cox-2Is would achieve higher growth inhibition than single agents at concentrations lower than the IC₅₀, the fixed concentrations of each drug were then tested in combined treatment of SCCHN cell lines. The concentrations of the three drugs were as follows: 10 μmol/L AG1478; 0.5 μmol/L ZD1839; and 25 μmol/L celecoxib (Fig. 3). In Tu177, Tu212, and 212LN cells, the combination of AG1478 with celecoxib inhibited cell growth > 60%. Similarly, the combination of ZD1839 with celecoxib showed up to 60 to 70% growth inhibition. Furthermore, both 686LN and 886LN cell lines were not appreciably inhibited by AG1478 or ZD1839 alone at the tested concentrations. However, the combined treatment of AG1478 with celecoxib inhibited 686LN and 886LN (40 to 50%). The combination of ZD839 with celecoxib also significantly inhibited the two cell lines (30 to 40%).

When statistically evaluating the percent inhibition of cells for the five cell lines, we found that the observed mean inhibition for the two drug combination was at least additive, i.e., the inhibition associated with the two-drug combination was at least as great as the sum of the percent inhibitions of cells for the two drugs considered separately. This was true for the combination of celecoxib with ZD1839 or AG1478. We also found that there was evidence of synergy for both drug combinations in the 686LN and 886LN cell lines and for the celecoxib and AG1478 in the Tu212 cell line. In all of these comparisons, we obtained P < 0.0001, which was well below the Bonferroni cutoff of P = 0.005.

Furthermore, clonogenic assays of two selected cell lines using both cell culture and soft agar showed that both celecoxib and ZD1839 inhibited colony formation from Tu177 and 686LN cells, but 686LN cell were less sensitive to the two agents than Tu177 cells. It is also illustrated that colony-forming capability was markedly suppressed by the combined treatment with ZD1839 and celecoxib as compared with each of the single agents (Table 1 and Fig. 4).

To study whether growth inhibition of SCCHN by AG1478, ZD1839, and celecoxib resulted from cell cycle delay or apoptosis, we examined the distribution of cell cycle and apoptotic cells by flow cytometry in the presence or absence of the three drugs either as single or combined agents in Tu177 cells. Table 2 shows that the combined treatment of AG1478 or ZD1839 with celecoxib for 72 to 96 hours slightly induced G₁ arrest in Tu177 cells as compared with the control.

Treatment for a total of 96 hours with the combination of AG1478 plus celecoxib induced 38% of apoptosis, although treatment with AG1478 or celecoxib alone resulted in only 10 to 20% apoptosis (Fig. 5,A). Treatment with the combination of ZD1839 plus celecoxib showed a greater induction of apoptosis (76%) than treatment with single agents (14% with ZD1839 and 26% with celecoxib; Fig. 5 B).

ZD1839 and celecoxib have been reported to suppress tumor vessel formation as single agents. To determine whether combination of ZD1839 and celecoxib more potently inhibited angiogenesis than single drug alone, we examined their effects on the ability of HUVECs to form capillary tube structures in vitro. Our results demonstrated that ZD1839 or celecoxib as single agents moderately inhibited HUVEC tube formation (Fig. 6, B and C, respectively), as compared with the DMSO-treated control (Fig. 6,A). However, a profound suppression and disruption of capillary tube formation was achieved when the HUVECs were treated with ZD1839 plus celecoxib (Fig. 6 D).

To elucidate the molecular mechanisms of the treatment effects of AG1478, ZD1839, and celecoxib on SCCHN, we examined whether AG1478, ZD1839, and celecoxib modulated expression levels of p-EGFR, p-ERK^1/2 [mitogen-activated protein kinase (MAPK)], and p-Akt. The levels of p-EGFR versus total EGFR in Tu177 and 686LN cells were studied by immunoblotting analyses with or without the treatment. As shown in Fig. 7, the inhibitory effects of AG1478 and ZD1839 on p-EGFR and p-ERK^1/2 were very similar; both drugs decreased p-EGFR in both Tu177 and 686LN cells. A combination of AG1478 or ZD1839 with celecoxib additionally reduced the level of p-EGFR without affecting total EGFR in Tu177 cells, but no additional reduction was observed in 686LN cells (Fig. 7,A). Celecoxib alone had minimal or no effects on both total EGFR and p-EGFR. Immunoprecipitation of EGFR was also performed to confirm the observation from immunoblotting with a phospho-specific antibody. As expected, p-EGFR was not detected after the combined treatment with ZD1839 and celecoxib (Fig. 7 D).

Similarly, p-ERK^1/2 was significantly reduced by AG1478 or ZD1839, whereas no reduction of p-ERK^1/2 was observed by treatment with celecoxib in Tu177 cells (Fig. 7 B). The combination of ZD1839 with celecoxib additionally reduced p-ERK^1/2, and the combination of AG1478 with celecoxib almost completely abolished p-ERK^1/2 in Tu177. In 686LN cells, none of the single drugs had much of an effect on the level of p-ERK^1/2, but significant reduction of p-ERK^1/2 was observed with the combined treatment.

Akt is an important signaling transducer of tumor cell growth and is regulated through comparable pathways, including EGFR signaling pathways (26). Thus, we examined both p-Akt/Akt in the presence or absence of ZD1839 and celecoxib in Tu177 and 686LN cell lines. It was found that ZD1839 significantly reduced p-Akt, and the combined treatment additionally reduced p-Akt in Tu177 cells (Fig. 7 C), whereas both single and the combined treatments showed no effect on p-Akt level in 686LN cells. The combined treatments also reduced total Akt level in both cell lines.

To examine whether AG1478, ZD1839 and celecoxib modulated Cox-2I-targeting protein Cox-2, we studied the expression of Cox-2 and its catalytic product PGE₂. Immunoblotting analysis showed that AG1478 significantly reduced Cox-2 expression in both Tu177 and 686LN cells, whereas ZD1839 slightly reduced Cox-2 only in Tu177 cells (Fig. 8, A and B). The combination of AG1478 with celecoxib did not additionally reduce Cox-2 levels in Tu177. However, the combination of ZD1839 with celecoxib significantly reduced Cox-2 in 686LN cells, although both ZD1839 and celecoxib as single agents at the treated concentrations showed no effect on Cox-2 protein expression in this cell line.

PGE₂ is a major product of Cox-2–catalyzed reaction. To examine the effect of celecoxib on an immediate target in SCCHN cells, we measured PGE₂ levels by enzyme immunoassay in cultured media collected from Tu177 and 686LN cells with and without treatment. Our results showed that celecoxib strongly inhibited PGE₂ production within 24 hours in Tu177 cells (Fig. 9), whereas this inhibitory effect was observed after 48 hours of incubation with the combination of ZD1839 and celecoxib in 686LN. In 72 hours, PGE₂ was no longer detectable in the presence of celecoxib. ZD1839 showed slight inhibitory effects on PGE₂ production only in Tu177 cells.

Although EGFR and Cox-2 play different biological roles in tumor cells, accumulating evidence suggests a direct interaction between EGFR signaling and Cox-2 activity. A recent study demonstrated that PGE₂ transactivated EGFR by induction of phosphorylation of EGFR and ERK (27). Additionally, overexpression of Cox-2 induces EGFR expression in colon cancer cell lines (28). On the other hand, transforming growth factor α, an EGFR ligand, can induce Cox-2 expression through activation of MAPK pathway (29). Therefore, targeting both EGFR and Cox-2 may be an effective approach to abrogate both pathways and their downstream targets. This notion has been supported by a study showing that using a combination of an EGFR TKI (EKB-569) and a Cox-2 inhibitor (Sulindac) significantly reduced intestinal polyps in APC^min/+ mice compared with the use of single agents alone (30). Tortora et al. (31) has recently reported that combination of an EGFR TKI (ZD1839), a Cox-2I (SC-236), and a protein kinase A antisense molecule achieved significant antitumor and antiangiogenic effects.

This in vitro study explored the combination of EGFR TKIs and Cox-2Is in SCCHN cells. The EGFR TKI ZD1839 and Cox-2I celecoxib were selected because both agents have been used in clinically for cancer treatment and chemoprevention. The recent clinical trial to use ZD1839 in SCCHN showed that the overall response rate was 10.6% (32). Phase III clinical trials of ZD1839 have been conducted for many other cancer types since 2000 (33). Other EGFR inhibitor trials in SCCHN have shown very similar results. For example, OSI-774 (erlotinib), another small-molecule TKI, produced 4.3% response rate in a similar patient population as in the ZD1839 trial (34). On the other hand, celecoxib has been used in chemoprevention trials not only for familial adenomatous polyposis but also other cancer types, including SCCHN (19). Therefore, a combination regimen using both ZD1839 and celecoxib, if it shows any higher anticancer effect than monotherapies, should have a potential for clinical application. To ensure biological function of the EGFR TKI, we also included another EGFR TKI, a frequently used laboratory agent AG1478, in our study.

We found that the combination of either AG1478 or ZD1839 with celecoxib additively/synergistically inhibited SCCHN cell growth. The growth inhibition resulted mainly from induction of apoptosis and a slight delay of cell cycle progression at G₁ phase. The remarkable apoptosis was observed in 96 hours after the treatments, implicating that blocking both EGFR- and Cox-2–mediated pathways may induce a secondary effect on cell growth regulation that requires additional investigation.

Both EGFR-signaling and Cox-2 promote tumor angiogenesis (for review, see refs. 35, 36). Previous studies have shown that angiogenesis in a variety of tumor types, including SCCHN, could be significantly suppressed by down-modulation of EGFR using EGFR-targeting strategies through inhibition of vascular endothelium growth factor and other angiogenesis factors (37, 38, 39). Furthermore, Cox-2 inhibition suppresses tumor angiogenesis by inhibition of blood vessel formation in corneal angiogenesis models (40). Dormond et al. (41) suggested this suppression might result from direct inhibition of proliferation and adhesion of endothelial cells through inhibition of α_Vβ₃ integrin-mediated and cdc42/Rac-dependent endothelial-cell activity. We found that the combined treatment with ZD1839 and celecoxib disrupted formation of endothelial capillary tubes more potently than either of the two drugs used as single agents. Proliferating, spreading, and restructuring of capillary blood vessels from endothelial cells in tumors is one of the key steps in tumor angiogenesis. As a unique tool, an in vitro capillary formation assay has been used to verify specific antiangiogenic activities of many agents with a good correlation to blood vessel formation in vivo (42, 43). Using this method, we observed that both ZD1839 and celecoxib as single agents inhibited capillary tube formation by HUVECs. It is not surprising to find that the two drugs are cooperatively antiangiogenic, which intensifies the antitumor effects on SCCHN.

As we illustrated, the main function of this combination was blocking EGFR/MAPK signaling pathways and reducing PGE₂ secretion. Moreover, EGFR TKIs reduce Cox-2 expression in some tumor cell lines. The down-regulation of Cox-2 by ZD1839 and AG1478 was consistent with articles by Tortora et al. (31) and Zakar et al. (44). They demonstrated that Cox-2 expression was regulated by the EGFR/MAPK signaling pathway. It was not unexpected to find that celecoxib alone did not reduce Cox-2 expression at the currently tested concentrations. Because the direct function of celecoxib is inhibiting enzymatic activity of Cox-2, it seems to inhibit PGE₂ production, as shown in our study. At concentrations ≥ 25 μmol/L, celecoxib induced Cox-2 in a SCCHN cell line, UM-SCC-1 (45), implicating a possible up-regulation of Cox-2 by celecoxib. Therefore, Cox-2 levels in certain SCCHN cells treated with the combination of TKI and Cox-2I, such as observed in 686LN cells, may depend on a balance between the contradictory effects from the two agents.

Obviously, reducing PGE₂ is not the only function of celecoxib. Of the five tested SCCHN cell lines, 686LN cells had the highest Cox-2 levels (Fig. 6,A), which may result in less sensitivity to celecoxib on PGE₂ production in 686LN cells than in Tu177 cells. However, all five cell lines showed a similar sensitivity to celecoxib in growth inhibition. Therefore, part of the inhibitory function of celecoxib on SCCHN cell growth is independent of Cox-2 activity and PGE₂ production. This observation has been highlighted by recent research. In 2002, the National Cancer Institute organized a special workshop to discuss the Cox-dependent and -independent mechanisms of nonsteroidal Cox inhibitors (46). It has been demonstrated that celecoxib induced cell cycle arrest and apoptosis at concentration higher than those observed at the clinically achievable level, <5 μmol/L. These celecoxib-mediated inductions were independent of Cox-2 (47, 48, 49). Similarly, in our study, celecoxib inhibited SCCHN cell growth at concentrations > 10 μmol/L (Fig. 1 A), which appeared to be independent of Cox-2 levels in these cells.

Activated EGFR rather that total EGFR contribute to EGFR-mediated signal transduction. However, we showed that blockage of EGFR activity alone did not lead to growth inhibition in some cancer cell lines such as 686LN and 886LN because we showed that 10 μmol/L AG1478 and 5 μmol/L ZD1839 almost completely abolished p-EGFR but had minimal inhibitory effects on cell growth in 686LN cell lines. This observation is supported by a recent publication showing that inhibition of proliferation and induction of apoptosis in breast cancer cells by ZD1839 was independent of EGFR expression levels (50). Cell proliferation signals should be provided also through other signaling pathways in these cell lines. For example, it was reported that 686LN cells expressed a higher level of interleukin-6 than Tu177 cells (51), whereas interleukin 6 contributed to EGFR-independent activation of signal transducers and activators of transcription 3 through gp130, an interleukin 6 receptor, which consequently conferred both proliferative and survival potential of SCCHN (52). Furthermore, PGE₂ also activates signal transducers and activators of transcription 3 through the same receptor (53). Although we are not certain why phosphorylations of ERK^1/2 and Akt were unaffected by ZD1839 at the tested concentration in 686LN, this phenomenon clearly supports heterogeneity of SCCHN cell lines and implies that a pathway other than EGFR/MAPK may play a role in supporting proliferation of these cells. This may be one explanation for the comparatively lower sensitivity of 686LN and 886LN to AG1478 and ZD1839. The selected concentration for ZD1839 in this in vitro study is based on the sensitivity of these SCCHN cell lines to the agent. Because of the various genetic backgrounds of individual cell lines, there is no simple translation from a concentration used in vitro to a clinical dosage. On the other hand, our current study was conducted in a medium containing 10% fetal bovine serum. It is possible that growth factors other than EGFR ligands, EGF or transforming growth factor α, in the serum may contribute to SCCHN growth. It is also possible that there are autocrine growth factors such as interleukins that support SCCHN cell growth. In the future, treating cancer by blocking several activated pathways simultaneously with combinational therapy may prove more effective than a single drug regimen.

Synergistic inhibition of SCCHN cell growth by the combination of AG1478 or ZD1839 with celecoxib provides a potential and novel strategy for cancer prevention and treatment. The combination of an EGFR TKI and a Cox-2 inhibitor definitely deserves additional in vivo and clinical studies.

We thank AstraZeneca Pharmaceuticals and Pharmacia Corporation/Pfizer, Inc., for providing EGFR TKI ZD1839 and Cox-2 inhibitor celecoxib, respectively.