Novel agents in the treatment of lung cancer: conference summary statement.

Abstract

A small, international, closed-door conference on Novel Agents in the Treatment of Lung Cancer, held in Cambridge, Massachusetts, October 17–18, 2003, was convened to present and discuss findings from recent and ongoing trials of investigational drugs for the treatment of lung cancer. Invited participants from the Netherlands, Italy, Spain, and the United States presented new data regarding the role of molecularly targeted agents in the treatment of advanced lung cancer and discussed their significance for clinical care. The conference format combined brief presentations with extended periods of open discussion. The conclusions reached over the course of the 2-day conference are summarized briefly below and presented at greater length in the individual papers and accompanying discussions that comprise the conference proceedings.Over the past decade we have witnessed a dramatic increase in our understanding of lung cancer biology. This greater knowledge has led to the identification of new therapeutic targets as well as the development of innovative preclinical model systems. Over the past year we have seen the fruit of this effort in the discovery of pemetrexed, gefitinib, and erlotinib as active agents in lung cancer. For the most part, these agents have modest activity when used as single agents in patients with previously treated lung cancer. Preclinical modeling of these agents and drugs such as the antisense molecule to protein kinase C (Affinitac) predicted that they would have synergy with chemotherapy. A major disappointment has been the failure of these novel agents to improve survival when added to standard chemotherapy, calling into question the validity of the preclinical model systems.Much discussion focused on understanding the reasons why so many compounds that appeared promising in preclinical and early-phase clinical studies did not fulfill that promise when taken to large-scale randomized trials. Participants identified several key issues that will need to be addressed in the investigation of other novel compounds still in early development, primarily concerning the selection, interpretation, and reporting of preclinical studies and the design and interpretation of Phase I/II studies. Fundamentally, participants felt that industry has moved too precipitously to bring novel compounds into Phase III clinical trials in a competitive push to be the first with a new class of agents. It is understandable that we are impatient with advanced lung cancer. In its advanced stages, this is a disease that is uniformly fatal with a median survival of 9 months. This accelerated advance to Phase III clinical studies without better understanding of who may benefit from a given novel agent has had the support of regulatory authorities and academic researchers anxious to find effective regimens. A major concern raised by participants in the conference was that the recent pattern of negative Phase III trials threatens the future of the field, if companies and other investors in research decide to reallocate resources to disease conditions with a higher likelihood of successful outcomes.The success rate in proceeding from the preclinical model to the patient in the clinic is markedly lower for oncology compared with other indications. The predictive value of preclinical studies needs to be improved. One issue identified by conference participants is the importance of defining the target for the novel molecular agents and demonstrating (a) that the target is relevant and (b) that the agent in fact has a clinically meaningful impact on the target. Cell lines and xenografts will continue to provide necessary but not sufficient data. At the conference the potential role of genetically engineered mouse models was explored, and great hope was expressed that in the future these might be able to better predict efficacy of novel treatments.In discussing the generally poor predictive value of the preclinical models, conference participants concluded that there is an obvious need not only for better models but also for better utilization of the existing models. Although not optimal, the available cell line and xenograft models provide useful information if they are rationally used and rigorously interpreted. In the past, positive results in selected cell lines have been presented without reporting whether other cell lines were also run and found negative. Many felt strongly that multiple cell lines should be looked at and the full data set should be used to influence the go/no go decision in clinical development. Positive results should be confirmed in different cell lines or different models before proceeding to clinical trial, and all data should be reported with SDs given.In addition to validating target, preclinical models should be used to explore schedule and dose questions. Early use and more thorough interpretation of preclinical pharmacokinetic and pharmacodynamic data might help to design Phase II trials that are more likely to be representative. Conference participants called for more complete and consistent reporting of preclinical data, with an end to the common practice of selective presentation of preclinical data to justify planned clinical trials. The standard for data reporting should be as stringent for preclinical as for clinical studies, with all responses and nonresponses reported.As with the preclinical studies, early-stage clinical trials should include more thorough analysis of clinical data, such as distinguishing the subsets of patients who did or did not respond to the investigational agent. While recognizing the inherent difficulty in obtaining tissue from lung cancer patients, the group felt that, wherever possible, study designs should include at least baseline and posttreatment biopsies to assess or confirm the agent’s efficacy against the molecular target and the tumor itself. Even if the potential targets are not well “flushed out” at the time of the study, tumor tissue should be archived so that it can be looked at in the future because newer technology might clarify potential mechanisms of response or resistance.A number of novel agents that appeared promising in Phase II studies, even those with single-agent activity, have proved disappointing when taken to larger randomized trials. To avoid expending resources in unproductive large trials, statistical models should be used to determine what type of early-phase data (response rate, time to progression, or other surrogate end points), with what magnitude of apparent effect, would be predictive of positive outcomes in larger randomized Phase III trials. Because the typical Phase II trial only enrolls up to 40–50 patients, the duration of follow-up and the number of events must be adequate to allow interpretation. Both Phase I and Phase II trials need to be better powered. Phase I trials should include a pharmacodynamic marker whenever possible to verify that the drug (and the dosing regimen) is in fact hitting the putative target.Clinical studies evaluating novel agents have been designed primarily to gain regulatory approval for an agent. Two study designs have predominated: (a) using the investigational agent as monotherapy and comparing this with standard chemotherapy; or (b) combining the novel agent with a standard chemotherapy regimen to produce a doublet or triplet. Whereas these are certainly reasonable approaches, new designs for Phase II trials should be considered, including adjuvant and preoperative studies, where tumor samples taken from a small number of patients before and after a brief neoadjuvant course of treatment may aid in determining whether the agent has a detectable impact on the tumor. In this setting, multicenter trials that give priority to the collection and analysis of surgical samples may be of service in developing a better understanding of what is happening at a molecular level and in correlating therapy response with population subgroups.There was much discussion about the merits of the randomized Phase II trial. Several conference participants argued in favor of randomized Phase II studies with an active control arm, whereas others cautioned against the risks of overinterpreting the data because the patient numbers are generally too low to permit confident interpretation of the results. Multiple arm early-phase trials were viewed as expending resources that should be reserved for Phase III trials of agents that have established their potential. Participants agreed that there should be more emphasis on multicenter Phase II studies rather than single-institution studies, in which the patient population may not be representative. Conference participants also agreed that an investigational agent should demonstrate antitumor activity as monotherapy in Phase II before it is further evaluated as combination therapy in large Phase III trials. However, exceptions may be possible when novel agents are working primarily as sensitizers of chemotherapy or radiotherapy, and their major role is expected only in combination.Ultimately, better surrogate markers of antitumor activity are needed than clinical response rate or time to progression (an end point that is dependent on the frequency of follow-up). Early-phase trials should include efforts to validate both biomarkers of optimal drug dosing and surrogate markers of drug efficacy. Too many agents have gone on to Phase II/III trials without resolving these issues, and agents that failed in large trials in the general population might have shown efficacy if the dosing and the proper surrogate markers had been determined in the preliminary trials. In this regard, a biomarker that measures pharmacokinetics should not be substituted for a surrogate marker. The latter must be validated as correlating with both target-related changes within the tumor and clinical outcomes.In vivo imaging in humans may prove to be a powerful tool in elucidating the effects of the targeted therapies, and incorporation of imaging studies such as positron emission tomography or dynamic contrast magnetic resonance imaging in early-phase trials may also prove of utility in determining which agents should move forward to randomized trials. One recently developed study design treats all patients for a predetermined time, at which point all patients are imaged to identify response or lack of response, permitting an early determination of whether the drug has an effect on time to progression.Obtaining biopsy tissue is a difficult goal in lung cancer trials. Patients with metastatic lung cancer often have a diagnosis made by just a fine-needle aspiration. Often there is no archival tissue. In breast cancer, by contrast, the majority of patients have had the removal of a primary tumor, and paraffin samples are available. Only with great commitment will it be possible to obtain this tissue. Surrogate tissue is also a possibility. In early-phase trials it is feasible to biopsy surrogate tissue to look for markers of response. This will be much harder to coordinate in larger Phase III studies involving multiple sites of care.Participants were asked to determine the relevant benchmarks that should serve as standards by which to compare new treatments for stage IV non-small cell lung cancer (NSCLC). The group agreed that two distinct standards had evolved for interpreting a Phase III study as positive. For industry-sponsored multi-institution studies, the benchmark was a 10-month median survival with a 35% response rate and a 4.5-month median time to progression. In contrast, for cooperative group studies, the benchmark has been an 8-month median survival, 25% response rate, and a 4-month median time to progression. These differences may largely be explained by different patient characteristics and, in particular, by the entry of many stage III patients in drug company-sponsored studies as compared with cooperative group studies, in which the patient population is often limited to patients with mainly stage IV disease.In designing trials of the novel targeted agents, there is the issue of whether enrollment should be restricted to patients whose tumor expresses the target. In principle, the answer to this question should be yes. However, often we do not appreciate the relevant target. In a disease like chronic myelogenous leukemia, the presence of the bcr-abl is pathognomonic for the disease. In most solid tumors the genetic changes that produce malignancy are complex and incompletely understood. An agent that is highly effective for a small subset of patients may go unappreciated if that subset is not included in the clinical study. This strategy becomes problematic with agents such as the epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors, for which the mechanism of action is, at best, poorly understood. In this example, clinical efficacy has not correlated with expression of the presumed target. Ultimately, there is no simple answer to this question. When it is clear what the target of interest is (which is the exception in lung cancer), it makes sense to enrich the population for such patients. When it is not known, it makes more sense to enroll broadly, collect tumor tissue, and ask correlative questions in a post hoc manner.Novel targeted agents are not the only drugs for which we need to improve our ability to predict response. Platinum-based combination therapy only helps a minority of patients with advanced NSCLC. Understanding chemotherapy sensitivity and resistance is a major goal as well. The mRNA expression levels of several genes (ERCC1, RRM1 and XPD) in the nucleotide excision repair pathway are implicated in cisplatin resistance, and RRM1appears to be a marker for gemcitabine/cisplatin resistance. Trials to further evaluate the prognostic significance of RRM1mRNA should be done to test the hypothesis that patients with low levels of RRM1 will benefit from gemcitabine/platinum chemotherapy.Genomic polymorphisms are another potential predictive marker of response and toxicity to chemotherapy. Pharmacodynamics is a potentially powerful tool for determining the effects of therapy and the tumor characteristics of those likely to respond. Predictive markers of therapeutic response need to be carefully distinguished from prognostic factors. For example, in untreated patients, expression of some poor prognosis markers, such as cyclooxygenase (COX)-2, may increase in response to therapy and serve as a valid target.For a marker to be predictive of response, a given drug has to have activity. In lung cancer, erlotinib and gefitinib showed activity in recent studies, but predictive markers of response have not yet been identified, although several leads have been reported. In terms of predictive markers, mitogen-activated protein kinase has potential and should be more carefully studied. It needs to be demonstrated that phospho-specific antibodies accurately detect phosphorylated receptor if there is a delay in processing the tissue specimen. More uniform methods of collecting and analyzing tissue for predictive markers should be developed before the validity of this and other markers can be confirmed. Another potential mechanism, PTEN loss and resultant phosphatidylinositol 3′-kinase/Akt pathway constitutive activation, may be important in mediating resistance to gefitinib. Given the redundancy of the pathways, it is likely that several markers, rather than one, may need to be elucidated.The participants were in agreement with regard to the current status of chemotherapy for metastatic NSCLC. All agreed that in metastatic disease, platinum-based combination chemotherapy provides a modest survival benefit compared with best supportive care. Two-drug regimens offer improved response and survival rates compared with one drug, whereas three-drug regimens do not improve survival in advanced NSCLC. Studies attempting to determine optimal therapy duration have found that prolonging treatment beyond three to four cycles increases toxicity with no increase in response or survival.During the 1990s, at least five “new” chemotherapeutic agents that had activity in lung cancer (paclitaxel, docetaxel, vinorelbine, gemcitabine, and irinotecan) have shown single-agent activity in advanced NSCLC. These agents are usually used in combination with a platinum compound, either cisplatin or carboplatin. Nonplatinum doublets have not been found to offer an advantage over platinum doublets, and decreased survival has been observed in two trials in the nonplatinum arms. In combination with a platinum agent, the newer agents appear to offer a slight improvement in outcome compared with older regimens. The randomized trials that have evaluated these modern platinum-based doublets have not shown that any specific combination is superior when survival is the primary measure of outcome. Some studies have suggested that cisplatin doublets are superior to carboplatin doublets, but such a difference has not been consistently observed. More research is needed to determine the optimal regimen for performance status 2 patients and for the elderly, a heterogeneous and understudied group of patients.Docetaxel remains the standard of care for second-line therapy for NSCLC. This agent has been shown to prolong survival compared with best supportive care. Recently, pemetrexed has been shown to have activity similar to that of docetaxel and is associated with less fever and neutropenia. Weekly administration of docetaxel is also a means to ameliorate neutropenia. The participants felt that second-line treatment for advanced NSCLC is an area that is rich for potential therapeutic advance. Both docetaxel and pemetrexed are rational platforms with which to combine novel agents.EGFR overexpression may portend a worse prognosis in lung cancer, but there are many conflicting reports, suggesting any effect is small. In lung cancer cells, EGFR activation has been shown to inhibit apoptosis and promote angiogenesis, invasiveness, metastasis, and tumor cell proliferation. Gene amplification and mutation are uncommon. This is the main rationale for pursuing EGFR as an important target in lung cancer. Although EGFR was found to be expressed in over half of the NSCLC specimens analyzed, the level of EGFR expression did not correlate with clinical response to EGFR tyrosine kinase inhibitors in the initial studies of gefitinib.In the IDEAL trials, the EGFR tyrosine kinase inhibitor gefitinib (Iressa) showed antitumor activity in between 10% and 20% of patients with previously treated advanced NSCLC. Another 20% of patients had stable disease as their best response. In both patients with stable disease and patients with response, tumor-related symptoms improved. Accordingly, gefitinib has been approved for the treatment of advanced NSCLC in the United States by the Food and Drug Administration and in Japan. The INTACT trials, however, failed to find a benefit in the first-line setting for gefitinib added to standard chemotherapy regimens (cisplatin/gemcitabine or carboplatin/paclitaxel). These major negative findings raised questions regarding the preclinical models that had shown synergistic or additive effects of chemotherapy and gefitinib.Although response to gefitinib did not correlate with EGFR expression, in the IDEAL trials a rapid improvement in disease-related symptoms was observed in patients who went on to respond, an observation that, if confirmed, may be useful in clinical decision-making. Women, patients with adenocarcinoma histology, and never-smokers were more likely to respond to gefitinib. No correlation between treatment response and the degree of skin toxicity was reported in the IDEAL studies, in contrast to the studies with erlotinib summarized below. Patients with bronchioalveolar carcinoma appear to constitute a distinct subpopulation who have only limited response to chemotherapy but possibly superior response rates to EGFR tyrosine kinase inhibitors. Studies of response in these patients may serve to identify predictive markers of efficacy for the anti-EGFR agents in other patient subgroups.Gefitinib is well tolerated, and Phase II studies have suggested it may be an alternative to docetaxel in the second-line setting. Randomized Phase III studies comparing gefitinib with docetaxel in terms of survival and quality of life are now ongoing. Additional studies are indicated to assess the possible role of gefitinib as maintenance therapy after chemotherapy for stage IV disease and after chemo/radiotherapy for stage III disease.Paralleling the experience with gefitinib, erlotinib (Tarceva) demonstrated activity in the second-line setting, but first-line trials showed no advantage to erlotinib in conjunction with standard chemotherapy. There are unresolved issues in dosing with both erlotinib and gefitinib because they are given as a fixed dose despite individual differences in gastrointestinal absorption. In the second-line trials, response to erlotinib was correlated with the occurrence and severity of rash, an observation that provides a rationale for titrating the dose in individual patients to a level that causes detectable skin rash.A large randomized Phase III trial of erlotinib as second/third-line therapy versus best supportive care is now completed, and the results of this trial will be important in clarifying whether the EGFR tyrosine kinase inhibitors can make a meaningful contribution in terms of survival improvement of lung cancer patients.Cetuximab (C225, Erbitux) is a monoclonal antibody that inhibits EGFR ligand binding, resulting in cell cycle arrest and increased expression of proapoptotic proteins. Preclinical studies indicated potential synergism between cetuximab and a number of chemotherapeutic agents, including cisplatin and paclitaxel. In fact, this synergy has been demonstrated clinically in colon cancer and head and neck cancer. Positive results have been reported in combination with chemotherapy. An ongoing study is now evaluating cetuximab as monotherapy in patients with NSCLC who have failed platinum-based chemotherapy. As with the EGFR tyrosine kinase inhibitors, molecular mechanisms predicting response to cetuximab therapy are currently not well understood. Some evidence suggests that anti-EGFR antibodies, which possess a different mechanism of action, may be more effective than the EGFR tyrosine kinase inhibitors when combined with chemotherapy. However, this is a very preliminary observation that has not yet been confirmed by clinical evidence.Bevacizumab (Avastin) is a monoclonal antibody directed against vascular endothelial growth factor. In a randomized Phase II trial in patients with advanced metastatic NSCLC, the addition of bevacizumab to standard carboplatin/paclitaxel chemotherapy significantly increased the time to progression with a nonsignificant increase in response rate as well. Several Phase II trials investigating bevacizumab in combination with chemotherapy or with a targeted agent, such as erlotinib, are now under way. A large randomized study comparing standard chemotherapy with standard chemotherapy plus bevacizumab is being undertaken by the Eastern Cooperative Oncology Group. Ongoing trials will examine issues of optimal dosage and scheduling with chemotherapy and with other targeted therapies, as well as minimizing the risk of bleeding episodes that have been the most serious safety issue in the lung cancer trials.Pemetrexed (Alimta) is an antifolate agent with antitumor activity demonstrated in a number of solid tumors, including NSCLC. Supplementation with folic acid and vitamin B12 has reduced the mucosal and bone marrow toxicity seen in early studies of pemetrexed, permitting more prolonged treatment. In NSCLC, single-agent activity has been demonstrated in the first- and second-line settings in Phase II trials. A randomized Phase III trial comparing pemetrexed with docetaxel as second-line therapy in advanced NSCLC has shown similar activity and less toxicity. Pemetrexed has also shown activity in Phase II combination trials with platinum compounds (cisplatin, carboplatin, and oxaliplatin), vinorelbine, and gemcitabine. Questions remain about the optimal sequencing of pemetrexed and gemcitabine, and a current Phase III trial is examining different dosing regimens of these two agents to resolve this issue. Additional trials with other novel agents (such as bevacizumab) are being planned.Oblimersen (G3139, Genasense) is a bcl-2 antisense oligonucleotide designed to bind to bcl-2 mRNA, causing its degradation and thus preventing protein translation. Bcl-2 is an antiapoptotic protein implicated in the development of resistance to both chemotherapy and radiation. However, conference participants felt that the clinical significance of bcl-2 as a target has not been sufficiently documented in lung cancer. Elevated bcl-2 expression appears to correlate with poor prognosis in a number of human cancers, but the data are unclear and contradictory on this point, particularly in NSCLC. Several participants also commented that there are unresolved safety issues with suspected immunomodulatory effects of antisense molecules. Results of small preliminary studies with oblimersen are promising. It is currently not clear whether oblimersen is best used as a single agent, given the probability that other signaling pathways can compensate for bcl-2 inhibition, or in combination with other targeted agents. It is being studied in combination with carboplatin and etoposide as first-line therapy in small cell lung cancer and with docetaxel as second-line therapy in relapsed or refractory NSCLC.Bortezomib (PS-341, Velcade) is a proteasome inhibitor that in preclinical models causes cell cycle arrest and increased apoptosis. It has been approved by the United States Food and Drug Administration for use in patients with refractory multiple myeloma. A randomized Phase II study is evaluating bortezomib with or without docetaxel in second-line therapy of advanced NSCLC. There is also a Phase II trial of single-agent bortezomib as second-line therapy of advanced small cell lung cancer. Several ongoing combination Phase I trials are evaluating bortezomib in combination with platinum-based chemotherapy. It is thought that bortezomib may make cells more sensitive to chemotherapy by inhibiting NF-κB. However, there are sequencing issues with the use of bortezomib and chemotherapy that require study because preclinical data suggest the possibility of antagonistic or synergistic interactions with chemotherapy, depending on how the two are administered concurrently or sequentially. As with pemetrexed, different cell lines have produced conflicting results.Flavopiridol is a cyclin-dependent kinase inhibitor that produces cell cycle arrest. In preclinical studies, flavopiridol provides synergistic cytotoxic effects when administered after a taxane, whereas it is antagonistic if given concurrently or prior to the taxane. Phase I trials have examined the effects of flavopiridol when administered sequentially with gemcitabine or with docetaxel. A randomized Phase II study of flavopiridol and sequential docetaxel is under way. There is uncertainty, however, regarding the optimal dosing for flavopiridol due to the lack of pharmacodynamic end points to confirm target inhibition. The conference participants felt that although flavopiridol may not have the pharmacological properties that make it an optimal agent, this target remains a valid one for experimental therapeutics in lung cancer. Newer, more potent cyclin-dependent kinase inhibitors are now entering clinical trials, and the results are eagerly anticipated.R115777 and L-778,123 are investigational FTIs designed to block ras signaling and thus inhibit cell proliferation and promote apoptosis. In preclinical studies, FTIs inhibited growth in small cell lung cancer and NSCLC cell lines and in xenograft models. Phase II trials have evaluated R115777 and L-778,123 in advanced untreated NSCLC and R115777 in relapsed small cell lung cancer. No objective responses were reported in any of these trials, and the drugs had no discernable single-agent activity, although the trial of R115777 in NSCLC showed some evidence of disease stabilization in a small minority of patients. The latter trial documented at least partial inhibition of farnesyl transferase activity in >80% of the patients, indicating that the drug was having the intended effect in surrogate tissues. FTIs have shown activity in other cancers, and biomarkers identified in these tumors may allow identification of a responsive lung cancer phenotype.The rationale for combining the molecularly targeted agents with standard chemotherapy needs to be reexamined. Combination therapy may have antagonist rather than additive effects if the regimens selected are not based on careful demonstration of the pharmacodynamics and mechanisms of action of both classes of agents. Additional preclinical studies are needed to explore sequencing versus concurrent administration of molecularly targeted agents with the taxanes. Targeted agents may in fact interfere with the cytotoxic effects of chemotherapy if their effect is to induce cell cycle arrest.Most chemoprevention trials with micronutrients such as β-carotene, α-tocopherol, or vitamin A derivatives have had disappointing outcomes. In fact, primary prevention with β-carotene or retinol appears to increase the risk of lung cancer in smokers. Currently, interest in chemopreventive strategies for lung cancer has focused on the selective COX-2 inhibitors celecoxib and rofecoxib. COX-2 is very highly expressed in preneoplastic lung lesions. In NSCLC, expression of COX-2 has been associated with decreased survival. Some epidemiological studies have reported a two-thirds reduction in lung cancer incidence with regular use of aspirin or nonsteroidal anti-inflammatory drugs, whereas other population-based studies have failed to find a protective effect for nonsteroidal anti-inflammatory drugs. It is unclear, however, what the optimal dosing is for the COX-2 inhibitors either as primary prevention or in conjunction with chemotherapy. A randomized, placebo-controlled trial of iloprost, a prostacyclin inhibitor, is also under way.EGFR expression is also very high in premalignant lesions, raising the possibility of using the EGFR-targeted agents for prevention. One trial has been proposed to test the EGFR tyrosine kinase inhibitor gefitinib in the reversal of premalignant lesions of the lungs, and another trial following the same study design would investigate the FTI tipifarnib. These two trials, the so-called STOP (Specialized Programs of Research Excellence Trials Of Prevention) Lung Cancer, have been put on hold pending resolution of safety issues raised by the United States Food and Drug Administration. Although these are both well-designed studies, the increased scrutiny and safety concerns for chemoprevention studies in healthy individuals make performing these trials difficult.There is also discussion of bcl-2 as a possible target for prevention, pending completion of current treatment trials with bcl-2 antisense with sufficiently positive results and acceptable safety data. All such chemoprevention trials need to address safety issues appropriate for a very high risk population.The disappointing trial results seen with a number of the molecularly targeted therapies have exposed the inadequacy of our current understanding of lung cancer biology. A more thorough comprehension of the multiple growth signaling pathways and of the roles of cross-talk, redundancy, and up-regulation of compensatory mechanisms is essential to making targeted therapies effective for more than small, difficult-to-define subgroups of patients. More research needs to be directed at understanding basic mechanisms of tumor growth and resistance. Newly developed knockout mouse models will be important tools for undertaking this research, in addition to a more determined effort to obtain serial biopsies and to study available archival tissue using microarray technology.The EGFR-targeted agents clearly have some, if limited, activity. They are well tolerated, and they are being prescribed in the clinic. The primary research issue that needs to be addressed is determining the clinical features or biological markers that predict which patients are likely to respond to these agents. There is a general consensus among conference participants that tumor tissue studies aimed at identifying determinants of response to EGFR inhibitors should be given a high priority to define specific patient populations with a greater probability of response. Although challenging to perform, the group felt strongly that studies which correlated outcome to biological end points had the greatest ability to improve our understanding of lung cancer biology and, ultimately, the treatment of patients with this dreaded disease.Presented at the First International Conference on Novel Agents in the Treatment of Lung Cancer, October 17–18, 2003, Cambridge, Massachusetts.Requests for reprints: Thomas J. Lynch, Massachusetts General Hospital, Division of Hematology/Oncology, 100 Blossom Street, Cox 210, Boston, MA 02114. Phone: (617) 724-1136; Fax: (617) 724-1137; E-mail: tlynch@partners.orgThe symposium and publication of these proceedings were made possible by unrestricted educational grants from AstraZeneca Pharmaceuticals, Aventis Pharmaceuticals, Genentech, Bristol Myers Squibb/ImClone, Eli Lilly & Co., and Pfizer Oncology. Editorial assistance and CME Sponsorship were provided by InforMEDical Communications, Inc. (Carlisle, MA).