Virtual Library

 x 
My Library Virtual Library FAQ

Start Your Search

G. Liu



Author of

  • +

    MO05 - Prognostic and Predictive Biomarkers II (ID 95)

    • Event: WCLC 2013
    • Type: Mini Oral Abstract Session
    • Track: Medical Oncology
    • Presentations: 1
    • +

      MO05.11 - The Effect of Two BRM Promoter Polymorphisms on the Risk of Advanced Non-Small Cell Lung Cancer (NSCLC) and Small Cell Lung Cancer (SCLC) in Smokers (ID 1987)

      16:15 - 17:45  |  Author(s): G. Liu

      • Abstract
      • Presentation
      • Slides

      Background
      BRM, an ATPase subunit of the SWI/SNF chromatin remodeling complex, is a putative tumor susceptibility gene in lung cancer. Loss of BRM expression occurs in 15% of lung cancers. Two BRM promoter insertion polymorphisms (BRM-741 and BRM-1321) lead to epigenetic silencing of BRM and highly correlate with loss of BRM expression and function in lung tumors. Pharmacologic reversal of the epigenetic changes of BRM is feasible. We previously demonstrated a strong risk association between these two polymorphisms and susceptibility to early stage NSCLC. Here, we evaluate the association between the two BRM polymorphisms and risk of developing: 1) advanced NSCLC, and 2) SCLC among smokers.

      Methods
      Genotyping for BRM promoter polymorphisms was performed using TaqMan. The cohorts analyzed were: 1) 417 stage III-IV NSCLC cases and 2) 111 SCLC cases treated at the Princess Margaret Cancer Centre (PMCC), Toronto; and 3) 43 SCLC cases from the University of Florida (U of F), all with a smoking history of ≥1 pack-year. Cases were matched to healthy controls by frequency distribution (1:2 for PMCC cases; 1:1 for U of F cases) based on age, gender, pack-year smoking history, and either current smoking status (PMCC) or ethnicity (U of F). Adjusted odds ratios (aORs) with their 95% confidence intervals (CI) of the association between polymorphisms and lung cancer risk were estimated by multiple logistic regression models.

      Results
      Of the 417 NSCLC cases, 59% were male; 41%, current smokers; 63%, adenocarcinoma; 51%, stage IV; median age, 63 years. The frequency of homozygosity was BRM-741, 21%; BRM-1321, 20%; both, 11%. The homozygous variants of BRM-741 and BRM-1321 were associated with an increased risk of advanced NSCLC compared to the wild types, with aOR’s of 1.6 (95% CI: 1.1-2.2; p=0.008) and 1.4 (95% CI: 1.0-2.0; p=0.04), respectively. Being homozygous for both BRM promoter variants carried an even greater risk (aOR 2.4 [95% CI: 1.4-4.0; p=0.0009]), with the strongest effect observed among current smokers (aOR 3.4; p=0.0005), and those with a histological diagnosis other than adenocarcinoma (aOR 3.2; p=0.0005). Among the 111 PMCC SCLC cases, 62% were male; 56%, current smokers; median age 65 years; of the 43 U of F SCLC cases, 35% were male; median age, 63 years. The presence of double homozygous variants of BRM-741 and BRM-1321 had no effect on the risk of developing SCLC in either of the two cohorts analyzed, with aOR’s of 1.1 (95% CI: 0.3-3.5; p=0.94) and 0.3 [95% CI: 0.04-2.41; p=0.27), respectively.

      Conclusion
      The presence of double homozygous variants of the BRM promoter polymorphisms, BRM-741 and BRM-1321, significantly increases the risk of advanced NSCLC among individuals with a smoking history greater than one year, with the strongest effect observed among current smokers. In contrast, the same two polymorphisms had no effect on the risk of developing SCLC in either of the two cohorts analyzed. Thus, this study offers further insight into potential mechanisms underlying the genetic susceptibility to developing advanced NSCLC among smokers. Validation in larger populations is warranted.

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    O01 - Prognostic and Predictive Biomarkers I (ID 94)

    • Event: WCLC 2013
    • Type: Oral Abstract Session
    • Track: Medical Oncology
    • Presentations: 2
    • +

      O01.01 - Genetic polymorphisms of inflammatory and DNA repair pathways, radiation-related esophagitis and pneumonitis in definitive chemoradiation treated non-small cell lung cancer patients. (ID 2997)

      10:30 - 12:00  |  Author(s): G. Liu

      • Abstract
      • Presentation
      • Slides

      Background
      The benefits of concurrent chemoradiotherapy in locally advanced non-small cell lung cancer (NSCLC) are tempered by treatment toxicity. Germline genetic variants have been associated with intrinsic radiosensitivity and radiotoxicity in various cancer settings. We investigated whether variants in genes involved in inflammation response and DNA repair pathways independently influence radiation-induced phenotypes of esophagitis and pneumonitis. From 19 candidate genes, 52 polymorphisms, directed by literature and by tagging procedures, were systematically selected for assessment. The candidate genes were involved in DNA repair (double-strand breaks, homology directed, nucleotide excision) and pro/anti-inflammatory signaling. The this investigation sought to evaluate the association of genetic sequence markers for two clinically significant radiation-induced toxicities - esophagitis and pneumonitis – seen in NSCLC patients treated with a curative intent.

      Methods
      From 312 patients treated at PMCC between 2005-12, a training cohort was defined consisting of 92 definitive concurrent chemoradiation/radiation-treated NSCLC patients with genotype information on the 52 polymorphisms. A second, validation cohort consisted of 209 patients. Multivariate logistic regression was performed for each polymorphism of interest, adjusting for known clinical and dosimetric prognostic factors on the dichotomized outcomes of radiation esophagitis (Grades 0-2 vs 3-5) and pneumonitis (Grades 0-1 vs 2-5). The CTCAEv4.03 grading criteria were used. Additive genetic models were used for genetic association analysis. In the training set, genetic variants, genotyped by IlluminaGoldenGate, with p<=0.05 were identified for validation; HWE was set at p>0.01, a criteria met by all polymorphisms with statistical significance.

      Results
      In the combined training and validation datasets, 63% were males, with median age of 65 years. Specifically in the training dataset, 65% were male, with median age of 62, median mean lung doses of 15.9, median max esophageal dose of 67.1 and median V20 of 27.6. For esophagitis, the final models were adjusted for concurrent chemotherapy, V20 and max esophageal dose. Five genetic variants linked to TNF and IL6 were significantly associated with outcome (each using wild-type genotype as reference) (Table 1). For pneumonitis, the final models adjusted for V20 and smoking status. Eight genetic variants found within four genes (ATM, BRCA2, IL1alpha, IL1RN) were associated with significant pneumonitis (Table 1).

      ESOPHAGITIS
      Function / Pathway Gene refSNP OR 95% CI P value
      pro-inflammatory cytokine TNF rs3093662 3.54 1.9-10.6 0.02
      pro-inflammatory cytokine TNF rs3093664 3.42 1.2-10.2 0.03
      pro-inflammatory cytokine TNF rs3093665 4.95 1.2-21.1 0.03
      anti-inflammatory cytokine IL6 rs1800797 2.53 1.0-6.2 0.04
      anti-inflammatory cytokine IL6 rs1800795 2.45 1.0-5.9 0.046
      PNEUMONITIS
      Function / Pathway Gene refSNP OR 95% CI P value
      double-strand break repair ATM rs664143 2.67 1.3-5.6 0.01
      double-strand break repair ATM rs664677 2.37 1.2-4.7 0.01
      homology-directed repair BRCA2 rs1799955 2.59 1.3-5.3 0.01
      homology-directed repair BRCA2 rs1801406 2.42 1.2-4.8 0.01
      homology-directed repair BRCA2 rs1799943 2.09 1.0-4.2 0.04
      anti-inflammatory cytokine IL1alpha rs17561 2.63 1.2-5.7 0.01
      anti-inflammatory cytokine IL1alpha rs2856863 2.60 1.1-5.9 0.02
      anti-inflammatory cytokine IL1RN rs3087263 0.17 0.04-0.8 0.04

      Conclusion
      In our 92 patient training set, genetic variations in TNF and IL6 are associated with radiation esophagitis, while genetic variations in ATM, BRCA2, IL1alpha and IL1RN are associated with pneumonitis. Results from the 209 patients in the validation dataset will be presented at the meeting (A.H. and G. L are co-senior authors).

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

    • +

      O01.03 - BRM Promoter Variants and Survival Outcomes of Advanced Non-Small Cell Lung Cancer (NSCLC) Patients: A Validation Study in the NCIC Clinical Trials Group (NCIC-CTG) BR24 Clinical Trial (ID 1999)

      10:30 - 12:00  |  Author(s): G. Liu

      • Abstract
      • Presentation
      • Slides

      Background
      BRM, an ATPase subunit of the SWI/SNF chromatin remodeling complex, is a putative tumor susceptibility gene in NSCLC. Loss of BRM expression occurs in 15% of NSCLC, and has been linked to adverse outcome. Two BRM promoter insertion variants (BRM-741 and BRM-1321) result in epigenetic silencing of BRM through recruitment of histone deacetylases. The presence of double homozygous BRM variants is linked to loss of BRM expression and function in lung tumors, and double the risk of lung cancer. Pharmacological reversal of the epigenetic changes of BRM is feasible. In this study we evaluated the association between the BRM promoter variants and survival outcomes of advanced NSCLC patients.

      Methods
      The training cohort consisted of 564 stage III-IV NSCLC patients treated at the Princess Margaret Cancer Centre, Toronto 2006-2010. The validation cohort comprised 219 stage IIIb-IV NSCLC patients from the NCIC-CTG BR24 clinical trial, a phase II/III double-blind randomized trial of cediranib versus placebo in patients receiving carboplatin/paclitaxel for the treatment of advanced or metastatic NSCLC. Genotyping for the BRM promoter variants was performed using Taqman. Associations of BRM promoter variants and overall (OS) and progression free survival (PFS) were assessed using Cox proportional hazard models adjusted for clinically relevant variables, and in the case of the BR24 population, stratified by treatment arm.

      Results
      Among the training cohort, 73% were Caucasian, 52% male, median age 63 yrs, 55% stage IV disease, and 67% adenocarcinoma. Median OS was 1.6yrs; median follow up, 3.6yrs. The frequency of homozygosity was BRM-741, 23%; BRM-1321, 21%; both 12%. Homozygous variants of BRM-741 were strongly associated with worse OS (adjusted HR [aHR] 2.5 [95% CI: 1.9-3.3; p=6x10E-10]) and PFS (aHR 2.0 [95% CI: 1.6-2.6; p=9x10E-8]) compared to the wild types. Similar findings were observed for the BRM-1321 homozygous variants (aHR for OS of 2.0 [95% CI: 1.5-2.6; p=2x10E-6]; aHR for PFS of 1.8 [95% CI: 1.4-2.4; p=3x10E-6]). The presence of double homozygous BRM-741 and BRM-1321 variants was strongly associated with worse OS (aHR 2.8 [95% CI: 1.9-4.0; p=7x10E-8]) and PFS (aHR 2.7 [95% CI: 1.9-3.8; p=1x10E-8]). Genotyping was possible for 219/296 BR24 participants. Of these, 59% were male, median age 59 yrs, 83% stage IV, 46% adenocarcinoma, with 50% receiving cediranib. Individuals carrying the homozygous variants of both BRM-741 and BRM-1321 (13% of cases) had a substantially worse OS (aHR 9.0 [95% CI: 4.3-18.5; p=1x10E-9]) and PFS (aHR 3.8 [95% CI: 1.9-7.3; p=3x10E-5]) compared to the wild types, irrespective of whether they were treated with cediranib (aHR for OS of 6.4; p=1x10E-4; aHR for PFS of 2.1; p=0.02) or placebo (aHR for OS of 16.8; p=2x10E-7; aHR for PFS of 8.3; p=1x10E-4).

      Conclusion
      The same two homozygous BRM promoter variants that are associated with increased risk of NSCLC are also strongly associated with adverse OS and PFS in this study of advanced NSCLC patients. We are completing additional studies focusing on the relationship between the BRM promoter variants and BRM protein expression; results will be presented at the meeting.

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    O06 - Cancer Control and Epidemiology I (ID 135)

    • Event: WCLC 2013
    • Type: Oral Abstract Session
    • Track: Prevention & Epidemiology
    • Presentations: 1
    • +

      O06.06 - Factors Associated with Smoking Cessation in Participants of The Pan Canadian Early Lung Cancer Study (ID 1469)

      10:30 - 12:00  |  Author(s): G. Liu

      • Abstract
      • Presentation
      • Slides

      Background
      Lung cancer screening programs provide unique opportunities to facilitate smoking cessation in smokers who participate in these programs. However, the effects of screening on motivation to quit might be mediated or modified by other variables. Identifying the participants more likely to quit will allow rapid application of smoking cessation resources to these participants, while those least likely to quit can be afforded experimental interventions. The aim of our study was to assess the impact of lung cancer screening on smoking cessation in current smokers at the time of enrollment and to identify factors that were associated with quitting smoking in this screening population.

      Methods
      Using data collected from the Pan-Canadian Study of Early Detection of Lung Cancer, both univariate and multivariable logistic regression analysis was used to identify predictors of smoking cessation among current smokers at enrolment. Smoking cessation was defined as quitting for at least a 6 month period, occurring anytime after enrolment.

      Results
      We analyzed baseline and follow-up questionnaires of 2320 participants, of which 1419 were current smokers. Of these 1419 patients, 392 (27.8%) met the definition of smoking cessation during a median of two annual follow-up visits. In both univariate and multivariable (MV) analysis, greater smoking cessation was associated with four factors: (i) having a diagnosis of lung cancer at any time during the screening process, with a MV Odds ratio (OR) of quitting of 2.4 (95%CI: 1.1-5.0); (ii) lower and medium nicotine addiction as assessed by the Fagerström Nicotine Dependence Scale Score, with MV-ORs of 3.2 (95%CI: 2.2-4.6) and 1.4 (95%CI: 0.9-2.0), respectively; (iii) having higher education, with MV-OR: 1.4 (95%CI: 1.1-1.9); and (iv) having an earlier age of onset of regular alcohol intake, with MV-OR of 1.11 (95%CI: 1.02-1.21) per 5 year decrease in age. Smoking cessation was also associated with (i) previous attempts of quitting [UV-OR 1.8 (95%CI: 1.2-2.7)], willingness to quit smoking within the next month (at baseline screening) [UV-OR 2.2 (95%CI: 1.8-2.9)] or within the next 6 months after baseline screening [UV-OR 1.8 (95%CI: 1.3.-2.4)]. Second-hand smoking exposure, including exposure as a child, or as an adult at work, at home, privately with friends, or in public settings, or a cumulative index of these different exposures, was not associated with smoking cessation. Presence of potential index symptoms for lung disease, including shortness of breath, cough (both dry and productive), hoarseness, audible wheezing or even chest pain, was not associated with an increased chance of smoking cessation.

      Conclusion
      The diagnosis of a new lung cancer had a major positive impact on screening participants quitting smoking, as were factors such as lower nicotine dependence, higher education, earlier starting alcohol drinking age, and willingness to quit. Whether a new lung cancer diagnosis triggered additional efforts by clinicians to help the person quit will be explored further. Individual lung symptoms and secondhand smoke exposure were not associated with smoking cessation. (Geoffrey Liu and Martin Tamemmagi are co-senior authors)

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    O08 - Preclinical Therapeutic Models I (ID 92)

    • Event: WCLC 2013
    • Type: Oral Abstract Session
    • Track: Biology
    • Presentations: 1
    • +

      O08.07 - Patient-derived primary non-small cell lung carcinoma (NSCLC) xenograft models for mechanistic studies of resistance to EGFR tyrosine kinase inhibitor therapy (ID 2380)

      16:15 - 17:45  |  Author(s): G. Liu

      • Abstract
      • Presentation
      • Slides

      Background
      Non-small cell lung cancer (NSCLC) patients with tumors bearing “driver” mutations in the epidermal growth factor receptor (EGFR) tyrosine kinase (TK) domain have very high response rates to small molecule EGFR TK inhibitors (TKIs). However, all patients eventually develop resistance to the TKIs, and more recent reports have shown that patients who have stopped TKI therapy may be sensitive again upon re-treatment. While several genetic mechanisms of resistance have been documented, including the gate keeper T790M mutation and Met amplification, cell line studies in vitro have also implicated alternate epigenetic mechanisms that may explain the clinical progression observed in patients with EGFR mutations treated by TKIs. Studies in vivo using patient-derived primary lung tumor xenograft models have not been reported.

      Methods
      Patient-derived primary tumor xenografts were established from surgically resected early stage NSCLC implanted subcutaneously in non-obese diabetic severe combined immune deficient (NOD-SCID) mice. Tumors were passaged after reaching the humane endpoint 1.5 cm maximum diameter. EGFR TKI therapy was initiated when tumors reached ~6 mm diameter. Treatment included daily oral gavage for erlotinib (50 mg/Kg) and dacomitinib (3 mg/Kg). Cetuximab was administered weekly intraperitoneally (50 mg/Kg).

      Results
      Among 33 tumors with EGFR mutations engrafted into the mice, only 6 (18.2 %) formed tumors that could be propagated beyond first passage. Three models have been studied for their responses to EGFR TKIs. Model 148 with L858R mutation showed intrinsic pan-resistance to erlotinib and dacomitinib, as well as to cetuximab. This model was derived from a patient who received pre-operative erlotinib in a window of opportunity trial and did not respond. The patient relapsed after surgery and did not receive additional TKI therapy. Model 137 with exon19 E746-A750 deletion mutation demonstrated complete response to both erlotinib and dacomitinib. However, microscopic examination of tissue from the implantation site revealed viable tumor cells, consistent with the inability of TKI to completely eradicate tumor cells even when complete response is observed clinically. The patient subsequently developed disease recurrence and responded to third line gefitinib treatment. Model 164 has double exon19 L747-T751 deletion/T790M mutations. As anticipated, the xenograft failed to respond to erlotinib but responded dramatically to cetuximab alone. Importantly, model 164 xenograft showed transient stabilization of the tumor growth when treated by dacomitinib, but eventually developed progressive growth after 2 weeks of treatment. Resistance was reversible each time the dacomitinib-resistant tumor was propagated, without drug in new mice. The reversibility of resistance observed upon re-initiation of dacomitinib treatment suggests an epigenetic mechanism for TKI resistance. This patient developed recurrence after surgery and failed to respond to second line erlotinib treatment.

      Conclusion
      Patient-derived primary lung cancer xenografts may provide important patient-like models to study mechanisms of resistance to targeted therapies, and to test novel treatment strategies that may improve further treatment efficacy.

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    O24 - Cancer Control and Epidemiology III (ID 134)

    • Event: WCLC 2013
    • Type: Oral Abstract Session
    • Track: Prevention & Epidemiology
    • Presentations: 1
    • +

      O24.05 - The impact of body mass index on survival in stage 3 and 4 lung cancer (ID 663)

      16:15 - 17:45  |  Author(s): G. Liu

      • Abstract
      • Presentation
      • Slides

      Background
      Obesity has been shown to be an adverse prognostic factor in several cancers, including breast, colorectal, endometrial, ovarian, pancreatic and prostate. However, studies of body mass index (BMI) and outcomes in lung cancer are lacking. Understanding the clinical impact of body weight on cancer outcomes is important given the high prevalence of obesity globally. We retrospectively evaluated the distribution of BMI at diagnosis and its effects on survival in stage 3 and 4 lung cancer patients.

      Methods
      1,121 patients with stage 3 or 4 lung cancer from a single institution were analyzed. Clinicopathologic data were collected retrospectively. Adjusted hazard ratios (aHR) for overall survival (OS) and progression-free survival (PFS) were generated by Cox regression for each BMI (kg/m[2]) category (underweight: <18.5, normal: 18.5-24.9, overweight: 25.0-29.9, obese: ≥30), after adjusting for age, gender, Charlson Comorbidity Index, performance status (PS), clinical stage and treatment regimen.

      Results
      In this cohort (n=1,121), the frequencies of stage 3A, 3B and 4 lung cancers were 35%, 32% and 33%, respectively. There were 633 (57%) adenocarcinomas, 238 (21%) squamous cell carcinomas, 38 (3%) small cell lung cancers, and 210 (19%) other histologies. Patients had variable BMI: 82 (7%) underweight, 550 (49%) normal weight, 333 (30%) overweight, 156 (14%) obese. Being overweight/obese was associated with older age (p=0.002) and stage 3A disease (p=0.001); underweight patients were more likely current smokers (p<0.001). OS was significantly decreased with age ≥65, males, PS 2-3, stage 4, and lack of systemic treatment (p<0.001). Median OS in underweight, normal weight, overweight and obese patients were 14, 23, 24 and 26 months, respectively. Compared with BMI ≥18.5, being underweight was associated with significantly poorer OS (aHR 1.33, 95% CI 1.01-1.77, p=0.045), but not PFS (aHR 1.12, 95% CI 0.86-1.46, p=0.414). The magnitude of this association was greatest among those aged less than 65 years (aHR 1.57, 95% CI 1.11-2.22, p=0.011).

      Conclusion
      In stage 3 and 4 lung cancer, being underweight at diagnosis is associated with significantly poorer OS, especially in patients younger than 65 years of age. Lower BMI is mostly observed in current smokers, while above normal BMI is seen in older patients and stage 3A disease. Unlike other malignancies, obesity does not increase mortality in this population. The BMI-survival relationship in lung cancer requires further study.

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    P1.05 - Poster Session 1 - Preclinical Models of Therapeutics/Imaging (ID 156)

    • Event: WCLC 2013
    • Type: Poster Session
    • Track: Biology
    • Presentations: 1
    • +

      P1.05-007 - Large scale establishment of genetically diverse patient-derived primary tumor xenografts from resected early stage non-small cell lung cancer (NSCLC) patients (ID 1539)

      09:30 - 16:30  |  Author(s): G. Liu

      • Abstract

      Background
      The fidelity of established NSCLC cell line models to reflect patient tumors has been challenged. Patient-derived primary tumor xenografts (PTXGs) established directly from patient tumors in immunodeficient mice reproduce closely the histology of the primary tumors, thus are potentially better preclinical models to investigate novel therapies. We previously reported that early stage NSCLC patients whose tumors form PTXGs have significantly greater risk of relapse after surgery (Clin Cancer Res 2011; 17: 134-141). We report here a more extended analysis of clinical-molecular-pathological features of early stage NSCLC that are associated with engraftment and its impact on patient outcome.

      Methods
      Resected NSCLC tumors were harvested within 30 minutes after surgery and were implanted into severely immunodeficient mice to establish PTXGs. Tumors that grew were propagated for up to 3 passages. The mutational profiles of the primary tumors were assessed by the MassARRAY platform that included 133 mutations with ‘putative’ driver function, which have been reported in COSMIC database as recurrent in NSCLC. All identified mutations were verified by direct sequencing in both the primary and PTXG tumors. Engraftment rate among clinical factors were tested using the Fisher’s exact or Mann-Whitney tests. The Kaplan-Meier method was used to estimate 3-year overall (OS) and disease-free survival (DFS) probabilities. The effect of engraftment on OS and DFS adjusting for clinical variables was assessed using a Cox proportional hazards model.

      Results
      From April 2005 to December 2010, 261 rigorously verified resected primary non-carcinoid NSCLCs were engrafted; 38 xenografts that were lymphoma were excluded from further analysis. For the remaining 223 primaries, 101 (45.3%) successfully engrafted and formed PTXG lines. Engraftment rates were 33.8% (48/142) for adenocarcinoma (AdC), 67.7% (42/62) for squamous cell carcinoma (SqCC), 66.7% (4/6) for large cell neuroendocrine carcinoma, and 53.8% (7/13) for others. The tumors forming PTXGs were more likely to be poorly differentiated (p=0.00012) and of larger tumor size and higher pT stage (p<0.0001), but were not correlated with the pN stage. Among 95/101 (94.1%) PTXG cases profiled for mutations, 6 had mutations in the EGFR tyrosine kinase domain, 18 in KRAS/HRAS, 5 in PIK3CA, 2 in paxillin and 1 in STK11/LKB gene; 56 (62.2%) were negative for mutations. The median follow-up time was 2.7 years (range 0.04 – 7.5 years). Patients whose tumors engrafted had decreased DFS (HR 2.68, 95% CI 1.16-4.60, Wald p<0.0001) and OS (HR 3.14, 95% CI 1.56-6.33, Wald p=0.0014). Significantly poorer survival was maintained in AdC. Among 33 patients with EGFR mutation, only 6 (18.2%) engrafted. Engraftment was associated with significantly poorer DFS (HR 4.76; 1.43-15.86, log-rank p=0.005) and OS (HR 8.55, 95% CI 0.77-94.3, log-rank p=0.035) in this population.

      Conclusion
      The ability to form PTXGs of early stage NSCLC is confirmed as a very strong poor prognostic marker. Although EGFR mutant tumors usually do not engraft, engraftment of EGFR mutant tumors is associated with poor patient survival. PTXGs appear to represent biologically aggressive NSCLC.

  • +

    P1.22 - Poster Session 1 - Epidemiology, Etiology (ID 166)

    • Event: WCLC 2013
    • Type: Poster Session
    • Track: Prevention & Epidemiology
    • Presentations: 1
    • +

      P1.22-011 - Assessment of the accuracy and reliability of health related behavioural data obtained from patient-reported surveys (PRS) compared with electronic patient records (EPR) in lung cancer patient population (ID 2948)

      09:30 - 16:30  |  Author(s): G. Liu

      • Abstract

      Background
      Cigarette smoking, alcohol consumption and presence of co morbidities are important factors that affect health status and mortality in patients diagnosed with lung cancer. While the gold standard for presence or absence of co-morbidities is EPR, the gold standard for obtaining accurate data pertaining to health-related behaviours is by PRS. The purpose of this study is to ascertain, whether in the absence of patient self-reported data, health related behavioural data pertaining to cigarette smoking and alcohol consumption abstracted from EPR provides an accurate and reliable surrogate.

      Methods
      731 lung cancer patients completed a PRS pertaining to information on their lifetime tobacco use, alcohol consumption and whether or not they had been diagnosed with certain co-morbid conditions. Relevant smoking, alcohol consumption and co-morbidity data was collected independently from EPR. Kappa coefficient analysis was used to assess the agreement.

      Results
      Results can be seen in Table 1. Ever/never status for smoking showed almost perfect agreement (k=0.95) between PRS and EPR and surpassed all other health behavioural measures and all co-morbidity agreement values. The calculation of pack-years from EPR and PRS showed substantial agreement (k=0.77); However, categorizing the smoking status into current/ former / never, resulted in only moderate agreement (k=0.47). Alcohol ever/ never status agreement was moderate (0.44) with high sensitivity (0.90) but low specificity (0.50). The lung related co-morbidities like emphysema (k=0.41) and chronic bronchitis (k=0.28) showed fair agreement but with substantial missing data through EPR.

      Table 1
      Health Behaviour N Missing Data in EPR Agreement (k) 95% CI (P value) Se Sp
      Smoking (E/N) 709 0 0.95 (0.79, 0.89) 0.995 0.94
      Smoking (Pkyrs)* 606 81(11%) 0.77 P<0.0001
      Smoking (C/F/N)** 705 4(0.5%) 0.47 (0.41, 0.51)
      Alcohol (E/N) 575 150(20.5%) 0.44 (0.36, 0.52) 0.9 0.5
      Comorbidity
      Emphysema 589 126(17.2%) 0.41 (0.33, 0.49) 0.41 0.95
      Chronic Bronchitis 601 94(12.8%) 0.28 (0.19, 0.37) 0.39 0.88
      *Spearman correlation coefficient
      **Weighted kappa

      Conclusion
      In the absence of PRS data, EPR provides a reliable surrogate for ever/ never smoking status and moderately reliable for lifetime smoking exposure in this lung cancer population. However current/ former/ never smoking status and ever/ never alcohol status cannot be reliably ascertained from medical records. Missing EPR data related to smoking pack years, alcohol consumption and lung co-morbidities is concerning and suggests more systematic or synoptic reporting by physicians would improve opportunities for research

  • +

    P3.12 - Poster Session 3 - NSCLC Early Stage (ID 206)

    • Event: WCLC 2013
    • Type: Poster Session
    • Track: Medical Oncology
    • Presentations: 1
    • +

      P3.12-015 - Surgery for Early Non-Small Cell Lung Cancer with Preoperative Erlotinib (SELECT): A Correlative Biomarker Study (ID 2529)

      09:30 - 16:30  |  Author(s): G. Liu

      • Abstract

      Background
      Erlotinib has demonstrated major activity in EGFR mutation positive NSCLC, but may also benefit those with wild type tumours. We conducted a single-arm trial of pre-operative erlotinib in early stage NSCLC to assess radiologic and functional response as well as correlation with known and investigational biomarkers.

      Methods
      Patients with clinical stage IA-IIB NSCLC received erlotinib 150 mg daily for 4 weeks followed by surgical resection. Tumor response was assessed using pre- and post-treatment CT and PET imaging. Tumor genotype was established using Sequenom MassARRAY analysis. EGFR, PTEN, cMET and AXL expression levels were determined by immunohistochemistry. Pre- and post-treatment circulating markers/ligands for EGFR activation (TGF-α, amphiregulin, epiregulin, EGFR ECD) were measured by ELISA. Tumor MET copy number by FISH and VeriStrat® analysis of pre-treatment serum samples is ongoing. Secondary endpoints included pathological response, toxicity and progression-free survival.

      Results
      Twenty-five patients were enrolled; 22 received erlotinib treatment with a median follow up of 4.4 years (range 2.2 to 6.4 years). Histology was predominantly adenocarcinoma (15) with smaller numbers of squamous cell carcinoma (7). PET response (25% SUV reduction) was observed in 2 patients (9%), both with confirmed squamous carcinoma histology. All patients met criteria for stable disease by RECIST and several experienced minor radiographic regression with histologic findings of fibrosis/necrosis, including 2 with squamous histology. The presence of an EGFR activating mutation was detected in two adenocarcinoma cases; one patient experienced minor radiographic response to treatment (exon 19 deletion) and the other stable disease (L858R). High pre-treatment serum levels of TGF- α correlated with tumor growth or primary resistance to erlotinib therapy (p=0.04), whereas high post-treatment soluble EGFR levels correlated with tumor response (p=0.02). Expression of EGFR, PTEN, cMET and AXL did not correlate significantly with tumor response.

      Conclusion
      Erlotinib appears to demonstrate some activity in EGFR wild-type tumors including those with squamous histology. These findings support that certain EGFR wild-type patients may respond to EGFR TKIs. Further research is needed to characterize these patients and elucidate the predictive ability of potential biomarkers such as TGF- α, EGFR copy number and others.