Associate Professor

Geoffrey Liu

MD, University of Toronto

Princess Margaret Cancer Research Tower
101 College St - 11-704 PMCRT, Toronto, Ontario Canada M5G 0A3
Research Interests
Cancer Diagnosis and Therapy, Cancer Mechanisms and Models

At a Glance

Overall research interests:

  • Cancer Diagnosis and Therapy
  • Cancer Mechanisms and Models
  • Machine Learning
  • Molecular Epidemiology

Research Synopsis

Pharmacogenomic Epidemiology

Pharmacogenomic epidemiology is a research area that utilizes epidemiologic methodologies to interrogate pharmacogenomic questions. This research focuses on the use of observational datasets or secondary analyses of clinical trials to evaluate the impact of genetic variation and tumour genomics within pharmacodynamic and pharmacokinetic pathways of therapies on important patient outcomes, such as treatment response, survival or toxicity.

The Liu laboratory, known as AMPPEL (Applied Molecular Profiling Pharmacogenomic Epidemiologic Laboratory) involves a wet component (cell line, primary human xenografts, and primary human tissues, functional assays) and a dry laboratory (clinic-epidemiologic recruitment, data collection, sample acquisition, statistical analysis, epidemiologic and health services outcomes research analyses). The application of molecular epidemiologic methods to pharmacogenetics has provided new opportunities to evaluate rigorously the role of genomic factors in cancer treatment outcomes and toxicity. When large clinical trials involving the drugs of interest are available, secondary analysis of such trials is the preferred method of studying pharmacogenetics. However, there are many instances, such as in rare tumours or when trying to evaluate pharmacogenomics of standard chemotherapy or radiotherapy treatments, where clinical trials are not available to answer important pharmacogenetic questions. Under these circumstances, the role of carefully planned prospective observational studies becomes instrumental to answer both pharmacogenomic and cancer prognosis questions. The quality of results obtainable from these studies is highly dependent on the methods used to recruit patients, to obtain and process samples, to accurately measure genetic markers, to determine accurate phenotypes and outcomes, and to perform appropriate statistical analysis. In addition, pharmacogenomic functional assays and testing are an important component. Therefore, PGE results from these high quality observational studies and clinical trials, the goal of much of AMPPEL's efforts, have strong potential to impact on patient risk stratification and in the choice of appropriate therapies. Dr. Liu is the overall lead PI of the CARMA BROS study (Canadian Rare Molecular Alteration Basket-umbrella Real-world Observational Study; NCT04151342), and site PI of PALEOS (Pan-Canadian Lung cancEr Observational Study; NCT04706754).

Research Interests


A major impediment to performing PGE research has been the lack of systematic collection of toxicity outcomes outside of the clinical trial setting. Dr. Liu is co-director of the ON-PROST ACRU (Ontario Patient Reported Outcomes of Symptoms and Toxicity Applied Clinical Research Unit) funded by Cancer Care Ontario (CCO). This clinical research unit is envisioned as a virtual laboratory for piloting different methods for systematically collecting patient symptoms and toxicity as part of routine care and for research purposes, developing province-wide consensus on what toxicities and symptoms to collect, to support the computer adaptive technology required for province-wide implementation. The majority of pilots are beginning at PMH (lung, head and neck, gastro-esophageal, testicular cancers), in areas that correspond to ongoing molecular epidemiological studies.

Evaluations of Clinical Utility and Clinical Uptake of pharmacogenomic testing are critical to translation from bench to bedside. In the era of personalized medicine, evaluating whether patients understand what pharmacogenomic testing means and involves, their attitude and preferences is important. Through interview-based research utilizing theoretical scenarios and either trade-off testing or conjoint testing, AMPPEL has been ascertaining PMH patient and physician attitudes, understanding and preferences for pharmacogenetic testing.

Dr. Liu is co-lead on 2BLAST (Biostatistics and bioinformatics for the longitudinal analysis of symptoms and toxicity) data science project at Princess Margaret Cancer Centre and co-lead on MBLAST (Machine Learning and Biostatistics for the longitudinal analysis of symptoms and toxicity). This research integrates use of the electronic health record with patient-reported data for analysis of patterns of care and outcomes, using machine learning and natural language processing methods. Some of these analyses are helping the 9th Edition Lung Cancer Staging Project. He is co-leading these efforts with Dr. Wei Xu (Biostatistics) and Dr. Robert Grant (Machine Learning).

AMPPEL Wet Lab section

AMPPEL is performing PGE evaluations in clinical trials and observational studies, using archival tissue, fresh tissue, blood and other surrogate tissues. The methodologic approaches used include: candidate-based analyses, pharmacokinetic (PK) and pharmacodynamic (PD) pathway analyses, genome-wide association studies (GWAS) and post-GWAS analyses (includes methods development), and in the next few years, Next Generation Sequencing (Whole Exome, Whole Genome Sequencing). In addition, AMPPEL has recruited hundreds of patients into molecular epidemiologic studies, including patients with Lung, Head and Neck, Gastro-esophageal, hepatobiliary, pancreatic, and testicular cancer patients, thymomas and mesotheliomas. The laboratory has helped to integrate and coordinate consenting, recruitment, data collection, data management, quality control and clinico-epidemiologic analyses for multiple cancer site

Dr. Liu has been PI of a five-year CCSRI Impact Grant, lung cancer primary derived xenograft program, and is currently PI of a subsequent CIHR-funded grant on primary derived organoids of lung cancer, OPTIMAL (Organoids to guide Post-resistance Therapy In driver MutAted Lung cancers (OPTIMAL). He is also co-lead of POLOR (Patient-derived Organoids of Lung cancer to understand Osimertinib Resistance), an industry sponsored organoid project. He co-leads these research programs with Dr. Ming Tsao at Princess Margaret Cancer Centre

Dr. Liu is lead of Princess Margaret CALIBRE program (Cancer And Liquid biopsy, Integrated analysis, Breathomics, Radiomics for Early Detection). In Lung-CALIBRE, the specific focus of this research is to identify biomarkers that can refine the selection for lung cancer CT screening, identify biomarkers that identify patients for specific interventions and develop risk stratification that incorporates biomarkers. He is Principal Investigator of breathomics as a predictive test for immune checkpoint inhibitor efficacy, which is funded through Industry support and the Canadian Cancer Society Research Institute (CCSRI). As co-Principal Applicant, he also co-leads the CIHR funded Lung cancer early detection and classification using methylome analysis of plasma cell free DNA.

Recent Publications


  1. Lam AC, Aggarwal R, Huang J, Varadi R, Davis L, Tsao MS, Shepherd FA, Lam S, Kavanagh J*, Liu G*. Point-Of-Care Spirometry Identifies High-Risk Individuals Excluded from Lung Cancer Screening. Am J Respir Crit Care Med . 2020 Jul 16;202(10). This publication provides data to support point-of-care spirometry to improve the selection of patients for lung cancer screening.
  2. Shepshelovich D, Xu W, Lu L, Fares A, Yang P, Christiani D, Zhang J, Shiraishi K, Ryan BM, Chen C, Schwartz AG, Tardon A, Schabath MB, Teare MD, Le Marchand L, Zhang ZF, Field JK, Brenner H, Diao N, Xie J, Kohno T, Harris CC, Wenzlaff AS, Fernandez-Tardon G, Ye Y, Taylor F, Wilkens LR, Davies M, Liu Y, Barnett MJ, Goodman GE, Morgenstern H, Holleczek B, Brown MC, Liu G*, Hung RJ. Body Mass Index (BMI), BMI Change, and Overall Survival in Patients With SCLC and NSCLC: A Pooled Analysis of the International Lung Cancer Consortium. J Thorac Oncol. 2019 Sep;14(9):1594-1607. This is the first publication of the Clinical Outcomes Studies of the International Lung Cancer Consortium (COS-ILCCO), where the coordinating centre is at Princess Margaret Cancer Centre (G Liu is the lead coordinator and data management group). The consortium now has over 35,000 patients across 24 studies harmonized in its database, and at the time of publication of this manuscript, ~25,000 cases had been harmonized over 16 studies for this particular analysis. *Corresponding author.
  3. Ren J, Xu W, Su J, Ren X, Cheng D, Chen Z, Bender N, Mirshams M, Habbous S, de Almeida JR, Perez-Ordonez B, Goldstein DP, Wang JR, Bratman SV, Huang SH, Jang R, Zhao Y, Waterboer T, Hung RJ, Liu G. Multiple imputation and clinico-serological models to predict human papillomavirus status in oropharyngeal carcinoma: An alternative when tissue is unavailable. Int J Cancer. 2020 Apr 15;146(8):2166-2174. This study utilized multiple imputation models to predict missing HPV status in older epidemiological studies.
  4. Shi R, Filho SNM, Li M, Fares A, Weiss J, Pham NA, Ludkovski O, Raghavan V, Li Q, Ravi D, Cabanero M, Moghal N, Leighl NB, Bradbury P, Sacher A, Shepherd FA, Yasufuku K, Tsao MS, Liu G. BRAF V600E mutation and MET amplification as resistance pathways of the second-generation anaplastic lymphoma kinase (ALK) inhibitor alectinib in lung cancer. Lung Cancer. 2020 Aug;146:78-85. This study identified specific resistance pathways for ALK mutated lung cancer after pharmacologic treatment using tyrosine kinase inhibitors, utilizing a patient derived xenograft model.
  5. Martins-Filho SN, Weiss J, Pham NA, Li Q, Cabanero M, Fares A, Stewart EL, Shi R, Patel D, Pal P, McConnell J, Bradbury PA, Sacher AG, Leighl NB, Grindlay A, Allison F, Li M, Yasufuku K, Shepherd FA, Moghal N, Tsao MS, Liu G. EGFR-mutated lung adenocarcinomas from patients who progressed on EGFR-inhibitors show high engraftment rates in xenograft models. Lung Cancer. 2020 Jul;145:144-151. This study identified primary derived xenograft factors of engraftment in EGFR-mutated lung cancer post-treatment models
  6. Aggarwal R, Lam ACL, McGregor M, Menezes R, Hueniken K, Tateishi H, O’Kane GM, Tsao MS, Shepherd FA, Xu W, McInnis M, Schmidt H, Liu G,* Kavanagh J*. Outcomes of Long-term Interval Rescreening With Low-Dose Computed Tomography for Lung Cancer in Different Risk Cohorts. J Thoracic Oncol. 2019 Jun;14(6):1003-1011. Epub 2019 Feb 13. Impact Factor 13.4. This manuscript demonstrates that lung cancer screening by low dose CT scan can be stratified by the underlying risk stratification process into different risk cohorts. The clinical impact is to demonstrate that continued screening after 4-5.5 years is still relevant, especially when using a risk stratification model.
  7. Kavanagh J, Liu G, Menezes R, O’Kane GM, McGregor M, Tsao M, Shepherd FA, Schmidt H. Importance of Long-term Low-Dose CT Follow-up after Negative Findings at Previous Lung Cancer Screening. Radiology. 2018 Oct;289(1):218-224. Epub 2018 Jul 10. Impact Factor 469. The only long-term follow-up study of lung cancer screening by CT scan after a gap in screening of over 5 years that shows the importance of continued longitudinal screening. The NELSON and NLST randomized trials had screening periods of only 3-4 years. 2. Demonstrates the impact of Lung-CALIBRE. Impact on discipline: Currently the Canadian recommendations for screening are for only three annual screens in participants who are between the ages of 50 and 74 years of age. These data should re-open the national discussion to consider screening over additional years.
  8. Ren J, Yang W, Su J, Ren X, Fazelzad R, Albert T, Habbous S, Goldstein DP, de Almeida JR, Hansen A, Jang R, Bratman SV, Hope A, Chen R, Wang J, Xu Y, Cheng D, Zhao Y, Xu W, Liu G. Human papillomavirus and p16 immunostaining, prevalence and prognosis of squamous carcinoma of unknown primary in the head and neck region. Int J Cancer. 2019 Sep 15;145(6):1465-1474. This study utilized our Princess Margaret Head and Neck Translational Study data, demonstrating the importance of HPV p16 staining in unknown primary cancers in the head and neck cancer region.
  9. Habbous S, Chu KP, Lau H, Schorr M, Belayneh M, Ha MN, Murray S, O’Sullivan B, Huang SH, Snow S, Parliament M, Hao D, Cheung WY, Xu W, Liu G. Human papillomavirus in oropharyngeal cancer in Canada: analysis of 5 comprehensive cancer centres using multiple imputation. CMAJ. 2017 Aug;189(32):E1030-E1040. Impact Factor 7.7 (Trainee publication, Trainee publication, GL supervised SH, KPC). Cross-Canada study of oropharyngeal cancer and Human Papilloma Virus status, that.highlighted the impact of the massive increase of HPV as a cause of head and neck cancer in Canada. The 2016 report on Canadian Cancer Statistics by the Canadian Cancer Society devoted a special section on HPV-related disease. Because there were no registry data on the proportion of oropharyngeal cancers that were HPV positive, the statistics guide made an assumption that all new oropharyngeal cancers were HPV positive. With the publication of our manuscript, we have demonstrated that this may be an underestimation of the rate of increase of HPV within oropharyngeal cancers in Canada. We demonstrated that the proportion of oropharyngeal cancers that were HPV positive continued to rise even in recent years; thus, the proportion of HPV-negative tumours were falling, which had not been taken into account statistically. This underestimation can have significant effects on resource allocation for long term morbidity and survivorship issues (e.g. HPV-positive cancer patients have significantly improved overall survival than HPV-negative cancer patients, but are often left with debilitating permanent swallowing and scarring issues that require health care resources).
  10. Liu, G, Cuffe S, Liang S, Azad AK, Cheng L, Brhane Y, Qiu X, Cescon D, Bruce J, Chen Z, Cheng D, Patel D, Tse BC, Laurie SA, Goss G, Leighl NB, Hung R, Bradbury PA, Seymour L, Shepherd FA, Tsao MS, Chen BE, Xu W, Reisman DN. BRM Promoter Polymorphisms and Survival of Advanced Non-Small Cell Lung Cancer Patients in the Princess Margaret Cohort and CCTG BR.24 trial. Clin Cancer Res. 2017 May;23(10):2460-2470. This study demonstrates the synergistic utilization of both clinical trial and observational data in the AMPPEL laboratory. This is the first study to have replication datasets showing the prognostic benefit of BRM polymorphisms. The magnitude of prognostic impact of this polymorphism is in the same range as that of clinical staging.
  11. Liu G, Tu D, Lewis M, Cheng D, Sullivan LA, Chen Z, Morgen E, Simes J, Price TJ, Tebbutt NC, Shapiro JD, Jeffery GM, Mellor JD, Mikeska T, Virk S, Shepherd L, Jonker D, O’Callaghan C, Zalcberg JR, Karapetis CS, Dobrovic A. Fc-y-Receptor Polymorphisms, Cetuximab Therapy, and Survival in the NCIC CTG CO.17 Trial of Colorectal Cancer Survival. Clin Cancer Res. 2016 May;22(10):2435-44. AACR Press Release. Impact Factor 10.2 (Trainee publication, GL supervised ZC, EM, and DC.). This study establishes an important role of the FCGR2 polymorphism for cetuximab therapy in colorectal cancer, and is an important step in demonstrating the role of germline polymorphisms in cancer drug selection for adult solid cancers.

Graduate Students

Khadija Jafarova