Oral Presentation Clinical Oncology Society of Australia Annual Scientific Meeting 2017

Patterns of multiple immunohistochemical biomarker expression predict survival after resection in pancreatic ductal adenocarcinoma: a retrospective cohort study and two-step cluster analysis. (#48)

Christopher B Nahm 1 2 3 4 , John Turchini 5 , Elizabeth Millar 2 , Malinda Itchins 2 3 , Emily Colvin 2 3 , Viive Howell 2 3 , Nick Pavlakis 2 3 , Stephen Clarke 2 3 , Thomas J Hugh 1 3 , Ross Smith 1 6 , Anthony J Gill 5 , Jaswinder S Samra 1 3 , Anubhav Mittal 1 3
  1. Upper Gastrointestinal Surgical Unit, Royal North Shore Hospital, St. Leonards, NSW, Australia
  2. Bill Walsh Translational Cancer Research Laboratory, Kolling Institute - University of Sydney, Sydney, NSW, Australia
  3. Sydney Medical School (Northern), University of Sydney, Sydney, NSW, Australia
  4. Sydney Vital, Sydney, NSW, Australia
  5. Cancer Diagnosis and Pathology, Kolling Institute - University of Sydney, Sydney, NSW, Australia
  6. Cancer Surgery, Kolling Institute - University of Sydney, Sydney, NSW, Australia


We previously published a systematic review of tumour biomarkers shown individually to have prognostic significance in pancreatic ductal adenocarcinoma (PDAC) [1], and identified a panel of those with the highest level of evidence that could be immunohistochemically detected. These were p53, CDKN2A, MUC16, S100A4, FOXC1, EGFR, Mesothelin, smad4, CD24, and UPAR. We aimed to determine whether simultaneous assessment of these proteins would reveal patterns of biomarker expression that better predicted survival outcomes than when these proteins were assessed individually.


Patients who underwent PDAC resection from 1996 to 2016 were included for analysis. Tissue microarrays (TMA) of formalin fixed paraffin embedded pancreatic tumour specimens were constructed for all patients, stained by immunohistochemistry using antibodies targeted against all ten proteins, and scored for immunolabelling intensity and percentage of tumour cells stained by two blinded observers. Survival analysis was undertaken by Kaplan-Meier method and difference in survival outcomes were assessed by log rank method. Two-step cluster analysis was performed using biomarkers individually shown to be of prognostic significance to identify patterns of biomarker expression associated with different survival outcomes.


249 patients were included for analysis. Individual biomarker expression associated with shorter mOS included: S100A4 (18 vs 30mths, p=0.001), EGFR (13 vs 25 mths, p=0.043), mesothelin (13 vs 26 mths, p=0.001), Ca125 (17 vs 37mths, p<0.001), and FOXC1 (7 vs 25mths, p=0.003). Two-step cluster analysis revealed four distinct patterns of biomarker expression, each associated with different survival outcomes: (1) S100A4+ Ca125+ Mesothelin+ (mOS 12mths); (2) S100A4+ Ca125+ Mesothelin- (mOS 17mths); (3) S100A4+ Ca125- Mesothelin- (mOS 33mths); (4) S100A4- Ca125- Mesothelin- (mOS 40mths); log-rank p<0.001.


S100A4, Ca125 and Mesothelin immunohistochemistry reveals patterns of biomarker expression that predict survival after resection. The prognostic utility of these biomarkers in tandem is superior than either of these biomarkers assessed individually.

  1. 1. Petrushnko W, et al. HPB (Oxford). 2016;18(8):652-663.