SESSION TYPE: Lung Cancer Posters I
PRESENTED ON: Wednesday, October 24, 2012 at 01:30 PM - 01:30 PM
PURPOSE: To evaluate risk assessment for lung cancer (LC) using demographic risk (DR) factors and EarlyCDT®-Lung.
METHODS: EarlyCDT-Lung is used by clinicians as an aid to early detection of LC in a high-risk population. The original two-stratum (positive/negative) EarlyCDT-Lung test for cancer risk (based around a single cutoff) optimized sensitivity for a specificity of ≥90%. Introduction of three sets of test cut-offs (high, middle, low) extended the test to four-strata (high-positive/moderate-positive/moderate-negative/low-negative). Validated in a case-control setting, the test was applied to the clinical setting (i.e., 811 normals and 36 LCs). DR was calculated using a modified version of the Spitz model where age, sex and smoking history were combined in a logistic regression model; the risk estimate can be used as a continuous variable or categorised into two- and four-risk strata. Performance was calculated separately for DR and EarlyCDT-Lung, and combined.
RESULTS: For the two-stratum test, DR and EarlyCDT-Lung were significant independent predictors of cancer risk (DR p<0.001, EarlyCDT-Lung p<0.001, interaction p>0.20). DR as a continuous or categorised variable gave similar results. There was no correlation between DR and EarlyCDT-Lung positivity (p=0.90). DR alone gave a specificity/sensitivity of 89.9%/27.8%, versus 91.4%/36.1% for EarlyCDT-Lung. This became 82.1%/55.6% for a combined test where either DR or EarlyCDT-Lung was positive. Results were similar for the four-stratum with no significant correlation between DR and EarlyCDT-Lung (p=0.50). In the highest-risk stratum, DR gave specificity/sensitivity of 94.2%/19.4% versus 98.2%/19.4% for EarlyCDT-Lung. This became 92.4%/36.1% for the combined test where either DR or EarlyCDT-Lung was positive. The new PPV (1 in 5.8) was midway between DR (1 in 7.7) and EarlyCDT-Lung (1 in 3.1) separately, while the new NPV was improved over both.
CONCLUSIONS: Using a two-stratum approach considering a combined positive DR or EarlyCDT-Lung test, specificity was reduced by 7.8% versus DR alone, but sensitivity was increased by 27.8%. Using a four-stratum approach, whilst specificity in the highest stratum was reduced by 1.8% over DR alone, the sensitivity was increased by 16.7%.
CLINICAL IMPLICATIONS: Adding EarlyCDT-Lung to a DR classifier can significantly improve the overall risk assessment performance.
DISCLOSURE: Graham Healey: Employee: Graham Healey is an employee of Oncimmune Ltd.
Laura Peek: Employee: Laura Peek is an employee of Oncimmune USA LLC.
Lynn Fredericks: Consultant fee, speaker bureau, advisory committee, etc.: Lynn Fredericks received consulting fees from Oncimmune USA LLC.
Geoffrey Hamilton-Fairley: Shareholder: Geoffrey Hamilton-Fairley is a shareholder of Oncimmune Ltd.
John Robertson: Shareholder: John Robertson is a shareholder of Oncimmune Ltd., Consultant fee, speaker bureau, advisory committee, etc.: John Robertson received consultant fees from Oncimmune.
No Product/Research Disclosure InformationOncimmune Ltd., Nottingham, United Kingdom