Virtual Library
Start Your Search
T. Feinberg
Author of
-
+
P2.06 - Poster Session/ Screening and Early Detection (ID 219)
- Event: WCLC 2015
- Type: Poster
- Track: Screening and Early Detection
- Presentations: 1
- Moderators:
- Coordinates: 9/08/2015, 09:30 - 17:00, Exhibit Hall (Hall B+C)
-
+
P2.06-009 - Oral Glucose Tolerance Test as a Diagnostic Tool in Lung Cancer (ID 1040)
09:30 - 09:30 | Author(s): T. Feinberg
- Abstract
Background:
Previous studies have demonstrated that volatile organic compounds (VOCs) in exhaled breath can distinguish between healthy and affected individuals, and can even discern between SCLC and NSCLC and within the subtypes of lung cancer (LC) and its mutations status. The current study assessed the differences in glucose metabolism on the volatile signature in LC through an oral glucose tolerance test (OGTT).
Methods:
This cohort included forty participants (22 control participants whom are at high risk for LC, 18 study participants whom have active, naïve lung cancer). Pre-OGTT and Post-OGTT blood glucose levels and exhaled breath samples were measured with a lay period of 90 minutes. A proton transfer reaction mass spectrometer (PTR MS) detected and measured the VOCs. The data was then analyzed using a series of feature selection methods to identify relevant inputs for multilayer perceptron (MLP) models to distinguish LC patients from controls, with and without the consideration of the glucose effect.
Results:
The feature selection method “infogain” revealed a combination of 14 masses (m/e) that were different between the two groups without considering the glucose effect. All the average values of these masses were higher in the LC group except for m/e 52, which was higher in the high-risk group. These 14 masses enable us to distinguish between the two groups with an average accuracy of 91.67% for three internal validation tests of a MLP (threshold set at 0.45). The analysis of the effect of glucose revealed that several m/e increased more for the control group whereas others increased more for the LC group. Moreover, three feature selections, each with a different combination of 4 masses, allowed the design of three MLPs that yielded 90% for K-fold cross-validation accuracy. Figure 1
Conclusion:
This study showed that breath analysis could discriminate between the high-risk and LC group. Furthermore, it demonstrated that glucose metabolism leaves a unique VOC pattern in the LC group. These findings may assist in the development of a non-invasive screening method for lung cancer.
