Lung Cancer: Novel Methods for Lung Cancer Diagnosis |

A Blood Based Non-small Cell Lung Cancer Diagnostic FREE TO VIEW

Andrew Kossenkov, PhD; Qureshi Rehman, PhD; Noor Dawany, PhD; Priyankara Wickramasinghe, PhD; Michael Showe, PhD; Margie Clapper, PhD; Gerard Criner; Jun-Chieh Tsay, MD; Harvey Pass, MD; Sai Yendamuri, MD; Thomas Bauer, MD; Anil Vachani, MD; William Rom, MD; Louise Showe, PhD
Author and Funding Information

The Wistar Institute, Philadelphia, PA

Copyright 2016, American College of Chest Physicians. All Rights Reserved.

Chest. 2016;150(4_S):734A. doi:10.1016/j.chest.2016.08.829
Text Size: A A A
Published online

SESSION TITLE: Novel Methods for Lung Cancer Diagnosis

SESSION TYPE: Original Investigation Slide

PRESENTED ON: Tuesday, October 25, 2016 at 08:45 AM - 09:45 AM

PURPOSE: To develop a non-invasive PAXgene derived NSCLC diagnostic that focuses on distinguishing primarily early stage lung cancer patients from “at risk” control groups of smokers and ex-smokers with lung nodules of various sizes and types, and to implement that on a platform appropriate for clinical testing.

METHODS: Peripheral blood samples were all collected in PAXgene RNA stabilizations tubes and RNA extracted according to the manufacturer. Samples were tested on a Nanostring nCounterTM against a custom panel of 559 probes: 432 were selected based on previous microarray data , 107 probes were selected from Nanostring studies and 20 were housekeeping genes. We analyzed 610 PAXgene RNA samples (278 cancers, 332 controls) derived from 5 collection sites. For QC, a Universal RNA standard (Agilent) was included in each batch of 36 samples tested. Probe expression values were normalized using the 20 housekeeping genes as well as spike-in positive and negative controls supplied by Nanostring. Z-scores were calculated for probe count values and served as the input to a Support Vector Machine (SVM) classifier using a polynomial kernel. Classification performance was evaluated by 10-fold cross-validation of the samples.

RESULTS: The classifier developed on all the samples showed a ROC-AUC of 0.81. With the Sensitivity set at 90%, the specificity is 46%. When nodule classification accuracy is assessed by size without using a specific threshold for sensitivity, we find that as nodules size and the cancer risk factor increases, the number of benign nodules classified as cancer increases. In this analysis nodules ≤8mm were correctly classified 88.9% of the time, for nodules >8,≤12mm accuracy was 75%, for nodules >12,≤16mm accuracy was 68%, for nodules >16mm accuracy is 53.6%. Since classification accuracy was found to be negatively correlated with benign nodule size, we reanalyzed the data using only nodules <10mm (n=244) and sensitivity fixed at 90%, in this case the specificity rises to 54% and the ROC-AUC to 0.85. For larger nodules, >10mm (n=88) the specificity drops to 24% and the ROC-AUC drops to 0.71.

CONCLUSIONS: These studies shows the we have successfully transitioned our microarruy blood based diagnostic that distinguishes NSCLCs from benign lung nodules, to the more stable Nanostring platform with simnilar accuracies. We also show that the 559 probe assay we developed can provide a preliminary stratification of patients by nodule size and thus cancer risk and can potentially identify those higher risk patients in need of further testing by LDCT, X-Ray, bronchoscopy or biopsy.

CLINICAL IMPLICATIONS: Because of the significant advantage of diagnosing NSCLC at an early stage, there is a need for the development of additional methods that do not require frequent exposure to radiation and that will allow monitoring of at risk individuals with heavy smoking histories and the presence of suspicious lung nodules. An accurate companion diagnostic that can help minimize radiation exposure, is minimally invasive and simple in application would be an important adjunct to lung patient care.

DISCLOSURE: Louise Showe: Grant monies (from industry related sources): Sponsored Research Agreement The following authors have nothing to disclose: Andrew Kossenkov, Qureshi Rehman, Noor Dawany, Priyankara Wickramasinghe, Michael Showe, Margie Clapper, Gerard Criner, Jun-Chieh Tsay, Harvey Pass, Sai Yendamuri, Thomas Bauer, Anil Vachani, William Rom

No Product/Research Disclosure Information




Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543