Lung cancer is a malignant lung tumor which is characterized by uncontrolled cell growth in tissues of the lung. Lung cancers typically start in the cells lining the bronchi and parts of the lung such as the bronchioles or alveoli and most cancers that start in the lung, known as primary lung cancers, are carcinomas which originate from epithelial cells. The two main types of lung cancer are non-small cell lung cancer and small cell lung cancer. The most common symptoms include a cough, breathlessness and weight loss. Lung cancer is the second most common cancer worldwide and is the leading cause of cancer deaths worldwide. Although lung cancer can occur in people who have never smoked, people who smoke have the greatest risk of lung cancer and the risk also increases with age.
Figure 1. The volcano plot shows the adjusted p-value compared to the difference in average protein expression (NPX) for all proteins in lung cancer compared to all other cancers. The lollipop plot shows the top 10 most important proteins resulting from the cancer prediction model with importance scores ranging between 0 to 100.
Pan-cancer protein panel
A total of 18 proteins were selected to predict patients with lung cancer (Table 1). The top protein (CEACAM5) is a plasma membrane protein expressed on epithelial cells, which has been shown to stimulate the progression of non-small-cell lung cancer by promoting cell proliferation and migration (Zhang X et al. (2020)). The second important protein is a matrix protein (MMP12) involved in tissue injury and has been shown to be upregulated in blood from lung cancer patients (Xu F et al. (2021)), and is associated with non-durable response of this cancer. The two cytokines IL6 and CXCL8 are shown to have higher plasma levels in our study, a finding supported by Dagnino et al (Dagnino S et al. (2021)). However, these two proteins were not selected by the prediction model to be included in the final pan-cancer panel, most likely since our model is based on a pan-cancer concept, in which the protein profiles of other cancers also are taken into account. In addition, another 16 proteins are selected by the prediction model, including PRDX5 and FKBP1B shared with colorectal cancer. The 18 panel proteins are able to predict patients with lung cancer with high specificity and sensitivity, although the performance is lower than most of the other cancers, despite the fact that a large number of proteins (n=18) are used in the model.