It has been reported that BMP4 is overexpressed in melanoma cell line and lung cancer. BMP4 plays an important role in bone metastasis of BAY 73-4506 datasheet prostate cancer [16], and BMP4 overexpression inhibits proliferation and induces apoptosis in many cancer cell line [15, 17]. This study also showed that BMP-4 expression was lower in primary tumors. Bone metastasis of lung cancer is a dynamic process involving bone resorption resulted from tumor cell-induced osteolysis and bone formation due to osteoblasts. This study didn’t show PTHrP and IGF-1R overexpression in NSCLC tissue related NSCLC bone metastasis. PTHrP is required

for colony of bone metastasis of cancer cells. It is a cytokine produced by the metastatic cancer cells [18]. But Henderson [19] had demonstrated that bone metastases that do not express PTHrP in primary breast cancer begin to do so when they reach bone. The bone microenvironment seems to provide what is needed for the breast cancer cells to produce PTHrP, even if they could not produce it before they got there. This study demonstrated that PTHrP was expressed only in 66.67% of the primary tumors. Breast cancer overexpress IGF-1R through promoting proliferation and reducing apoptosis to increase bone metastasis [20], the effects of IGF-1R have been confirmed in bone metastasis of prostate cancer [21] but the role of IGF-1R overexpress in NSCLC bone metastasis is

not clear, it still needs to be further investigated. Multivariate Logistic regression GSK1210151A datasheet Epothilone B (EPO906, Patupilone) has successfully established a model for predicting the risk of bone metastasis

in resected Stage III NSCLC: logit (P) = − 2.538 +2.808 CXCR4 +1.629 BSP +0.846 OPG-2.939BMP4. The area under the ROC curve was 81.5%. When P = 0.408, the sensitivity was up to 71%, specificity 70%. The model has successfully validated in 40 patients with resected stage III NSCLC from 2007 to 2009 whole cohort in clinic trial, who were followed up for 3 years. The model showed a sensitivity of 85.7% and specificity of 66.7%, Kappa: 0.618. The results are highly consistent. The model based on bone metastasis-associated biomarkers established in this study is useful in providing rationale for the screening, intervention and targeted therapy of bone metastasis in lung cancer. Although the results are interesting, the limitations of this study should be acknowledged. The patients enrolled into the prediction model and validation model were whole cohort of completed resected stage III patients, not including patients from other groups. Therefore, there might be selection bias in the model construction and results interpretation. The results might be more suitable to clinically stage III patients. Any generalization to other stages should not be expected. In the future, a bigger study with larger sample size with different stages, could help more objectively judge the value of this prediction model.