Development of a radiosensitivity gene signature for patients with soft tissue sarcoma

Abstract

Adjuvant radiotherapy is an important clinical treatment option for the majority of sarcomas. The motivation of current study is to identify a gene signature and to predict radiosensitive patients who are most likely to benefit from radiotherapy. Using the public available data of soft tissue sarcoma from The Cancer Genome Atlas, we developed a cross-validation procedure for identifying a gene signature and predicting radiosensitive patients through. The result showed that the predicted radiosensitive patients who received radiotherapy had a significantly better survival with a reduced rate of new tumor event and disease progression. Strata analysis showed that the predicted radiosensitive patients had significantly better survival under radiotherapy independent of histologic types. A hierarchical cluster analysis was used to validate the gene signature, and the results showed the predicted sensitivity for each patient well matched the results from cluster analysis. Together, we demonstrate a radiosensitive molecular signature that can be potentially used for identifying radiosensitive patients with sarcoma.

Keywords: gene signature; radio-sensitivity; radiotherapy; sarcoma; survival prediction.

Figure 1. The survival curves under radiotherapy and nonradiotherapy for both predicted radiosensitive (RS) and nonradiosensitive (NRS) patients

The colored areas denote the 95% confidence intervals for survival rate.

Figure 2. The HR estimation for radiotherapy (RT) verse nonradiotherapy (NRT) and radiosensitive (RS) verse nonradiosensitive (NRS)

These p values here are estimated by wald test. The adjusted factors are gender, age, chemotherapy, histologic type, and residual rumor (the significant factor in multivariable analysis).

Figure 3. The comparisons among different groups for the rate of new tumor event and progressive disease

The rates for different groups are compared by Fisher exact test. RT: radiotherapy; NRT: nonradiotherapy; RS: radiosensitive; NRS: nonradiosensitive.

Figure 4. The survival curves under radiotherapy and nonradiotherapy for predicted radiosensitive and nonradiosensitive patients with different histologic types

For DLS, MFS, MPNST, and SS, the proportion of predicted radiosensitive and nonradiosensitive are very similar and the sample sizes are also small for these groups. Therefore, they are combined together for logrank test (Figure 4e and 4f).

Figure 5. Hierarchical clustering analysis

Hierarchical clustering was used to determine the expression pattern of selected 26 genes. The top blue and yellow bands denote the predicted radiosensitive and nonradiosensitive patients, respectively. Totally, 83 out of 101 predicted radiosensitive patients are classed at the left branch, and 126 out of 152 predicted nonradiosensitive patients are classed at the right branch.