Cox Regression Survival Analysis with Only Tumor Data?

Question

[menon_ankita]

Hello,

I am a student conducting a gene expression analysis project using the gene expression data and the event free survival time for 249 cancer patients. I am applying the Cox Regression method to this data. However, I only have expression data from the tumor tissue (I do not have "normal" or "control" expression values).

I was under the impression that this would not be an issue and that I would still be able to conduct the survival analysis because I have the event-free survival times, state (dead or alive), and gene expression tumor data for all these patients and I am just looking for genes whose expression seems to affect the survival time.

However, as time goes on I am feeling a bit more doubtful and I would really appreciate it if someone would be able to tell me whether it is still alright to conduct a survival analysis as I have been doing without "control" data or two different expression groups (tumor and normal).

Thank you very much!

~Ankita

Answer 1

[dsull]

If you're assessing how gene expression in tumor affects survival (e.g. tumors with high expression of gene X are those that have worse prognosis), then just tumor data is fine.

You only need normal if you're actually interested in the normal tissue gene expression (e.g. if you want normal tissue gene expression or gene expression fold change to be an explanatory variable).

[menon_ankita]

@dsull Okay, thank you!!