question about TCGA survival data

Question

[Johnny R]

Many TCGA samples have three rows in their clinical data that may be of interest for survival analysis: days to death, days to last follow-up, and days to sample procurement. My questions are, in this data, what does day 0 refer to, and what is sample procurement?

For instance, consider SKCM patient "tcga-fs-a1zc". His days to death value is "10870", his days to last follow up value is "NA", and his days to sample procurement value is "10158". If I were plotting his information on a Kaplan Meier plot, would his survival time be 10870, or 712? (10870-10158)

Thanks!!

[js1234]

Does anyone know to answer the question "in this data, what does day 0 refer to"?

At the TCGA dictionary i saw

"days_to_last_followup / last_contact_days_to – Time interval from the date of last follow up to the date of initial pathologic diagnosis, represented as a calculated number of days."

"death_days_to / days_to_death – Number of days between the date used for index and the date from a person's date of death represented as a calculated number of days."

Are Date for index and date of initial pathologic the same? In my survival annalysis i want to use the "death_days_to" as the event and "days_to_last_followup" for censoring, and i don't know if this is correct since i suspect that the initial time point, or "day 0" for both cases is different

Thanks

Answer 1

[naranoth]

Survival time is 10870 days. As the columns vary between TCGA studies I've found http://www.cbioportal.org to be a better resource as the clinical data is in a more user-friendly format with better column names (and it is pulled directly from the latest Firehose version). Getting the info using their R package 'cgdsr' is really easy or just download it from: http://www.cbioportal.org/study?id=skcm_tcga#summary

[Johnny R]

But... survival time since what? The sample hasn't sat in a freezer for 10,870 days. That has to be the survival time for the initial diagnosis. But if I'm doing any sort of biomarker study, it has to be from when the sample was taken... and is that 712?

[naranoth]

https://www.genome.gov/multimedia/slides/tcga3/25_watson.pdf

Look at slide 22. It seems likely that 712 is the survival time since sample procurement/surgical resection of the metastasis...this is supported by the difference in the disease free and overall survival as well. Tough to be certain though.