How to Interpret p-value from multi-curve Kaplan-Meier Graph
1
0
Entering edit mode
4 months ago
andrewnrdoig ▴ 10

I've generated a few KM graphs from TCGA data. I am unsure of what the p-value provided by the function and the graphed chart indicates. I just looked through the survival.pdf from cran but am still unsure.

I currently believe that the p-value indicates with a 0.95 confidence that there is statistically significant variance of survival probabilities over time between the curves.

Here is a representative graph: Example graph of multicuve km plot iwth p-value

The data is being run through the following code to generate this graph:

#Generate survival object
survival_object <- Surv(time = cesc_survival_data$survival_time, event = cesc_survival_data$survival_status)
#Fit the survival data to a curve that is defined by the quartiles of zscore SLC2A2
quartiles_SLC2A2_expression_survival_curve <- survfit(survival_object ~ zscore_value_SLC2A2_quartile, data = cesc_survival_data)
#Generate a graph of survival object/expression quartile curve of zscore SLC2A2 median quartiles
ggsurvplot(quartiles_SLC2A2_expression_survival_curve, data = cesc_survival_data, 
           #conf.int = TRUE,
           pval = TRUE,
           fun = "pct",
           risk.table = TRUE,
           size = 1,
           linetype = "strata",
           #palette = c("#E7B800", "#2E9FDF"),
           legend = "bottom",
           legend.title = "Expression Quartiles",
           legend.labs = c("Low Exp.", "Med. Low Exp.", "Med. High Exp.", "High Exp."),
           caption = "SLC2A2 Expression Quartile Survival Curve"
)

If it is helpful the complete code can be found at: https://github.com/andrewnrdoig/RIndyResearch

R survival analysis kaplan-meier graph • 584 views
ADD COMMENT
2
Entering edit mode
4 months ago

The default p-value that is calculated is the log-rank p-value. It is testing the null hypothesis that each of your strata (survival groups) has the same survival probability.

I show this in a previous answer, here: A: survfit(Surv()) P-value interpretation for 3 survival curves?

In my other answer, I also show how you can use the coxph() function to generate p-values for each pairwise stratum / curve.

Looking at your data, it p-value is not statistically significant and your data-points are sparse after ~75 days.

Kevin

ADD COMMENT
1
Entering edit mode

Thanks Kevin, I appreciate the links also!

ADD REPLY

Login before adding your answer.

Traffic: 1863 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6