How to do a survival analysis on tumour data when there are multiple tumours per patient?

Question

[psyc_biostars]

I have tumour data for a group of individuals. I want to do a survival analysis, using "number of lymphocytes per tumour" as the feature, ie. it will be split by median into two groups, low and high (please note this is not my actual feature, but just an example).

Eg.

PATIENTID    TUMOR_SITE    VALUE    TIME    STATUS
Patient 1    Kidney            23    172        1
Patient 2    Brain             18    280        0
Patient 3    Liver             32    185        1

So I would then do a survival analysis on this data, with the grouping of VALUE either "high" or "low".

My question is - how do I do the survival analysis if there are multiple tumours per patient? eg:

PATIENTID    TUMOR_SITE    VALUE    TIME    STATUS
Patient 1    Kidney           23    172        1
Patient 1    Lymph            13    280        0
Patient 1    Skin             11    280        0
Patient 2    Brain            18    280        0
Patient 3    Kidney           34    185        1
Patient 3    Liver            32    185        1

[Hamid Ghaedi]

You should remove those patients from your analysis. Suppose your research question is how would be the association of tumor site with survival probability? (right?) For those patients you have more than one tumor , survival probability is associated with both tumors. Actually there is no way to measure each tumor association with survival probabilty seperately. I