One thing I don’t understand is why patients coming off drug due to toxicity need to be censored.they should still be followed on study for relapse/ survival .
The thing is that censoring rules are made in order to be sure that any progression/relapse event is occuring under the treatment under investigation. If this cannot be ascertained, the patient will be censored, even though the patient is found to progress after some time. That's why PFS and DFS are more prone to informative censoring. However, in OS, even though patients who dropped off should ideally be followed to capture their survival status, this is not always the case and informative censoring can also happen.
Below are the rules as described in the protocol of the CONTACT-2 trial. As one can see, if a patient misses two CT scans (which can happen if the patient feels too weak to come), and is later found to progress, the patient will be censored at the time of the prior CT scan. Also, if a patient starts a new therapy (because of disappointment or toxicity) before the event is captured, again, the patient will be censored.
"General censoring rules for the primary analysis of PFS are described below:
• Subjects who receive systemic NPACT (non-protocol anticancer therapy), nonprotocol radiation therapy (other than to bone), or surgery to resect target lesions before experiencing an event will be right censored at the date of the last tumor assessment prior to the date of initiation of subsequent therapy/surgery. If there is no such tumor assessment post randomization, the subject will be right censored on the date of randomization.
• Subjects who have not experienced an event (and are not otherwise censored) at the time of data cutoff will be right censored on the date of their last tumor assessment post
randomization that is on or prior to the data cutoff. If there is no such tumor assessment post randomization, the subject will be right censored on the date of randomization.
• Subjects who miss 2 or more scheduled tumor assessments followed by an event will be right censored on the date of their most-recent tumor assessment prior to the missing assessments. If there is no such tumor assessment post randomization, the subject will be right censored on the date of randomization."
One thing I don’t understand is why patients coming off drug due to toxicity need to be censored.they should still be followed on study for relapse/ survival .
I’ll be sure to read the protocol of the next manuscript on a phase 3 study I read to see if that information is in the statistics section.
Showing the numbers censored in the Kaplan Meier curves would be very useful, but it’s not often shown.
Yes, it's not often show, nor the "reasons" for censoring.
That's where digitizing curves is really great:
- they don't want to give you the data?
- no worries, you can have very good estimates minutes after the curves are published!
(PS : if you need a trial to be digitized and you don't know how to do it, feel free to send me an email, contact details are on the website)
Ok thanks for the explanation
Thank you for your interest.
The thing is that censoring rules are made in order to be sure that any progression/relapse event is occuring under the treatment under investigation. If this cannot be ascertained, the patient will be censored, even though the patient is found to progress after some time. That's why PFS and DFS are more prone to informative censoring. However, in OS, even though patients who dropped off should ideally be followed to capture their survival status, this is not always the case and informative censoring can also happen.
Below are the rules as described in the protocol of the CONTACT-2 trial. As one can see, if a patient misses two CT scans (which can happen if the patient feels too weak to come), and is later found to progress, the patient will be censored at the time of the prior CT scan. Also, if a patient starts a new therapy (because of disappointment or toxicity) before the event is captured, again, the patient will be censored.
The censoring rules may differ, potentially affecting estimates, see our work here : https://www.ejcancer.com/article/S0959-8049(24)00678-6/fulltext
Rules from CONTACT-2 protocol:
"General censoring rules for the primary analysis of PFS are described below:
• Subjects who receive systemic NPACT (non-protocol anticancer therapy), nonprotocol radiation therapy (other than to bone), or surgery to resect target lesions before experiencing an event will be right censored at the date of the last tumor assessment prior to the date of initiation of subsequent therapy/surgery. If there is no such tumor assessment post randomization, the subject will be right censored on the date of randomization.
• Subjects who have not experienced an event (and are not otherwise censored) at the time of data cutoff will be right censored on the date of their last tumor assessment post
randomization that is on or prior to the data cutoff. If there is no such tumor assessment post randomization, the subject will be right censored on the date of randomization.
• Subjects who miss 2 or more scheduled tumor assessments followed by an event will be right censored on the date of their most-recent tumor assessment prior to the missing assessments. If there is no such tumor assessment post randomization, the subject will be right censored on the date of randomization."