Ever planned RCT w/ PFS and OS as formal endpoints?

Ever simulated PFS and OS for same set of patients?

In both cases: How did you model the fact that PFS and OS are highly dependent? E.g. we know that PFS <= OS and P(PFS = OS) > 0.

A (long) thread - but it' s worth it!

1/n
We are very used to model dependency of an endpoint over time (= group-sequential designs) or over nested populations (= enrichment designs), b/c then it's not difficult to write down correlation between test statistics.

However, typically in trial planning PFS and OS...

2/n
are treated as if they were independent and proportional hazards (PH) is assumed for both.

We propose to model dependency between PFS and OS using an illness-death multistate model (IDM), building on the work of Meller et al (2019) doi.org/10.1002/sim.82…

3/n
Making assumptions on 6 transition-specific hazards (3 in control + 3 in tmt arm) we *induce* survival functions for PFS and OS, and parsimoniously model their dependence in mathematically appropriate way.

Turns out that for constant transition-specific hazards...

4/n
we always have PH for PFS, but that for OS hazards are only proportional under *very restrictive* assumptions that are typically not met in a clinical trial.

--> So, assuming PH for PFS and OS independently when planning RCT is too simplistic! <--

What is the implication?

5/n
Accounting for dependency can increase power. In one of our scenarios we can reduce number of necessary OS events to perform final analysis at from 770 to 600.

However, since shape of OS survival functions is induced through assumption on transition-specific hazards...

6/n
modeling based on IDM can also *reduce* power for OS. So, making simplifying assumption of independence of PFS and OS can lead to RCT that is underpowered for OS.

Our paper describes all this and illustrates how RCT can be planned based on simulating from an IDM.

7/n
R package provides easily accessible implementation.

Curious to hear comments!

#oncology #multistate #survivalanalysis #biostatistics

Link to paper: arxiv.org/abs/2301.10059

Link to R package: cran.r-project.org/web/packages/s…

The end.
Link to R package on github: github.com/insightsengine…

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Kaspar Rufibach

Kaspar Rufibach Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @numbersman77

Mar 10, 2021
1/n

Stop the abuse: plea for more principled approach to analysis of AEs.

A (long) thread. Please RT.

@tim_friede @ADAlthousePhD @VR_Cornelius @RachPips @statsepi

How are AEs typically reported in an RCT? Along the lines of the table below.

From: nejm.org/doi/full/10.10…
2/n

So what is the purpose of such tables? I see two:

1) "Signal detection" --> find AEs that differ between arms.

2) Estimate true probability of an AE of interest, P(AE).

I will focus on 2) below.
3/n Assume we are interested in P(AE) - this is our estimand.

What do proportions in above table estimate? The incidence proportion:

IP = P(AE happening in [0, t] and that this AE is observed).
Read 12 tweets
Mar 10, 2021
1/n

Clinical trials in patients with hematological malignancies often present unique challenges for trial design due to complexity of tmt options and existence of potential curative but highly risky procedures, for example, SCT or tmt sequence across different phases.
2/n

I am very happy to see our paper

Estimands in hematologic oncology trials

now online at Pharm Stat: doi.org/10.1002/pst.21…

The paper illustrates how to apply the estimand framework in hematological clinical trials and how the estimand framework can address...
3/n

... potential difficulties in trial result interpretation. Three phase 3 RCTs are used to illustrate different scientific questions and the consequences of the estimand choice for

- trial design,
- data collection,
- analysis,
- and interpretation.
Read 4 tweets
Nov 11, 2020
1) Why do we run group-sequential trials in drug development?

2) How does the effect we power at relate to the effect(s) needed to stop?

3) Is stopping such a trial early (like the one for the Pfizer vaccine) "cheating"?
4) If stopping early the effect estimate may be biased. Is this an issue?

5) What happens operationally if a trial stops early?
A thread from a pharma statistician who has developed, run, analyzed, and taught courses about trials with such designs.

@statsepi @lakens @MaartenvSmeden @stevesphd @ADAlthousePhD @DominicMagirr @thomas_jaki
Read 20 tweets
Nov 9, 2020
1/n I was asked to give an industry statistician's view on A) below. Disclaimer: I do not know about the exact regulations (which might also be region-specific). What I offer is a 1st hand experience of what happens around a trial stopping early. A thread.
3/n I was the trial statistician for this trial: nejm.org/doi/full/10.10…

Planned efficacy interim after ~245 PFS events.
1) Data with sponsor, except for tmt assignment (=rando codes).
2) Rando codes with IxRS vendor.
3) Indep. Stat. Reporting Group (ISRG) coordinates.
Read 11 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(