Thread by @ManniMD1

How I view PFS and OS and discuss it to patients-

Using myeloma as an example with references.

An educational thread!

/1

#mmsm

I explain what PFS is.

PFS is time from start of treatment to either disease progression or death.

I explain how we don’t have great data from trials on how relapse happens, but retrospective data indicates majority of progressions are biochemical.

onlinelibrary.wiley.com/doi/full/10.10…

I describe that for newly diagnosed MM, PFS does not correlate well with OS due to effective salvage therapies.

I mention attrition here as well, but at least in trials despite well-documented attrition, PFS still hasn’t correlated to OS.

onlinelibrary.wiley.com/doi/full/10.11…

I describe that for newly diagnosed MM, to show whether a new intervention today that improves a surrogate like MRD or PFS also improves OS will take a long time.

And thus-I weigh each surrogate in context, analyzing how much additional toxicity pursuit of the surrogate brings

For example, when analyzing the triplet versus quad question in MM, the addition of cd38 therapy improves short-term surrogate outcomes without much additional toxicity and without worsening of quality of life.

ashpublications.org/blood/article/…

Thus when discussing 4 versus 3 drug induction- even if on long term follow-up, we find out that overall survival is no better, I would not have brought additional toxicity (on average) to my patient.

(Costs to society different but very important issue).

Another key thing that I discuss is opportunity costs of improving the surrogate.

In this case, the addition of fourth drug does not add extra visits. Patients are coming in anyways for bortezomib injections.

What if drug improves PFS but requires months/years of additional visits relative to standard?And OS is unknown.

This is serious opportunity cost and needs to be transparently discussed.

Case in example: Long-term PI + len maintenance versus just len.

thelancet.com/journals/lanon…

In both these examples, OS was unknown.

What if we know PFS is better but on long term follow-up OS is same.

In these cases, OS isn’t “confounded” but post-protocol therapy, it simply means that giving the treatment at the time you gave it didn’t translate to increased survival

Example here- ixazomib/len/dex versus len/dex at first relapse.

Better PFS with ixazomib/len/dex but same OS.

Much worse toxicity with ixazomib.

We wrote a paper on this:

ascopubs.org/doi/full/10.12…

If on long term follow-up with appropriate post protocol therapy, an intervention improved PFS but not OS- I discuss this with the patient, highlighting the side effects, opportunity cost.

I personally wouldn’t choose such an intervention. More toxicity for no extra longevity

Now what about PFS better, but worse OS.

This situation happens too.

These are failed treatments, and highlight the issues with using surrogates.

We wrote about this here, from a patient perspective with @MyelomaAmateur

thelancet.com/pdfs/journals/…

So what have we learnt?

Better PFS, OS unknown. Look at strength of surrogacy and costs of pursuing additional PFS. May consider depending on context.

Better PFS, OS same. The intervention wasn’t sufficient to improve longevity and came at a cost.

Better PFS, OS worse. Avoid.

I think of this as Common Sense.

But in the literature and lay media, PFS is over-blown and hyped up. It doesn’t necessarily translate to survival. It doesn’t translate to quality of life as shown by @oncology_bg

pubmed.ncbi.nlm.nih.gov/30374970/

What we must also recognize, that imbalances in censoring can lead to a completely inert drug showing a PFS benefit, and that any PFS result should be taken with a grain of salt due to this reason.

jamanetwork.com/journals/jamao…

I recognize there are different ways individuals handle this situation, and how patients perceive this information.

Many nuances cannot be captured due to Twitter format either.

Thanks for reading!

/END