NCCN adds AV and AVO as Preferred Regimens

[SeymourB]

In February of this year (2025), the NCCN added AV (Acalabrutinib/Calquence plus Venetoclax) and AVO (AV plus Obinutuzumab/Gazva) for first line treatment to its Clinical Guidelines for CLL, page CSLL-D, despite the fact that these 2 treatment regimes are not yet approved by the FDA or any other regulatory body in the world. They labeled it as a Preferred Treatment with highest level evidence , category 1.

As we saw with Zanubrutinib/Brukinsa, NCCN recommendation sometimes preceeds and may even inspire FDA approval. This comes after presentation of results from the AMPLIFY trial at ASH 2024. Even though AMPLIFY did allow del17p/mutated TP53 patients, the NCCN included that high risk group in its recomendation because of another AVO trial with results at ASH 2024.

Those guidelines are available for a free download if you create a patient account at the first attemmpt to download:

nccn.org/professionals/phys...

The Clinical Guidelines are over 100 pages long and cover things not included in the NCCN CLL Guidelines for Patients, which is still waiting for an update to reflect the Clinical Guideline changes,

The Patient Guidelines can be downloaded free:

nccn.org/patients/guideline...

The full AMPLIFY results are yet to be published, but this abstract at ASH 2024 has:

ash.confex.com/ash/2024/web... Oral Abstracts 642.Chronic Lymphocytic Leukemia: Clinical and Epidemiological

Fixed-Duration Acalabrutinib Plus Venetoclax with or without Obinutuzumab Versus Chemoimmunotherapy for First-Line Treatment of Chronic Lymphocytic Leukemia: Interim Analysis of the Multicenter, Open-Label, Randomized, Phase 3 AMPLIFY Trial

AMPLIFY is an international trial that's been running since February 2019, and comapres A&V with AVO, and FCR/BR. It has 984 patients enrolled, and no longer accepting new patients. It's National Clinical Trial number is NCT03836261:

clinicaltrials.gov/study/NC...

=seymour=

[EDITED to correct information about AMPLIFY and TP53]

[SeymourB]

I should add that I've heard of several people on Facebook whose insurance covered either A&V or AVO in the U.S.

=seymour=

[bennevisplace]

Thanks Seymour. I guess this is v good news for CLL patients in the USA.

A quirk in the results of the AMPLIFY trial: while the OS for AV was best, the OS for AVO was not significantly better than for FCR/BR. Possible reason? Many of the deaths were attributed to Covid, and subjects receiving O or R may have had greater exposure to the virus in the outpatient setting.

[Zweistein in reply to bennevisplace]

It is too early to tell. See statistical error margins/bands in subfigure C:

(taken from nejm.org/doi/full/10.1056/N... or x.com/NEJM/status/188762234... )

To the right of subfigure C the overall survival at 36 months is given (AV: 94.1%, AVO: 87.7%, FCR/BR: 85.9%). After that time, the uncertainties are likely too big to state "official" values.

survival in amplify trial

[SeymourB in reply to Zweistein]

Zweistein -

I finally downloaded the full paper because of the limit of 2 downloads per month just expired. I must say they present much more data than the usual trial reports. I hope that subsequent annual updates continue in the same format so that we can observe changes over time.

I think we have to consider that the patients with the longest survival so far were the ones that first signed up for the trial, and the demographics of the early patients may differ from the overall demographics by favoring more risky patients. I think they consider the demographic balance as time goes on rather than denying some early patients a chance at the trial.

=seymour=

[Zweistein in reply to SeymourB]

You are absolutely right. The data on the very right of the curves should not be trusted too much. In particular, the steep drop for AV in OS is very likely a statistical artifact. Kindly, the uncertainties in this study are shown in the KM graphs, which is rarely the case.

[bennevisplace in reply to Zweistein]

Could these be subjects who left the trial, rather than deaths or statistical artefacts.

[SeymourB in reply to bennevisplace]

bennevisplace -

I don't think so. People who leave are censored from that point in the statistics. It thins the population for survival calculations, though. So that's a statistical artifact. Cause of death is often hard to assess, too. If a person dies in a car accident, that's unrelated to treatment. But death due to an heart attack, or stroke might be. COVID is controversial. Treatment may make a survivable case of COVID fatal. But it took awhile for people and their families to adapt to COVID by masking and avoiding.

"The sensitivity analyses for progression-free survival and overall survival with censoring for deaths due to Covid-19 are shown in Figure S4A and Figure S4B, respectively."

Since the trial started in 2019, and the pandemic had its worst mortality in the first few years, it's likely the COVID deaths would skew results.

Time will heal statistical oddities.

=seymour=

[bennevisplace in reply to SeymourB]

I agree that the easterly cliff on the OS curve is artefact, and I can't see why it has been included in the figure. Also, the number at risk tabulation under the graph makes no sense after month 36. I think more than one contributor was asleep at the wheel here.

[Skyshark in reply to Zweistein]

The data is for 36 months. Anything after 36 months is still developing.

If the lead subject progressed/died the KM PFS/OS line would plummet to Zero. A few months later the next subject surpasses that point and then the line is halfway back to where it was, as can be seen in the OS line for AV. Next subject surpasses that time then it's 2/3rds. By the time 100 have exceeded that time point the reduction is 1/100th and looks just like all the other events. Or in the case of events before 36 months about 1/3rd of a percent as there are nearly 300 in each arm.

[bennevisplace in reply to Zweistein]

Zweistein, thanks for the figure from the paper in NEJM. Statistical error bands are all very well, and are statisticians' way of telling us "don't read too much into the numbers" or "we can always use more data". Still... it appears the OS for AVO and FCR/BR are beginning to diverge after 36 months as the Covid effect (if that's what it is) washes out. "It's too early to tell" - Hmm, I'm not sure, with this data set in this trial, that we'll ever be able to tell.

The quirk in OS that I noted came from Seymour's cited ash.confex.com/ash/2024/web... in which "AV demonstrated an OS benefit trend over FCR/BR (HR 0.33, nominal P<0.0001)". From the summary given, deaths up to cutoff date, other than Covid-related, were AV 8/291, AVO 12/286, FCR/BR 21/290.

[Zweistein in reply to bennevisplace]

Whether the curves diverge depends on the chosen confidence level. It is common to require 95% certainty. As the error bands overlap the region where the curves diverge one can not be sure if the effect really is due to covid. With e.g. 50% certainty the interpretation may be different.

[BigfootT in reply to bennevisplace]

Here's a short video where Dr. Jennifer Brown speaks to this trial results. She talks about the COVID deaths in AVO arm at time 4:04 and shows a slide with COVID deaths censored out.

youtu.be/wnigPsXiwso?si=MHA...(Not working as expected?)

Bigfoot

[Skyshark in reply to Zweistein]

What I see there is that uIgHV are near parity with mIgHV for PFS on AVO. While there is a big gap in PFS for AV, which is similar to the gaps seen in V+O, V+I and VenR.

[Zweistein in reply to Skyshark]

Even an apparently bigger gap may not be significant if the number of patients who progressed to this point is still small. I can not really distinguish all the error bands in Figure B.

[Skyshark in reply to Zweistein]

Error bands, "confidence index" doesn't change much as the KM line develops, they are far more dependant on the total number of subjects. The error bands will all be about the same for the mIgHV as there are about 120 in each arm. The error bands will be tighter for uIgHV as there are around 170 in each arm.

This is shown by the red and orange bands on the CLL14 plots.

The CLL14 results I present show that for mIgHV with TP53del they are practically worthless with n=5. But CAPTIVATE FD also has very close correspondence with n=10.

CLL14 72 month results with error bands

[Zweistein in reply to Skyshark]

Do not forget that points within a curve are strongly correlated because the same patients contribute to every point until they drop out. One can only state that two curves are different if the error bands of the curves do not overlap. Curves may just seem to be different because of their intrinsic correlations.

[Skyshark in reply to Zweistein]

But overlap doesn't mean it's not statistically significant.

medium.com/data-science/why...

[SeymourB in reply to Skyshark]

Skyshark -

Thanks for this!

I really need to take a statistics course.

=seymour=

[Skyshark in reply to SeymourB]

That make two of us!

Waiting to enter the examination hall for 2nd Year Mech Eng Maths exam and the lecturer runs up in a panic and tells us not to attempt the 3 stats questions, they are for production engineers. So three hour exam, answer 5 from 6 instead of 9. We hadn't covered any stats in 2nd year.

[SeymourB in reply to Skyshark]

Skyshark -

I'm an undergrad history major. When I'm done with that (2 more classes till graduationm I think), I'm gonna load up on science courses for fun. My wife also wants in on the fun, too. She teaches 8th grade science, and longs for more stimulating reading. I took a lot of MOOC (Massively Open Online Courses) science classes since my dx in 2011. But some of that is probably obsolete by now - genetics and immunology certainly are.

=seymour=

[Zweistein in reply to Skyshark]

Since our daughter was born at the time you answered, I missed your reply. The argument in the link you cited is incorrect in our case. The error bands in Kaplan-Meier curves represent the uncertainty in the estimator, i.e., the uncertainty in the mean of the distribution, not its width. The relationship between the standard deviation and the uncertainty of the mean is expressed by the factor 1 / sqrt(n), where n, in this specific case, is the number of remaining patients.

[Skyshark in reply to Zweistein]

Does that also mean that each CI only relates it's KM line and overlap with other lines and their CI isn't relevant.

[Zweistein in reply to Skyshark]

The KM lines are not based on an (unknown) true distribution, but only on an (unknown) true value that the estimator "guesses" with an associated uncertainty.

One could create distributions for the period of progression-free time. These would then have a true width, and the cited argument would apply to them.

[Justasheet1 in reply to Zweistein]

Z,

I was reading through and saw you mention the birth of your daughter. Congratulations!

Jeff

[Zweistein in reply to Justasheet1]

Thank you. We almost gave up our hopes to have a child at all, but that is a different story. The joy that comes with the birth of our daughter even outweighs the recent CLL diagnosis. Nevertheless, it was a diagnosis at the wrong time and I hope to see my daughter growing up for as long as possible.

[Justasheet1 in reply to Zweistein]

Z,

Even 10 years ago things looked less hopeful than they do now. Try to stay out of your head and if you don’t already have an expert CLL doc you need one because this can be a long and complicated road and every decision today affects your decisions tomorrow.

Congratulations to you again.

Jeff

[SeymourB in reply to Skyshark]

Skyshark -

A further nuance is whether both copies of 17p or TP53 are missing or mutated. Having a single copy del17p or mutated TP53 is not as bad. But FiSH testing of TP53 is not adequate, and most previous studies did not look at homozygous vs heterozygous deletions and mutations. Even sequencing TP53 brings further questions about VAF (Variant Allele Frequency) interpretation. It's possible to have a low or mid range VAF that affects all or most cells vs an identical VAF that represents a single copy of TP53. Changes in sequencing methods might solve the ambiguity.

ERIC recommendations for TP53 mutation analysis in chronic lymphocytic leukemia—2024 update Leukemia. 2024 May 16;38(7):1455–1468.pmc.ncbi.nlm.nih.gov/articl...

"However, since in many published studies del(17p) and TP53 mutations were not distinguished [32, 33], and in some only del(17p) was included [38], this relevant issue is currently inconclusive. Moreover, the presence of homozygous mutations has not been considered at all. Thus, it is now imperative to include definitions of the type, clonal burden, and number of TP53 defects in clinical trials and academic studies in order to be able to provide a uniform classification, similar to myeloid neoplasms [39]."

So, patients should look closely at test results or ask their doctor whether FiSH reported on both copies or single copies, and whether TP53 was sequenced instead of tested by FiSH.

I wish I had MRD graphs showing differences, but I haven't seen them yet.

=seymour=

[Zweistein in reply to Skyshark]

In Figure B many curves are close. FCR/BR uIGHV is significantly worse than the rest except AV uIGHV, AV uIGHV is significantly worse than AVO. The other curves are too close to be really sure which one is the best. Only those with uIGHV clearly have a benefit from AVO in terms of progression free survival .

[Skyshark in reply to bennevisplace]

Also the anti-CD20 antibody destroys a lot more B-cells, with and without CLL. That suppresses the immune system.

GAZVYA B-cell CD20 targets

[SeymourB in reply to Skyshark]

Skyshark -

All of our therapies destroy normal B-cells. The anti-CD20 drugs suppress B-cells a lot longer, and I've never read a good explanation of why.

Venetoclax suppresses neutrophils and platelets in some people, but they recover within days of stopping treatment. Obinutuzumab and Rituximab suppress antibody production for 6 months or more after last infusion, according to the COVID vaccine studies. In my case, I finally saw a little recovery of B-cells in blood 18 months after last infusion, and 12 months after Venetoclax and Pirtobrutinib. I haven't seen antibody levels recover yet. In other diseases with more infusions, anti-CD20s have sometimes led to perment loss of new antibodies due to B-cell suppression. I don't think they kill plasma B-cells, because those don't have CD20. But since evennormal and long-lived plasma B-cells eventually die, memory B-cells need to develop into plasma B-cells to replace them for new disease exposure.

=seymour=

[TLSAtlanta]

Thanks for this post! I am hoping to start AV this week but have been confused over whether its FDA vs NCCN approved. Which your post cleared up. I am waiting for prior authorization, but I dont think its the insurance company causing the delay - I think the prior authorization team at DFCI haven't gotten it in yet. Also my doctor says we can look at adding in the Obin at the end of the 14 AV cycles if I'm not sufficiently MRD- although she thinks I will be since I am mCLL which she says responds well to the AV protocol.

[SeymourB in reply to TLSAtlanta]

TLSAtlanta -

From the supplemental data published on NEJM, it looks like uMRD4 (flow cytometry) rates in peripheral blood were:

AV 26.8% (22.0-32.1%) at Cycle 9 Day 1

AVO 66.4% (60.8-71.7%) at Cycle 10 Day 1

FCR/BR 51.0% (45.3-56.8%) at Cycle 6 Day 1 plus 12 weeks

uMRD6 rates (ClonoSEQ) at EOT (End of Treatment - Cycle 14 Day 28 for AV/O, Cycle 6 Day 28 for FCR/BR ) in peripheral blood were:

AV 5.2%, 1% indeterminate

AVO 40.9%, 2.4% indeterminate

FCR/BR 14.5%, 5.2% indeterminate

uMRD6 rates at EOT+12 Weeks in peripheral blood:

AV 5.5%, 0.3% indeterminate

AVO 43.4%, 2.3% indeterminate

FCR/BR 12.8%, 6.9% indeterminate

-----

The slight increase in the uMRD6 rates from EOT to EOT+12 weeks may be due to continued B-cell suppression long after last Obin infusion, or recovered T and NK cell function, or something else. It's not clear to me if doing Obin early vs late is worse. All the data is on early Obin.

I'm not sure what the indeterminate is all about. The lack of bone marrow results is a concern for people with SLL, I think.

You can also look into ZV or ZVO to see if insurance covers it. A 5 year followup abstract on the Phase 2 BOVen were published at ASH, too, and on paper look better than AMPLIFY.

ash.confex.com/ash/2024/web... 1867 Multicenter Phase II Trial of Zanubrutinib, Obinutuzumab, and Venetoclax (BOVen) in Treatment-Naïve Chronic Lymphocytic Leukemia: 5-Year Follow up, Retreatment Outcomes, and Impact of MRD Kinetics (ΔMRD400).

That was an MRD driven study, 8 to 24 cycles ZVO depending on MRD status. Phase 2 trials don't aim for balanced demographics, but they did include Del17p/mutated TP53,

With a median treatment duration of 10 cycles (interquartile range [IQR] 8-14) uMRD4 by flow cytometry:

96% in peripheral blood

92% in bone marrow and peripherla blood

The most common grade ≥3 AEs were neutropenia (26.9%), thrombocytopenia (7.7%), lung infection (5.8%)." AMPLIFY saw AE≥3 neutropenia in AV of 32.3%, AVO 46.1%, FCR/BR 43.2%. Oddly, AMPLIFY did not report thrombocytopenia.

=seymour=

[TLSAtlanta in reply to SeymourB]

Thanks for all this. The ZV studies are not far enough along to have made it to the NCCN guidelines so not yet an option unless you get into the study. The stats you quoted don't distinguish the unmutated vs mutated, is that right? Its my mutated status that makes my doc confident I will get good results with the AV protocol.

[SeymourB in reply to TLSAtlanta]

TLSAtlanta -

The NCCN endorsement is huge. But doctors in the U.S. can request any combination of drugs that have already been approved singly or in combination as off-label. They typically write a letter to the insurance company providing evidence to justify it. So it's not out of the question. On Facebook, we saw patients doing AV and AVO before AMPLIFY results and before the NCCN.

Likewise, you could also try for a Pirtobrutinib combo. There's preliminary evidence from the Phase 2 trial that I participated in (PVO) at M.D.Anderson. I haven't heard of anyone doing that yet, though. Before I did the trial in 2023, my own local doctor had offered me Pirtobrutinib monotherapy based on its approval for relapsed/refractory Mantle Cell Lymphoma. But I wanted the combo and testing from MDA.

=seymour=

[Skyshark in reply to SeymourB]

I'm not sure why it would be any more used in US than Ven + Ibr which is also in the NCCN guidelines and both separately approved.

[julius_the_cat]

I believe AMPLIFY did not allow del17p/mutated TP53 patients - perhaps a typo?

[bennevisplace in reply to julius_the_cat]

You are correct.