Mistergern, here's the clarification, thank you for pointing this out

[biopearl123]

The abstract you're referring to appears to be the Kuykendall et al. real-world analysis using the Flatiron Health database (published in Leukemia & Lymphoma in early 2026, also presented at ASH 2024, and likely appearing at ASCO 2026). The OS in that table — 37.4 months (with baseline anemia) and 64.9 months (without baseline anemia) — is measured from ruxolitinib initiation, not from rux failure.
That's the key reason the numbers look so long compared to prior studies. When OS is counted from rux failure (as in most second-line trial analyses like IMbark or PERSIST-2 comparators), the median is typically in the 12–16 month range. Here, because the clock starts at rux start, you capture the entire duration of rux therapy plus whatever follows, naturally inflating the OS figure substantially.
There are a few other reasons these numbers may look longer than expected:
The Flatiron database captures a broad real-world population, which likely includes lower-risk patients (intermediate-1) who were started on rux but have an inherently better prognosis. Clinical trials like COMFORT-I/II enrolled predominantly intermediate-2 and high-risk patients, so their OS benchmarks reflect a sicker population.
There may also be some immortal time bias — patients in the database had to survive long enough to be captured and to receive rux, which selects for a healthier cohort.
The "no anemia" group's 64.9-month median OS is striking, but again, patients without anemia at rux start tend to be earlier disease, lower-risk patients who would be expected to do better regardless.
So in short: the OS is from rux start, not rux failure, which explains why it appears so much longer than what you'd expect from prior post-rux failure studies.
Sources:

Ruxolitinib in MF with anemia: Real-world treatment patterns and survival outcomes (MPN Hub)
Full article: Ruxolitinib treatment for myelofibrosis with anemia — Leukemia & Lymphoma
Retrospective Real-World Analysis (ASH 2024 abstract)
Imetelstat Sustains Significant Survival Benefit in R/R Myelofibrosis
AskSonnet 4.6

[biopearl123]

So we have to subtract out Rux duration to get to R/R and then run the numbers. (mOS still meaningfully longer.) Not clear about the language stating that patients then cycle to other agents or clinical trials. Since mOS requires that they die, it would be good to get clarification e.g. if the clinical trial they cycled into included Imetelstat.

[Ryan]

The article points out that patients are starting Rux way earlier.

[biopearl123]

And keeping patients on it for a much shorter treatment period.

[Ryan]

Here's a thorough breakdown of the article and its key investment/clinical narratives:

---

## The Core Study: IMbark Post-Hoc Analysis (2026 ASCO)

An updated post-hoc analysis from the Phase 2 IMbark trial has reaffirmed that imetelstat delivers a clinically meaningful and statistically significant overall survival (OS) benefit versus best available therapy (BAT) in post-JAK inhibitor relapsed/refractory intermediate-2 or high-risk MF — even as real-world outcomes in this population have improved substantially over the past decade. This analysis will be presented at the 2026 ASCO Annual Meeting.

The comparison included 59 IMbark patients and 54 closely matched real-world patients. At a median follow-up of ~47–48 months, median OS with imetelstat was **30.7 months** vs **15.4 months** with BAT — roughly a **doubling of survival** (HR ~0.51, p=0.003).

---

## The Ruxolitinib Usage Shift — The Key Nuance

This is the most strategically important finding in the article, and it actually creates a **bull-bear tension** for imetelstat's Phase 3.

**What changed with ruxolitinib use:**

Patients diagnosed after 2016 achieved an mOS of 34.2 months compared with 15.4 months for those diagnosed before 2016. Stratifying around 2019 showed an even larger gap — reaching 39.8 months. In other words, the real-world comparator arm (BAT patients) is getting much better over time.

Earlier ruxolitinib initiation and shorter ruxolitinib exposure suggest more rapid switching to next-line therapy or clinical trials, while stable transplant rates (5.6%–9.5%) do not explain the survival gains.

What this means in plain terms: In the earlier era (pre-2016), patients tended to stay on ruxolitinib longer — exhausting the drug more thoroughly before discontinuing. In the more recent era, physicians are initiating rux earlier in the disease course and **discontinuing sooner**, pivoting patients more quickly into clinical trials or next-line options. This behavioral shift — **shorter rux exposure, faster switching** — is improving real-world outcomes independent of any new drug.

**Why this matters for imetelstat's narrative:**

The bear case would be: if real-world BAT patients now live 34–40 months post-rux discontinuation, imetelstat's 30.7 months no longer looks like a win. The bull rebuttal from the study is that:
- The head-to-head comparison was deliberately **restricted to the pre-2016 cohort** to match the IMbark era, minimizing temporal bias
- Propensity-weighted approaches (ATO and sIPTW) reproduced the OS benefit, arguing against baseline imbalances as the primary driver
- The OS gap holds up even after accounting for the improving real-world environment

## The IMpactMF Phase 3 — Where Things Stand

**Trial design:**
IMpactMF is a randomized Phase 3 trial enrolling patients who have progressed on ruxolitinib or another JAK inhibitor. Patients are randomized to imetelstat or BAT, with JAK inhibitors specifically excluded from the control arm. The primary endpoint is OS — unique in MF, where regulatory endpoints have historically focused on spleen and symptom response.

**Enrollment complete:**
Geron completed enrollment of 320 patients in IMpactMF in September 2025.

**Interim readout timing — slipped from earlier guidance:**
Based on current assumptions for death rates in the trial, the interim OS analysis is expected in the second half of 2026, with the final analysis expected in the second half of 2028. This is a meaningful delay from the earlier guidance of H1 2025 for interim and H1 2026 for final. At the 2025 ASCO Annual Meeting, investigators stated the interim analysis (at ~35% of enrolled patients having died) was expected in early 2026. The slippage to H2 2026 implies death rates are running slower than projected — which could mean patients are living longer (potentially good for imetelstat's survival thesis) but delays the data event.

**The high-stakes endpoint:**
The boldness of IMpactMF — a randomized trial in the post-JAK inhibitor setting asking specifically whether imetelstat improves survival — and the clarity of that endpoint is why the field is eagerly awaiting results.

## Has the Narrative Changed Going into the Interim?

**Strengthened:**
- The IMbark post-hoc now uses more rigorous propensity-score matching and a larger real-world dataset (126 patients from Moffitt, 2010–2025), and the survival benefit holds up
- Imetelstat already has FDA approval in lower-risk MDS (Rytelo, June 2024) and EU approval (March 2025), giving the drug established commercial and regulatory credibility
- The biomarker story is strengthening — post-hoc IMbark analyses showed dose-dependent reductions in IL-8 and TNF-alpha with the 8.9 mg/kg dose, and these cytokine reductions corresponded with improvements in symptom scores, spleen volume, and longer survival trends

**Complicated:**
- The real-world BAT arm is improving on its own due to changed rux usage patterns, making the Phase 3 comparator potentially tougher than when the trial was designed
- The interim readout has slipped ~12–18 months from original guidance, extending investor uncertainty
- The field is also exploring imetelstat in combination with ruxolitinib earlier in the disease course (IMproveMF), suggesting the post-rux monotherapy slot may not be the drug's ultimate commercial home

**Bottom line:** The survival signal from IMbark remains intact and has actually been reinforced by more rigorous matching methodology. The conceptual risk heading into the Phase 3 interim is not that the drug doesn't work — it's that the real-world control arm may be a harder hurdle than it was in the Phase 2 era, precisely because ruxolitinib is now being used and discontinued differently. The trial's OS-primary design remains the cleanest possible question in this space, and the H2 2026 interim is the defining moment for the Geron story.

[Ryan]

The DSMB's Math Problem — Reconstructing the Calculus
Step 1: What the Trial Was Designed Around
IMpactMF was powered assuming a hazard ratio of 0.60 for imetelstat versus BAT, with ~88% power at a one-sided significance level of 0.025, using a group sequential design. Taylor & Francis
The trial enrolled ~320 patients: approximately 214 to imetelstat and 106 to BAT (2:1 ratio). The interim analysis triggers when ~35% of planned enrolled patients have died — so roughly 112 deaths. The final analysis triggers at >50% — roughly 160+ deaths. Q4cdn
The alpha spend at interim is ~0.01 (one-sided), meaning the DSMB needs to see an extremely strong signal to stop early for efficacy — this is a very conservative, O'Brien-Fleming-style boundary. Crossing it would require a p-value well below 0.01. StreetInsider

Step 2: What the Trial Assumed About the Control Arm (BAT)
This is the crux. When the trial was designed and powered, the reference point for what "BAT" produces in post-rux MF was the historical data: median OS of roughly 14–16 months post-JAK inhibitor discontinuation. That was the real-world benchmark, and it's what the HR of 0.60 was calibrated against.
The IMbark phase 2 showed imetelstat at ~29–30 months mOS, which against a BAT of ~15 months produces an HR approaching 0.45–0.50 — far better than the 0.60 assumed. So the trial was, on paper, conservatively powered.

Step 3: What This New Article Changes for the DSMB
Here is the paradigm shift embedded in the ASCO 2026 post-hoc analysis:
Real-world patients diagnosed after 2016 achieved a mOS of 34.2 months, and those diagnosed after 2019 reached 39.8 months — versus the ~15 months seen in the pre-2016 era. Targeted Oncology
Earlier ruxolitinib initiation and shorter ruxolitinib exposure suggest more rapid switching to next-line therapy and clinical trials. Stable transplant rates of 5.6–9.5% do not explain the survival gains. Targeted Oncology
Translation for the DSMB: The BAT arm in IMpactMF — which enrolled from 2020 through September 2025 — is not the historical BAT that the trial was powered against. These are post-2019-era patients. If the real-world BAT mOS for that cohort is now 34–40 months, and imetelstat in IMbark produced ~30–31 months, the HR could theoretically be above 1.0 — meaning BAT might be outperforming imetelstat in the Phase 3.

Step 4: The Three Scenarios the DSMB is Now Weighing
Scenario A — The Bull Case: The RCT randomization insulates the data
The study's 2:1 randomization means the same temporal era affects both arms equally. Both the imetelstat arm and the BAT arm are drawn from the same post-2019 patient pool — patients who are being switched off rux earlier, are healthier at switching, have better supportive care, and have access to more clinical trial crossover options. If imetelstat's mechanism genuinely confers a survival advantage, that advantage should persist regardless of era improvement. The HR of 0.60 the trial was powered for could still be achieved — or even bettered — if imetelstat's 30-month phase 2 mOS translates to, say, 40+ months in the Phase 3 era (as both arms improve but the HR holds).
Scenario B — The Bear/Futility Case: Converging survival curves
This is the new risk that the ASCO 2026 analysis makes explicit. If the BAT arm in the Phase 3 is now achieving 30–40 month mOS due to the changed rux usage patterns and better supportive care — and imetelstat's arm is achieving something similar — then the HR narrows toward null. At the interim (112 events), if the curves are running close together, the DSMB faces a futility signal, not an efficacy signal. With only 35% of events, even a genuine HR of 0.75 may be hard to distinguish statistically from 1.0, and the conditional power going forward collapses.
Scenario C — The Crossover Confound
The protocol permits crossover to imetelstat for BAT patients who meet progressive disease criteria (≥25% increase in spleen volume from baseline) after 6 months of BAT. If the BAT arm is doing well for 6+ months and then crossing over to imetelstat on progression, that crossover inflates BAT arm OS and compresses the HR — exactly the wrong direction for demonstrating imetelstat superiority at the interim. American Society of Clinical Oncology

Step 5: What the Timing Delay Actually Signals
The interim was originally guided for H1 2025, then shifted to early 2026, and is now guided for H2 2026.
Based on current assumptions for event rates in the trial, the interim OS analysis is expected in the second half of 2026, with the final analysis expected in the second half of 2028. Stocktitan
The fact that death events are accruing more slowly than projected — hence the repeated slippage — is itself data. It means patients in this trial (both arms) are living longer than the original design assumed. That is consistent with the ASCO 2026 paper's finding that real-world BAT survival has improved materially in the post-2019 era. The trial is playing out in an era where the control arm is much harder to beat than when it was designed.

The DSMB's Actual Decision Framework at Interim
At ~112 events, the DSMB sees three possible paths:
HR ≤ ~0.60, p < 0.01 | Strong early efficacy signal| Stop for superiority
HR ~0.65–0.85, p > 0.01 | Trending positive but not crossing boundary | Continue to final
HR ≥ ~0.90, conditional power <15–20% | Futility signal | Recommend stopping

The critical new wrinkle from the ASCO analysis is that the probability mass has shifted toward the middle and right buckets compared to what was anticipated at trial design. The IMbark HR of ~0.51 against a 2010s-era BAT is not necessarily the same HR that will emerge when both arms benefit from 2020s-era care standards, earlier rux cycling, and trial access.

The bottom line — and what the article is really about: The ASCO 2026 post-hoc was not simply a confirmatory victory lap. It is implicitly a preemptive framing exercise — making the case that imetelstat's survival benefit holds even against a better-performing control arm. The investigators are trying to front-run the narrative concern that the Phase 3's BAT arm will be stronger than the Phase 2 comparator, and to argue that propensity-matched real-world data still shows a durable HR. Whether the DSMB's actual randomized data — where both arms came from the same contemporary patient pool — reflects that same HR is the question $GERN investors are waiting to have answered in H2 2026.

[Secret Third Arm]

“ The conceptual risk heading into the Phase 3 interim is not that the drug doesn't work — it's that the real-world control arm may be a harder hurdle than it was in the Phase 2 era, precisely because ruxolitinib is now being used and discontinued differently.‘

Couldn’t it be argued that patients discontinuing Rux sooner could see an improved benefit from Imetelstat? The longer Rux is used and its effect diminishes then the sicker the patients are who are getting Imetelstat in the r/r population. Getting off Rux sooner should therefore potentially increase the potential of Imetelstat.

[biopearl123]

MG and Ryan thanks for the posts. I do think we are getting some much needed clarification,

STA, yes exactly. This is what I was getting at with two posts:
1. “What’s good for the goose is good for the gander” and 2. “A rising tide lifts all boats”.

Was basically trying to get at exactly your point.

[mistergern]

I shared this conversation with Claude - this was his reply: Ok here is the latest blurbs from Imetelchat -10:40 AMClaude responded: This is genuinely the most substantive and balanced analysis of the IMpactMF situation I've seen in this conversation.This is genuinely the most substantive and balanced analysis of the IMpactMF situation I've seen in this conversation. Let me evaluate it honestly.
What Ryan's analysis gets right:
The core tension is real and well-articulated. The ASCO paper simultaneously strengthens the IMbark evidence base AND raises a legitimate question about whether the Phase 3 BAT arm is harder to beat than when the trial was designed. That's an honest read of what the data shows.
The three DSMB scenarios are correctly framed. The HR thresholds are consistent with what we calculated.
Where I want to push back on Ryan's bear case:
The 34-40 month real-world OS figures are measured from ruxolitinib initiation — not from rux failure. You correctly identified this earlier. These numbers are not valid comparators for IMpactMF patients whose OS is measured from JAK inhibitor discontinuation. Ryan's scenario B essentially relies on this inflated figure without adequately flagging that distinction.
Secret Third Arm and BP's point is analytically correct:
Patients discontinuing ruxolitinib earlier and in better condition at discontinuation should theoretically respond better to imetelstat — because they haven't been exhausted by prolonged JAK inhibitor exposure. If true, the imetelstat arm benefits from the same trend that improves the BAT arm. The HR could hold or improve.
What I genuinely don't know:
Whether the IMpactMF BAT arm OS is tracking at 15 months or something materially higher. That's the question the interim answers. The enrollment timing arithmetic we built remains valid — but Ryan is right that if BAT arm OS in the actual trial is substantially higher than 15.4 months, the HR compresses.
The honest bottom line:
Ryan's analysis is the most intellectually honest bear case I've seen. It doesn't say the drug doesn't work. It says the hurdle may be higher than anticipated. That deserves to be taken seriously rather than dismissed.
Our model assumed 13-22 months BAT OS. The confirmed matched comparator from Kuykendall is 15.4 months. The question Ryan raises is whether the actual IMpactMF BAT arm — enrolled 2021-2025 — performs at 15.4 months or materially better due to era effects.
That is a genuinely open question the interim will answer. I don't think it changes the probability distribution dramatically given the enrollment timing analysis we built — but it's a legitimate uncertainty I should have weighted more carefully rather than progressively dismissing throughout this conversation.You said: My calculation is that the HR will be below .My calculation is that the HR will be below .60 if the mOS for BAT patients is =< 30 and that we have not seen any evidence yet that makes a mOS > 30 at all probable.10:44 AMClaude responded: Your calculation is mathematically correct — if BAT median OS stays at or below 30 months the HR remains below 0.Your calculation is mathematically correct — if BAT median OS stays at or below 30 months the HR remains below 0.60.
But I want to be honest about something before agreeing too readily.
The evidence we have for BAT OS in this population is 15.4 months from the Kuykendall matched cohort. That's the best available published figure for post-JAK failure intermediate-2 or high-risk MF patients. That's what we should anchor to.
However Ryan raises a legitimate point we shouldn't dismiss. The IMpactMF BAT arm enrolled from 2021-2025 — a different era than the Kuykendall matched cohort which used pre-2016 patients specifically to match the IMbark population. The question of whether contemporary BAT patients in IMpactMF perform at 15.4 months or somewhat better due to era effects is genuinely open.
The honest answer to your specific claim — that we haven't seen evidence making BAT OS above 30 months probable — is correct. The 34-40 month figures in the paper are from rux initiation not rux failure, and include lower-risk patients not in IMpactMF. Those numbers don't apply.
But "not probable above 30 months" is different from "confirmed at 15.4 months." The actual IMpactMF BAT arm performance is unknown until the interim reads.
What I'm confident in is that at any BAT OS below approximately 22-24 months — which covers all published real-world evidence for this specific population — the HR crosses the OBF boundary in our model. That's the range supported by evidence.
Above 24 months for BAT OS there isn't published support. Ryan's concern requires assuming something we don't have evidence for.

[biopearl123]

Making sense of the Kuykendall numbers (corrections welcome):

1. They start the clock at the onset of therapy with Rux not the end (MG has pointed this out)

2. It appears to include migration to next line therapy including investigational therapy.

3. Makes comparisons with numbers we are used to seeing murky.

[Secret Third Arm]

My question is: What patient would be allowed into an investigational therapy while simultaneously being in the IMpact trial? Doesn't being in an ongoing trial preclude a patient from participating in another? If most of the impact on mOS for BAT patients is investigational treatments then I would think that would not have an effect on this trial other than affecting the dropout rate in the comparator arm. For all that we know, the investigational treatments being referred to is Imetelstat itself.

[biopearl123]

STA, I think that's a hard no. Dr. K's abstract is a registry outlining how Rux has been used and is changing with practice evolution. So if you are an MF patient at a respected institution with a strong MF program upon diagnosis you would enter a registry and start on a front line drug like Rux. When that failed you would be offered several options, a different line of therapy (other JAKi?) or entry into an investigational study (aprox 5 ongoing at Moffitt currently) with the understanding that some have a placebo or BAT arm that you could be randomized to. That's how it would work. You would remain in the registry. Only one study would be allowed at a time. So yes one of those options (after wash out and fulfillment of entry criteria) could be Impact near as I can tell.

[Secret Third Arm]

So I'm trying to understand why new investigational therapies would have any impact on mOS for the BAT arm. It seems to me that the new survival numbers post-2016 are affected mainly by investigational therapies that would not be available to patients in the BAT arm of our trial. As far as I know, no new therapies have been approved (and therefore are not investigational) in the time since 2019 that would extend the life of BAT patients. As per the trial design, BAT patients can receive:

Drug: Best Available Therapy (BAT)
Non-JAK-inhibitor treatment will be given, which may include but is not limited to hydroxyurea, thalidomide or an analog of thalidomide, interferon, danazol, hypomethylating agents, chemotherapy or radiotherapy.

Even with enhanced disease and Rux understanding, we can't possibly need to worry about some black swan change to R/R treatment that would upset the predicted survival for BAT.

[biopearl123]

STA, I have to admit I was really confused after reading (and re reading) the Kuykendall abstract. Your point is a good one. The way I tried to make sense of it may not be correct but here's what I came up with (with the help of MG, Ryan and others.). Initially I saw the RW mOS numbers and could not believe that RW OS had improved so dramatically. So here's my (and I emphasize my) understanding. Median OS for the purposes of the Impact study in the BAT arm is measured from the enrollment ITT (intention to treat) time when a patient is enrolled in the study until the time they die. When they enrolled they have already failed Rux or other therapies in order to qualify for enrollment. This arm is referred to as the RW OS BAT arm. The Impact OS clock starts. In the Kuykendall abstract, it has already been going since the patient started Rux and will continue to go when they stop either for good or when they die after trying other lines of therapy and study drugs. This very confusingly has also been labeled as the BAT RW arm. in the abstract. They are not the same. When a patient dies in the BAT Impact arm their clock stops. Most of these patients have failed Rux or other non Imetelstat treatments. In the Kuykendall registry the clock is running through Rux failure and also running through other therapies which I am presuming also could include Imetelstat randomization. It stops when they die. The patient would have to know he might be randomized to the non Imetelstat arm to enter the Impact study. So I think the answer to your concern is that yes one of the pathways (and only one Impact pathway could be chosen upon Rux failure would be to be randomized in the study with a 2:1 chance you could get Imetelstat.) You are still in the registry itself which now includes a record of time from diagnosis to Rux start, time through Rux treatment and cessation, time to death after entering other treatment lines or a study (which could include imetelstat--I think most patients would choose the latter.). This to me explains the very long RW OS in the abstract. It is a registry of all treatments start to finish, lumped. The abstract doesn't appear to break them out, if it did your question could be definitely answered!

[biopearl123]

It's not a black swan or new treatment we have to worry about but better supportive care and a marked change in treatment patterns with Rux that present patients to the study with less advanced disease. This may affect both arms in a similar fashion, we shall see. There is a possibility that this pattern change could preferentially be beneficial to the Imet arm. To clarify, a patient could not be simultaneously on an investigational drug beside Imetelstat while in the trial.but they would all be included in the registry. So this latter possibility might help explain the extended numbers.

[Secret Third Arm]

“This may affect both arms in a similar fashion, we shall see.”

Of course the problem with this statement is that if both arms see improvement in mOS then we will not have enough divergence at the interim analysis.

Can someone please explain to me though why earlier discontinuation from RUX could lead to longer mOS for patients? I mean in general, not in this trial. I can understand perhaps how an earlier start could help, but not how earlier discontinuation can.

[mistergern]

STA - from Claude - STA's question is a good one and deserves a direct answer.
Why would earlier ruxolitinib discontinuation lead to longer mOS for patients after rux failure?
The most logical explanation is patient fitness at the time of discontinuation, not anything about what happens afterward.
Patients who stay on ruxolitinib for 30+ months are typically exhausted patients who have been on a failing therapy for a long time — their disease has progressed further, their bone marrow is more fibrotic, their performance status has declined, and they have fewer therapeutic options remaining. They are sicker when they finally come off rux.
Patients who discontinue ruxolitinib earlier — after 13-15 months as the post-2019 pattern shows — are discontinuing before the drug is fully exhausted. They are less sick at the point of discontinuation. They have better performance status. They are more eligible for subsequent therapies including clinical trials.
So the survival improvement is not because subsequent therapies are better — it's because the patients entering subsequent therapy lines are in better shape to tolerate and respond to them.
The implication STA is circling:
If IMpactMF BAT arm patients are the healthier post-2019 phenotype — less sick at rux discontinuation — they may survive longer on BAT than the 15.4 month historical figure suggests. That's the legitimate concern.
But BP's point is also correct — those same healthier patients entering the imetelstat arm should respond better to imetelstat too, since earlier discontinuation means less clonal exhaustion and potentially more telomerase-dependent malignant cells remaining. The benefit may accrue to both arms proportionally.
Whether it accrues equally or differentially is what the randomized data will show. That's genuinely unknown until the interim reads.You said: Only the Intermediate-2 would be eligible for IMPactMF , so they still would likely conform more to the historical mOS than to the extended 34m - 39m new RW d…Only the Intermediate-2 would be eligible for IMPactMF , so they still would likely conform more to the historical mOS than to the extended 34m - 39m new RW data mOS, correct?3:06 PMClaude responded: That's a reasonable point but I want to be careful before fully agreeing.That's a reasonable point but I want to be careful before fully agreeing.
What's correct:
IMpactMF requires intermediate-2 or high-risk disease plus JAK inhibitor failure. That's a sicker population than the broad Kuykendall registry which includes intermediate-1 patients. The 34-39 month figures from the post-2019 cohort almost certainly include a meaningful proportion of intermediate-1 patients who don't qualify for IMpactMF and have inherently better prognosis. So yes — those extended figures are not directly applicable to the IMpactMF population.
Where I want to be honest about uncertainty:
The 15.4 month matched cohort from Kuykendall used pre-2016 patients specifically to match the IMbark era. IMpactMF enrolled 2021-2025. Even restricting to intermediate-2 and high-risk patients, there may be some era effect from earlier rux discontinuation producing healthier patients at enrollment — just not as dramatic as the 34-39 month figures suggest.
The honest range for IMpactMF BAT arm OS is probably 14-20 months for the intermediate-2 and high-risk population specifically. The lower end is well-supported by published data. The upper end represents the era effect concern without overstating it.
At any point in that range our model crosses the OBF boundary. But I want to be straightforward — I don't know precisely where in that range the actual IMpactMF BAT arm will land. Neither does anyone outside the DSMB.
That uncertainty is real and worth holding honestly rather than dismissing.

[Secret Third Arm]

“…there may be some era effect from earlier rux discontinuation producing healthier patients at enrollment —”

I don’t understand this statement at all. The enrollment criteria and the definition of Intermediate 2 are scientifically established. A patient ‘healthier’ than Intermediate 2 is by definition not Intermediate 2.

[huntingonthebluffs]

From my layman's perspective, are we over complicating due to the confusion around the Kuykendall et al. real-world analysis, etc.? Obviously that analysis is important but does it say anything about why patients with intermediate-2 or high-risk MF are becoming R/R to Jak inhibitors sooner? And if they don't meet the R/R criteria, they can't qualify for the IMpactMF CT. So who are these new age patients and why are they becoming R/R to jaki sooner with less advanced disease and/or healthier after 2016. Have the end posts moved?

I would assume once being R/R to JAK inhibitors an intermediate-2 or high-risk Myelofibrosis (MF) patient would not just continue on without benefit. Were the oncologists and insurance companies asleep at the wheel?

Has the definition of R/R changed? I don't think so, at least not in the CT. Are the oncologists fudging on the eligibility criteria? Have the oncologist been enlightened somehow? Just doesn't make sense that patients are now somehow R/R sooner and therefore healthier coming into the study.

[biopearl123]

I think doctors and patients were desperate and the drug may have been used beyond its utility and there was no place else to go. (Rux was approved in 2011. )