- The following is my opinion of the briefing documents' content. No-one should use the following to make investment decisions without conducting their own research.
Glossary of Terms:BLA Biologics License ApprovalCQA Critical Quality Attributes (the things MSB claim make Rem-L have potency and that they intend to control in manufactureDP Drug Product
MAGIC. Mount Sinai Acute GVHD International ConsortiumMSB-GVHD001MSB single-arm pediatric SR-aGVHD clinical trial for Remestemcel-LODAC Oncological Drug Advisory CommitteeORR. Overall Response Rate
RCT Randomised Clinical TrialRem-L Remestemcel-L
SAP. Statistical Analysis PlanSR-aGVHD Steroid Refractory acute Graft Versus Host DiseaseI thought it would be interesting to have another looking at the FDA briefing documents that were presented prior to the ODAC meeting. If I have time I'll go through each of the docs and highlight the issues raised by the FDA and what I think might be reasonable responses to those issues.If I get the time, I'll go through all the docs (I haven't read them all yet!), and the MESO presentation at the ODAC. My comments inItalics.Following on from the previous post, I'll now examine
FDA-ODAC-PM the briefing document for ODAC afternoon session. This is a long document (38 pages), so I'll likely break this into separate posts.
Section 1 Is and intro to MSB-GVHD001 - the single-arm pediatric SR-aGVHD trial that the PM session is seeking to discuss.
Section 1.2 introduces why the purpose of the meeting, to discuss the adequacy of the trial design and whether the total evidence supports a conclusion that Rem-L has efficacy in treating pediatric SR-aGVHD, it also includes a number of statements that reflect the FDA's thinking:
- Protocol MSB-GVHD001 was a prospective, multi-centre, single-arm trial of remestemcel-L for treatment of pediatric patients with SR-aGVHD grades B-D (excluding grade B skin alone).
- Primary endpoint is 28-day ORR
- Protocol designed to see if proportion with ORR >45%
- Null Hypothesis prespecified
- Justification of Null-Hypothesis was that it was 20% lower than Rem-L obtained from post hoc analyses of pediatric subgroups of prior RCT
- Null Hypothesis further justified by post hoc analysis of SoC patients in prior RCT and examination of MAGIC database and published papers.
- FDA asks for input from ODAC as to whether this was a persuasive way of establishing null hypothesis
- FDA asks about whether single-arm is suitable given subjective grading of aGVHD, comparisons with prior treatments in prior RCT
Phew - that's a lot of stuff for 2 paragraphs and there is more to come in this section, but let's deal with the important points.
The main issue centres around the selection of the Null Hypothesis. The FDA rather clumsily characterises the selection of the null hypothesis in such a way that it almost sounds like scientific fraud - i.e. you work out what your drug does, then subtract 20% from it and claim that's the SoC rate and bingo - your drug is 20% better. While the SAP does discuss this process as "Rem-L gets 65%+ so if Null is as 45% this would be 20% better which would be significant" it was more of a discussion of what the upper limit of SoC could be to still show Rem-L as significantly more effective, and hence even if SoC turns out to be lower, then using 45% as the null hypothesis would be a conservative choice.
The MAGIC database and prior trials confirmed that 45% is a good upper bound to the ORR for SoC in SR-aGVHD, there really isn't any better information available and importantly the FDA doesn't propose any alternative null hypothesis. You'd think they would be more professional than saying "we don't think the null hypothesis is robust " without actually having a clearly better approach (beyond a full placebo controlled RCT).
It is particularly bizarre give that the FDA approved Jakafi in a single-arm trial that had no null hypothesis. The trial just arrived at a 54.1% ORR from a smaller number of pediatric patients mainly driven by a 100% ORR from grade 2 patients (the least severely affected) and somehow this was enough to convince the FDA, no quibbling about subjective grading etc. They just seemed to say Jakafi worked on some other diseases (a blood cancer and a bone marrow cancer) so it must be working here ... look 57.1% ORR! Jakafi also has quite a list of side-effects which aren't present in Rem-L.
I hope MSB bring this up in their Type A Meeting, it's highly inconsistent and clearly having them meet a much higher bar for approval.
With all this said, I think it was a big mistake to mention the 65%-20% approach to establishing the null hypothesis upper bound in their trial SAP.
Instead they should have just stated that they arrived at the null hypothesis through MAGIC stats and the stats for SoC in their other RCT and the literature on SR-aGVHD. It unnecessarily muddied the waters, making the 45% appear contrived. MSB should challenge the FDA to produce a better null hypothesis for standard of care, noting that if they go much beyond 45% their Jakafi approval looks even more unsafe. I expect that the 69.1% ORR with Rem-L should suffer from the same "subjective grading of aGVHD" as Jakafi, so it seems weird that the much smaller ORR is sufficient to receive approval.The remaining paragraphs outline the stats behind the 69.1% ORR figure and how similar results were obtained by post hoc analysis of prior pediatric patients of earlier RCTs. They also discuss a number of earlier trials and state that the didn't meet their primary objectives.
Once again this section rather flippantly deals with prior trial data, mentioning the overall result (i.e. not meeting primary objectives) but fthey ail to explain that they were not pediatric only trials, and therefore irrelevant to this BLA and that post hoc analysis produced some statistically meaningful positive outcomes, particularly for pediatric subsets of the patient populations. These positive outcomes should at least inform them that there is some efficacy being shown in a placebo controlled RCT.Section 2This section begins by describing the results of Jakafi. Interestingly they mention that "success" was measured by an ORR that was above 40%.
They go on to mention some of the reasons why it was approved even though it was a single-arm trial:
- no available therapies
- effect of response attributable to treatment
- life-threatening disease
- no approved therapies
- the activity of the drug is established in other diseases that shared similar pathophysiology
- there was a substantial safety database
They then go on to say that no safe dosage exists for children under 12 years of age.
Clearly the first 4 apply equally to Rem-L, I'm struggling to find "similar pathophysiology" between cancers and an immune response and the "substantial safety database" actually outlines quite a few very serious (some would be life threatening to aGVHD patients) side-effects. But apparently the FDA thought these were "pluses".The sub-section ends with the results of REACH-2 a RCT for Jakafi in adults that showed efficacy in treating aGVHD, however it was released after the Jakafi approval.
It should be noted that the ORR for the control arm of the REACH-2 trial was 39%, well below the 45% null hypothesis for Rem-L. The FDA document does not provide this data which - given it was an RCT with 309 patients, clearly destroys any claims about the null hypothesis being poorly constructed.I'll take up the rest of this section in a later post as we've covered quite a bit of ground here.