Quick thoughts post-EHA abstract

I posted some thoughts last night before bluebird released their EHA abstract. Bluebird did, in fact, release preliminary data from their first SCD patient today. Here is the abstract, which contains a bit more detail as well.

First thing to note is the T87Q percentage, which was already up to 24% at 4.5 months post-transplant. Importantly, this number is increasing, as it was up from 9% at 3 months. I would expect this percentage to continue to increase in the more mature data bluebird will present at EHA. This slower increase versus B-thalassemia is not surprising considering the lower cell dose in the LentiGlobin product infused into this patient, as mentioned yesterday. The patient is still receiving blood transfusions, which the abstract says they are tapering. This means that most likely the %T87Q would be even higher if the patient were not also receiving transfusions (not as diluted with transfused blood). It's also good to see that the patient has not had any post-treatment hospitalizations for SCD-related events, but it's probably too early to draw a strong conclusion yet on that, in addition to the fact the patient is still receiving blood transfusions.

In terms of the stock, it is reacting positively. I think seeing the increasing %T87Q is important, which I was not sure if they would report the changes over time, or just mention the current %T87Q, which on its own, I was worried might look disappointing. However, it looks like management has done a good job recalibrating expectations for the market in terms of the amount of %T87Q to expect in SCD, as this number was almost surely going to be significantly lower than in B-thal patients. Combining the T87Q & HbF percentages was over 30% of total hemoglobin, which seemed to be the threshold people were hoping to see them cross. Again, this 30% level comes from the amount of fetal hemoglobin (HbF) that is present in patients who also have hereditary persistence of fetal hemoglobin, who have a very mild disease, or are even asymptomatic - paper. I mentioned yesterday, however, why things might be a bit more complicated in directly translating this to %T87Q, but would think this level should still have therapeutic benefit.

One other thing that was a bit surprising to me from the abstract was the peripheral blood vector copy number (VCN), which was measured as an average of 2.4 at 4.5 months. This was surprisingly high, considering the infused CD34+ product only had about 1.1 average VCN. In the B-thal patients, normally the VCN in peripheral blood was fairly correlated with the amount in the CD34+ product, although the patient with the lowest VCN in CD34+ cells had even lower in peripheral blood, and the patient with highest VCN in CD34+ cells had even higher in peripheral blood:

I think it's difficult to say, at this point, why the increase in VCN is happening, or how much it's helping generate more T87Q. Presumably, since I think these numbers are from nucleated peripheral blood cells (and not red blood cells), this reflects more an increase in the average VCN at the stem/progenitor state, than representing the selective advantage of non-sickling red blood cells, which is also probably happening. One cannot probably make too much of this piece of data yet, but it probably does bode well for having a good response in this first SCD patient.

The results in this abstract were only for the HGB-205 study, which had the 2 longest follow-up B-thal patients, as well as the first SCD patient. The two B-thal patients remain transfusion independent, but not much other data was provided. I am also very interested in seeing the follow-up data on the additional B-thal patients treated in the Northstar study, particularly subject 1102, who was a "slow-responder" in terms of slowly increasing T87Q production. It will be interesting to follow this patient to get a preliminary sense of kinetics and maximal T87Q production for someone treated with a lower cell dose, which might have some relevance to the lower cell doses that will be given to patients with SCD.

Happy to continue the discussion on twitter or in the comments below.

Edit: Some additional comments based on responses:
Bluebird previously announced only HGB-205, not Northstar data will be presented at EHA, so unfortunately no follow-up on the other B-thalassemia patients. HT: @jq1234t

Also, @Sharma1981N asked: Do you think much of the increase in %T87Q from one point to the next could be attributed to decreasing % of transfused RBCs?
This is a good point, considering the HbS levels also increased to 33% from 14.7% total hemoglobin at 4.5 from 3 months, so maybe by this timepoint the patient had less transfused blood diluting both HbS and T87Q compared to 3 months. However, the total hemoglobin levels were 12 g/dl at 4.5 months, which seems in the normal range, so most likely the T87Q levels did increase considerably from 3 to 4.5 months. However, it will be good to see everything spelled out in terms of g/dl T87Q at the different timepoints, which we will presumably see in the EHA presentation. 24% of 12 g/dl, suggests they are currently at ~2.9 g/dl T87Q.

Quick thoughts on bluebird pre-EHA abstract release

Just some quick thoughts before EHA abstracts get released tomorrow, where presumably we'll see some first data for LentiGlobin in sickle cell disease (SCD), as well as continued follow-up data in B-thalassemia. I'll also try to write up some impressions tomorrow if substantive data is released. To reiterate, none of this is intended as investment advice, more of a point to start a conversation on bluebird, and I do own some shares of bluebird currently. Generally, I am concerned that the massive run-up post-ASH data is a bit overdone, and might be pricing in basically perfect results, considering we're still relatively early in understanding the effects of bluebird's therapy. There are a few potential risks that I'll outline below:

Sickle Cell:
I co-authored an article on LentiGlobin in SCD with Zack @BioTerp, which can be found here. I basically feel the same way now as I did then, so, for a more in depth look at my feelings on this approach in SCD, definitely check that out. I expect the primary outcome we will see are the T87Q hemoglobin levels in the patient. My concerns are more that the therapy will take a longer time to produce maximal T87Q levels rather than never producing sufficient T87Q for clinical benefit. The reason it may take longer has more to do with the cell dose the first SCD patient received - 5.6 x 10^6 CD34+ cells. This is lower than B-thal subject 1102 (6.5 x 10^6 CD34+), who has taken a longer time than previous patients to achieve maximal levels of T87Q expression (also had lower VCN than other treated patients).

One reason for the low cell dose, and the reason all other treated SCD patients will also likely have lower cell doses is because SCD patients have to have CD34+ cells harvested from bone barrow as opposed to mobilized peripheral blood, and this typically gives lower cell yields. So if there is data in the abstract for only about 4+ months post-infusion, I would think there is a good chance the patient is not producing maximal T87Q levels yet.

A second point is the amount of maximal T87Q levels to eventually expect to see in this patient, or other patients with SCD. I expect this will be lower than what is seen in B-thal, but it would be difficult to accurately say how much. The reason I think we'll see lower total levels of T87Q expression, but even more so lower %T87Q relative to total hemoglobin, is due to the fact that in SCD, the patient has 2 copies of beta-globin (both with sickle mutations), whereas in B-thalassemia the patient only has 1 or 0 partially functional copies of beta-globin (βE in some of the first patients treated). Considering the LentiGlobin vector uses the transcriptional control elements of normal beta-globin, I would expect feedback within the cell will keep total levels of beta-globin relatively normal, and thus T87Q Hb has to compete against 2 expressing copies of HbS versus 1 or 0 copies of HbE. The sell-side analyst projections I see for %T87Q levels they are hoping to see (normally 30% T87Q) is considerably lower than what has been seen in B-thal patients so far, and I would say more reasonable with what one might expect.

For the reasons above, I don't think we'll see 70% T87Q in SCD patients dosed at ~1VCN, and definitely not this quickly. 30-50% T87Q I think we could see eventually, as the maximal levels, perhaps depending on the VCN of the infused product. I am also not as convinced that 30% T87Q is guaranteed to be functionally curative, or even equivalent to 30% HbF (fetal hemoglobin). My main concern here is that the fetal hemoglobin is spread more evenly throughout all the red blood cells, whereas probably only around 30% of stem cells will likely contain the T87Q vector, and make T87Q producing red blood cells. Now, there is data from "mixed chimerism" stem cell transplants in SCD that suggests a relatively small percentage (~10-30%) of normal stem cells can mediate complete protection, and that there is a survival advantage of non-sickling red blood cells, so that the vast majority (>90%) of red blood cells are actually normal, even if coming from only 30% of stem cells. So this may help a greater percentage of red blood cells to be T87Q protected, but I think it's more complicated than just saying 30% T87Q hemoglobin levels is the magic number that will be curative, and we'll have to see how it plays out clinically. I do think that these levels of T87Q would be very likely to be clinically beneficial, based on the relatively small increase in HbF seen in patients treated with hydroxyurea, but too early to say whether functionally curative or not.

B-Thal:
In B-thal I expect to continue to see durable transfusion independence in the patients that are already independent. I think the durability will likely not be a big issue considering the durability of expression seen in subject 3 dosed with the original lentiviral T87Q hemoglobin expressing vector.

Considering the dose (CD34+ cell number) & quality (VCN) of the drug varies among patients, I would consider the whole clinical experience more like many n of 1 trials than a single trial with a single drug. Thus, I am a bit concerned that patients receiving lower VCN like subject 1104 (0.7 VCN) might produce lower maximal levels of T87Q than previous patients.

 
I would be more unsure if a patient with that low of a VCN were B0/B0, who would have very little/no hemoglobin of his/her own to rely on to supplement the T87Q produced, would reach transfusion independence within 12 months. Perhaps the lower VCN patients will still produce sufficient levels of T87Q for transfusion independence, but I think it's something to keep an eye on, and one should not assume yet that transfusion independence will be always be achieved in every patient.

So to summarize, my concerns have more to do with the market's reaction to the data, than with the long term viability of the approach. I still think, in the long run, LentiGlobin will be very effective in B-thalassemia, and is likely to have clinical benefit, although way too early to say if curative, in sickle cell disease as well. As an aside, I also feel their potential in the CAR-T field is under-appreciated, given some of the people involved in their program. We can only speculate on what their initial approaches will be, but I will definitely be interested in listening as we hopefully learn more about their CAR-T plans later this year and early next year.

Edit:
Here is the follow-up post with my comments after the EHA abstract was released.

Acute vs Chronic: Pondering siRNA liver indications

by Robert Kruse

The field of siRNA therapeutics has been slow in development after its original discovering by Fire and Mello in 1998, but has made recent significant progress. Various siRNA companies have mostly focused on chronic diseases and indications, seeking to use siRNA as a weekly (or monthly) treatment much like current drugs. siRNA therapies could target cholesterol levels, HBV, cancer, or transthyretin all of which would need to be continuously dose over months of time or longer.

Given the potentially high cost of scaling siRNA manufacturing, as well the need for intravenous routes for some formulations, it is suggested that alternative indications may be useful, where the power of siRNA knockdown could be leveraged against current undruggable targets. In particular, siRNA delivery to the liver seems to have reached optimized levels that are efficient enough for human therapies. This should be considered a relatively safe bet for companies and investors. Focusing on the liver then, acute fulminant hepatitis (acute liver failure) is an important disease treated at the intensive care units of hospitals around the world. It can be caused by drugs and viruses among other ailments. The disease can be fatal if not properly treated, and even with proper treatment, may result in the need for liver transplant.

While an acute disease, there is a significant potential value being added in preventing extremely expensive downstream therapies. In the past, the death rate approached 80% before the advent of newer strategies of supportive care. Still, a significant portion of patients die from acute liver failure. Many others depend on liver transplant for survival, which isn't available for many patients with surgery carrying its own risks. It is suggested here that this represents an excellent indication for siRNA then, where only 2 or 3 doses over 1 week could save a person's life. One caveat could be the potential toxicity of the delivery vector in the background of fulminant hepatitis, so this should be considered when companies develop strategies to treat liver failure. Another caveat is the kinetics of siRNA knockdown, which could take days, or more, for significant knockdown. Thus, siRNA would need to be administered to patients before symptoms become progressively worse, and thus might not be applicable for clinical scenarios where a patient's liver is unrecoverable within a single day, in which case, liver transplant would remain the gold standard.

In screening the literature for work done previously for acute hepatic indications, I found a Nature Medicine paper from 2003 where they knocked down the Fas protein with siRNA in order to prevent fulminant hepatitis. Fas is the key mediator of hepatic apoptotic signaling, as mediated through immune cells expressing Fas ligand. T cells and Kupffer cells (liver macrophages). This strategy was very successful in preventing liver failure induced by concavalin A activation of immune cells, or with an agonistic Fas antibody, and should be broadly applicable to any mechanisms where immune infiltration is the cause of acute liver failure, such as acute viral hepatitis.

While useful for preventing immune reaction mediated acute liver failure, other indications such as acetaminophen toxicity or alcohol poisoning would likely need other targets in order to stabilize hepatocytes from liver death. One example might be targeting the Bax protein, which promotes apoptosis via the intrinsic apoptosis pathway, promoting cytochrome c release. Careful examination of various hepatic mRNA's during liver failure could identify more targets that might be useful in preserving hepatocytes. Furthermore, while the discussion here focuses on hyperacute presentations, there are acute and subacute presentations of hepatic failure that could present over several weeks, if the goal of biotech companies would be to increase doses for more profit.

While I haven't done an exhaustive search of all siRNA company's portfolios, I don't think a line of programs for acute liver failure has been openly disclosed by them. Given the limited ability of small molecules to currently rescue hepatocytes from death, it appears that this might be a relatively safe indication where hepatocytes could be efficiently targeted and modulated by siRNAs in unique ways. I would recommend scientists at Arrowhead, Alnylam, and others to consider this possibility going forward, given its important medical need.

Papers of the Week 4/27-5/3

 
Allogeneic IgG combined with dendritic cell stimuli induce antitumour T-cell immunity
This paper started with the goal of trying to better understand why even HLA-matched recipients can still completely reject tumors. They hoped that understanding this mechanism might allow implementation of the pathways involved against a patient's own tumor. As a model, they used B16 melanoma cells, which are syngeneic to C57BL/6 mice, and LMP cells which are syngeneic to 129S1 mice. When the cells were implanted into their respective syngeneic hosts, they grew fine, as expected, but when implanted into the other mouse background (allogeneic) the tumor cells were rejected. They found NK cells were dispensable for rejection, but both CD4 & CD8 T cells, as well as B cells were necessary. Further, they found increased mature dendritic cells at allogeneic tumors, and these dendritic cells had internalized more tumor-derived molecules. However, co-culture of dendritic cells with tumor cells didn't cause tumor antigen uptake. They found allogeneic tumor cells became covered with IgM and IgG antibodies when injected into mice. Additionally, they found that the surface proteins allogeneic antibodies were able to bind more tightly to were typically ones that had polymorphisms that distinguished the proteins between different mouse strains.  

Excitingly, if you took antibodies from allogeneic mice, and injected them into mice that were then injected with syngeneic tumors, they could now reject syngeneic tumors. These effects were all dependent on the Fc receptor, which binds to the Fc on the antibodies. They found that if they incubated tumor cells with allogeneic antibodies, now co-cultured dendritic cells would be activated & be able take up tumor antigens, and present them on MHC. Further, these dendritic cells, if adoptively transferred back into mice whose syngeneic tumors had been removed, could prevent relapse of those tumors. 

Importantly, it seemed that it was not something special about allogeneic IgG, but just the fact the tumor cells were coated in antibodies, that stimulated the dendritic cells. They tested this by showing forced binding of syngeneic IgG onto tumor membrane proteins could similarly mediate dendritic cell uptake. However, they found direct injection of allogeneic IgG into tumors to be insufficient to generate an anti-tumor immune response. They found that the tumor associated dendritic cells were unable to be stimulated by the allogeneic IgG allone. They tested if they could better stimulate an immune response by simultaneously injecting TNF-alpha and anti-CD40 along with the allogeneic IgG, and found this combination could generate complete or near complete responses in a variety of tumor models, both in the injected tumor, and at metastatic sites. 

There are a number of potential applications of this approach. Potentially one could do an adoptive transfer of bone marrow-derived dendritic cells after incubation with autologous tumor coated in allogeneic or other tumor-binding IgG. Alternatively, one could do try their intratumoral injection route of allogeneic or other tumor-binding IgG with anti-CD40 and TNF-alpha. As an aside, CD19 CAR-T patients are already treated with allogeneic IgG (IVIG) due to their B cell aplasia, however allogeneic IgG alone has been tried in cancer, with little benefit observed on its own.

Immunosuppressive plasma cells impede T-cell-dependent immunogenic chemotherapy

Unlike the allogeneic IgG paper, this paper found an immunosuppressive role of an immunoglobulin producing cell, IgA+ B cells. They focused on a mouse model of prostate cancer where they had previously shown B cells could aid in tumor progression and resistance to therapies. Here they look at their role in resistance to oxaliplatin, a platinum based DNA damaging agent, thought to not only kill cells, but also cause an immunogenic cell death. They found B cell deficient mice had an improved response to oxaliplatin, and had increased numbers and activation of CD8+ T cells at the tumor site. They found oxaliplatin induced infiltration of both T cells and B cells, particularly an IgA+ PD-L1+ and IL-10 secreting subset of B cells, with evidence that these cells may also be present in human tumors. Further mechanistic studies showed that TGF-beta signaling was important for oxaliplatin induction of IgA+ B cells, and that blocking either PD-L1 or IL-10 production significantly prevented the tumor-protective effect of the B cells. This potentially suggests a combination of PD-L1/PD-1 blockade with oxaliplatin.

Due to the incredibly exciting data generated in human trials that use different ways to stimulate the immune system, there has been a renewed interest in seeing if any target or therapy also positively affects the immune system, and would make for a rational combination. A previous papers of the week discussed the potential immunomodulatory roles of radiation therapy. The problem is figuring out if these are mild immunomodulatory effects, or if they legitimately can affect treatment outcomes. Most likely further understanding of the mechanisms involved, whether they appear to be at play in human tumors, and the further development of biomarkers of the immune response, will help sort out what's really important.


Other interesting papers this week:

Targeting cancer with kinase inhibitors
Blueprint Medicines ($BPMC) IPO'd this past week, and they also came out with a review on the main focus of their research - kinase inhibitors for cancer therapy. This is a good general introduction to kinase inhibitors and historical successes, with obviously an optimistic viewpoint. It also discusses the variety of mechanisms of resistance that can occur, and the importance of combination therapy to delay onset of resistance. It mentions improving kinase inhibitor specificity as well as a bit about covalent and allosteric kinase inhibitors. Not surprisingly, they also have a section on kinases in immunotherapy. Since kinases are important signaling proteins in many cells, it would not be surprising to see increased attempts at targeting kinases with the goal of immunomodulation.