Here’s What bluebird bio Investors Should Know about Sickle Cell Disease

By Zack Fink and guest contributor Dan Marks

Bluebird bio (BLUE) will release data from the first Sickle Cell Disease patient treated with its LentiGlobin gene therapy product sometime in 2015. To emphasize, sickle cell patients in bluebird bio’s ongoing trials are receiving the same exact drug product, LentiGlobin, as previously treated ß-thalassemia individuals. Sickle Cell Disease has a much higher prevalence than ß-thalassemia globally, and thus could significantly expand LentiGlobin’s market opportunity. Despite this rather large market opportunity, most Wall Street analysts covering bluebird bio attribute zero value to this program – for good reason given a lack of clinical data.

(On December 29, Roth Capital’s Debjit Chattopadhyay became the only analyst on the street to include SCD in his valuation, increasing Roth’s price target by $26/share based on a 25% probability of success for LentiGlobin in SCD.)

By the same token, this program represents significant upside potential for BLUE, and with many investors expecting LentiGlobin Sickle Cell Disease data in early 2015, we found it prudent to review the pathophysiology of Sickle Cell Disease; the rational for testing LentiGlobin in individuals diagnosed with Sickle Cell Disease; and ultimately, why we believe recent LentiGlobin ß-thalassemia data support the theory that LentiGlobin could provide a clinical benefit for individuals diagnosed with Sickle Cell Disease.

BLUE is up 140% from PropThink’s original recommendation.

What is Sickle Cell Disease?

Red blood cells (erythrocytes) are the most common blood cell in humans and are responsible for carrying oxygen (O₂) from the lungs to the rest of the body. Hemoglobin is a tetramer protein responsible for carrying O₂ inside red blood cells. The most common form of hemoglobin, adult HbA, consists of 2 α-globin chains bound to 2 β-globin chains to form the HbA O₂-carrying protein.

Sickle cell disease, similar to β-thalassemia, is an inherited blood disorder and was the first genetic disease to be characterized at a molecular level. Individuals with sickle cell disease (SCD) have a mutated form (Glu6Val substitution) of the β-globin chain, and this results in a sickled HbA phenotype referred to as HbS. More specifically, SCD refers to homozygous and heterozygous β^s mutation genotypes that cause clinical outcomes [1].

Sickle cell disease, similar to β-thalassemia, leads to dysfunctional HbA and red blood cells (RBCs) as a result of the mutated β-globin chains. Individuals with sickle cell anemia – the most common form of SCD – possess the β^s/β^s genotype, meaning β-globin chains produced in these individuals will be primarily β^s-globin, producing ‘sickled’ Hemoglobin (HbS). Unlike normal adult hemoglobin (HbA), HbS can polymerize, leading to extreme cell rigidity and disruption of normal cell homeostasis (stability). It is the polymerization of HbS that leads RBCs to change from their normal circular shape to the distinct ‘sickled’ crescent moon (shown below).

The symptoms of SCD include chronic anemia, hemolysis, and vasculopathy as a result of ‘sickled’ RBCs. Sickled RBCs clog blood vessels when they clump together, which sets off a chain reaction of clinical symptoms: vaso-occlusive pain (‘pain crises’), anemia, and damaged organs (shown above). Individuals with SCD often have a shortened lifespan and poor quality of life (QoL), exemplified anecdotally by one patient who had been hospitalized 38 times in the previous year for severe pain [2].

Individuals with SCD in developed nations typically have lifespans into their 50’s; however, those living in low-income countries have a significantly shorter lifespan, with up to 90% dying before the age of 5 [2]. It’s important to note that the course and severity of SCD varies significantly from patient to patient because of the disease’s heterogeneity.

SCD patients have a dearth of viable treatment options, and most are subject to chronic therapies to alleviate or reduce the symptoms of the disease.

Current treatments for Sickle Cell Disease, and how they can inform our view of LentiGlobin.
Blood Transfusions. Red Blood Cell transfusions, similar to β-thalassemia, have been established as a cornerstone management option for both the acute and chronic complications of SCD. RBC transfusions are performed with the goal of correcting anemia, decreasing the percentage of HbS, and reducing hemolysis [1].

One of the primary goals of blood transfusions is to reduce and maintain HbS concentrations below 30% of total hemoglobin. There is clinical support for this rationale, coming from SCD individuals who possess a single copy of the β^s mutation and one normal, “wild type”, copy of β-globin. These individuals naturally have HbS at 30-40% of their total Hb, and in most cases have few complications attributable to SCD [3]. In theory, if there is enough (~60-70%) non-mutated Hb (like the HbA transferred in RBC transfusions), SCD symptoms can be completely eliminated. This theory does not hold up for blood transfusions, because as we discuss later on, there is not an adequate level of sickling-protected RBCs. This theory does hold up for allogeneic hematopoietic stem cell transplantation because it permanently reduces the production of RBCs containing HbS.

Allogeneic Transplant and resulting mixed chimerism. Currently, the only curative treatment for SCD is an allogeneic hematopoietic stem cell (HSC) transplant. Patients have their bone marrow depleted (myeloablation) and have donor HSCs (HLA-matched) transplanted to repopulate the patient’s bone marrow. There is the potential for rejection of the graft, and graft versus host disease (GVHD), which can be fatal, but prevention and management of these complications appears to be improving with recent event-free survival reported to be over 95% [4].

A major reason for transplants not being performed in patients with severe sickle cell disease is the lack of an HLA-matched sibling donor, with one study demonstrating that only 14% of patients that are eligible for transplant have a confirmed HLA-matched sibling [5]. Additionally, the myeloablative preparative regimen needed to perform these transplants carries potential toxicities that can be considered “unduly toxic” in patients with SCD. This generally limits the use of transplants to younger patients, typically under the age of 16 [6].

An important coincidental discovery: even when the patient’s bone marrow is not completely replaced by the donor’s (i.e. there are HSCs from both the patient and the donor making their blood cells) patients were still functionally cured [7]. This state is called mixed chimerism. This led physicians to attempt to use milder preparative regimens (non-myeloablative) to expand potential transplants to those older than 16, with the goal of generating mixed chimerism. The study of mixed chimerism is directly relevant to the rationale for gene therapy approaches like bluebird’s LentiGlobin.

While it is not surprising that complete replacement of a patient’s HSCs with a donor’s is unnecessary to cure the patient of symptoms, it is surprising how small of a portion of the reconstituted stem cell pool needs to come from the donor. In one early study, a patient with only 11% of cells from the donor (as measured in peripheral blood nucleated cells, not from mature red blood cells) had stable mixed chimerism [8]. In a series of studies, it has been shown that HSC chimerism, even with donor cell levels of 10-30%, is sufficient to greatly improve clinical outcomes in sickle cell disease as well as β-thalassemia [9-12]. While only a small percentage of stem cells may come from the donor, the majority, sometimes greater than 90%, of red blood cells will be from the donor. This is a result of healthy red blood cells, produced by the donor’s stem cells, ability to survive and outcompete the patient’s own red blood cells [reviewed in 6, 13, 14].

So what does this mean for a gene therapy like bluebird’s LentiGlobin? Even though all of the patient’s HSCs are transduced (infected) with the LentiGlobin viral vector, not all of them actually have the vector permanently integrated into the cell’s genome. Therefore, the 10-30% donor cell HSC levels might represent a threshold for the percentage of CD34+ cells that would need to be LentiGlobin-modified to achieve similar efficacy. This assumes that the T87Q hemoglobin expressed by the LentiGlobin construct will be expressed in amounts and with anti-sickling properties sufficient to protect modified RBCs from sickling, which will be discussed later on.

So what is the likelihood that there will be sufficient numbers of modified cells in the LentiGlobin product? Looking at the previous gene therapy trial using a lentiviral vector in X-linked adrenoleukodystrophy, 9-14% of peripheral blood cells were modified [15]. In the original β-thalassemia trial with the older parental lentiviral vector, up to 15% of peripheral blood cells were modified, although this number increased over time and may have been aided by a proliferative advantage of where the vectors integrated [16].

Bluebird has not explicitly stated what percentage of peripheral cells are vector modified by LentiGlobin in their ongoing β-thalassemia trials; however, we can make some educated guesses. Bluebird has provided the average vector copy numbers (VCN) of the CD34+ cells in previous patients, before they are infused, and the average vector copy numbers of peripheral blood cells. While bluebird’s average VCN of the infused CD34+ product hover around 1 vector per cell, this does not mean that every cell is modified. Another group, using a lentiviral modified-hemoglobin-expressing construct, recently published a paper where they transduced CD34+ cells and looked at the cells’ progeny to see how many were actually vector modified [17]. While the average VCN was around 0.92, the distribution of vector integration in CD34+ cell progeny was not uniform, as shown in the figure below. In total, 30% of the cells contained at least one copy of the vector, with 26% containing 1-2 vectors, 3% containing 3-6, and 1% containing 7-9. Considering bluebird has an average VCN of around 1 or higher in CD34+ cells, it is reasonable to believe these would have a good chance of being within the 10-30% “corrected” HSC range suggested to be therapeutically beneficial based on allogeneic transplant data. Again, this assumes enough vector-modified cells are resistant to sickling – discussed below –to provide a similar survival advantage as an allogeneic donor cell to cause the vector-modified cells to make up the bulk of mature red blood cells in patients.

Additionally, it is likely that bluebird will need to continue to use myeloablative conditioning for the best chance of repopulating the bone marrow with sufficient numbers of LentiGlobin-modified cells. Considering the previously mentioned concerns about myeloablation in older patients with SCD, this may limit LentiGlobin’s usage to only younger patients.

With the significant caveat that we are starting with the relevant phenomenon of mixed chimerism, and looking at the integration abilities of other lentiviral vectors and extrapolating, we believe that bluebird’s LentiGlobin CD34+ cell product has a reasonable chance at producing a sufficient number of vector modified cells so as to have the potential for clinical benefit. To consider some of the factors involved in whether the LentiGlobin Hb-β^A-T87Q vector will have sufficient strength to correct an individual red blood cell, let us turn to a discussion of fetal hemoglobin, and how the T87Q mutation is expected to reduce sickling by mimicking its structure.

Fetal Hemoglobin. Until a few months after birth when adult HbA begins to dominate, fetal hemoglobin (HbF) is the most common hemoglobin in infants. HbF serves the same function as HbA, except it has a slightly different structure: composed of 2 α-globin and 2 γ-globin chains. Researchers believe that HbF is produced as a result of red blood cell stress, consistent with the fact that HbF is expressed at greater concentrations in sickle cell anemia individuals than in normal adults (5-8% vs <1% HbF of total Hb). HbF, albeit at a much lower level, dilutes the concentration of HbS, similar to the high HbA levels that dilute HbS in sickle cell trait individuals. HbF also has the added benefit for SCD patients of preventing the polymerization of HbS, resulting in HbF having ‘anti-sickling properties’.

SCD patients who produce enough HbF, due to the cells’ anti-sickling property, do not exhibit typical SCD symptoms, leading to a much-improved QoL and lifespan. Platt et al reported that SCD individuals producing >8.6% HbF of total Hb had a significantly improved probability of survival compared to those producing <= 8.6% HbF (n = 3386) [19].

In addition, incremental increases in HbF have been associated with a decrease in SCD symptoms and reduced disease severity [18-20]. Some of the strongest clinical evidence for the ability of HbF to ameliorate SCD symptoms can be seen in the examination of individuals co-harboring mutations for SCD and hereditary persistence of fetal hemoglobin.

Hereditary persistence of fetal hemoglobin (HPFH) characterizes a group of genetically heterogeneous conditions marked by an elevated production of HbF. Ngo et al studied individuals who are compound heterozygous for SCD and HPFH and found that those who produced HbF at a concentration of ~30% of total Hb possess near normal hematological values and are asymptomatic for SCD [21]. 

In theory, a ~30% HbF concentration could be adequate to ameliorate SCD symptoms and improve QoL. About 30% HbF potentially provides ~70% of RBCs with adequate concentrations of HbS-polymerization-inhibiting fetal hemoblogin. RBCs with adequate levels of HbF (~10 pg/cell) are termed ‘protected F-cells’. We believe that the percentage of F-cells with adequate levels of polymer-inhibiting HbF (% of protected F cells) is a more important and accurate outcome determinant than concentration of HbF or number of F-cells [22]. An individual with an inadequate percentage of protected F-cells will still demonstrate SCD symptoms because non-protected RBCs will continue to clump inside blood vessels. This logic is further supported by individuals with the sickle cell trait and only one copy of HbS. Assuming these individuals produce relatively homogeneous levels of HbA inside RBCs, the concentration (~60-70%) of HbA is adequate to dilute HbS (~30-40%), enabling the prevention of HbS polymerization in any given RBC. In other words, sickle cell trait individuals are not asymptomatic for SCD because of the amount of non-sickling Hb (HbA): they are asymptomatic because enough RBCs contain sufficient normal hemoglobin within each cell to dilute HbS and prevent polymerization. This pancellular protection thus prevents the downstream consequences of sickling.

Hydroxyurea: The only FDA approved drug for Sickle Cell Disease
The rationale of attempting to prevent polymerization of HbS as a mechanism to induce a clinical benefit in individuals diagnosed with sickle cell disease is supported by the regulatory approval of hydroxyurea for SCD. Hydroxyurea is the only approved drug for the treatment of sickle cell disease, and at least part of its mechanism of action is to induce increased production of HbF. As a reminder, HbF prevents the polymerization of HbS; thus if expressed at high enough levels inside cells, it can ‘protect’ cells from sickling.

Charache et al reported sickle cell anemia individuals administered hydroxyurea had a 44% reduction in the median annual rate of painful crises. In addition, this randomized controlled trial also demonstrated hydroxyurea’s ability to induce a clinically beneficial response as seen by the reduction in sick cell related complications and reduction in the number of required blood transfusions [25]. Hydroxyurea’s safety and ability to improve quality of life in sickle cell anemia individuals was confirmed in two long-term clinical trials [26-27].

Hydroxyurea in the above trials also demonstrated that modestly increasing HbF can lead to regulatory approval of a drug.

Across three clinical trials, baseline HbF percentage was approximately 5%-10% of an average baseline total Hb of 8-9 [g/dl] in SCD patients. Essentially, SCD patients produce, on average, around 0.4-0.9 [g/dl] of HbF [20, 27-28]. SCD individuals receiving hydroxyurea enrolled in clinical trials demonstrated a mean change in absolute production of HbF over 2 years of 3.6% +/- 5.4% [28]. Using the average baseline of total Hb, we can estimate the mean change in HbF production over 2 years is around 0.3 to 0.81 [g/dl]. The 2-year mean change in HbF production for the best responders – the upper 25% quartile – was ~12%. Again, using the estimate for average total Hb at baseline, this corresponds to the best hydroxyurea responders having a mean change in baseline HbF of approximately 0.96-1.08 [g/dl]. Although we caution against using HbF percentage or absolute HbF production as a parameter to predict and measure efficacy, we believe hydroxyurea’s very modest ability to increase HbF for clinical benefit (~1 [g/dl]) demonstrates the low bar of success for drugs in SCD. In addition, this also gives some insight into why it is extremely rare for an individual diagnosed with SCD to become asymptomatic after hydroxyurea treatment. Even if a SCD patient had above average baseline HbF (>10%) and an exemplary response to hydroxyurea (increase >12% HbF), the total HbF production (>22%) would still in most cases fall short of the 30% HbF threshold commonly cited as the cut-off for significant amelioration of SCD symptoms. This variation in percentage of protected RBCs can be seen in the below graphs which model different ratios of protected F cells with a total HbF concentration of 20% in each graph.

Those who have followed bluebird are probably familiar with HbF and HPFH since bluebird has used HPFH, and individuals who are asymptomatic for SCD with ~30% HbF, to support the theory that LentiGlobin could ‘cure’ SCD. Understanding this rationale requires an understanding of the β^A-T87Q-globin transgene that LentiGlobin delivers.

T87Q β-globin mutation. β^A-T87Q-globin is simply the natural human β-globin gene, with codon 87 modified to glycine. This modification mimics the anti-sickling properties of HbF because the anti-polymerization properties of HbF are primarily the result of glycine at codon 87 (below right) and aspartic acid at codon 80. Pawliuk et al demonstrated that Hb incorporating β^A-T87Q-globin (Hb-β^A-T87Q) is about as potent of an inhibitor of HbS polymerization as HbF (below left) [23]. In addition to demonstrating in pre-clinical testing that Hb-β^A-T87Q has similar anti-polymerization properties as HbF, bluebird also confirmed Hb-β^A-T87Q anti-sickling properties in in-vivo mouse models. Since no transgeneic mouse model perfectly models the exact disease characteristics of human SCD patients, bluebird conducted testing in two mouse models: SAD and BERK mice. The SAD and BERK bone marrow was transduced with a lentiviral vector containing the β^A-T87Q-globin transgene, and after 3 months the percentage of β^A-T87Q-globin was up to 52% and 12% of the total Hb in the BERK and SAD mice, respectively. As a reminder, this in-vivo experiment was conducted almost 15 years ago with a primitive lentiviral vector compared to that used in LentiGlobin.

Since Hb-β^A-T87Q is about as potent as HbF at inhibiting HbS polymerization, bluebird has hinted that if SCD patients are producing Hb-β^A-T87Q >30% of total Hb, this could potentially ameliorate SCD symptoms.  We caution against believing that 30% Hb-β^A-T87Q is the best efficacy bar because, as outlined above with HbF and F-cells, it is not the percentage of ‘anti-sickling’ Hb that is the relevant parameter, but the percentage of RBCs protected from HbS polymerization. 

Our Outlook for LentiGlobin in SCD
At the recent American Society of Hematology (ASH) Annual Meeting, bluebird presented the most recent interim data from individuals diagnosed with β-thalassemia treated with LentiGlobin. As shown below, β-thalassemia patients treated with LentiGlobin that became transfusion independent appear to possess ~70% Hb-β^A-T87Q once reaching steady production (Subject 1102 needs longer follow-up) [24].

Additionally, bluebird has given the averages for VCN in peripheral blood cells from their β-thalassemia trials, so we can also analyze how well the CD34+ VCN in their infused product is translating into the VCN in the CD34+ cells’ progeny once inside the  patients post-myeloablation.

As shown below, from the same ASH presentation, the average CD34+ drug product VCN have translated reasonably well into average peripheral blood VCN. Only subject 1102’s VCN in peripheral blood is still relatively low at 6 months, even though she is producing 3.8 g/dl Hb-β^A-T87Q [24]. This again suggests that LentiGlobin transduced cells, at least in terms of percentage of modified cells, have a chance at reaching the 10-30% “corrected” amount suggested to be efficacious by mixed chimerism-post-allogeneic transplant.

Bluebird also released early outcomes for the first SCD individual treated with LentiGlobin. As shown below, the first patient has the genotype β^S/ β^S, and therefore is diagnosed with sickle cell anemia.

This individual and those with sickle cell anemia, similar to β-thalassemia individuals, require chronic blood transfusions. Bluebird revealed that this individual had successful engraftment with LentiGlobin-infused hematopoietic stem cells (HSCs); however, it was too early for any additional data. It is important to note that unlike in β-thalassemia patients receiving the LentiGlobin product, CD34+ cells come directly from bone marrow, and not from mobilized peripheral blood HSCs. It is necessary to use bone marrow because the mobilization process is potentially dangerous in sickle cell patients. The speed of recovery, based on platelet and neutrophil engraftment, is generally slower when bone marrow is used as the HSC source [29]. It is thus not surprising that bluebird’s first sickle cell patient’s neutrophil engraftment was slower than their previously treated β-thalassemia patients. One could also speculate this may cause a less rapid increase in the amount of Hb-β^A-T87Q produced versus some of the previous β-thalassemia patients.

We believe that LentiGlobin could be a treatment option for sickle cell anemia and SCD because of the robust levels (~70% of total) of anti-sickling Hb-β^A-T87Q potentially produced in individuals with SCD. The bar for providing a clinical benefit – similar to β-thalassemia – is rather low because of the extremely poor QoL in Sickle Cell Disease, requirement of chronic blood transfusions, and a lack of adequate treatment options.

LentiGlobin’s ability to meet/exceed this bar can be seen by comparing the levels of anti-sickling Hb in LentiGlobin-treated individuals to SCD individuals with high HbF production, who have much milder SCD symptoms [21]. Although we caution against stratifying data in this manner, the best responding LentiGlobin treated β-thalassemia patients are producing 70% anti-sickling Hb-β^A-T87Q of total Hb, significantly higher than the 15-25% of anti-sickling HbF needed to produce a milder SCD clinical course.
In order to make an accurate prediction in SCD individuals from the percentage of Hb-β^A-T87Q being produced in β-thalassemia treated subjects, we must consider the present HbS. This is apparent when analyzing an ‘example’ sickle cell anemia patient producing hemoglobin that is 100% HbS at 8.5 g/dl.
(We use 8.5 g/dl of HbS as a conservative estimate for the average HbS in sickle cell following a hydroxyurea trial in which the mean total of total Hb for enrolled subjects was 8.5 [g/dl] [25].) The second value needed is an estimate of Hb-β^A-T87Q (in g/dl) that is being produced in this example patient. We use an estimate of 7 [g/dl] because this is the amount of Hb-β^A-T87Q being produced on average by the four transfusion-independent β-thalassemia subjects in ongoing LentiGlobin studies. By using these two values, this example LentiGlobin-treated sickle cell patient – if producing the same amount of Hb-β^A-T87Q (in g/dl) on average as LentiGlobin β-thalassemia subjects – would have ~ 45% Hb-β^A-T87Q of total Hb.

This essentially offers a hint that if LentiGlobin SCD individuals are producing similar levels of β^A-T87Q-globin as β-thalassemia individuals treated with LentiGlobin, that LentiGlobin could be poised to become a treatment option for SCD. This is also apparent when comparing the average change in anti-sickling Hb between those treated with hydroxyurea and individuals treated with LentiGlobin. As discussed above, the upper quartile (25%) of hyroxyurea responders had an approximate change in HbF of 1 [g/dl] in a 2 year trial studying changes in HbF. LentiGlobin treatment has led to the production of 7 [g/dl] (on average) of Hb-β^A-T87Q in the first four transfusion-independent β-thalassemia subjects. This is significantly higher than the 1 [g/dl] produced by the best responding hydroxyurea patients, and supports the theory that LentiGlobin could provide a clinical benefit for individuals diagnosed with Sickle Cell Disease.

Despite this, we believe investors should have tempered expectations and caution against setting concrete expectations for LentiGlobin in SCD. This is because of the extreme heterogeneity in the pathophysiology of individuals with SCD, causing these patients to have variations in clinical outcomes even if treatment parameters are identical. Suppose an ‘example SCD patient’ produces Hb-β^A-T87Q at 70% of total Hb. This patient could either possess:

• As high as 100% of RBCs protected with each producing around 70% of Hb-β^A-T87Q inside each individual cell
• As low as 70% of RBCs protected with each producing around 100% Hb-β^A-T87Q inside each individual cell

It is difficult to predict outcomes in SCD patients by stratifying current LentiGlobin β-thalassemia data. As we’ve highlighted, it is not the Hb-β^A-T87Q as a percentage of total Hb that is accurate in predicting SCD treatment outcomes, but the percentage of RBCs protected from HbS polymerization (by having adequate levels of Hb-β^A-T87Q/cell). One thing that should aid the percentage of RBCs protected is the expected survival advantage of “corrected” RBCs as seen in the related mixed chimerism data. Bluebird has not yet provided any public information on the percentage of RBCs protected from HbS-polymerization; therefore it is difficult to make concrete predictions for LentiGlobin-treated SCD patients.

Nevertheless, we remain optimistic that early and forthcoming SCD LentiGlobin data will be positive. Below is a summary of important components of any efficacy signal:

• Reduction in need for blood transfusions.
• Percent of protected RBCs
• Signs of pancellular RBC Hb-β^A-T87Q production
• Reduction in frequency and severity of SCD related ‘pain crises’ and other SCD symptoms.
• Our Bar for asymptomatic ‘cured’ SCD: >60-70% RBCs protected from HbS polymerization.

Below, we reference SunTrust’s SCD model to consider potential upside as LentiGlobin in SCD is de-risked, based on modeled probability of success and a current per share price of $91.26. Again, forward modeling is an imperfect approach, but the market opportunity for an efficacious, let alone alone curative, product in SCD is tremendous.